1: % The source file is: Pt2IN.tex
2: % The source file is: Pt2.tex
3: \title{Towers of recollement and bases for diagram algebras:
4: planar diagrams and a little beyond}
5: %{{{ preamble+macros
6: \documentclass[12pt]{article}
7: \usepackage{latexsym}
8: \usepackage{amssymb}
9: \usepackage[all]{xy}
10: \UseComputerModernTips
11: \usepackage{graphicx}
12: \usepackage{epsf}
13: \usepackage[dvips]{rotating}
14: \textheight 24cm \topmargin -2cm \oddsidemargin1cm \textwidth 15cm
15: %{{{ %reminder junk
16: %}}}
17: %{{{ global macro glossary making
18: \providecommand{\noglossaryignore}[1]{}
19: \newcommand{\globalglossaryentry}[3]{\makebox[1.5in][l]{\tt $\backslash${#1}}
20: \makebox[1.1in][l]{{$#2$}} \makebox[2.5in][l]{{#3}}\newline}
21: \newcommand{\newcommandabbreviation}[3]{\newcommand{#1}{#2}%
22: \noglossaryignore{\globalglossaryentry{#3}{#2}{}}}
23: \newcommand{\renewcommandabbreviation}[3]{\renewcommand{#1}{#2}%
24: \noglossaryignore{\globalglossaryentry{#3}{#2}{}}}
25: \newcommand{\newcommandmacro}[4]{\newcommand{#1}{#2}%
26: \noglossaryignore{\globalglossaryentry{#3}{#2}{#4}}}
27: \newcommand{\gge}[3]{\noglossaryignore{\globalglossaryentry{#1}{#2}{#3}}}
28: %{{{ make glossary macros (local version, not used here)
29: %}}}
30: %}}}
31: %{{{ global macro defns
32: %{{{ addresses (city etc)
33: \newcommand{\myaddress}%
34: {\parbox{3in}{\footnotesize \begin{center}
35: Mathematics Department, City University, \\
36: Northampton Square, London EC1V 0HB, UK.\end{center}}}
37: %}}}
38: %{{{ xfig patches
39: \newcommand{\twlrm}{} \def\tenrm{} \def\ninrm{} \def\sixrm{} \def\sevrm{}
40: %}}}
41: %{{{ minidef and minicapt
42: \newcounter{minidef}[section]
43: \renewcommand{\theminidef}{\thesection.\arabic{minidef}}
44: \newcommand{\mdef}{\refstepcounter{minidef}
45: \medskip \noindent ({\bf \theminidef}) }
46: \newcounter{minicapt}
47: \newenvironment{capt}[1]
48: {\refstepcounter{minicapt}\stepcounter{figure}
49: \begin{center}\sf Figure \theminicapt. }{ \end{center}}
50: %}}}
51: %{{{ theorems
52: \newtheorem{theo}{Theorem}[subsection] \newtheorem{cor}{Corollary}[theo]
53: \newtheorem{de}[theo]{Definition}
54: \newtheorem{pr}[theo]{Proposition}
55: \newtheorem{co}[theo]{Corollary}
56: \newtheorem{rem}[theo]{Remark}
57: \newtheorem{lem}[theo]{Lemma}
58: \newtheorem{claim}{Claim}
59: %}}}
60: %{{{ greek and other abbreviations
61: \noglossaryignore{GREEK ETC.\newline}
62: \newcommandabbreviation{\e}{\epsilon}{e}
63: \newcommandabbreviation{\lam}{\lambda}{lam}
64: \newcommandabbreviation{\la}{\langle}{la}
65: \newcommandabbreviation{\ran}{\rangle}{ran}
66: \newcommandabbreviation{\ha}{\#}{ha}
67: \newcommandabbreviation{\rmap}{\rightarrow}{rmap}
68: \newcommandabbreviation{\aaa}{\alpha}{aaa}
69: \newcommandabbreviation{\ab}{\alpha,\beta}{ab}
70: \newcommandabbreviation{\aab}{a(\ab )}{aab}
71: %}}}
72: %{{{ rings
73: \noglossaryignore{\newline RINGS\newline}
74: \newcommandabbreviation{\HH}{H \!\!\! I}{HH}
75: \newcommandabbreviation{\C}{\mathbb C}{C}
76: \newcommandabbreviation{\N}{\mathbb N}{N}
77: \newcommandabbreviation{\Z}{\mathbb Z}{Z}
78: \renewcommandabbreviation{\Re}{\mathbb R}{Re}
79: \newcommandabbreviation{\R}{{\mathbb R}}{R}
80: \newcommandabbreviation{\Q}{\mathbb Q }{Q}
81: \renewcommandabbreviation{\H}{\mathbb H }{H}
82: %}}}
83: %{{{ symmetric group
84: \noglossaryignore{\newline SYMMETRIC GROUP\newline}
85: \def\Sym(#1){\Sigma(#1)}
86: \gge{Sym(-)}{\Sym(-)}{symmetric group on - objects}
87: \def\Sy(#1){\Sigma_{#1}}
88: \gge{Sy(-)}{\Sy(-)}{symmetric group irreducible -}
89: \def\sym(#1){\mbox{\LARGE s}(#1)}
90: \gge{sym(-)}{\sym(-)}{symmetric group on - objects (variant)}
91: \def\sy(#1){\mbox{\LARGE s}({#1})}
92: \gge{sy(-)}{\sy(-)}{symmetric group irreducible - (variant)}
93: \newcommandmacro{\cs}{\C \, \sy(n)}{cs}{symmetric group algebra over $\C$}
94: %}}}
95: %{{{ partitions/sets
96: \noglossaryignore{\newline PARTITIONS/SETS\newline}
97: \newcommand{\Nset}[1]{\underline{#1}}
98: \gge{Nset\{-\}}{\Nset{-}}{set of natural numbers to -}
99: \def\nset(#1){ \{ #1 \}_{ \underline{n} }}
100: \gge{nset(-)}{\nset(-)}{a set $-\times\Nset$}
101: \def\ul(#1){_{\underline{#1}}}
102: \gge{ul(-)}{{}\ul(-)}{subscript underline -}
103: \def\Ee(#1){{\bf E}_{#1}}
104: \gge{Ee(-)}{\Ee(-)}{set of equivalence relations on set -}
105: \def\Eee(#1){{\bf E}_{\{ #1 \}_{\underline{n}}}}
106: \gge{Eee(-)}{\Eee(-)}{ditto for nset}
107: \def\Een(#1,#2){{\bf E}_{\{ #1 \}_{\underline{#2}}}}
108: \def\Ssn(#1,#2){{\bf S}_{\{ #1 \}_{\underline{#2}}}}
109: \def\Ss(#1){{\bf S}_{#1}}
110: \def\Sss(#1){{\bf S}_{\{ #1 \}_{\underline{n}}}}
111: \def\bbc(#1){((\beta_1)(\beta_2)...(\beta_{#1}))}
112: \newcommandmacro{\Ln}{{\Gamma}^{n}}{Ln}{large index set}
113: \newcommandmacro{\LnQ}{{\Gamma}^{n}_Q}{LnQ}{index set}
114: \newcommandmacro{\Zz}{\zeta}{Zz}{`shape' function}
115: \newcommand{\Ww}[1]{\mbox{\LARGE $w$}_{#1}}
116: %}}}
117: %{{{ partition algebra
118: \noglossaryignore{\newline PARTITION ALGEBRA\newline}
119: \def\ka(#1){\kappa_{#1}}
120: \def\Sm(#1){\Sigma_{#1}}
121: \newcommandmacro{\com}{\bullet}{com}{bullet composition}
122: \newcommandmacro{\enm}{\; e^n(\! m\! ) \;}{enm}{product of idempotents}
123: \def\Ai(#1){ A^{ #1 \cdot } }
124: \def\Aij(#1,#2){ A^{ #1 #2 } }
125: \newcommandmacro{\One}{\mbox{\bf $1 \!\!\! 1$}}{One}{algebra unit 1}
126: \newcommand{\Ff}[2]{F_{#1}^{(#2)}}
127: \newcommand{\ef}[2]{\prod_{j=1}^{#2-#1}e_j}
128: \newcommand{\efbr}[2]{\left(\prod_{j=1}^{#2-#1}e_j\right)}
129: \newcommand{\Ef}[2]{E_{#1}^{(#2)}}
130: \def\fu{
131: \left( \prod_{i=1}^{n} \left( \One - \Aij(i, \; n+1) \right) \right)
132: }
133: \def\Fu{\mbox{\Large $\chi$}}
134: \newcommandmacro{\Bp}{B_p}{Bp}{partition basis}
135: \def\Bb(#1){B_p[#1]}
136: \def\Pp(#1){P_n[#1]}
137: \def\Ps(#1){P_n[#1] \! /}
138: \newcommandmacro{\Ph}{\hat{P}}{Ph}{P hat algebra}
139: \def\Is(#1){\sim^{#1}}
140: %}}}
141: %{{{ modules
142: \noglossaryignore{\newline MODULES\newline}
143: \def\Wm(#1){{\cal S}_{#1}}
144: \gge{Wm(-)}{\Wm(-)}{Weyl module with index -}
145: \def\wm(#1,#2){{}_{#1}{\cal S}_{#2}}
146: \gge{wm(-1,-)}{\wm(-1,-)}{Weyl module with index -}
147: \def\Ind(#1,#2,#3){\mbox{Ind}_{#1}^{#2}#3}
148: \gge{Ind(-1,-2,-)}{\Ind(-1,-2,-)}{induction}
149: \def\Res(#1,#2,#3){\mbox{Res}_{#1}^{#2}#3}
150: \gge{Res(-1,-2,-)}{\Res(-1,-2,-)}{restriction}
151: \newcommandabbreviation{\weyl}{standard}{weyl}
152: \newcommandabbreviation{\mod}{\mbox{mod}}{mod}
153: \newcommandabbreviation{\head}{\mbox{head }}{head}
154: \newcommandabbreviation{\Weyl}{Weyl}{Weyl}
155: \def\SS(#1){{\cal S}_{#1}}
156: \gge{SS(-)}{\SS(-)}{Specht/Weyl module index -}
157: \def\LL(#1){{\cal L}_{#1}}
158: \gge{LL(-)}{\LL(-)}{simple module index -}
159: %}}}
160: %{{{ functors/maps
161: \noglossaryignore{\newline FUNCTORS/MAPS\newline}
162: \newcommandmacro{\Gg}{{\cal G}}{Gg}{G Functor}
163: \newcommandmacro{\Fg}{{\cal F}}{Fg}{F Functor}
164: \newcommandmacro{\ra}{\rightarrow}{ra}{}
165: \def\ses(#1,#2,#3){0\ra #1 \ra #2 \ra #3 \ra 0}
166: \gge{ses(1,2,-)}{\ses(1,2,-)}{\hspace{.5in} short exact sequence}
167: \def\starr(#1){ \stackrel{ #1 }{\longrightarrow} }
168: \gge{starr(-)}{\starr(-)}{}
169: \newcommandmacro{\doublerightarrow}{\; -\!\!\! -\!\!\!\!\!\! \gg \;}
170: {doublerightarrow}{}%{ $---->>$ }
171: %}}}
172: %{{{ partition algebra maps
173: \noglossaryignore{\newline PARTITION ALGEBRA MAPS\newline}
174: \newcommandmacro{\smap}{s}{smap}{`inclusion' map}
175: \newcommandmacro{\tmap}{t}{tmap}{$ P_n -> S_n$}
176: \newcommandmacro{\pmap}{\psi}{pmap}{$ S_n -> P_n $}
177: %}}}
178: %{{{ miscellaneous
179: \noglossaryignore{\newline MISC.\newline}
180: \def\Amap(#1){{\cal A}_{#1}}
181: \gge{Amap(-)}{\Amap(-)}{}
182: \def\Rr(#1){R_{#1}}
183: \gge{Rr(-)}{\Rr(-)}{restriction of E}
184: \def\Cr(#1){C_{#1}}
185: \gge{Cr(-)}{\Cr(-)}{restriction of E to N}
186: \newcommandmacro{\Tm}{{\cal T}}{Tm}{Transfer Matrix}
187: \def\On(#1){{\cal I}_{#1}}
188: \gge{On(-)}{\On(-)}{}
189: \newcommandmacro{\UU}{\underline{\sqcup}}{UU}{}
190: \newcommandmacro{\UUU}{\sqcup}{UUU}{}
191: \newcommandmacro{\Vq}{V_Q^{\otimes n}}{Vq}{Potts config. space}
192: \def\bs(#1,#2){\mbox{{\Large $\ast$}}^{#1}_{#2}}
193: \gge{bs(-,-)}{\bs(-,-)}{general plumbing multiplier}
194: \newcommand{\ignore}[1]{}
195: %}}}
196: %{{{ math stuff
197: \def\choo(#1,#2){ \left( \begin{array}{c} #1 \\ #2 \end{array} \right) }
198: \gge{choo(-1,-)}{\choo(-1,-)}{choose}
199: \newcommand{\Qed}{$\Box$}%{Qed}{QED}
200: \gge{Qed}{\mbox{\Qed}}{QED}
201: \def\staq(#1){\stackrel{#1}{=}}
202: \gge{staq(-)}{\staq(-)}{}
203: \def\stam(#1){\stackrel{#1}{\rightarrow}}
204: \gge{stam(-)}{\stam(-)}{}
205: \def\mat{ \left( \begin{array} }
206: \def\tam{ \end{array} \right) }
207: \gge{mat/tam}{...}{matrix delimiters}
208: %}}}
209: %{{{ equation environments etc (draft/final)
210: \newcommand{\beq}{\begin{equation} }
211: \def\eql(#1){ \begin{equation} \label{#1}
212: }
213: \newcommand{\eq}{\end{equation} }
214: \def\eqal(#1){\begin{eqnarray} \label{#1} }
215: \def\eqa{\end{eqnarray} }
216: \def\lab(#1){\label{#1}
217: }
218: \def\prl(#1){ \begin{pr} \label{#1}
219: }
220: \def\del(#1){ \begin{de} \label{#1}
221: }
222: \newcommand{\smeq}[1]{\[ \mbox{\small $ #1 $} \] }
223: \gge{smeq\{-\}}{...}{small equation}
224: \newcommand{\fneq}[1]{\[ \mbox{\footnotesize $ #1 $} \] }
225: \gge{fneq\{-\}}{...}{very small equation}
226: %}}}
227: %}}}
228: %{{{ local defns(Hecke/blob)
229: \noglossaryignore{\newline HECKE/BLOB\newline}
230: \newcommandmacro{\Hnq}{H_n(q)}{Hnq}{ * freestanding symbol}
231: \newcommandmacro{\Hn}{H_n}{Hn}{ *-mod etc.}
232: \newcommandmacro{\A}{{\cal A}}{A}{}
233: \newcommandmacro{\Cwts}{C}{Cwts}{}
234: \newcommandmacro{\CA}{{\cal A}}{CA}{}
235: \def\words#1{\mbox{ #1}}
236: \newcommandmacro{\calA}{{\cal A}}{calA}{}
237: \newcommandmacro{\modi}{\mbox{Mod} }{modi}{was mod not modi!}
238: \newcommandmacro{\Wgen}{{\Bbb S}}{Wgen}{}
239: \def\ol(#1){\overline{#1}}
240: \newcommandmacro{\st}{\mbox{St}}{st}{}
241: \def\twiddle{\~}
242: \def\CMult(#1,#2){(#1:#2)}
243: \def\CM(#1,#2){( #1 : #2 )}
244: \def\FMult#1,#2{(#1:#2)}
245: \def\CF#1,#2{(#1:#2)}
246: \def\RMod#1{#1-\!\mbox{mod}}
247: \def\LMod#1{\mbox{mod}\!-#1}
248: \newcommandmacro{\Top}{\mbox{Top}}{Top}{}
249: \newcommandmacro{\Soc}{\mbox{Soc}}{Soc}{}
250: \newcommandmacro{\Head}{\mbox{Head}}{Head}{}
251: \newcommandmacro{\Filt}{{\cal F}}{Filt}{}
252: \newcommandmacro{\Mod}{\mbox{mod}}{Mod}{}
253: \newcommandmacro{\Resi}{\mbox{Res }}{Resi}{was without i!}
254: \newcommandmacro{\Indi}{\mbox{Ind }}{Indi}{was without i!}
255: \newcommand{\II}[2]{I^{#1}_{#2}}
256: \def\RR(#1,#2){R^{#1}_{#2}}
257: \def\TT(#1,#2){T^{#1}_{#2}}
258: \def\implies{\Rightarrow}
259: \def\qSchur{S_q(n,N)}
260: \def\Uq{U_q(sl_N)}
261: \def\Usl{U(sl_N)}
262: \def\qGroup{$q$-group}
263: \def\VV{V_N^{\otimes n}}
264: \def\Hom{\mbox{Hom}}
265: \def\v{v}
266: \def\len{\mbox{len}}
267: \def\bigplus{\mbox{\LARGE $+$}}
268: \def\dom{dominance }
269: \def\ldom{\unlhd}
270: \def\gdom{\unrhd}
271: \def\laffine{$l$-affine }
272: \newcommand{\nt}[1]{{\em #1}}
273: \def\Sp{\Delta}
274: \def\Stan{\Delta}
275: \def\Ysym{Y}
276: \def\eqn#1{equation (\ref{#1}) }
277: \newcommand{\eqr}[1]{equation~(\ref{#1})}
278: \def\HP#1{P'_{#1}}
279: \def\UP#1{P_{#1}}
280: \def\HSp#1{\Sp'_{#1}}
281: \def\USt#1{\Stan_{#1}}
282: \def\HL#1{L'_{#1}}
283: \def\UL#1{L_{#1}}
284: \def\Chi{\chi}
285: \def\Cartan{C}
286: \def\Rad{\mbox{Rad }}
287: \def\facets#1{{\cal A}_{#1}}
288: \def\St{\st}
289: %}}}
290: %{{{ local defns(partition)
291: \newcommandmacro{\Ann}{\mbox{Ann}}{Ann}{}
292: \newcommandmacro{\Cen}{\mbox{Cen}}{Cen}{}
293: \newcommandmacro{\End}{\mbox{End}}{End}{}
294: \newcommandabbreviation{\semisimple}{semisimple}{semisimple}
295: \newcommandabbreviation{\Bratteli}{Bratteli}{Bratteli}
296: \newcommandabbreviation{\JBC}{Jones Basic Construction}{JBC}
297: \newcommandabbreviation{\pa}{partition algebra}{pa}
298: \newcommandabbreviation{\TM}{transfer matrix}{TM}
299: \newcommandabbreviation{\PM}{Potts model}{PM}
300: \newcommandabbreviation{\QSC}{quantum spin chain}{QSC}
301: \newcommandabbreviation{\Hamiltonian}{Hamiltonian}{Hamiltonian}
302: \newcommandabbreviation{\YS}{Young symmetrizer}{YS}
303: %}}}
304: %{{{ %local defns (blob)
305:
306:
307: %}}}
308: %{{{ Local vars
309:
310:
311: %}}}
312:
313: %{{{ new commands/local macros
314: \newcommand{\be}{\begin{eqnarray}}
315: \newcommand{\eeq}{\end{eqnarray}}
316: \newcommand{\non}{\nonumber}
317:
318: \newcommand{\n}{\ensuremath{\mathcal{N}}}
319: \newcommand{\XR}{\ensuremath{\mathcal{R}}}
320: \newcommand{\tr}{\mathop{\rm tr}\nolimits}
321:
322: \newcommand{\TL}{Temperley--Lieb}
323: \newcommand{\YB}{Yang--Baxter}
324:
325: \newcommand{\rank}{N}
326: \newcommand{\sites}{n}
327:
328: \newcommand{\lateral}{lateral}
329:
330: \newcommand{\braid}[1]{{\cal B}_{#1}}
331: \newcommand{\braido}[1]{{\cal B}^{\circ}_{#1}}
332: \newcommand{\Artin}[1]{{\cal A}_{#1}}
333:
334: \newcommand{\xfigin}[2]{\begin{figure}\input{xfig/#1.eepic}%
335: \caption{\label{#1} #2}\end{figure}}
336: \newcommand{\xfiginps}[3]{\begin{figure}\includegraphics[width=#3in]{xfig/#1.eps}%
337: \caption{\label{#1} #2}\end{figure}}
338:
339: \newcommand{\affine}[1]{\hat{#1}}
340: \newcommand{\twist}{\tau}
341: \newcommand{\squash}{\sigma} \newcommand{\sm}{l}
342: \newcommand{\glue}{\iota}
343: \newcommand{\cable}{\gamma}
344: \newcommand{\calR}{{\cal R}}
345: \newcommand{\rr}{R_}
346: \newcommand{\U}{U_}
347: \newcommand{\ee}{e}
348: \newcommand{\Uone}{\U1}
349: \newcommand{\Uzero}{\ee}
350: \newcommand{\kap}{\kappa}
351: \newcommand{\kk}{K}
352: \newcommand{\T}{\theta_}
353: \newcommand{\tlambda}{\T}
354: \newcommand{\D}{\delta}
355: \newcommand{\sh}[1]{sh(#1)\;}
356: \newcommand{\ch}[1]{ch(#1)\;}
357: \newcommand{\Tr}{\mbox{Tr}}
358: \newcommand{\pvac}{\; |\;\ran}
359: \newcommand{\defsign}{-}
360: \newcommand{\minusdefsign}{+}
361: \newcommand{\yy}{m}
362: \newcommand{\yz}{y}
363: \newcommand{\sep}{\zeta}
364: \newcommand{\mT}{\!\theta_}
365:
366: \newcommand{\abU}{{\mathcal U}}
367: \newcommand{\abe}{{\bf e}}
368: \newcommand{\Uqsl}[1]{U_q(sl(#1))}
369: \newcommand{\AUqsl}[1]{U_q(\widehat {sl(#1)})}
370:
371: \newcommand{\Id}{{\mathbb I}}
372: \newcommand{\Perm}[1]{{\mathcal P_{#1}}}
373: \newcommand{\KI}[1]{K^{(#1)}}
374: \newcommand{\rrep}[1]{{\cal R}_{#1}}
375: \newcommand{\ij}{i \! + \! 1}
376:
377:
378: \newcommand{\myquery}[2]{ \hrule #1 \\ {\Large #2} \hrule . \\ }
379: \newcommand{\pdef}[1]{\medskip \noindent {\small {\bf #1 }} \newline }
380: %}}}
381:
382: %{{{ Local Variables
383:
384:
385: %}}}
386: \begin{document}
387: \newlength{\fred}
388: \setlength{\fred}{10pt}
389: \newcommand{\aar}{\ar@{-}}
390: \newcommand{\headroom}{80}
391: \newcommand{\raised}{-\headroom}
392: \newcommand{\pp}[2]{\begin{picture}(#1,\headroom)(0,\raised)
393: \put(-20,0){#2} \end{picture}}
394: \newcommand{\ppp}[4]{\begin{picture}(#1,#2)(0,-#3) \put(-20,0){#4}
395: \end{picture}}
396:
397: \newcommand{\putch}{\put(-.3,-26)}
398: \newcommand{\poutch}{\put(-.3,-20)}
399: \newcommand{\hash}{\#}
400: \newcommand{\resp}{respectively}
401: \newcommand{\parker}[1]{\[ \framebox[7in]{ \mbox{\Large #1}} \]}
402: %}}}
403: %{{{ MACROS AND NOTATION
404: \newcommand{\aipic}{isotopic}
405: \newcommand{\aipy}{isotopy}
406: \newcommand{\pseudo}{ur}
407: \newcommand{\RX}{C}%%% contour algebra
408: \newcommand{\DD}{{D}}%%% diagram set hook
409: \newcommand{\JJ}{{J}}%%% unbeaded diagram set hook
410: \newcommand{\Jn}[1]{{\JJ_{#1}}}%%% unbeaded diagram set (pair partitions)
411: \newcommand{\DV}{{\DD(V)}}%%% diagrams (fixed n, S)
412: \newcommand{\DVe}[2]{{\DD(V^{#1}_{#2})}}%%% diagrams (fixed n, S)
413: \newcommand{\DVnm}{{\DD(V^n_m)}}%%% diagrams (fixed S)
414: \newcommand{\DSVnm}{{\DD_S(V^n_m)}}%%% diagrams
415: \newcommand{\Do}{{\DD^o}}%%% pseudodiagram category root
416: \newcommand{\DoV}{{\DD^o(V^{}_{})}}%%% pseudodiagrams (fixed n, S)
417: \newcommand{\DoVe}[2]{{\DD^o(V^{#1}_{#2})}}%%% pseudodiagrams (fixed S)
418: \newcommand{\DoVnm}{{\DD^o(V^n_m)}}%%% pseudodiagrams (fixed S)
419:
420: \newcommand{\Dz}[1]{{\DD^{z}_{#1}}}%%% planar diagrams (any n)
421: \newcommand{\Dnz}{{\Dz{n}}}%%% planar diagrams
422: \newcommand{\Dnzl}[2]{{\DD_{#1}^{z,#2}}}%%% planar l-exposed diagrams
423: \newcommand{\Dnp}{{\DD^p_{n}}}%
424: \newcommand{\Dnpc}{{\DD^{pc}_{n}}}%
425: \newcommand{\Dppc}[1]{{\DD^{pc'}_{#1}}}%
426: \newcommand{\Dph}{\DD^{\phi}}
427: \newcommand{\Dzo}[1]{{\JJ^{z}_{#1 }}}%%% planar unbeaded diagrams
428: \newcommand{\Jp}[1]{{\JJ_{(#1)}}}%%% periodic unbeaded diagrams
429: \newcommand{\Jpp}[1]{{\JJ'_{(#1)}}}%%% periodic unbeaded diagrams with ncloops
430: \newcommand{\DB}[1]{{\JJ^{B}_{#1}}}%%% symmetric planar unbeaded diagrams
431: \newcommand{\DBe}[1]{{\JJ^{Be}_{#1}}}%%% symm. planar unbeaded ede diagrams
432: \newcommand{\Beta}{{\mathfrak B}}
433: \newcommand{\pseud}{{\mathcal H}}%%% set H(n,m) of LR pseudodiagrams
434: \newcommand{\achira}{symplectic}
435: \newcommand{\achiral}{achiral}%%% deformed LR symm TLD
436: \newcommand{\Achira}{Symplectic}
437: \newcommand{\Achiral}{Achiral}%%% deformed LR symm TLD
438: \newcommand{\achiralb}{\achira\ blob}%%% deformed LR symm blob
439: \newcommand{\Achiralb}{\Achira\ blob}%%% deformed LR symm blob
440: \newcommand{\ASTL}{Affine symmetric TL}
441: \newcommand{\aSTL}{affine symmetric TL}
442: \newcommand{\CC}{CC}
443: \newcommand{\Deltag}{\Gamma}
444: \newcommand{\sS}{\mathcal{S}}
445: \newcommand{\hati}{\tilde}
446: \newcommand{\V}{V}
447:
448: %}}}
449: %{{{ titles
450: \author{Paul Martin%
451: \footnote{Centre for Mathematical Science, City University, Northampton
452: Square, London EC1V 0HB, UK},
453: R. M. Green%.
454: \footnote{Department of Mathematics, University of Colorado,
455: Campus Box 395, Boulder, Colorado 80309-0395, USA},
456: and Alison Parker%
457: \footnote{Department of Mathematics, University of Leicester,
458: Leicester, LE1 7RH, UK}.
459: }
460: \date{}
461: %}}}
462: \maketitle
463: %{{{ abstract
464: \vspace{5cm} \begin{abstract}
465: The recollement approach to the representation
466: theory of sequences of algebras is extended to pass basis information
467: directly through the globalisation functor.
468: The method is hence adapted to treat sequences that are not
469: necessarily towers by inclusion, such as \achiralb\ algebras
470: (diagram algebra quotients of the type-$\hati{C}$ Hecke algebras).
471:
472: By carefully reviewing the diagram algebra construction,
473: we find a new set of functors interrelating
474: module categories of ordinary blob algebras
475: (diagram algebra quotients of the type-${B}$ Hecke algebras)
476: at {\em different} values of the algebra parameters.
477: We show that these functors generalise
478: to determine the structure of
479: \achiralb\ algebras, and hence of certain
480: two-boundary Temperley-Lieb algebras
481: arising in Statistical Mechanics.
482:
483: We identify the diagram basis with a cellular basis for each
484: symplectic blob algebra, and prove that these algebras are
485: quasihereditary over a field for almost all parameter choices,
486: and generically semisimple.
487: (That is, we give bases for all cell and standard modules.)
488: \end{abstract} \newpage
489: %}}}
490: {\small \tableofcontents}
491: \section{Introduction}
492: %{{{ intro
493: \newcommand{\Ring}{K}
494: %{{{ INTRO
495:
496: The idea of recollement \cite{BeilinsonBernsteinDeligne81}
497: is applied to categories of modules in
498: \cite{ClineParshallScott88}.
499: Iterated `towers' of recollement are used in
500: algebraic representation theory in \cite{Martin91} and
501: formalised, for example, in \cite{CoxMartinParkerXi03}.
502: (The {\em tower} here refers to the algebraic structure needed for
503: statistical mechanics \cite{Martin91}, although an elementary connection can be made
504: in the semisimple case to Jones basic construction \cite{GoodmanDelaharpeJones89}.)
505:
506: If $A$ is an algebra, and
507: $e \in A$ an idempotent, then the category $eAe$-mod
508: of left $eAe$-modules fully embeds in $A$-mod.
509: At its most basic the idea is that if $eAe$-mod
510: may be relatively simply analysed, the embedding then gives partial
511: knowledge of $A$-mod.
512: The standard tower picture has $A$ as
513: part of a tower of algebras by inclusion, such that $eAe$ may be identified
514: isomorphically with one of the subalgebras.
515: The interplay of induction/restriction and
516: globalisation/localisation functors facilitates representation
517: theory in such a tower.
518: This is the device discussed in \cite{CoxMartinParkerXi03}.
519:
520: %}}}
521: %{{{ INTRO I
522:
523: All this would be of academic interest only, were it not for the
524: ubiquity of such towers `in nature'.
525: Transfer matrix algebras are algebras whose
526: representation matrices build statistical mechanical transfer
527: matrices \cite{Martin91}. The stability of the thermodynamic limit corresponds
528: (to cut a long story short) to the
529: existence of a tower of module-categorical embeddings.
530: However \cite{CoxMartinParkerXi03} addresses only one way
531: among many in which
532: such a tower could occur.
533:
534: A further limitation
535: in the formalisation of \cite{CoxMartinParkerXi03} is
536: that it concentrates on the abstract module category tower, and does
537: not incorporate the tower of special module bases found in concrete
538: examples (such as in \cite{MartinSaleur94a,Martin96},
539: and cf. \cite{GrahamLehrer96}).
540: This paper
541: describes two ways in which the framework formalised in
542: \cite{CoxMartinParkerXi03}
543: may be extended, so as to treat
544: the representation theory of a wider class of algebras.
545: Firstly we integrate the module category tower framework
546: with special algebra bases
547: (as in diagram algebras, for example).
548: This allows us to enumerate explicit bases for modules and
549: algebras, rather than simply to generate structure theorems.
550: Secondly we show that the framework is useful even when the tower is
551: not (necessarily) a tower of inclusion.
552: Indeed the control of basis compensates for the lack of induction and
553: restriction functors, so the framework will work for algebra towers
554: without induction and restriction.
555: This latter is important for treating
556: our motivating examples:
557: families of algebras arising
558: recently in the Physics of systems with special boundaries \cite{DeGier02},
559: which do not include
560: (generalising the ordinary Temperley-Lieb
561: algebras \cite{TemperleyLieb71}
562: and their immediate family, which do include \cite{Martin91}).
563: We demonstrate the method by determining the structure of these algebras.
564: (In the process we also introduce and make use of functors relating
565: module categories for algebras differing by the choice of
566: specialisation of a deformation parameter --- certain choices of
567: specialisation then being treatable together in `meta-categories'.)
568:
569: %}}}
570: \bigskip
571: %{{{ INTRO II (algebra specifics) One may deform
572:
573: \noindent
574: One may
575: deform the ordinary \TL\ diagram algebra by two-colouring the diagrams
576: (seen as maps --- see later) and assigning different parameters to
577: shaded and unshaded loops.
578: The result is isomorphic to the original.
579: However deforming the B-type (left-right symmetric) subalgebra similarly
580: constructs an algebra with a true extra parameter --- the blob algebra.
581: Varying the extra parameter, the blob algebra may be used to build
582: the representations of the periodic \TL\ algebra
583: (\cite{MartinSaleur93,MartinSaleur94a,FanGreen99,Green98,GrahamLehrer03}).
584: Thus all these algebras can be analysed using the included-tower
585: technology \cite{CoxMartinParkerXi03}.
586: The next natural generalisation is the \achiralb\
587: algebra
588: (left-right symmetric {\em and} periodic).
589: This sequence of algebras cannot be made to include in an appropriate way,
590: and so presents a suitable challenge for our method.
591:
592: %}}}
593: %{{{ INTRO III
594:
595: In this paper we implement a {\em towers of recollement} programme,
596: and variations, to determine the structure
597: (bases and representation theory) of three interesting algebras.
598: Many workers have considered wreath-like extensions for Brauer,
599: Temperley--Lieb diagram
600: \cite{JonesPlanar,MartinSaleur94a,Green98b,FanGreen99} and even
601: partition algebras \cite{Bloss04}.
602: These extensions %%have been considered
603: are of interest
604: as testing grounds for techniques intended to be applied in the representation
605: theory of more classical objects, such as the symmetric group.
606: For example, one approach to systems of algebras with
607: Jucys--Murphy-like elements \cite{Jucys74,Murphy81,Cherednik91}
608: is to consider extensions of the
609: algebras by new commuting generators which obey relations emulating
610: identities obeyed by the Murphy elements.
611: With careful preparation these extensions behave like wreaths
612: (confer \cite{Nazarov96,HaringOldenburg99,CoxMartinParkerXi03}).
613: Here, considering the most general case of nonabelian wreath algebras,
614: we also explore a new and intriguing set of interconnections,
615: taking us to the \achiralb\ algebra.
616: This leads in particular to applications in boundary integrable
617: statistical mechanics \cite{DeGier02}.
618:
619: %}}}
620: %{{{ Planar diagram algebras
621:
622: Planar diagram algebras (such as ordinary \TL) have been much studied
623: (with useful consequences in both representation theory and physics),
624: as have `fully non-planar' algebras such as the partition algebra.
625: These represent simplifying extremes in a range of generally harder
626: problems.
627: As mentioned above,
628: certain annular algebras can be brought into the
629: planar framework, using the blob algebra \cite{MartinSaleur94a}.
630: The algebras we focus on
631: here are (in a suitable sense --- see later) mildly non-planar diagram
632: algebras.
633: These are harder to treat, but not intractable, as we shall
634: demonstrate.
635:
636: %}}}
637: %{{{ aims
638:
639: The motivating aims of the paper are:
640: \\
641: 1. To provide an organisational framework for unifying the
642: representation theory of various
643: forms of periodic/annular/type-B/boundary TL algebras
644: studied in the literature
645: \cite{MartinSaleur93,MartinSaleur94a,tomDieck94,DeGier02}.
646: (The ordinary \TL\ algebra is a nexus for many branches of mathematics,
647: with isomorphic algebras constructed in areas such as:
648: factors \cite{JonesPlanar},
649: representation theory, combinatorics,
650: statistical mechanics \cite{TemperleyLieb71,Baxter82}.
651: Variants appear in these contexts, but are no longer all isomorphic,
652: and connections between them are not yet fully understood.)
653: \\
654: 2. To provide the representation theoretic formalism for analysing these
655: algebras.
656:
657:
658:
659:
660: %}}}
661: %{{{ layout
662:
663: The layout of the paper is as follows.
664: In section~\ref{cat} we introduce the general theory necessary to
665: pass specific basis elements, for special types of module,
666: between layers of a globalisation tower
667: (irrespective of inclusion).
668: We also discuss how certain other features
669: of modules which we shall use later
670: behave under globalisation.
671: \\
672: In section~\ref{s diag} we collect the definitions of
673: one of the families of diagram algebras which we shall need.
674: All of these are based on Weyl's diagrams
675: for the Brauer algebra \cite{Weyl46}.
676: To help prepare the ground for later more elaborate constructions
677: we also point out some paradigms for diagram algebra construction.
678: For example:
679: combinatorial sets with diagrammatic realisations
680: (which form bases for algebras via diagram concatenation),
681: containing topologically characterised subsets respected by concatenation;
682: with the resultant subalgebras amenable to deformations not tolerated
683: by the original combinatorial algebra.
684: \\
685: In section~\ref{blob'} we focus on deformations of Temperley--Lieb
686: algebras --- again looking at subalgebras and deformations.
687: We use these to establish homomorphisms between various families of
688: algebras.
689: \\
690: In section~\ref{recol b} we use the alternate realisations established above
691: to construct new functors in the tower of blob algebras, and hence to
692: relate categories of modules for blob algebras with different
693: values of the defining parameters (and hence to
694: analyse their representation theory).
695: \\
696: In sections~\ref{ASTLA} and~\ref{ASTLA'}
697: we define affine symmetric Temperley--Lieb
698: algebras and \achiralb\ algebras, and relate their categories of
699: modules.
700: \\
701: In section~\ref{rep ASTLA} we investigate the representation theory of affine
702: symmetric Temperley--Lieb algebras, using results from all the
703: previous sections. The main results here are perhaps the simple indexing
704: Theorem~\ref{index wangy}, and the generic structure results of
705: section~\ref{wa}. Beyond the semisimple cases, we show that the algebra
706: is almost always quasihereditary, and give bases for the standard modules.
707:
708: %}}}
709: %{{{ endbit
710:
711: Several of the incarnations of the \TL\ algebra have
712: an interesting `periodic' generalisation, as noted above,
713: but these are much harder to treat.
714: The blob algebra is a device that largely solves this problem
715: --- casting the representation theory of (infinite) periodic \TL\
716: algebras into the
717: setting of a (finite) generalisation of TL with properties much closer
718: to the original.
719: The blob algebra suggests various generalisations of its own, but
720: these
721: are once again rather harder to treat, and have (until now) lacked the
722: motivations of the blob algebra (i.e.
723: its simple but beautiful representation theory;
724: its application to periodic \TL\ and hence affine Hecke algebras).
725: Recently, though, both the blob algebra and its two-boundary \TL\
726: algebra generalisation have arisen in the treatment of boundary
727: integrable systems in Statistical Mechanics
728: \cite{DeGier02,NicholsRittenbergdeGier05},
729: suggesting that category embedding methods should work here.
730:
731: %}}}
732: \subsection{Preliminary definitions}\label{INot}
733: %{{{ coxeter/weyl/hecke
734:
735: A Coxeter graph is any finite undirected
736: graph without loops (that is, without edges that begin and end on the
737: same vertex).
738: For example:
739: \[
740: A_n: \;\; \raisebox{-.1in}{\includegraphics{xfig/AnGraph.eps}}
741: \qquad\qquad
742: B_n: \;\; \raisebox{-.1in}{\includegraphics{xfig/BnGraph.eps}}
743: \]
744: The Coxeter Artin system of Coxeter graph $G$ is a pair $(B,S)$
745: consisting of a group $B$ and a set of pairs
746: $g_v, g_v^{-1}$ of generators of $B$ labelled by
747: the vertices $v$ of $G$, with relations
748: \[
749: g_v g_{v'} ... \; = \; g_{v'} g_v ...
750: \]
751: where the number of factors on each side is two more than the number
752: of edges between $v$ and $v'$.
753:
754: The Coxeter system of $G$ is a pair $(W,S)$ where $W=W(G)$ is the quotient
755: of $B$ by the relation $g_v = g_v^{-1}$.
756:
757: For $K$ a ring, let $q_v$ be an invertible element in $K$, for each
758: $v \in G$, such that $q_v = q_{v'}$ if $g_v$ conjugate to $g_{v'}$ in $B$.
759: Let $Q_v = (g_v - q_v)(g_v + q_v^{-1})$.
760: The generic Hecke algebra of $G$ is
761: \[
762: H(G) = K B / \langle Q_v \; | \; v \in G \rangle
763: \]
764: Examples: If $G=A_n$ then all generators are conjugate and we have the
765: (one-parameter) Hecke algebras of type-$A$.
766: If $G=B_n$ there is one generator not conjugate to the rest and we have the
767: (two-parameter) Hecke algebras of type-$B$.
768: If $G=\hati{C_n}$
769: there are two generators not conjugate to the rest,
770: or to each other, and we have the
771: (three-parameter) Hecke algebras of type-$\hati{C}$.
772:
773: %}}}
774: %{{{ presentation
775:
776: Each of these algebras has an algebra homomorphism onto $K$ defined by
777: \[
778: \rho_+ (g_v) = q_v
779: \]
780: Note that for $q_v=1$ the relation $Q_v=0$ is $g_v^2=1$ and hence a
781: group relation, so that $H(G)$ is the group algebra of $W(G)$ in this case.
782: It will be convenient to write $\sigma_v$ for $g_v$ in the group
783: case. We have \cite[Ch.7]{Humphreys90} that if
784: $w= \sigma_{i_1} ...\sigma_{i_l}$
785: is a reduced expression in $W(G)$ then
786: $\{ T_w = g_{i_1} ...g_{i_l} \; | \; w \in W(G) \}$ is a basis for $H(G)$ in general.
787: Define the {\em symmetrizer}
788: \[
789: E_G = \sum_{w \in W(G)} \rho_+(w) T_w
790: \]
791: in $H(G)$.
792: For example if the vertices of $A_2$ are labelled from $\{1,2\}$ we
793: have
794: \[
795: E_{A_2} = 1+ q (g_1 + g_2) + q^2 (g_1g_2 + g_2 g_1) +q^3 g_1 g_2 g_1 .
796: \]
797: Defining $u_v = (g_v + q_v^{-1})$, so that $g_v u_v = q_v u_v$, we have
798: \[
799: E_{A_2}
800: = q^3 ( u_1 u_2 u_1 - u_1 ) .
801: \]
802: If the vertices of $B_2$ are labelled from $\{ 0,1 \}$ we have
803: \[
804: E_{B_2} = 1 + q_1 g_1 +q_0 g_0 + q_0 q_1 (g_0 g_1 + g_1 g_0 )
805: +q_0 q_1 (q_0 g_0 g_1 g_0 + q_1 g_1 g_0 g_1 ) +(q_0q_1)^2 g_0 g_1 g_0 g_1
806: \]
807: \eql(uuuu)
808: = u_0 u_1 u_0 u_1 - \frac{q_0^2 + q_1^2}{q_0 q_1} u_0 u_1 .
809: \eq
810:
811:
812: Note that the relations of $H(G)$ are invariant under the parameter transformation
813: \[
814: s_v : q_v \mapsto -q_v^{-1}
815: \qquad \mbox{ and }
816: s_v : q_w \mapsto q_w \qquad \mbox{ for $g_v,g_w$ not in the same class}
817: \]
818: (that is, there is one such transformation for each parameter).
819: For each parameter transformation $s_w$ there are, in addition to $\rho_+$, further
820: homomorphisms of $H(G)$ onto $K$:
821: \[
822: \rho_w (g_v) = s_w (q_v) .
823: \]
824:
825: %}}}
826: %{{{ parabolic
827:
828:
829: Any subset of $S$ generates a parabolic subalgebra of $B$ or of $H(G)$.
830: If $v,v' \in G$ have at least one edge between them define
831: $E_{vv'}$ as the symmetrizer on their parabolic subalgebra of $H(G)$;
832: else $E_{vv'} =0$.
833: Then define
834: \[
835: T(G) = H(G)/ \langle E_{v,v'} \; | \; v,v' \in G \rangle .
836: \]
837: Example: $T(A_n)$ is the ordinary Temperley--Lieb algebra
838: \cite{TemperleyLieb71}.
839:
840: Not much is known about $H(G)$ or $T(G)$ for general $G$ (although see
841: \cite[Ch.9]{Martin91}), but the cases in which $G$ is positive
842: definite or positive semidefinite are relatively tractable (although
843: still interesting)
844: \cite{MartinSaleur94a,GrahamLehrer98,FanGreen99,ErdmannGreen99,Green03}.
845:
846:
847: %}}}
848: \medskip
849: %{{{ presentation.tex
850:
851:
852: \newcommand{\vardelta}{\delta_e}% ???
853: \newcommand{\vargamma}{\gamma}%
854: \newcommand{\uU}{U}
855: \newcommand{\ue}{e}
856:
857:
858: For $K$ a ring, $x$ an invertible element in $K$,
859: $q=x^2$, and $\vargamma,\vardelta \in K$,
860: define $TLb_n^K$ to be the $K$--algebra with
861: generators $\{ 1,e,\uU_1,\ldots,\uU_{n-1} \}$ and relations
862: \begin{eqnarray}
863: \uU_i \uU_i &=& (q+q^{-1}) \uU_i \label{TL001} \\
864: \uU_i \uU_{i\pm 1} \uU_i &=& \uU_i \label{TL002} \\
865: \uU_i \uU_j &=& \uU_j \uU_i \hspace{1in} \mbox{($|i-j|\neq 1$)} \label{TL003}
866: \end{eqnarray}
867: \begin{eqnarray}
868: \uU_1 e \uU_1 &=& \vargamma \uU_1 \label{TL004} \\
869: e e &=& \vardelta e \\
870: \uU_i e &=& e \uU_i \hspace{1in} \mbox{($i \neq 1$)} . \label{TL006}
871: \end{eqnarray}
872:
873:
874: %}}}
875: %{{{ rescaled
876:
877: Note that $e$ can be rescaled to change $\vargamma$ and
878: $\vardelta$ by the same factor.
879: Thus, if we require that $\vardelta$ be invertible, then we might as well
880: replace it by 1.
881: This brings us to the original two--parameter presentation of the
882: algebra,
883: sometimes known as
884: the blob algebra by presentation, or
885: the two--parameter \TL\ algebra of type B.
886: The subalgebra generated by $\{ 1,\uU_1,\ldots,\uU_{n-1} \}$
887: is $T(A_n)$, the \TL\ algebra of type A,
888: sometimes denoted $TL_n^{K}$.
889:
890: For $k$ a field that is a $\Ring$--algebra define
891: $TLb_n = k \otimes_{\Ring} TLb_n^{\Ring}$.
892:
893: %}}}
894: %{{{ isomorphisms
895:
896: The type-A algebra is isomorphic to the well known
897: ordinary \TL\ diagram algebra \cite{Martin91,Graham95};
898: and the algebra $TLb_n$ is isomorphic to the blob diagram algebra $b_n$
899: \cite{MartinSaleur94a,Graham95,Green98,FanGreen99}
900: (see section~\ref{dir homs}).
901: Because of these isomorphisms
902: it is common to use generators and diagrams interchangeably.
903:
904:
905:
906: %}}}
907: %{{{ %opposite
908:
909:
910: %}}}
911: %{{{ duality
912:
913: There is also
914: a {\em periodic} \TL\ diagram algebra (TLDA).
915: That is, a TLDA
916: defined using certain periodic TL diagrams.
917: Continuing the above `duality',
918: the {\em periodic TLA},
919: on the other hand, is defined by abstract generators and relations.
920: See
921: \cite{PasquierSaleur90,Levy91,Martin91,MartinSaleur93,MartinSaleur94a,%
922: FanGreen99,DeGier02}
923: and references therein for details of both.
924: The relationship between the two versions is not so
925: straightforward as in the ordinary TLA case.
926: See also \cite{Jones94b,GrahamLehrer98}.
927:
928: To summarize the
929: naming conventions: \TL\ algebras are defined by
930: generators and relations. Blob, contour, partition algebras (and
931: others with diagram suggestive names) are defined via bases of
932: diagrams and diagrammatic composition rules.
933:
934: %}}}
935: %{{{ pr hecke quotient
936: \prl(hecke)
937: The map $u_i \mapsto U_i$ ($i>0$), $u_0 \mapsto e$,
938: extends to an algebra homomorphism $\phi$ from $T(B_n)$
939: to $TLb_n^K$ in the case
940: $q=q_1$,
941: $\delta_e= q_0+q_0^{-1}$
942: and $\gamma = \frac{q_0^2 + q^2}{q_0 q}$.
943: \end{pr}
944: {\em Proof:} The relation checking is largely routine.
945: Note from (\ref{uuuu}) that
946: $\phi(u_1u_0u_1 - \frac{q_0^2 + q_1^2}{q_0 q_1} u_1)$ vanishing is
947: sufficient to ensure
948: $\phi(E_{B_2})=
949: \phi(u_0(u_1u_0u_1 - \frac{q_0^2 + q_1^2}{q_0 q_1} u_1))=0$.
950: \Qed
951:
952: \medskip
953:
954: Several authors have used this so-called `blob' homomorphism to investigate
955: Hecke algebra representation theory in the type-$B$ and type-$\hati{A}$
956: cases
957: \cite{GrahamLehrer03,CoxGrahamMartin03,MartinWoodcock03}.
958: One final way to view the present paper is as
959: a similar `blob' approach to type-$\hati{C}$.
960: (The three parameter affine-$C$ Hecke algebra itself is of interest for a
961: variety of reasons ---
962: see for example \cite{Lusztig89C,Sahi99} and references therein.)
963:
964: %}}}
965: \subsection{Summary of notations}\label{SINot}
966: %{{{ index of notations
967:
968: For the reader's reference we summarize here the
969: notations for algebras used in this paper
970: (and indicate the section in which each is defined):
971: \\
972: $b_n$ blob algebra, section~\ref{dir homs} \\
973: $b'_n$ \achiral\ algebra
974: section~\ref{blob'} \\
975: $b_n^x$ \achiralb\ algebra, section~\ref{ASTLA} \\
976: $b_n^{x'}$ big \achiralb\ algebra, section~\ref{ASTLA} \\
977: $b_{2m}^{\phi}$ affine symmetric \TL\ diagram algebra, section~\ref{ASTLA'} \\
978: $\Beta_n$ Brauer algebra, section~\ref{s diag} \\
979: $C_{n,m}(l)$ contour algebra, section~\ref{s contour} \\
980: $C_{n}^{\sim}(l)$ generalised contour algebra, section~\ref{s contour} \\
981: $H(G)$ generic Hecke algebra of graph $G$, section~\ref{INot} \\
982: $TL_n$ \TL\ algebra of type-A, section~\ref{INot} \\
983: $TLb_n$ \TL\ algebra of type-B, section~\ref{INot} \\
984: $T(G)$ a quotient of $H(G)$, section~\ref{INot} \\
985:
986: %}}}
987: %{{{ figure schemat
988:
989: The relationship between these algebras
990: is indicated by the
991: schematic in figure~\ref{schemat}.
992:
993: \begin{figure}
994: \[
995: \xymatrix{
996: && \ar@/_1pc/@{.}[ddddd]
997: \\
998: & \txt{Defined by \\presentation:} && \txt{Defined by \\diagram:}
999: \\
1000: & \txt{Ordinary TL \\ $TL_n$} \ar@{<->}_{\sim}[rr] \ar@{^{(}->}[d]
1001: && \txt{Ordinary TL \\ diagram} \ar@{<->}_{\sim}[r] \ar@{^{(}->}[d]
1002: & \txt{shaded loop \\ deformed \\ TLDA} \ar@{^{(}->}[d]
1003: \\
1004: & \txt{$TLb_n$} \ar@{<->}_{\sim}[rr] \ar@{^{(}->}[d]
1005: && \txt{blob $b_n$} \ar@{<->}_{\sim}[r] \ar@{^{(}->}[d]
1006: & \txt{\achiral\ algebra \\ $b'_{2n}$}
1007: \ar@{^{(}->}[d] \ar@{<->}_{\sim}[r]
1008: & \txt{contour \\ $\RX^{}_{n,2}(1)$} \ar@{^{(}->}[dd]
1009: \\
1010: & \txt{2-Boundary TL} \ar@{->>}_{}[rr]
1011: && \txt{\achiralb\ \\ $b^x_n$} \ar@{<->}_{\sim}[r]
1012: & \txt{affine \\ symmetric \\ TLDA \\ $b^{\phi}_{2n}$}
1013: \\
1014: &&& \txt{$b^{x'}_n$} \ar@{->>}_{}[u] \ar@{^{(}->}[rr]
1015: && \txt{generalised \\ contour \\ $\RX^{\sim}_{n}(l,r)$} }
1016: \]
1017: \caption{\label{schemat} Algebra relationships}
1018: \end{figure}
1019:
1020: %}}}
1021: %{{{ glossary
1022:
1023:
1024: \newcommand{\gloss}[3]{$#1 \;$ #2, section~\ref{#3}\\ }
1025:
1026: \noindent
1027: A glossary of other notations used for sets may also be useful:
1028: \\
1029: \gloss{B_n}{set of blob diagrams}{dir homs}
1030: \gloss{B_n^x}{set of left-right blob diagrams}{pdcat}
1031: \gloss{B_n^{\phi}}{set of left-right symmetric reduced periodic pseudodiagrams}{reduc}
1032: \gloss{\DV}{set of beaded diagrams (given vertex set $V=V_n$ and bead set $S$)} {brauer}
1033: \gloss{\DoV}{beaded pseudodiagrams}{brauer}
1034: \gloss{\Dnz}{planar diagrams}{s contour}
1035: \gloss{{\Dnzl{n}{l}}}{planar $l$-exposed diagrams}{s contour}
1036: \gloss{D_n=D_{n,m}}{planar diagrams with $\Z_m$-beads}{s contour}
1037: \gloss{\Dnp}{cylinder embeddable diagrams}{deform}
1038: \gloss{\Dnpc}{isotopy classes of concrete cylinder diagrams}{deform}
1039: \gloss{\Dppc}{%isotopy
1040: classes of concrete cylinder diagrams
1041: including non-contratible loops}{deform}
1042: \gloss{\Dph_n}{classes of left-right symmetric concrete cylinder diagrams}{Dph}
1043: \gloss{{\Jn{n}}}{unbeaded diagrams}{brauer}
1044: \gloss{{\Dzo{n}}}{planar unbeaded diagrams}{dir homs}
1045: \gloss{\Jp{n}}{periodic unbeaded diagrams}{deform0}
1046: \gloss{\DB{n}}{symmetric planar unbeaded diagrams}{deform2}
1047: \gloss{\DBe{n}}{symmetric planar unbeaded $ede$ diagrams}{recol b}
1048:
1049:
1050: %}}}
1051: %}}}
1052: \section{Category theory preliminaries} \label{cat}
1053: %{{{ cat thy
1054: %{{{ category theory
1055:
1056: The starting point at the category theory level is as follows.
1057: Given an algebra
1058: $A$ and an idempotent $e \in A$ then $eAe$ is also an algebra, and
1059: $Ae$ is a left $A$ module and a right $eAe$ module.
1060: Thus we may define functors between the category $A$-mod,
1061: of left $A$-modules, and $eAe$-mod:
1062: \newcommand{\glob}{G}
1063: \newcommand{\loc}{F}
1064: \begin{eqnarray}
1065: \label{glob}
1066: \glob: \mbox{$eAe$-mod} & \longrightarrow & \mbox{$A$-mod}
1067: \\ \non
1068: M & \mapsto & Ae \otimes_{eAe} M
1069: \end{eqnarray}
1070: \begin{eqnarray}
1071: \label{loc}
1072: \loc: \mbox{$A$-mod} & \longrightarrow & \mbox{$eAe$-mod}
1073: \\ \non
1074: N & \mapsto & eN
1075: \end{eqnarray}
1076: with various powerful properties
1077: (summarized in \cite{Green80,MartinRyom02}).
1078: In particular if $A$ is an algebra over a field then
1079: $\loc$ is exact and $\glob$ is right exact.
1080: Further, the image of a simple
1081: module under $\glob$ has simple head.
1082: \prl(grebo)\mbox{\rm \cite{Green80}}
1083: Let $\{ S_{\lambda} \; | \; \lambda \in \Lambda \}$ be a complete set
1084: of inequivalent simple (left) modules of $eAe$ over some field.
1085: Then $\{ \head(G(S_{\lambda})) \; | \; \lambda \in \Lambda \}$
1086: is a set of inequivalent simples of $A$, and any simple $A$-module $S$
1087: in an equivalence class not represented in this set obeys $eS=0$.
1088: \end{pr}
1089:
1090: %}}}
1091: %{{{ prestandard
1092: \newcommand{\prestandard}{prestandard}
1093: \newcommand{\Prestandard}{Prestandard}
1094: \subsection{\Prestandard\ modules}
1095:
1096: The functor $\loc$ is called localisation, and $\glob$ is
1097: globalisation, with respect to $e$.
1098: (We may write $\loc_{e},\glob_{e}$ where convenient.)
1099: Suppose that we are given an idempotent $e$ in an algebra $A$, and
1100: that $S$ is a simple module of $eAe$.
1101: Then $\glob(S)$ is called the {\em \prestandard} module of $A$ associated
1102: to $S$ by $e$.
1103:
1104: Indeed, suppose that $e=e_1e_2=e_2e_1$ ($e_1,e_2$ also
1105: idempotent). Then the sequence of idempotents $1,e_1,e$ defines a
1106: sequence of algebras $A,e_1Ae_1,eAe$.
1107: A \prestandard\ module $M = \glob_{e_2}(S)$ of $e_1Ae_1$
1108: will globalise to a \prestandard\ module of $A$.
1109: On the other hand if $M \neq \head\!({M}) $ then these two
1110: will not necessarily globalise to the same module
1111: in $A$
1112: (although both are of \prestandard\ type by construction).
1113: This makes the \prestandard\ notion less canonical,
1114: although more general, than the {\em standard} modules of
1115: quasihereditary algebras, for example.
1116:
1117: %}}}
1118: %{{{ prl orestes
1119: \prl(orestes)
1120: If $M$ is a \prestandard\ $A$-module then
1121: \\
1122: (i) it has simple head
1123: $L_M$ (say), and if $M_0$ is the maximal proper submodule then $M_0$
1124: does not contain $L_M$ as a simple composition factor;
1125: \\
1126: (ii) if $A$ has an involutive antiautomorphism defined on it
1127: fixing $e$
1128: then $M$ has at most one
1129: contravariant form defined on it,
1130: up to scalars,
1131: and the rank of any such form is the dimension of $L_M$.
1132: \end{pr}
1133: {\em Proof:} (i): Only the last claim remains to be proven. Suppose that
1134: $M=G(S)$ for simple $eAe$-module $S$, then $F(M)=eM \cong S$.
1135: In particular $e$ acts as zero on all but one simple factor in $M$.
1136: Now suppose there exists a proper submodule $M'$ of $M$.
1137: If $eM' \neq 0$ then $eM'=S=eG(S)$, so $AeM'=AeG(S)=G(S)$,
1138: which would imply $M' \supseteq G(S) \supset M'$ --- a contradiction.
1139: Thus $F$ kills every proper submodule of $M$, so $e L_M \neq 0$.
1140: \\
1141: (ii): There is a one-to-one correspondence between such forms and
1142: homomorphisms from $M$ to its contravariant dual $M^{\times}$,
1143: but by (i)
1144: there is at most one such homomorphism (up to scalars),
1145: whose image is $L_M$.
1146: To see this note that by the proof of (i)
1147: $eL_N=0$ for every simple factor $L_N$ of $M$ not in the head.
1148: Let $L_M^{\times}$ denote the contravariant dual of $L_M$, a simple module.
1149: Since $eL_M \neq 0$ we have $eL_M^{\times}\neq 0$ so neither $L_M$ nor
1150: $L_M^{\times}$ appears below the head in $M$.
1151: Thus $L_M$ does not appear above the socle in $M^{\times}$ and a homomorphism
1152: $M \rightarrow M^{\times}$ is only possible if it maps the head $L_M$ of $M$
1153: to the socle $L_M^{\times}$ of $M^{\times}$, with $L_M \cong L_M^{\times}$.
1154: \Qed
1155:
1156:
1157: Propositions~\ref{grebo} and~\ref{orestes}
1158: and the exactness properties make \prestandard\
1159: modules potentially useful modules to study,
1160: in representation theory.
1161: In this paper we will encounter various
1162: modules (for algebras) with useful natural bases.
1163: It will be convenient, where possible,
1164: to be able to identify these as \prestandard.
1165:
1166: %}}}
1167:
1168: \subsection{Globalisation and balanced maps}
1169: %{{{ unpack a bit (balanced maps)
1170:
1171:
1172: The exactness properties of $\glob$ and $\loc$ are standard
1173: results (see \cite{MartinRyom02,Green80}).
1174: However it is worth unpacking a little before we go on,
1175: since some of the mechanics will be used later.
1176: The first thing to recall is the notion of balanced map
1177: \cite{Chevalley56,CurtisReiner62}. For $M$ a right module and $N$ a
1178: left module of a ring $R$ with 1, a balanced map $f$ of $M \times N$ into
1179: an additive abelian group $P$ is a map such that
1180: $f (m+m',n)=f(m,n)+f(m',n)$,
1181: $f(m,n+n')=f(m,n)+f(m,n')$,
1182: $f(m,rn)=f(mr,n)$.
1183:
1184: The map that takes $(m,n)$ to $m \otimes n \in M \otimes_R N$ is a
1185: balanced map.
1186: If $f:M \times N \rightarrow P$ is balanced then there is a
1187: homomorphism $f^* : M \otimes_R N \rightarrow P$ such that
1188: $f^*(m \otimes n) = f(m,n)$
1189: (in fact $f^*$ is uniquely determined by $f$) \cite{CurtisReiner62}.
1190:
1191: Now consider
1192: \[
1193: \loc( \glob (N)) = eAe \otimes_{eAe} N \stackrel{\mu}{\rightarrow} N
1194: \]
1195: where $\mu$ is derived from the (NB, balanced) map
1196: \[
1197: (a,n) \mapsto an .
1198: \]
1199: We may define a homomorphism $\nu: N \rightarrow eAe \otimes_{eAe} N$
1200: by
1201: \eql(inve)
1202: \nu (n) = e \otimes n
1203: \eq
1204: Obviously $\mu \nu$ is the identity map on $N$; and
1205: \[
1206: \nu ( \mu ( a \otimes n) = \nu( an) = e \otimes an = a \otimes n
1207: \]
1208: since $a \in eAe$, so $\mu$ is an isomorphism.
1209:
1210: %}}}
1211: %{{{ multiplication map / return ideal
1212:
1213: Similarly we have that
1214: \eql(iso x)
1215: G(eAe) = Ae \otimes_{eAe} eAe \cong Ae .
1216: \eq
1217: More generally, suppose that $S$ is a left sub-$eAe$-module of $eAe$
1218: (i.e. a left ideal),
1219: then there is a multiplication map
1220: \begin{eqnarray*}
1221: \mu : Ae \otimes_{eAe} S & \rightarrow & AeS \\
1222: ae \otimes s & \mapsto & aes
1223: \end{eqnarray*}
1224: (in the rest of this section, $\mu$ applied to a tensor product
1225: of this form will
1226: always be the appropriate multiplication map)
1227: however, this surjection need not be an injection in general.
1228: (There is a grotesque example in \cite{CurtisReiner62}.)
1229: The issue is the construction of the `inverse' as in (\ref{inve}).
1230: The `identity element' $e \in eAe$ will not generally lie in $S$.
1231: On the other hand, suppose that there are $f,g \in eAe$ such that
1232: $S=eAef$ and $fgf=f$.
1233: (Such an $f$ is said to satisfy the {\em return condition}.
1234: We call a left $eAe$-ideal $S$ of form $eAef$ with $f$ satisfying
1235: the return condition a {\em return ideal}.)
1236: Then there is a map
1237: $\nu: AeS \rightarrow Ae \otimes_{eAe} S$ given by
1238: \[
1239: \nu(x) = x \otimes gf
1240: \]
1241: so that $\mu(\nu(x)) = xgf=x$ and
1242: $\nu(\mu(a \otimes s))= \nu(as) = as \otimes gf= a \otimes s$.
1243: Therefore
1244: \prl(lemmin)
1245: If $S=eAef$ is a left ideal of $eAe$ generated by $f \in eAe$ such
1246: that $fgf=f$ for some $g \in eAe$, then the multiplication map $\mu$
1247: is an isomorphism
1248: \[
1249: G(S) \cong AeS = AS = Af
1250: \]
1251: (NB, $\mu$ and its inverse are given explicitly).
1252: In particular the set inclusion of $S$ in $AS$ passes to an injection
1253: $\nu$ of $S$ into $G(S)$. This is not an algebra-module map,
1254: but if $D$ is a linearly independent set in $S$ then it is linearly
1255: independent in $AS$ and $\nu(D)$ is in $G(S)$.
1256: \end{pr}
1257: Note that $fg$ is idempotent, so
1258: \[
1259: S = eAef \doublerightarrow eAefg
1260: \]
1261: is a surjective map to a projective $eAe$-module.
1262:
1263: %}}}
1264:
1265: \subsection{Module bases under globalisation}
1266: %{{{ bases
1267:
1268: The functors $\loc,\glob$ are tools for analysing categories of
1269: modules,
1270: rather than specific bases or representations.
1271: However,
1272: following the discussion above, there are realistic cases in which
1273: one can use a basis for $S$ to construct a basis for $\glob(S)$.
1274: (This is particularly so for diagram algebras, which come with a
1275: diagram basis for elements $f$ of which the return condition $fgf=f$ is
1276: always true for some algebra element $g$. Indeed $g$ can usually be
1277: chosen a basis element, or else a scalar multiple thereof.)
1278:
1279: %}}}
1280: %{{{ prl basis thang
1281:
1282: \prl(basis thang)
1283: Let $S$ be a submodule of the (left) regular module of $eAe$,
1284: and suppose that this submodule has basis $D$.
1285: Then the concrete set of elements $e \otimes D$
1286: is independent in and
1287: will also generate $\glob(S)$.
1288: Further,
1289: $$
1290: \glob(S) \stackrel{\mu}{\doublerightarrow} AS \hookrightarrow \glob(eAe)
1291: \stackrel{(\ref{iso x})}{\cong} Ae \hookrightarrow A
1292: $$
1293: as left $A$-modules.
1294: \end{pr}
1295: {\em Proof:}
1296: For $d \in D \subset S$ then $e \otimes d \in Ae \otimes_{eAe} S = \glob(S) $.
1297: The image $\mu(e \otimes d) = d$, so $\mu(e \otimes D) = D$.
1298: On the other hand, the direct set map $D \hookrightarrow AD$ is an inclusion,
1299: so $D$ is linearly independent in $AS$.
1300: The multiplication map
1301: $\mu : \glob(S) \rightarrow AS$
1302: is surjective, not necessarily bijective,
1303: but it is still a module homomorphism.
1304: Thus if $e \otimes D$ {\em were} to be linearly {\em dependent} in $\glob(S)$,
1305: the image $D$ would be linearly dependent in $AS$ --- a contradiction.
1306: So the set $e \otimes D$ is linearly independent.
1307:
1308: On the other hand $D$ spans $S$, so $Ae\otimes D$ spans $\glob(S)$.
1309: Thus $e\otimes D$ extends to a basis of $\glob(S)$ by the exchange
1310: theorem.
1311: If the multiplication map $\mu$ is bijective then $\glob(S)$ has a
1312: natural isomorphic image in $A$, with a basis which contains $D$ as a
1313: subset.
1314:
1315:
1316: Finally
1317: $eD=D \subset eAe$, so $AeD \subset Ae$.
1318: \Qed
1319:
1320: %}}}
1321: %{{{ left pseudo exact
1322:
1323: \prl(flat)
1324: Suppose that $ S_1 \stackrel{\psi}{\hookrightarrow} S_2 $
1325: is an inclusion of return ideals.
1326: Then $G(S_1) \stackrel{G(\psi)}{\hookrightarrow} G(S_2)$,
1327: i.e. $G$ behaves as if left exact.
1328: \end{pr}
1329: {\em Proof:} Unpacking the assumptions
1330: then $eAef_1 \hookrightarrow eAe f_2$, so $f_1 \in eAe f_2$.
1331: Let us say (WLOG) $f_1 = f f_2$.
1332: We have
1333: \[
1334: \xymatrix{
1335: eAe f f_2 & \stackrel{\psi}{\hookrightarrow} & eAe f_2 \\
1336: \downarrow G & & \downarrow G \\
1337: Ae \otimes_{eAe} f f_2 &
1338: \stackrel{G(\psi)}{\rightarrow} & Ae \otimes_{eAe} f_2 \\
1339: \downarrow \mu_1 & & \downarrow \mu_2 \\
1340: A f f_2 & \hookrightarrow & A f_2
1341: }
1342: \]
1343: Since both $\mu$-maps are isomorphisms,
1344: there are two ways of constructing a homomorphism in the middle:
1345: via the functor $\glob$; or via the bottom row inclusion
1346: $\nu_2(\mu_1(a\otimes ff_2)) = \nu_2(aff_2)
1347: = aff_2 \otimes g_2f_2 = a \otimes ff_2 = a f \otimes f_2$.
1348: Again since the multiplication maps are isomorphisms, the latter
1349: construction
1350: is an injection, i.e. $\glob(S_1) \hookrightarrow \glob(S_2)$.
1351: On the other hand
1352: $G(\psi)(a \otimes ff_2) = a \otimes \psi(ff_2)
1353: =a\otimes ff_2 \in G(S_2)$,
1354: so $G(\psi)$ and $\nu_2 \circ \mu_1$ are the same map.
1355: (NB, for $M \hookrightarrow N$ the element
1356: $a \otimes m \in G(M)$ is not the same thing as
1357: $a \otimes m \in G(N)$ in general --- a set of such objects can
1358: be independent in $G(M)$ and not in $G(N)$, but here
1359: we can also build such objects on the $G(S_2)$ side
1360: by the kernel-free $\nu_2 \circ \mu_1$ route.)
1361: \Qed
1362:
1363: Indeed, suppose that
1364: \eql(inc seq1)
1365: S_1 \hookrightarrow S_2 \hookrightarrow \ldots \hookrightarrow eAe
1366: \eq
1367: is a nested sequence of return ideals,
1368: and $D_i$ a basis for each such
1369: that $D_i \subset D_{i+1}$. Consider the sections defined by this
1370: sequence:
1371: \[
1372: 0 \rightarrow S_{i} \stackrel{\psi}{\rightarrow} S_{i+1}
1373: \rightarrow S_{i+1}/S_{i} \rightarrow 0
1374: \]
1375: then $\glob(S_{i}) \hookrightarrow \glob(S_{i+1})$
1376: and $\glob(S_{i+1}/S_{i}) = \glob(S_{i+1}) / \glob(S_{i})$.
1377: Thus in particular if $S_{i+1}/S_{i}$ is simple then
1378: $\glob(S_{i+1})/\glob(S_{i})$ is \prestandard.
1379:
1380: \medskip
1381:
1382: It may be that (\ref{inc seq1}) is valid over a ground ring
1383: that
1384: specialises to a field in a number of different ways.
1385: Then the sections of the sequence make sense (both before and after
1386: globalisation) over the ring, and \prestandard\ modules have the flavour of
1387: Specht modules \cite{JamesKerber81}.
1388:
1389: Diagram algebras come with bases with special properties.
1390: We will use this concrete construction to decompose the regular module
1391: explicitly throughout entire towers of algebras.
1392:
1393: %}}}
1394: %}}}
1395: \section{Diagram algebras: initial examples} \label{s diag}
1396: %{{{ Diagram algebras
1397: %{{{ ignore footnote V-diagram
1398:
1399: %}}}
1400: %{{{ preamble
1401:
1402: In order to define diagram algebra quotients of Hecke algebras later,
1403: we start by defining related algebras which have both diagram and
1404: `linear' realisations.
1405:
1406: %}}}
1407: \subsection{Brauer algebra wreaths} \label{brauer}
1408: %{{{ brauer wreaths
1409: %{{{ V-diagram
1410:
1411: Fix $n,m \in \N$ with $n+m$ even, and let
1412: $\V^n_m=\{1,2,\ldots,n,1',2',\ldots,m'\}$,
1413: called the set of vertices.
1414: Write $\V=\V_n$ for $\V^n_n$.
1415: Write $J^n_m$ for the set of pair partitions of $\V^n_m$
1416: (so $J_n := J^n_n$ is the usual basis of the Brauer algebra $\Beta_n$
1417: \cite{Brauer37,Weyl46}).
1418: For $S$ a set,
1419: an $S$-decorated pair partition is
1420: an element of $J^n_m$
1421: together with a map from the set of pairs to the set of words in $S$.
1422: Fixing $S$,
1423: write $\DVnm = \DSVnm$ for the set of such objects.
1424: Thus $\DD_{\emptyset}(\V^n_n) \cong J_n$.
1425:
1426: Let
1427: \[
1428: p = \{\{ i, j \},\{ k, l \}, \ldots \}
1429: \]
1430: be a pair partition of $\V^n_m$. Then we may write $d \in \DVnm$ as
1431: \[
1432: d = \{\{ i, j \}_{w_1}, \{ k, l \}_{w_2}, \ldots \}
1433: \]
1434: where $w_l$ is the word in $S$ associated to the $l$-th pair.
1435: (We adopt the convention of omitting the subscript when the word is
1436: the empty word.)
1437: We call this the {\em serial} realisation of $d \in \DVnm$.
1438: We next describe two further useful realisations.
1439:
1440: %}}}
1441: %{{{ string/bead realization
1442:
1443: It will be helpful to think of the following mild extension to Weyl's
1444: \cite{Weyl46} diagram realisation of the pair-partition basis in the
1445: Brauer algebra. Consider:
1446: \newline (i)
1447: the vertices as arranged on a rectangular frame,
1448: $1$ to $n$ across the top edge, $1'$ to $m'$ across the bottom;
1449: \newline (ii)
1450: the pairings as pieces of string (called {\em lines})
1451: appropriately connecting vertices;
1452: \newline (iii)
1453: the accompanying elements of $S$ as threaded beads,
1454: threaded in the order indicated by the word
1455: (reading from vertex $i$ to $j'$, or from $i$ to $j$ if $i<j$,
1456: or from $i'$ to $j'$ if $i<j$).
1457: \newline
1458: (Note that any tangling of strings,
1459: perhaps arising from some perceived embedding in an underlying space,
1460: is irrelevant here --- it is only
1461: the pairings they define that matter.)
1462:
1463: %}}}
1464: %{{{ third realisation
1465:
1466: A third realization is achieved by arbitrarily embedding
1467: \cite{Armstrong79}
1468: each string $\{i,j \}$ as a line from $i$ to $j$
1469: in the plane region bounded by the frame rectangle. For example
1470: \[
1471: \{\{1,3'\}_{ab}, \{2,3 \}_{c}, \{1',2' \}_{} \}
1472: =
1473: \raisebox{-.21in}{
1474: \includegraphics{xfig/abcdiag1.eps}
1475: }
1476: \]
1477: NB, it is not possible in general to do this without distinct lines
1478: crossing (see later, and cf. \cite{JonesPlanar}).
1479:
1480: In summary: we will call objects in the first realisation
1481: decorated pair partitions;
1482: objects in the string/bead realisation diagrams;
1483: and objects in the third realisation concrete diagrams.
1484:
1485: %}}}
1486: \medskip
1487: %{{{ pseudodiagrams
1488:
1489: \noindent {\bf Pseudodiagrams}
1490: \\
1491: Consider the idea of closed loops of string in the string/bead picture
1492: --- that is, strings that do not end on any of the vertices.
1493: Corresponding to this,
1494: it will be convenient to extend the notion of
1495: decorated pair partitions to include
1496: (possibly multiple) copies of the empty set in the partition.
1497: (Hence we make mild abuse of this terminology.)
1498: Following \cite{MartinWoodcock2000} we call such diagrams augmented by
1499: zero or more loops {\em (Brauer) pseudodiagrams},
1500: and the extended partitions {\em pseudopartitions}.
1501: For example (with $\{\}^l$ denoting $l$ copies)
1502: \[
1503: d = \{\{1,3'\}_{ab}, \{2,3 \}_{c}, \{1',2' \}_{},
1504: \{\}_{def}, \{\}^2
1505: \}
1506: = \{\{1,3'\}_{ab}, \{2,3 \}_{c}, \{1',2' \}_{},
1507: \{\}_{edf}, \{\}^2
1508: \}
1509: \]
1510: As before, any perceived embedding of these loops in an underlying
1511: space is irrelevant --- thus in particular a loop does not have an
1512: orientation or starting point for the reading off of bead sequences.
1513: Thus $\{ \}_{def} = \{ \}_{edf}$.
1514:
1515: \del(c_d)
1516: We write $\DoVe{n}{m}$ for the set of $(n,m)$-pseudodiagrams.
1517: If $d \in \DoVe{n}{m}$ is a pseudodiagram
1518: let $c_d \in \DVe{n}{m}$ denote the underlying diagram,
1519: that is, the diagram obtained by omitting any loops;
1520: and $ c^o_d \in \DoVe{0}{0}$ the complement, obtained by keeping only loops.
1521: \end{de}
1522:
1523: %}}}
1524: %{{{ loop equiv
1525:
1526: For $w$ a word in $S$ let $w^o$ denote the opposite word
1527: (the
1528: word with the same letters, but written in the reverse order).
1529: In the serial (pair-partition) realisation,
1530: if we write a vertex pair in a definite
1531: (not necessarily canonical) order: $(i_1, i_2)$, then the string/bead
1532: datum for this string obeys
1533: \eql(string op)
1534: (i_1, i_2)_{w} = (i_2, i_1)_{w^o}
1535: \eq
1536: On the other hand if a pseudodiagram has a closed loop
1537: then this may again have beads on it, but reading the
1538: sequence of beads depends on an arbitrary choice of starting point and
1539: direction round the loop.
1540: We say two words are {\em loop equivalent}
1541: if one can be changed to the other or its opposite
1542: by any cyclic permutation. (Note that loop equivalence is an
1543: equivalence relation on the set of words.)
1544: Thus the bead sequence on a loop is only defined up to loop equivalence.
1545:
1546: %}}}
1547: \medskip
1548: %{{{ compose
1549:
1550: We {\em concatenate} pseudodiagrams
1551: $d_1 \in \DoVe{n}{m}$,
1552: $d_2 \in \DoVe{m}{l}$
1553: to form a pseudodiagram
1554: $d_1d_2 \in \DoVe{n}{l}$
1555: as follows.
1556: Pass to the string/bead realization and there juxtapose the vertices
1557: $1',2',\ldots,m'$ in $d_1$ with the corresponding unprimed vertices in
1558: $d_2$.
1559: Some of
1560: the chains of string resulting from this concatentation
1561: will connect pairs among the
1562: unprimed vertices in $d_1$ and the primed vertices in $d_2$
1563: (defining a new diagram $c_{d_1d_2}$ on these vertices),
1564: and some will form closed loops.
1565: Because of the string/bead realisation we call this the
1566: {\em abacus product} on pseudodiagrams.
1567:
1568: %}}}
1569: %{{{ example
1570: For example
1571: \eql(frubeddd)
1572: d_1 . d_2 . d_3 = \;
1573: \raisebox{-.521in}{
1574: \includegraphics{xfig/abcdiag3.eps}
1575: }
1576: \eq
1577: is
1578: $$
1579: \{\{1,3'\}_{ab}, \{2,3 \}_{c}, \{1',2' \}_{} \}
1580: . \{\{1,2'\}_{ef}, \{3,3' \}_{g}, \{2,1' \}_{d} \}
1581: . \{\{1,2\}_{hi}, \{3,3' \}_{}, \{1',2' \}_{} \}
1582: $$ $$
1583: = \{\{1,3'\}_{abg}, \{2,3 \}_{c}, \{1',2' \}_{}, \{ \}_{efihd} \}
1584: .$$
1585:
1586: To confirm that this product is well defined note that
1587: in the ordered pair form of the serial realisation (\ref{string op})
1588: composition is given by a sequence of $m$ replacements of form:
1589: \[
1590: \underbrace{\{ \ldots (i_1, i_2')_{w} \ldots \}}_{d_1}
1591: \underbrace{\{ \ldots (i_2, i_3)_{w'} \ldots \}}_{d_2} \; \leadsto
1592: \{ \ldots (i_1, i_3)_{ww'} \ldots \}
1593: \]
1594: and
1595: \[
1596: (i_1, i_1)_{w} \leadsto ()_{w} = \{ \}_{w}
1597: \]
1598: the order of application of which, when not forced, produces no ambiguity.
1599:
1600: %}}}
1601: %{{{ prop
1602:
1603: \prl(abacus) The abacus product is associative.
1604: The closed case with $n=m=l$ is unital.
1605: Hence $\Do$ is a category with object set $\N$ and $(n,m)$ morphism set
1606: $\DoVe{n}{m}$.
1607: \end{pr}
1608: {\em Proof:} Associativity follows from the construction.
1609: For an example, consider $(d_1 d_2) d_3$ and
1610: $d_1 (d_2 d_3)$ in equation~(\ref{frubeddd})
1611: --- one draws the same picture in each case.
1612:
1613: The identity element is the pair partition $\Id$:
1614: \eql(Id)
1615: \Id = \{\{1,1'\},\{2,2'\},\ldots,\{n,n'\}\}
1616: \eq
1617: (all words empty).
1618: \Qed
1619:
1620: %}}}
1621: %{{{ ignore loop equiv
1622: %}}}
1623: %{{{ diagram reduction
1624:
1625: Fix $\Ring$ a ring and $\delta_w \in \Ring$ for each $w$ a
1626: loop class representative word in $S$.
1627: \del(k_d)
1628: For each $d \in \DoVe{n}{m}$ define a scalar $k_d \in \Ring$ by
1629: $$
1630: k_{d} = \prod_{l} \delta_{w_l}
1631: $$
1632: with a factor
1633: $\delta_{w_l} $ for each closed loop $l$ in $d$ with bead sequence $w_{l}$.
1634: \end{de}
1635: An example follows shortly.
1636:
1637: %}}}
1638: %{{{ diagram comp
1639:
1640: Define a map
1641: \begin{eqnarray} \label{compose}
1642: \DVe{n}{m} \times \DVe{m}{l} &
1643: \rightarrow & \Ring \times \DVe{n}{l} \rightarrow \Ring\DVe{n}{l}
1644: \\ \non
1645: (d_1,d_2) & \mapsto & (k_{d_1 d_2}, c_{d_1 d_2})
1646: \mapsto k_{d_1 d_2} c_{d_1 d_2}
1647: \end{eqnarray}
1648: (recall that the strings of diagram $c_{d_1 d_2}$ are the open chains
1649: from the abacus product,
1650: each carrying the accumulated beads of this chain in the natural order).
1651:
1652: Since the underlying abacus product is associative,
1653: this (\ref{compose}) extends to
1654: an associative unital product on $\Ring\DVe{n}{n}$.
1655: For example,
1656: in equation~(\ref{frubeddd}) $(d_1 d_2) d_3 = \delta_{w}
1657: \{\{1,3'\}_{abg}, \{2,3 \}_{c}, \{1',2' \}_{} \} $
1658: where $w$ represents the class containing $efihd$.
1659:
1660: Remark:
1661: This product is amenable to massive generalisation,
1662: which we will largely ignore,
1663: but see for example \cite{Martin94,JonesPlanar,AlvarezMartin05}.
1664: A milder generalisation is the {\em cyclotomic} variant,
1665: in which
1666: \eql(t def)
1667: (i_1,i_2)_{s_1 s_2 \ldots} = (i_2,i_1)_{\ldots s_2^t s_1^t} ,
1668: \eq
1669: where $t$ is an involutive map on $\langle S \rangle$ that does not
1670: necessarily fix elements of $S$.
1671:
1672: %}}}
1673: %}}}
1674:
1675: \subsection{Periodic pair-partitions} \label{deform0}
1676: %{{{ periodic brauer (wreath)
1677:
1678: Consider the `infinite' rectangular frame in which the primed and
1679: unprimed vertices are labelled by corresponding
1680: copies of the set of integers.
1681: The set of arbitrary pair-partitions of this vertex set
1682: (call it $V_{\infty}$) is rather
1683: unmanageable, but there are a number of more manageable subsets which
1684: are closed under composition (ignoring loops).
1685: A pair-partition is said to be $n$-periodic if for every pair
1686: $\{i,j\}$ there are pairs $\{ i \pm n , j \pm n \}$
1687: (with $m' \pm n \; := \; (m\pm n)'$).
1688: It follows that there are only $n$ distinct orbits of pairs.
1689: An $n$-periodic pair-partition can be specified by listing a
1690: fundamental subset of $n$ pairs.
1691:
1692:
1693: The first element in each such pair (at least) can be chosen to lie in the
1694: fundamental set $\V^n_n$.
1695: Then if $(i_1,i_2)=(i_1, \bar{i_2}+mn)$
1696: where $\bar{i_2}$ also lies in $\V^n_n$
1697: we might write $(i_1,\bar{i_2})_m$ for $(i_1,i_2)$.
1698: Using this notation we can demonstrate
1699: a map from decorated pair partitions on $\V^n_n$
1700: with bead set $S=\{ L_+, L_- \}$,
1701: and $L_+^t = L_-$ the involution as in equation~(\ref{t def}),
1702: to $n$-periodic pair partitions. We take a string with $m$ beads
1703: $(i_1,i_2)_{L_{\pm}^m}$ to $(i_1,i_2)_{\pm m}$ ($m \geq 0$).
1704: In other words each bead $L_+$ corresponds to winding once clockwise
1705: round the period, and $L_-$ is anticlockwise (thus we
1706: take the quotient with $L_+ L_- =L_-L_+=1$).
1707:
1708: We write $\Jp{n}$ for the set of $n$-periodic pair-partitions.
1709: There are infinitely many of these.
1710: For example with $n=1$
1711: we have
1712: $\{\{\{1,m'\}\} \; | \; m \in \Z \}$
1713: (writing only a fundamental subset for each partition).
1714:
1715: For later convenience it will be useful sometimes to index vertices by odd
1716: integers rather than all integers.
1717: When we do this we will write the pair partition $\{\ldots \}_{o}$.
1718: For $n=2$ examples include
1719: \[ \{\{ 1,7 \},\{ 1',7' \}\}_{o} .
1720: \]
1721: Note that this pair-partition can be realised by vertex-connecting lines
1722: embedded in the infinite rectangular interval, as in the finite case.
1723: But note that in this example these lines necessarily cross
1724: (the orbit of $\{ 1,7 \}$ includes $\{-3,3\}$ and $\{5,11\}$ for example).
1725: For $n=4$ examples include
1726: \[ \{\{ 1,-1 \},\{ 3,5 \},\{1',-1' \},\{3',5' \}\}_{o}
1727: \qquad \{\{ 1,3 \},\{ -1,-3 \},\{1',3' \},\{-1',-3' \}\}_{o} .
1728: \]
1729: Note that these particular examples can be realised by non-crossing lines.
1730:
1731: %}}}
1732: %{{{ composition
1733:
1734: Under composition, ignoring loops for a moment,
1735: it is a simple exercise to show that $n$-periodicity is preserved.
1736: Two types of loops can appear: an `orbit' of loops individual members
1737: of which are periodic images of one another;
1738: and an individual `non-contractible' loop which is
1739: mapped into {\em itself} by periodicity.
1740: Finitely many instances of each type may be created in composition.
1741: We will see
1742: in section~\ref{deform}
1743: a natural way to keep track of these
1744: (and any possible decorations).
1745: Thus, given two different types of loop, we may introduce a set of
1746: periodic pseudodiagrams, which is then closed under composition.
1747: It will also be convenient to introduce the set
1748: $\Jpp{n}$
1749: of periodic diagrams
1750: augmented just by the non-contractible type of loops.
1751: A suitable collection of relations removing the orbits of loops then
1752: makes $\Jpp{n}$ a basis for an algebra generalising the Brauer algebra
1753: (cf. \cite{OrellanaRam01}).
1754:
1755: There is an injective homomorphism from $\Jn{n}$ into $\Jp{n}$ which
1756: simply uses $p \in \Jn{n}$ as the fundamental subset. This can be
1757: extended to an algebra map.
1758:
1759: We write $\Jp{n}^{\!\! S}$ for the beaded version of $\Jp{n}$.
1760:
1761: %}}}
1762:
1763: \subsection{Planar embeddings and contour algebras} \label{s contour}
1764: %{{{ planar (was precompose)
1765:
1766: \begin{de}\label{dplanar}
1767: (1) A diagram in $\DVe{n}{m}$ is called
1768: {\em planar} if it is possible to embed the
1769: strings in the plane interior to the rectangle
1770: (touching the boundary only at the vertices)
1771: in such a way that they do
1772: not (self-intersect or) touch one another.
1773: \\
1774: (2) Any specific such embedding is called a {\em concrete} planar diagram.
1775: \\
1776: (3) If one concrete planar diagram may be continously deformed into another,
1777: with all the intermediate stages concrete planar diagrams, then the diagrams
1778: are said to be {\em \aipic} \cite{Moise77}.
1779: \end{de}
1780:
1781: \prl(post3)
1782: Two concrete planar diagrams are isotopic if and only if they are
1783: realisations of the same underlying diagram.
1784: \Qed
1785: \end{pr}
1786:
1787: \begin{rem} \label{cave} {\em
1788: If we consider planar pseudodiagrams in the same way, then a given
1789: Brauer pseudodiagram may have more than one isotopy class of planar
1790: embeddings. Note however that both $c_d$ and $k_d$ can be considered
1791: as applying to planar pseudodiagrams via their underlying Brauer
1792: pseudodiagrams.
1793: }
1794: \end{rem}
1795:
1796: Write $\Dnz \subset \DVe{n}{n}$
1797: for the set of planar diagrams. It will be evident that
1798: the restriction of diagram composition (\ref{compose})
1799: to $\Dnz$ closes on $\Ring\Dnz$.
1800:
1801: %}}}
1802: %{{{ 0-covered
1803:
1804: \del(covered)
1805: A string in a planar diagram is called {\em exposed} (or 0-covered) if it may
1806: be deformed \aipic ally to touch the western frame edge,
1807: and $l$-covered if it may be deformed to touch an $(l-1)$-covered line
1808: and no lower.
1809: \end{de}
1810:
1811: Let $\Dnzl{n}{l} \subset \Dnz$
1812: denote the subset of planar diagrams in which only the
1813: $l'$-covered lines with $l' \leq l$ may be decorated
1814: (i.e. beaded, i.e. map to other than the empty word).
1815: \begin{pr}
1816: The restriction of diagram composition (\ref{compose})
1817: to $\Dnzl{n}{l}$ closes on
1818: $\Ring\Dnzl{n}{l}$.
1819: \end{pr}
1820: {\em Proof:} Composition may expose new line segments, but it cannot
1821: cover any that were previously exposed (since the relevant part of the
1822: western frame is still in place). Thus any decorated (hence no more
1823: than $l$-covered) line remains no more covered (hence decorable) in
1824: composition. \Qed
1825:
1826: %}}}
1827: %{{{ mod m
1828:
1829: \medskip
1830: \noindent
1831: We now note some specialisations with interesting finite dimensional quotients.
1832:
1833: Fix bead set $S$ of order one
1834: ($S= \{ L \}$, say).
1835: It follows that words in $S$ are all of form $L^i$,
1836: and that each is in a separate loop class.
1837: Write $\delta_i \; := \; \delta_{L^i}$.
1838: Fix $m \in \N$ and
1839: let $D_{n}=D_{n,m}$ denote the subset of $\Dnz$ in which no
1840: string carries more than $m-1$ beads.
1841: Consider the $\Ring$-algebra with basis $\Dnz$
1842: and $\delta_{i+m}=\delta_i$.
1843: With this specialisation of the parameters we may
1844: impose the quotient relation
1845: that $m$ beads together may be cancelled ($L^m =1$).
1846: This produces an algebra $\RX_{n,m}$ with basis $D_{n,m}$
1847: ($D_{n,m}$ is clearly spanning; to see that it is independent
1848: note that the relation cannot be used to change the {\em shape} of a
1849: diagram).
1850: \del(contour)
1851: Define $\RX_{n,m}(l)$ as the subalgebra
1852: of $\RX_{n,m}$
1853: spanned by $\Dnzl{n}{l}$
1854: (and hence with basis $\Dnzl{n,m}{l} \; := \; \Dnzl{n}{l} \cap D_{n,m}$).
1855: These are called contour algebras (see \cite{CoxMartinParkerXi03}).
1856: \end{de}
1857:
1858: %}}}
1859: %{{{ More generally
1860:
1861: More generally, the `algebra' taking place on a single string is the
1862: free monoid on the generators $S$.
1863: Every quotient by some set of relations $\sim$ to a finite monoid
1864: (or even $\Ring$-algebra with basis a finite subset of the free
1865: monoid),
1866: together with a consistent specialisation of the parameters,
1867: induces a finite generalised contour algebra $\RX^{\sim}_{n}(l)$.
1868: (Here $\Ring$-linear combinations of words on a string pass linearly to
1869: corresponding $\Ring$-linear combinations of diagrams.)
1870:
1871: %}}}
1872: %{{{ special elements
1873: As usual the identity in these contour algebras is the pair partition
1874: $\Id$.
1875: Suppose that $L \in S$.
1876: We define $L_i \in \DVe{n}{n}$ as the diagram that is $\Id$
1877: as a pair partition, but has the single letter word $L$
1878: on the $i$-th string, with all other words empty.
1879:
1880: Define
1881: $U_i$ as the diagram differing from $\Id$ in having the pairs
1882: $\{i,i+1 \}, \{ i', (i+1)' \}$.
1883:
1884: \prl(claim1)
1885: The algebra $\RX_{n,m}(l)$ with $m>1$, $l<n$, $S=\{ L \}$, is generated by the set
1886: $$
1887: \{ \mathbb{I} \} \cup \{ L_i \}_{i=1}^{l+1}
1888: \cup \{ U_i \}_{i=1}^{n-1} .
1889: $$
1890: \end{pr}
1891: {\em Proof:} See Appendix. \Qed
1892: %{{{ \stuffb
1893: \newcommand{\toappendix}[2]{\newcommand{#1}{{ #2 }}}
1894: \toappendix{\stuffb}{{
1895: %{{{ claim1
1896: \section{Proof of Proposition~\ref{claim1}}
1897:
1898: To prove: that
1899: the set $D_{n,m}^{z,l}$, can be generated by the
1900: set
1901: $$B := \{ \mathbb{I} \} \cup \{ L_i \}_{i=1}^{l+1}
1902: \cup \{ U_i \}_{i=1}^{n-1} .$$
1903:
1904: Clearly the set of diagrams generated by $B$ is contained in
1905: $D_{n,m}^{z,l}$, since concatenation can never increase the level of
1906: coveredness of a particular string.
1907: So we need only prove that $D_{n,m}^{z,l}$ is contained in the set of
1908: diagrams generated by $B$.
1909: We sketch a proof of this by induction on $l$.
1910:
1911: If $l=-1$ and there are no decorated lines, then this is just the
1912: result for the diagram version of the Temperley--Lieb algebra
1913: \cite{Martin91}.
1914:
1915: Now suppose $l=0$ and we have a diagram with a decorated $0$-covered
1916: line with $j$ beads.
1917: Since the decorated line is $0$-covered there are two
1918: possibilities. Either the line is the string starting at the first
1919: position or ending at the first position
1920: - in which case we can decompose in the diagram into a
1921: product of $j$ $L_0$'s together with a smaller diagram with one less
1922: $0$-decorated line,
1923: or vice versa,
1924: or the line is a starting at a greater position
1925: than
1926: the first. In this case we get a diagram that looks like figure
1927: \ref{fig:gens0} where we have only decorated
1928: the $0$-covered line with one bead rather than $j$ beads for simplicity.
1929: We have drawn a propagating $0$-covered line; the dashed line
1930: represents a non-propagating $0$-covered line.
1931: \begin{figure}[ht]
1932: \begin{center}
1933: \epsfbox{gens0.eps}
1934: \end{center}
1935: \caption{\label{fig:gens0}
1936: }
1937: \end{figure}
1938: Now note that the grey regions to the left of the decorated line
1939: cannot contain any propagating lines - since the decorated line is
1940: $0$-covered.
1941: But both these regions must contain at least one string that is
1942: $0$-covered and so we can deform the diagram to look like that on the
1943: right hand side of figure \ref{fig:gens0}.
1944:
1945: We can arrange it so the two grey regions on the right hand
1946: side are only joined by
1947: propagating lines and the number of these lines $x$, say will be the
1948: same (\resp\ different) parity as $n$ if the decorated $0$-covered line is
1949: non-propagating (\resp\ propagating). Thus the difference $n-x$
1950: is even if the $0$-covered line is non-propagating and odd if the
1951: $0$-covered line is propagating.
1952: We now ``wiggle'' the $0$-covered line enough times
1953: so that we get the right number
1954: of lines so that the
1955: middle section of the diagram enclosed in dotted lines is now the
1956: diagram product $U_1L_1U_2U_1$ (which has $n-3$ propagating lines).
1957:
1958: So we can decompose the diagram
1959: into a product of three smaller diagrams, the outside diagrams having
1960: a smaller number of $0$-covered lines.
1961:
1962: The case with $l \ge 1$ is similar and is illustrated
1963: in figure~\ref{fig:gens1} below.
1964: We have drawn a propagating $l$-covered line; the dashed line
1965: represents a non-propagating $l$-covered line.
1966: We can assume that the $l-1$-covered line (which may be decorated or
1967: not) is propagating, for otherwise the $l$-covered line would not be
1968: $l$-covered.
1969:
1970: \begin{figure}[ht]
1971: \begin{center}
1972: \epsfbox{gens1.eps}
1973: \end{center}
1974: \caption{\label{fig:gens1}
1975: }
1976: \end{figure}
1977: Note that now when we ``pull apart'' the grey regions we can always
1978: stretch then so that we can get only propagating lines joining the two
1979: smaller
1980: grey regions.
1981: Also note that the number of propagating lines in the grey region on the
1982: left hand side is at exactly $l-1$, for otherwise the $l-1$-covered
1983: line would not be $l-1$-covered.
1984: We again get the right parity, so that we can ``wiggle'' the
1985: $l$-covered line so that
1986: the middle section of the diagram enclosed in dotted lines is now the
1987: diagram product $U_{l+1}L_{l+1}U_{l+2}U_{l+1}$,
1988: and so we can decompose the diagram
1989: into a product of three smaller diagrams, the outside diagrams having
1990: a smaller number of $l$-covered lines.
1991:
1992:
1993:
1994: %}}}
1995:
1996: }}
1997:
1998:
1999: %}}}
2000:
2001: Note that in case $m=2$ we may replace
2002: the relation set $\sim = \{ LL = 1 \}$
2003: (giving $L_i L_i = \Id$)
2004: with $\sim = \{ LL=L \}$
2005: (giving $L_i L_i = L_i$)
2006: and obtain an isomorphic algebra.
2007: Indeed we may deform the monoid (algebra) via $LL=\kappa L$ similarly
2008: (giving $L_i L_i = \kappa L_i$).
2009: Only the case $\kappa=0$ departs from the rest.
2010:
2011: %}}}
2012: %{{{ left-right case
2013:
2014: It will be evident that similar definitions
2015: to $\RX^{\sim}_{n}(l)$ may be contructed for
2016: {\em eastern} exposure, and for composites
2017: $\RX^{\sim}_{n}(l,r)$.
2018: Also of interest, as it turns out, are subalgebras of the case
2019: $S=\{ L, R \}$
2020: generated by
2021: $$
2022: \{ \Id, \; L_1, \; R_n \} \cup \{ U_i \}_{ i=1}^{ n-1} .
2023: $$
2024: where $\sim$ defines a certain noncommutative monoid.
2025: (Note that this choice of generators
2026: prescribes the way in which $L$ and $R$ can meet,
2027: which leads to some interesting topological effects --- see later.)
2028:
2029: %}}}
2030:
2031: \subsection{Direct homomorphisms with known algebras} \label{dir homs}
2032: %{{{ direct homs
2033:
2034: %{{{ relationships
2035: \del(TL)
2036: A Temperley--Lieb (TL) diagram is an isotopy class of concrete planar
2037: diagrams, or any representative thereof. Ordinary TL diagrams are beadless.
2038: \end{de}
2039:
2040: By Proposition~\ref{post3}
2041: the subset $\Dzo{n} \hookrightarrow \Dnz \hookrightarrow \DVe{n}{n} $
2042: with no beads
2043: is in bijection with the set of ordinary
2044: Temperley--Lieb
2045: diagrams on two rows of $n$ vertices.
2046:
2047: %}}}
2048: %{{{ blob
2049:
2050: \del(blob)
2051: {\rm \cite{MartinSaleur94a}}
2052: The set $B_n$ of {\em blob diagrams}
2053: is the set of decorated
2054: TL diagrams on two rows of $n$ vertices
2055: in which western exposed lines, only, may be
2056: decorated, with at most a single bead (`blob') on each.
2057: \end{de}
2058: For example, for any given $n$ the diagram $e \in B_n$ has the shape
2059: of the identity diagram ($n$ vertical lines), but with the leftmost
2060: line decorated with blob.
2061:
2062: The blob algebra
2063: $b_n = b_n(\delta=q+q^{-1},\delta_e,\gamma)$
2064: (as in \cite{MartinSaleur94a}, but as parameterised in
2065: \cite{MartinRyom02}),
2066: is generated by TL diagrams and $e$,
2067: with two blobs on a line appearing in composition replaced via:
2068: \eql(ee=e)
2069: e e = \delta_e e
2070: \eq
2071: and a loop decorated by a blob replaced by a factor of $\gamma$.
2072: Thus $b_n(\delta,\delta_e,\gamma)$ has basis $B_n$.
2073:
2074: %}}}
2075: %{{{ prop
2076:
2077: It is easy to show that
2078: \begin{pr} For all $n$:
2079: \newline
2080: (i) The subset $\Dzo{n}$ of $\Dnz$
2081: generates a finite dimensional algebra
2082: isomorphic to
2083: the ordinary Temperley--Lieb algebra $TL_n(\delta_0=q+q^{-1})$
2084: \cite{TemperleyLieb71}.
2085: \newline
2086: (ii) The set $\Dnzl{n,2}{0}$ is
2087: essentially identical to the set of blob diagrams $B_n$,
2088: with $L_1=e$.
2089: The algebra $\RX_{n,2}(0)$ is isomorphic to the blob
2090: algebra $b_n$ \cite{MartinSaleur94a}
2091: and to $TLb_n$
2092: (more generally, $\RX_{n,m}(0)$ is the coloured blob algebra mentioned in
2093: \cite{MartinWoodcockLevy00}).
2094: \newline
2095: (iii) The case $\RX_{n,m}(n)=\RX_{n,m}(\infty)$ is isomorphic to the cyclotomic
2096: Temperley--Lieb algebra \cite{CoxMartinParkerXi03}.
2097: \Qed
2098: \end{pr}
2099:
2100: %}}}
2101: %{{{ cf cmpx
2102:
2103: Note (from section~2 of \cite{CoxMartinParkerXi03})
2104: that the tower of recollement framework applies to all of the above algebras.
2105:
2106: %}}}
2107: %{{{ remark
2108: Following the largely physically motivated
2109: investigation of the ordinary Temperley--Lieb algebra in the 1980s,
2110: the blob algebra was introduced in order to allow use of
2111: the western edge of the frame (actually either one) as a cohomology
2112: seam,
2113: and hence to address the {\em periodic} Temperley--Lieb algebra
2114: (again, originally, with physical motivation).
2115: The very first level of exposure is sufficient for this purpose (see
2116: the literature, for example \cite{MartinSaleur94a,GrahamLehrer03}).
2117: Interest in the `homogeneous' case (i.e. not filtered by exposure) has been
2118: slower to arise, but now see \cite{RuiXi04} and \cite{JonesPlanar}
2119: (whose primary interest is in subfactors).
2120: The general
2121: intermediates have yet to find a physical application.
2122:
2123: %}}}
2124:
2125: %}}}
2126:
2127: \subsection{Diagram embeddings, subalgebras and deformations} \label{deform}
2128: %{{{ subsection
2129: %{{{ preamble
2130: This section describes `topological' realisations of
2131: certain subsets of diagrams,
2132: generalising the planar embedding of definition~\ref{dplanar},
2133: and the deformations of the algebra product possible in these cases.
2134:
2135: %}}}
2136: %{{{ planar/isotopy
2137:
2138: \pdef{Isotopy:}
2139: A brief remark is in order on the general notion of isotopy
2140: following from the benign paradigm in definition~\ref{dplanar}.
2141: A realisation of a (pseudo)diagram is called a picture if it is an arbitrary
2142: choice among a continuum of such realisations of the same diagram.
2143: (The existence of such realisations ---
2144: faithful but non-canonical {\em drawings} of the diagram ---
2145: is at the heart of the use of the word diagram to describe these objects.)
2146: Suppose we have a subset of a set of diagrams characterised by
2147: the existence of pictures satisfying certain properties
2148: (such as the concrete planar embedding in definition~\ref{dplanar}).
2149: Then given a picture of a diagram,
2150: another diagram realisation is said
2151: to be {\em isotopic} to it if they belong to a continuum of pictures
2152: all satisfying the characterising property.
2153:
2154: We note as a paradigm for later reference that when restricted to
2155: $\Dnz$ the product in (\ref{compose}) is amenable to deformation.
2156: This is firstly because the orientation of loops becomes an invariant of
2157: isotopy (a loop cannot be flipped without some intermediate
2158: crossing). Thus $\delta_w=\delta_{w'}$ is only necessary if $w,w'$
2159: related by a cyclic permutation.
2160: \[
2161: \includegraphics{xfig/loopcross.eps}
2162: \]
2163: Secondly, the nonisotopic placement of loops noted in
2164: remark~\ref{cave}
2165: gives scope for further deformation (see section~\ref{blob'}).
2166:
2167: %}}}
2168: %{{{ periodic
2169:
2170: \begin{de}\label{dperiod}
2171: (1) Consider the manifold constructed from the plane interior to the
2172: boundary rectangle of a diagram by identifying the eastern and western
2173: edges (a cylinder).
2174: A diagram in $\DVe{n}{m}$ is called {\em periodic}
2175: if it is possible to embed the
2176: strings in this manifold
2177: (touching the northern and southern boundaries only at the vertices)
2178: in such a way that they do not touch one another.
2179: \\
2180: (2) Any specific such embedding is called a {\em concrete} periodic diagram.
2181: \\
2182: (3) If one concrete periodic diagram may be continously deformed into another,
2183: with all the intermediate stages concrete periodic diagrams, then the diagrams
2184: are said to be {\em \aipic} \cite{Moise77}.
2185: \end{de}
2186: For example, every planar diagram is periodic, while
2187: \[
2188: \tau \; = \; \{\{1,2'\},\{2,3'\},\ldots, \{n,1'\}\}
2189: \]
2190: is periodic but not planar.
2191:
2192: Write $\Dnp$ for the set of such periodic diagrams,
2193: so that
2194: $\Dnz \subset \Dnp \subset \DVe{n}{n}$.
2195:
2196: %}}}
2197: %{{{ periodic isotopy waffle
2198:
2199: \toappendix{\stuffbb}{{
2200: \section{On constrained isotopy} \label{stuffbb}
2201: It may be helpful to elaborate on the meaning of isotopy in
2202: Definition~\ref{dperiod}. For planar diagrams, once the vertices are
2203: labelled on the frame there is no isotopy that can obfuscate this
2204: order. The precise location of individual vertices on the frame is not
2205: of any concern in converting between (non-unique) concrete diagrams
2206: and their (unique) underlying diagrams.
2207: For periodic diagrams, if any movement on the frame is allowed then
2208: it might seem that
2209: isotopy can untwist the full twist (on the identity element for
2210: example).
2211: However, the intermediate objects in the associated
2212: continuum would be formally ill-defined as concrete diagrams
2213: (in that it would not be possible to regard them
2214: as concrete realisations of the original diagram, or indeed of any
2215: particular diagram).
2216: It follows that the `natural' embedding of the identity element is not
2217: isotopic to a twisted one in our definition of isotopy.
2218: A more general algebra arises, therefore, if we consider
2219: the fundamental objects to be (frame-fixing)
2220: isotopy classes of concrete periodic diagrams, than if we regard the
2221: corresponding diagrams as fundamental.
2222: }}
2223: %}}}
2224: %{{{ periodic isotopy
2225:
2226: Proposition~\ref{post3} notes that
2227: two concrete planar diagrams are isotopic if and only if they are
2228: realisations of the same diagram.
2229: This is not true in general for periodic diagrams.
2230: In particular there are non-isotopic embeddings of the identity
2231: diagram.
2232: The picture of $\tau^n$ obtained by composing pictures is an embedding of
2233: $\One$ not isotopic\ to the obvious embedding, for example.
2234: See appendix~\ref{stuffbb} for a fuller discussion.
2235:
2236: Let $\Dnpc$ denote the set of periodic isotopy classes of concrete diagrams
2237: associated to $\Dnp$ (hence a set of periodic TL diagrams).
2238:
2239: %}}}
2240: %{{{ more waffle
2241:
2242: Because of the non-isotopic embedding possibility mentioned above the
2243: set $\Dnpc$ is larger than the underlying set of diagrams,
2244: as defined by their serial realisations.
2245: It is possible (and useful) to have a serial realisation of isotopy classes,
2246: however.
2247: Note that embedded periodic diagrams may be drawn as period-$n$ periodically
2248: repeating {\em planar} diagrams in the infinite frame
2249: (a string coming out of vertex 1 produces strings coming out of all
2250: vertices congruent to 1 modulo $n$, and so on),
2251: and hence as a subset of $\Jp{n}^{\!\! S}$:
2252: \[
2253: \Dnpc \hookrightarrow \Jp{n}^{\!\! S} .
2254: \]
2255: If we use {\em this} labelling in writing down the serial realisation of
2256: diagrams (and treat vertices on different `sheets' as distinct, even
2257: if they are congruent), then we recover the situation that concrete
2258: periodic diagrams are isotopic if and only if they are realisations of
2259: the same diagram.
2260: (There are still infinitely many such diagrams however, even without beads.
2261: Our set of examples
2262: $\{\{\{1,m'\}\} \; | \; m \in \Z \}$
2263: are all in $\Dnpc$ with $n={1}$.)
2264:
2265: The restriction of diagram composition (\ref{compose})
2266: to $\Dnp$ closes on $\Ring\Dnp$.
2267: This product is amenable to deformation through the non-isotopic
2268: embeddings mentioned above
2269: (see \cite{MartinSaleur93,FanGreen99} for example).
2270: In particular let $\Dppc{n}$ denote the augmentation of $\Dnpc$
2271: in case $S=\emptyset$ by classes of concrete diagrams
2272: including (non-crossing)
2273: non-contractible loops in the manner of $\Jpp{n}$.
2274: (NB, There is a drawing error in Fig.10 of \cite{MartinSaleur93}.
2275: This diagram should have 4 non-contractible loops, not 3.)
2276:
2277: %}}}
2278: %}}}
2279: %}}}
2280: \section{The blob algebra $b_n$ and the \achiral\ algebra $b_n'$} \label{blob'}
2281: %{{{ TLAs: b,b' AND section recollement
2282:
2283: %{{{ some TLAs
2284: %{{{ two colour
2285:
2286: \subsection{Two-coloured diagrams}
2287: Concrete TL diagrams may be thought of as partitions of
2288: the plane interior to the frame rectangle, with the lines being the
2289: boundaries of parts.
2290: The non-crossing rule means that these diagrams may be two-coloured
2291: (in the four colour theorem sense).
2292: For the sake of definiteness let us say that the part whose closure
2293: includes the interval on the frame between the vertices 1 and 2 is
2294: coloured black.
2295: It follows that the part between 1' and 2' is also black,
2296: and that concatenation preserves this canonical colouring.
2297: It also follows that closed loops formed in concatenation may have
2298: either a black or a white interior (or to be more precise, immediate
2299: interior, since they may be nested).
2300:
2301: Note that although colouring requires embedding, which is
2302: noncanonical, the number of loops of each colour formed in
2303: concatenation is invariant under plane \aipy\
2304: (note that this is not true for the larger classes of Brauer \aipy).
2305: It follows that we can generalise the composition rule in
2306: (\ref{compose}) by generalising $k_{a,b}$.
2307:
2308: By rescaling the generators $U_i$ one can see that
2309: this generalisation is isomorphic to the ordinary case
2310: (excepting the specialisation in which one of the parameters is not
2311: invertible, which is not appropriate for the context of computation
2312: for Potts and vertex models, where the algebra has its origin
2313: \cite{Baxter82}).
2314: However, we shall now show that it leads the way to some
2315: further rather more useful generalisations
2316: (in the spirit of the blob algebra
2317: \cite{MartinSaleur94a} viewed as a generalisation of
2318: the type-B algebra of \cite{tomDieck94}).
2319:
2320: %}}}
2321: %{{{ reflection \subsection{Subalgebras and deformations}
2322:
2323: \subsection{Subalgebras and deformations} \label{deform2}
2324: Let $\DB{n} \subset \Dzo{n}$ be
2325: the subset of TL diagrams that are (\aipic\ to concrete diagrams that
2326: are) invariant under reflection in a central vertical line.
2327: Putting aside for a moment any closed loops that might arise,
2328: this subset is obviously fixed under concatenation.
2329: That is, the concatenation of two concrete representatives of elements
2330: of $\DB{n}$ is a representative of an element of $\DB{n}$
2331: (ignoring closed loops).
2332:
2333: \noindent {\em Remark:}
2334: Indeed we could consider a version of $\DB{n}$ that is not a strict subset of
2335: $\Dzo{n}$ (a set whose elements are isotopy classes, where even
2336: classes containing symmetric elements
2337: include elements which are not concretely symmetric) but to be such
2338: that the elements are `symmetric isotopy' classes - i.e. classes whose
2339: elements maintain exact symmetry.
2340: So long as we define our algebra composition without reference to
2341: pseudodiagrams this distinction is academic
2342: (but see section~\ref{reduc}).
2343:
2344: If $n$ is odd then $\DB{n}$ is a rather uninteresting subset.
2345: If $n$ is even, redefine canonical colouring to be that in which
2346: the part whose closure includes the central northern interval
2347: is coloured white.
2348: Note that within this subset the property that the concatenation of
2349: two diagrams forms a loop that {\em crosses} the centre line
2350: (indeed white,
2351: or black, loop that crosses)
2352: is invariant under \aipy.
2353: Thus we may deform the algebra spanned by these diagrams by
2354: generalising the scalar factor $k_{d_1 d_2}$ so that
2355: \eql(new k)
2356: k_{d_1 d_2} = \delta^{l_0} \delta_e^{l_w} \kappa^{l_b}
2357: \eq
2358: where $\delta, \delta_e, \kappa \in \Ring$, $l_0$ is the number of
2359: (pairs of) noncrossing loops, $l_w$ is the number of white crossing
2360: loops and $l_b$ the number of black crossing loops.
2361: \del(b')
2362: Let $b'_n(\delta, \delta_e, \kappa)$ denote the algebra which is
2363: $\Ring \DB{n}$ with product defined by (\ref{new k}).
2364: \end{de}
2365:
2366: %}}}
2367: %{{{ identity and e=/abe
2368:
2369: As usual the identity in this algebra is the pair partition $\Id$.
2370: For any given $n=2m$ let $\abe'$ denote the diagram corresponding to the
2371: pair partition differing from 1 only in $\{m,m+1\},\{m',m+1'\}$,
2372: that is (in case $m=3$):
2373: \[
2374: \abe' = \;
2375: \raisebox{-.21in}{
2376: \includegraphics{xfig/eTLB.eps}}
2377: \]
2378:
2379: %}}}
2380: \subsection{The unfolding map $\mu$}
2381: %{{{ ur-diagrams
2382:
2383: \begin{de}
2384: An {\em \pseudo-diagram} is a TL diagram in which strings may end on
2385: the east or west edge of the rectangle as well as north and south.
2386: \end{de}
2387:
2388: Recall that $B_n$ denotes the set of blob diagrams.
2389: Given a blob diagram $d$, we may define a \pseudo-diagram from it by
2390: deforming every string with a blob until an arc in the immediate
2391: neighbourhood of the blob is just on the outside of the western edge
2392: of the rectangle, and then discarding this arc.
2393: If we compose this \pseudo-diagram with its mirror
2394: image in the western edge
2395: (NB, this is a well defined construct on isotopy classes)
2396: we have a left-right symmetric TL diagram.
2397: \newcommand{\Mu}{{\Large \mu}}%
2398: Let us call it $\Mu(d)$.
2399: For example $\Mu(e)=\abe'$.
2400: \prl(blob base)
2401: The map $\Mu:B_m \longrightarrow \Dzo{2m}$ defined above is an injection,
2402: and the range is the set $\DB{2m}$ of left-right symmetric diagrams.
2403: \Qed
2404: \end{pr}
2405: (This generalises the combinatorial map described in
2406: \cite{CoxGrahamMartin03}.)
2407: %}}}
2408: %{{{ blob iso
2409: \prl(blob iso)
2410: The map $\Mu$ extends to an algebra homomorphism, so that
2411: the algebra $b'_{2m}(\delta, \delta_e, \kappa)$
2412: is isomorphic to the blob algebra $b_m(\delta, \delta_e, \kappa)$.
2413: \end{pr}
2414: {\em Proof:} We need to check that $\Mu(a) \Mu(b) = \Mu(ab)$,
2415: where $a,b$ are blob diagrams. If we consider $ab$ as a concatenation,
2416: without (for a moment) imposing the blob relations,
2417: it can have two blobs on the same line, and it can have closed loops,
2418: with and without blobs. The map $\mu$ makes sense on such an $ab$, and
2419: commutes with concatenation. It is thus necessary to check that the
2420: imposition of the blob relations on $ab$ produces the same factor as
2421: $k_{\Mu(a), \Mu(b)}$.
2422: If two blobs come together we use the blob relation $ee=\delta_e e$
2423: from (\ref{ee=e}).
2424: The image under $\Mu$ is (locally):
2425: \[
2426: \includegraphics{xfig/eedee.eps}
2427: \]
2428: which has one white loop, so $k_{\Mu(a), \Mu(b)}=\delta_e$ as
2429: required.
2430: A decorated loop is replaced by a factor $\kappa$ in the blob algebra
2431: $b_m(\delta, \delta_e, \kappa)$ (cf. section~\ref{dir homs}),
2432: and passes to a black loop under $\Mu$.
2433: From (\ref{new k}) this again gives a factor $\kappa$.
2434: An undecorated
2435: loop ($\delta$ in the blob algebra) passes to a pair of off-axis loops
2436: under $\Mu$.
2437: \Qed
2438:
2439: %}}}
2440: %}}}
2441:
2442: \section{Recollement and $b_n'$ representation theory} \label{recol b}
2443: %{{{ recollement
2444: \begin{figure}
2445: \[
2446: \includegraphics{xfig/eTLB2.eps}
2447: \]
2448: \caption{\label{foobre} Removing the central cup and cap.}
2449: \end{figure}
2450:
2451: Suppose that $\delta_e$ is invertible.
2452: Then the subset of $\DB{2m}$ consisting of
2453: symmetric diagrams in which the inner central `cup' and `cap' appear
2454: (i.e. diagrams that contain
2455: $\{m,m+1\},\{ m',m+1' \}$
2456: as pair partitions) are a basis for a subalgebra of $b'_{2m}$.
2457: Since $\abe' \abe' = \delta_e \abe'$ then $\frac{1}{\delta_e} \abe'$
2458: is idempotent,
2459: and is the unit in this subalgebra.
2460: It may be identified with the idempotent subalgebra
2461: $b''_{2m} = \frac{\abe'}{\delta_e} b'_{2m}(\delta,\delta_e,\kappa)
2462: \frac{\abe'}{\delta_e}$.
2463: Since a white loop appears automatically in every composition in this
2464: algebra, a better basis is the set
2465: $\DBe{2m}$
2466: of such diagrams each multiplied by
2467: $\frac{1}{\delta_e}$.
2468: Then if no {\em other} loops appear in composition the
2469: product of two basis elements is another basis element.%
2470: \\
2471: {\em Remark:}
2472: Actually the algebras $b'_{2m}(\delta,\delta_e,\kappa)$
2473: and $b'_{2m}(\delta,\alpha \delta_e, \alpha \kappa)$ are readily seen to be
2474: isomorphic for any invertible $\alpha$
2475: (consider the isomorphic \cite{Martin95pres}
2476: presentational form (\ref{TL001}-\ref{TL006}), for example).
2477: It is thus possible to replace
2478: $b'_{2m}(\delta,\delta_e,\kappa)$ with
2479: $b'_{2m}(\delta,1,\delta_e^{-1}\kappa)$ without loss of generality.
2480: \medskip
2481:
2482: %}}}
2483: %{{{ almost a map
2484:
2485: It will be evident from figure~\ref{foobre} that
2486: there is
2487: a set map
2488: \[
2489: \rho^- : \DBe{2m} \rightarrow \DB{2m-2}
2490: \]
2491: defined by simply removing the central upper `cup' and lower `cap' from
2492: the diagram underlying each basis element (and discarding the factor
2493: $\frac{1}{\delta_e}$). Indeed $\rho^-$ is a bijection.
2494: We next extend this to an algebra map.
2495:
2496: %}}}
2497: %{{{ prop
2498:
2499: \prl(recool)
2500: The map $\rho^-$ extends $K$--linearly to an algebra isomorphism
2501: $$
2502: \rho^-:
2503: \frac{\abe'}{\delta_e} b'_{2m}(\delta,\delta_e,\kappa) \frac{\abe'}{\delta_e}
2504: \longrightarrow b'_{2m-2}(\delta,\kappa,\delta_e) .
2505: $$
2506: \end{pr}
2507: {\em Proof:}
2508: Consider $\rho^-$ extended in the obvious way to pseudodiagrams.
2509: Concatenation of diagrams may be thought of as
2510: the first step in computing composition
2511: on both sides, and commutes with $\rho^-$
2512: (given the automatic cancellation of one loop with a normalising
2513: factor on the domain side).
2514: The final step is interpretation of the pseudodiagram as a scalar
2515: multiple of the underlying basis element.
2516: This differs on the range side,
2517: precisely in that the colour assigned to each loop is reversed
2518: (by the cup/cap removal).
2519: \footnote{
2520: Consider the colouring of a pseudodiagram
2521: on the domain side of $\rho^-$, which
2522: determines the scalar factors there.
2523: If we think of trying to keep this colouring, then
2524: once the cup and cap are removed the colour of
2525: the upper central interval (as used to determine scalar factors)
2526: is black, not white.
2527: Obviously then, {\em all} colours are inverted,
2528: compared to what they would normally be on the target side.
2529: Thus in particular central loops which would properly be
2530: white will be black, and vice versa.
2531: Thus the map $\rho^-$ reverses all colours
2532: and will not extend naively to an algebra homomorphism.
2533: Swapping the colours exchanges the roles of $\delta_e$ and $\kappa$,
2534: so we can fix this by defining the extension
2535: as above for basis elements,
2536: but which maps scalar coefficients by exchanging the
2537: roles of $\delta_e$ and $\kappa$.
2538: }
2539: This difference is thus itself reversed by exchanging the roles of
2540: $\delta_e$ and $\kappa$.
2541: \\
2542: \Qed
2543:
2544:
2545: %}}}
2546: %{{{ eg
2547:
2548: For example, let $a$ be a basis element with underlying diagram
2549: as shown in the upper left of:
2550: \[
2551: \includegraphics{xfig/eTLB3.eps}
2552: \]
2553: We see that the product on the left is $a.a = \kappa a$
2554: (taking account of factors of $\frac{1}{\delta_e}$).
2555: The image $\rho(a)$ is shown on the right, together with
2556: $\rho(a).\rho(a) = \tilde \delta_e \rho(a)$
2557: (writing $\tilde \delta_e$ for the $\delta_e$ parameter in the image).
2558: Thus in order for $\rho(a.a)=\rho(a).\rho(a)$ we require
2559: $\tilde \delta_e = \kappa$.
2560:
2561: On the other hand, with the basis elements indicated on the left in:
2562: \[
2563: \includegraphics{xfig/eTLB4.eps}
2564: \]
2565: we have $b.c = \delta_e c$, while on the left
2566: $\rho(b).\rho(c) = \tilde \kappa \rho(c)$.
2567: So we require $\tilde \kappa = \delta_e$, as stated.
2568:
2569: %}}}
2570: %{{{ recollement2
2571:
2572: For the moment let us use the shorthand $A_m$ for
2573: $b'_{2m}(\delta,\delta_e,\kappa)$ and $B_{m}$ for
2574: $b'_{2m}(\delta,\kappa,\delta_e)$,
2575: and $\Lambda^A_m$ for an index set for simple modules of $A_m$
2576: (and similarly for $B_m$).
2577: We may now bring to bear the recollement part of the machinery
2578: described in \cite{CoxMartinParkerXi03}
2579: (summarized in section~\ref{cat}).
2580: In particular note that
2581: Proposition~\ref{recool} tells us the following.
2582: \begin{theo} \label{b' fullembed}
2583: The category of left
2584: $b'_{2m-2}(\delta,\delta_e,\kappa)$-modules
2585: is {\em fully embedded} in the category of left
2586: $b'_{2m}(\delta,\kappa,\delta_e)$-modules.
2587: \Qed
2588: \end{theo}
2589: It follows (via proposition~\ref{grebo}) that the simple modules $S$
2590: of the latter {\em not} obeying $\abe'S=0$ may be indexed by
2591: $\Lambda^B_{m-1}$.
2592: The simple modules obeying $\abe'S=0$ are also simple modules of the
2593: quotient
2594: $A_m/A_m \abe' A_m$, but it is easy to see that this has precisely one
2595: simple module
2596: (since $A_m \abe' A_m$ contains every diagram with fewer than $2m$ propagating lines).
2597:
2598: In this way we derive the (well known) index set for simple modules of
2599: $A_m$. We also derive some striking results relating the homomorphisms
2600: between standard modules for $A_m$ and $B_{m-1}$ which, while not
2601: revealing any homomorphisms which were not already known, do reveal a
2602: layer of symmetry in the organisation of these homomorphisms which has
2603: not been noted before.
2604: This structure is not needed in the analysis of the blob algebra's
2605: representation theory, but it raises the very intriguing possibility
2606: of similar symmetries in (affine) Hecke algebra representation theory.
2607: We will discuss this further elsewhere.
2608:
2609: %}}}
2610: \subsection{The $b_n$ version} \label{remi}
2611: %{{{ remarks / the rho_1 map
2612:
2613: With the benefit of hindsight we see that the recollement can be
2614: invoked directly in $b_n$.
2615: Let $B_n'$ be the subset of $B_n$ consisting of elements in which both
2616: the string containing vertex 1 and that containing vertex $1'$ are decorated
2617: (of course this could be the same string).
2618: Let $B^e_n = \{ \frac{1}{\delta_e} d \; | \; d \in B_n' \} \; \subset b_n $.
2619:
2620: Define a map
2621: \eql(rho_1)
2622: \rho_1: B^e_n \longrightarrow B_{n-1}
2623: \eq
2624: as follows.
2625: For $d \in B_n'$,
2626: consider the region of $d$ with the western edge in its closure:
2627: we have a sequence of one or more decorated lines reading clockwise around
2628: this region, starting from the vertex 1
2629: (ignore the undecorated ones).
2630: This sequence is of the general form
2631: $\{ 1 , i_1 \}, \{i_2,i_3\},\ldots, \{ i_l, 1' \}$ (some possibly
2632: primed vertices $i_1,\ldots,i_l$),
2633: or simply $\{ 1, 1' \}$.
2634: In the latter case, simply erase the (decorated) line $\{ 1, 1' \}$.
2635: Otherwise, erase the sequence and replace with
2636: decorated lines $\{i_1,i_2 \}, \ldots, \{ i_{l-1}, i_l \}$.
2637: After suitable renumbering we have an element of $B_{n-1}$.
2638: This is $\rho_1( \frac{1}{\delta_e} d)$.
2639:
2640: The map is illustrated by the following example:
2641: \[
2642: \includegraphics{xfig/eTLB15.eps}
2643: \]
2644: It will be evident that this map is a bijection.
2645:
2646: %}}}
2647: %{{{ CLAIM -> alg bijection
2648:
2649: \prl(b case) %%\begin{claim}
2650: The map $\rho_1$ extends $K$--linearly
2651: to an algebra isomorphism from the subalgebra $b^e_n$
2652: of $b_n(\delta,\delta_e,\kappa)$
2653: spanned by $B^e_n$ (with unit $\frac{1}{\delta_e} e$) to
2654: $b_{n-1}(\delta,\kappa,\delta_e)$.
2655: That is, $\rho_1$ is an algebra isomorphism with $\delta_e$
2656: and $\kappa$ interchanged.
2657: \end{pr}
2658: {\em Proof:} Consider Proposition~\ref{blob iso} and
2659: Proposition~\ref{recool}.
2660: A short manipulation of diagrams (in the diagram algebra
2661: sense) confirms that the following diagram is commutative:
2662: \[
2663: \xymatrix{
2664: b^e_{n} \ar[d]_{\rho_1} \ar[r]^{\mu} & b''_{2n} \ar[d]^{\rho^-} \\
2665: b_{n-1} \ar[r]_{\mu} & b'_{2n-2}
2666: }
2667: \]
2668: for the appropriate parameter values.
2669: \Qed
2670: %}}}
2671: %{{{ Alternatively
2672: \\
2673: The parameter change can be seen
2674: directly
2675: by considering multiplications in low $n$ cases.
2676: For example:
2677: \eql(pc1)
2678: \xymatrix@R=1pt%
2679: @C=25pt@M=0pt{
2680: \includegraphics{xfig/eTLB151.eps}
2681: \raisebox{.21in}{$\rightarrow$}
2682: \includegraphics{xfig/eTLB153.eps}
2683: \\
2684: \includegraphics{xfig/eTLB152.eps}
2685: \raisebox{.21in}{$\rightarrow$}
2686: \includegraphics{xfig/eTLB154.eps}
2687: }
2688: \eq
2689: To implement the proposed normalisation, let us call the top left
2690: diagram $d_1$ and the bottom left $d_2$. Then on the left,
2691: implementing the normalisation, we want to consider
2692: $\frac{d_1}{\delta_e} \frac{d_2}{\delta_e}
2693: = \frac{\delta_e^3}{\delta_e^2} e U_2 U_4
2694: = \delta_e^2 \frac{e U_2 U_4}{\delta_e} $.
2695: The figure shows that the left-hand side of this identity passes to
2696: $\kappa^2 U_1 U_3$ under $\rho_1$, while the right-hand side passes
2697: directly to $ \delta_e^2 U_1 U_3$.
2698: Here we see that, allowing for the normalisation of the first blob as
2699: an idempotent, on the left we pick up a factor $\delta_e^2$,
2700: and on the right a factor $\kappa^2$.
2701: Meanwhile
2702: \eql(pc2)
2703: \xymatrix@R=1pt%
2704: @C=25pt@M=0pt{
2705: \includegraphics{xfig/eTLB155.eps}
2706: \raisebox{.21in}{$\rightarrow$}
2707: \includegraphics{xfig/eTLB156.eps}
2708: \\
2709: \includegraphics{xfig/eTLB155.eps}
2710: \raisebox{.21in}{$\rightarrow$}
2711: \includegraphics{xfig/eTLB156.eps}
2712: }
2713: \eq
2714: has $\kappa^2$ on the left but $\delta_e^2$ on the right.
2715:
2716:
2717: %}}}
2718:
2719: %}}}
2720: \section{\Achiralb\ algebras} \label{ASTLA}
2721: %{{{ Affine symm
2722: %{{{ preamble
2723: In this section we define several new algebras with the same flavour
2724: as $b_n$, and relate them to other algebras studied in the literature
2725: (cf. \cite{OrellanaRam01,NicholsRittenbergdeGier05}).
2726: \\
2727: The first step is to define an appropriate class of pseudodiagrams,
2728: which compose by concatenation. Then we define
2729: a reduction of pseudodiagrams into scalar multiples of basic diagrams,
2730: so that composition reduces to an algebra multiplication.
2731:
2732: %}}}
2733: \subsection{Pseudodiagram categories}\label{pdcat}
2734: %{{{ para
2735:
2736:
2737: A TL {\em pseudodiagram} is a TL diagram possibly including closed loops.
2738: NB, loops cannot move isotopically over lines, so the set of $(n,m)$
2739: TL pseudodiagrams is larger than the subset
2740: of $\DoVe{n}{m}$ with no beads.
2741:
2742: \del(pseud)
2743: {\em \cite[\S2.3]{MartinWoodcock2000}}
2744: A (left) {\em blob pseudodiagram} is a TL pseudodiagram in which any
2745: left 0-covered arc may be decorated with a (left-)blob.
2746: \end{de}
2747: There is a corresponding notion of right blob pseudodiagrams.
2748: A {\em left-right blob pseudodiagram} is a pseudodiagram which may have
2749: left and right-blob decorations, so long as every decorated arc may be
2750: deformed to touch its appropriate edge {\em simultaneously}.
2751: Write $\pseud(n,m)$ for the set of $(n,m)$-pseudodiagrams of this
2752: kind.
2753:
2754: %}}}
2755: %{{{ compose
2756:
2757: Provided they have the right number of vertices,
2758: these planar
2759: pseudodiagrams may be composed by extending the usual diagram concatenation
2760: (\ref{frubeddd}).
2761: That is, concatenate concrete representations to give a concrete
2762: representative of the composite.
2763: \begin{lem} \label{weeee}
2764: This composite is well defined
2765: (i.e. independent of the choice of representatives),
2766: associative and unital.
2767: \end{lem}
2768: {\em Proof:}
2769: The argument of Proposition~\ref{abacus} is not affected by the need to take
2770: account of isotopy classes distinguished by the embedding of closed loops.
2771: \Qed \\
2772: Further
2773: \prl(logic)
2774: The composition of an $(n,l)$-- and an $(l,m)$--pseudodiagram of the same
2775: type (ordinary, blob, left-right blob) is an $(n,m)$--pseudodiagram
2776: of that type.
2777: For each type the triple $(\N,\pseud,\circ)$
2778: is a category, where $\pseud(n,m)$ is the set of morphisms for
2779: $n,m \in\N$.
2780: \end{pr}
2781: %}}}
2782: %{{{ features
2783: Note that there are unboundedly many pseudodiagrams of each type.
2784: Various features can appear (repeatedly) in pseudodiagrams, such as:
2785: \newline
2786: ($\delta$) undecorated loops;
2787: \newline
2788: ($\delta_L$, $\delta_R$)
2789: consecutive runs of two left or right-blobs on the same arc ($LL$ or $RR$);
2790: \newline
2791: ($\kappa_L$, $\kappa_R$) loops decorated with a left or right-blob;
2792: \newline
2793: ($\kappa_{LR}$) loops decorated with a left and a right-blob
2794: (NB, LRLR sequences are not possible on loops, so it is always possible to
2795: arrange blobs on loops into at most two same-type runs);
2796: \newline
2797: ($k_L$) in a pseudodiagram with unique propagating line; this line can have
2798: LRL (resp. RLR) sequences on it.
2799: \newline
2800: It will be convenient to be able to refer to these features by
2801: the set of shorthand names indicated:
2802: $P_B \; := \; \{\delta, \delta_L, \delta_R, \kappa_L, \kappa_R,
2803: \kappa_{LR}, k_L \}$.
2804:
2805: \prl(fin of)
2806: For given $(n,m)$, there are only finitely many
2807: diagrams with {\em none} of the features in $P_B$.
2808: \end{pr}
2809: {\em Proof:} Note that there are only finitely many underlying
2810: (undecorated) TL
2811: shapes possible, since all possible decorated loops are excluded.
2812: Then only finitely many decorations of the
2813: lines in these loop free shapes remain.
2814: \Qed
2815:
2816: For example in case $(1,1)$ every diagram has underlying TL diagram
2817: with just one string. The $R,L$-word on this string is one of
2818: $\{ \emptyset, L, R, LR, RL \}$.
2819:
2820: %}}}
2821: %{{{ lr blob diagram set defn
2822:
2823: \del(Bx')
2824: A left-right blob diagram is a pseudodiagram in which none of the
2825: features above occur.
2826: The set of left-right blob diagrams with
2827: $m$ vertices on the northern edge and $m$ vertices on the southern edge
2828: is denoted $B^{x'}_{m}$.
2829: \end{de}
2830: This set $B^{x'}_{m}$
2831: is thus the set of diagrams whose underlying TL diagram has no
2832: loops, and where each western exposed line may be decorated with at
2833: most one left blob, and each eastern exposed line may be decorated
2834: with at most one right blob, subject to the condition that it must be
2835: possible to deform each blob to its appropriate edge simultaneously
2836: without lines crossing.
2837: Each diagram has a representation as a pair partition with
2838: decorations, much as in the $b_n$ case except that decorations $R$ and
2839: even $LR$ and $RL$ may be possible. For example
2840: \[
2841: \{ \{1,2 \}_{LR} , \; \{1',2' \}_{LR} \} \in B^{x'}_{2}
2842: \]
2843: More generally, focusing on $L$-decorated pairs, the simultaneous
2844: deformation requirement gives the following.
2845: \prl(LR)
2846: If $d \in B^{x'}_{m}$ and $i,j \in \{1,\ldots,m\}$ then:
2847: \\
2848: $(i,j)_{LR}$ a pair part in $d$ implies $i<j$ are the two largest
2849: unprimed numbers in the list of $L$-decorated pairs;
2850: \\
2851: $(i,j')_{LR}$ or $(i,j')_{RL}$
2852: a pair part in $d$ implies $i$ (\resp\ $j'$) is the largest
2853: unprimed (\resp\ primed) number in the list of $L$-decorated pairs;
2854: \\
2855: $(i',j')_{LR}$ a pair part in $d$ implies $i<j$ are the two largest
2856: primed numbers in the list of $L$-decorated pairs.
2857: \end{pr}
2858: Thus
2859: \begin{co} \label{ok}
2860: At most two $R$s can appear in the list of $L$-decorated pairs
2861: in $d \in B^{x'}_{m}$,
2862: and then precisely in the situation of the following figure.
2863: \end{co}
2864: \eql(tq RHS)
2865: \includegraphics{xfig/eTLB8xRHS.eps}
2866: \eq
2867: Note that in pictures we use a solid blob for $L$ and $\circ$ for
2868: $R$.
2869:
2870: \del(Bxm)
2871: The set $B^{x}_{m}$ is obtained from $B^{x'}_m$ by discarding the
2872: diagrams of the type shown
2873: in (\ref{tq RHS}).
2874: \end{de}
2875:
2876:
2877:
2878: %}}}
2879: \subsection{Initial pseudodiagram reduction}
2880: %{{{ leadsto
2881:
2882: \newcommand{\mysim}[1]{\stackrel{#1}{\leadsto}}%
2883: Let $d$ be a pseudodiagram. Write
2884: $$
2885: d \mysim{x} d'
2886: \qquad
2887: (x \in P_B)
2888: $$
2889: if $d,d'$ differ by removal of one corresponding loop
2890: or, respectively, $L,R$-string replacement:
2891: $LL \leadsto L$,
2892: $RR \leadsto R$,
2893: $LRL \leadsto L$,
2894: $RLR \leadsto R$.
2895:
2896: %}}}
2897: %{{{ more
2898:
2899: It will be evident that all maximal chains of relations
2900: $\mysim{\delta}$ starting from $d$ end in a diagram with no
2901: $\delta$-loops, and are of the same length --- call this length
2902: $\hash_{\delta}(d)$;
2903: and that $\hash_{\delta_L}(d)$, $\hash_{\delta_R}(d)$ may be defined
2904: similarly.
2905: Indeed
2906: \prl(reducer)
2907: For any $d$ there is always a chain of relations, with each relation some
2908: $\mysim{x}$ ($x \in P_B$),
2909: ending in a diagram with {\em none} of the identified features.
2910: There are in general multiple such chains from $d$, but every
2911: one ends in the same `reduced' diagram --- call it $r(d)$.
2912: Each such chain for $d$ has the same number of links of the form $\mysim{x}$
2913: for {\em given} $x \in P_B$.
2914: \end{pr}
2915: This number of links defines $\hash_x(d)$ for each $x \in P_B$.
2916: \\
2917: {\em Proof:} Note that the reductions are of two types: those that
2918: shorten the sequence of decorations on some arc of some line; and
2919: those that remove a loop.
2920: Both types are localised to individual lines
2921: (leaving all other structure unchanged),
2922: so we may talk of individual lines as being `locally' reduced.
2923: Note also that loop removals only apply to
2924: loops that are locally reduced.
2925: Thus we may consider the reduction of each individual line.
2926: On each line we have a simple Bergman diamond
2927: \cite{Bergman} for the reduction
2928: of sequences of Ls and Rs.
2929: The only ambiguity is the reduction of LRLR to LR via LRL or RLR
2930: replacement, but both of these has $x=k_L$.
2931:
2932: Once line-local reductions are complete, the loop removal process is
2933: immediate.
2934: \Qed
2935:
2936: %}}}
2937: \subsection{Towards finite and localisable algebras}
2938: %{{{ hash
2939:
2940: Note that if $\hash(a)$ is the number of occurences of any given one
2941: of the features $P_B$ in pseudodiagram $a$, then
2942: \[
2943: \hash(ab) \geq \hash(a) + \hash(b)
2944: \]
2945: With this in mind, we can set out to define a finite dimensional
2946: algebra of diagrams (with fixed number of northern and southern
2947: vertices) by applying a quotient rule which equivalences a
2948: pseudodiagram with some such feature with a scalar multiple of the
2949: same diagram but with this feature
2950: excised or replaced as in $\mysim{x}$.
2951: Thus
2952: \eql(LL)
2953: L L = \delta_L L
2954: \eq
2955: \eql(LRL)
2956: LRL=k_L L
2957: \eq
2958: and so on, replacing each feature in $P_B$ with
2959: a correspondingly named scalar:
2960: \eql(Loop)
2961: d=k_d r(d)
2962: \eq
2963: where $k_d = \prod_{x \in P_B} x^{\hash_x(d)}$.
2964: (Note that this is not a unique finitising procedure
2965: --- for example we could omit
2966: the excision of $RLR$, to leave a slightly larger but still finite
2967: quotient.)
2968:
2969: The set $B^{x'}_{m}$ is a basis for the proposed
2970: finite-dimensional quotient algebra.
2971: To see that this quotient is internally consistent
2972: note that composition proceeds by concatenation to produce a
2973: pseudodiagram, which is then reduced using the relations.
2974: This reduction is consistent by Proposition~\ref{reducer}.
2975:
2976: %}}}
2977: %{{{ ab init
2978:
2979: Ab initio one might try to assign a different scalar parameter to
2980: the given LRL and RLR reductions,
2981: but they are related by commuting diagram:
2982: \[
2983: \xymatrix{
2984: & LRLR \ar@{->}[dl] \ar@{->}[dr]_{RLR=k_R R}
2985: \\
2986: k_L LR \ar@{=}[rr] & & k_R LR
2987: }
2988: \]
2989:
2990: Note that it is not possible to reduce the occurrence of LRL (or RLR)
2991: in a loop using $k_L$ ($=k_R$) since this requires that the LRL reside
2992: on the only propagating line, so there is no return route to complete
2993: the loop.
2994: Indeed $k_L$ never occurs in even index
2995: (even $m$)
2996: algebras,
2997: while $\kappa_{LR}$ {\em only} occurs in even index algebras
2998: (no propagating line can pass either to the left or to the right of
2999: the loop).
3000: It will turn out to be appropriate to set
3001: \eql(KLR=KL)
3002: \kappa_{LR} = k_L
3003: \eq
3004: as can thus be done without loss of generality.
3005:
3006: %}}}
3007: \newcommand{\abparams}{\delta,\delta_L,\delta_R,\kappa_L,\kappa_R,\kappa_{LR}}%
3008: %{{{ pre affine blob and a question
3009:
3010: At this stage we define algebra
3011: $b^{x'}_{m}(\abparams)$
3012: to be the quotient of the linear extension of the pseudodiagram
3013: composition by the relations (\ref{LL}), (\ref{LRL}) and so on
3014: associated to the $\mysim{x}$ relations.
3015: The following is clear.
3016:
3017: \prl(blob is sub)
3018: There is an algebra monomorphism
3019: $$
3020: b_{m}(\delta,\delta_L,\kappa_L) \hookrightarrow b^{x'}_{m}(\abparams)
3021: $$
3022: that takes $e \mapsto L_1$. \Qed
3023: \end{pr}
3024:
3025: We want to define an algebra (the {\em \achiralb\ algebra})
3026: that will be the quotient of $b^{x'}_m$ by some
3027: small set of further relations,
3028: chosen so as to make the representation
3029: theory of the algebra tractable by localisability.
3030: (These relations will also turn out
3031: to make contact with some other interesting algebras;
3032: see section~\ref{ASTLA'}.)
3033: That is, following the remarks in section~\ref{remi} we may
3034: determine the structure of such an algebra at level $m$ by constructing an
3035: idempotent subalgebra isomorphic to some known algebra
3036: (i.e. some version of the level $m-1$ algebra, known by inductive
3037: hypothesis)
3038: --- the localisation of the algebra under consideration.
3039:
3040: %}}}
3041: \medskip
3042: %{{{ affine-C
3043:
3044: Before passing to a localisable quotient,
3045: note that $b_n^{x'}$ is a quotient of the affine-$C$ Hecke algebra
3046: defined in section~\ref{INot}:
3047: \prl(hecke C)
3048: The map $u_i \mapsto U_i$ ($n>i>0$),
3049: $u_0 \mapsto e$ (left-blob),
3050: $u_n \mapsto f$ (right-blob),
3051: extends to an algebra homomorphism $\phi$ from $T(\hat{C}_n)(q_0,q_1,q_x)$
3052: to $b_n^{x'}$ in case
3053: $q=q_1$,
3054: $\delta_L= q_0+q_0^{-1}$,
3055: $\delta_R= q_x+q_x^{-1}$,
3056: $\kappa_L = \frac{q_0^2 + q^2}{q_0 q}$,
3057: and $\kappa_R = \frac{q_x^2 + q^2}{q_x q}$.
3058: \end{pr}
3059: {\em Proof:} The relation checking is largely routine.
3060: Note from (\ref{uuuu}) that
3061: $\phi(u_1u_0u_1 - \frac{q_0^2 + q_1^2}{q_0 q_1} u_1)$ vanishing is
3062: sufficient to ensure
3063: $\phi(E_{B_2})=
3064: \phi(u_0(u_1u_0u_1 - \frac{q_0^2 + q_1^2}{q_0 q_1} u_1))=0$,
3065: and similarly with $u_0,q_0$ replaced by $u_n,q_x$.
3066: \Qed
3067:
3068: \medskip
3069:
3070: This is closely analogous to the connection between the blob algebra
3071: and type-$B$ Hecke algebra (Proposition~\ref{hecke}), which has
3072: allowed the localisability of the {\em blob} to be used to
3073: investigate the representation theory of that Hecke algebra
3074: \cite{CoxGrahamMartin03}.
3075:
3076: %}}}
3077: \subsection{Combinatorial localisation}
3078: %{{{ localisation
3079:
3080: By analogy with section~\ref{remi},
3081: we may form the subset $B^{x'e}_{m}$ of diagrams
3082: (normalised by $1/\delta_e = 1/\delta_L$ as before)
3083: in which the
3084: string(s) involving vertices 1 and $1'$ are decorated (with $L$).
3085: \\
3086: The set $B^{xe}_{m}$ is the subset of $B^{x'e}_{m}$
3087: whose underlying diagrams are elements of
3088: $B^{x}_{m}$.
3089: \medskip
3090:
3091: Our relations imply that this subset $B^{x'e}_{m}$ spans a subalgebra.
3092: The question is:
3093: {\em How do we construct an isomorphism between this subalgebra
3094: (or a `large' quotient) and the algebra at level $m-1$?}
3095: (Note that this choice of idempotent subalgebra
3096: breaks the left-right symmetry of the algebra
3097: --- so there is a corresponding R--based formulation.)
3098:
3099: To address this we first construct set maps between these algebras'
3100: basis diagrams.
3101:
3102: %}}}
3103: %{{{ affine blob (extend rho1)
3104:
3105: \del(def pre tilde rho)
3106: Let $\pseud(m,m)^e$ denote the subset of $\pseud(m,m)$
3107: in which the string(s) involving vertices 1 and $1'$ are $L$-decorated,
3108: and $\pseud_o(m,m)$ the subset with no loops and no
3109: multiple $L$-decorations on the same segment,
3110: and $\pseud_o(m,m)^c = \pseud_o(m,m) \cap \pseud(m,m)^e$.
3111: \end{de}
3112:
3113: Let $\tilde\rho$ denote the direct extension of the $\rho_1$ map
3114: (\ref{rho_1})
3115: for $b_n$ to $ B^{x'e}_{m}$.
3116: That is
3117: \[
3118: \tilde\rho : B^{x'e}_{m} \rightarrow \pseud(m-1,m-1)
3119: \]
3120:
3121: Examples:
3122: \eql(top q1)
3123: \includegraphics{xfig/eTLB160.eps}
3124: \eq
3125: \eql(top q1+)
3126: \includegraphics{xfig/eTLB15x.eps}
3127: \eq
3128:
3129:
3130: Note that the range is not $B^{x'}_{m-1}$, because segments with
3131: decoration RLR are possible, as (\ref{top q1}) illustrates.
3132:
3133: %}}}
3134: %{{{ rho*
3135:
3136:
3137: \del(def rho*)
3138: Let
3139: \[
3140: \rho^* : B^{x}_{m-1}
3141: \rightarrow \{ \frac{1}{\delta_L} p \; | \; p \in \pseud(m,m)^e \}
3142: \]
3143: be defined as follows. If $d$ has no $L$ decorations then
3144: it maps to the same diagram but with a propagating $L$-decorated line
3145: on the left. Otherwise deform all blobs to just outside the western
3146: edge; cut the blobs off ---
3147: so some positive even number of lines now end at the western edge;
3148: deform the top and bottom-most of these lines so that they end on the
3149: top and bottom edge respectively;
3150: $L$-decorate these lines;
3151: and reclose the remaining western endpoints with $L$-decorated arcs in
3152: adjacent pairs in the only possible way.
3153: \end{de}
3154:
3155: %}}}
3156: %{{{ p*p
3157:
3158: \prl(p*p)
3159: The map $\tilde\rho$ restricts to a bijection:
3160: \[
3161: \tilde\rho : B^{xe}_{m} \rightarrow B^{x}_{m-1}
3162: \]
3163: with inverse $\rho^*$.
3164: \end{pr}
3165: {\em Proof:}
3166: First note that if $d$ has $\{1,1'\}$ as $L$-decorated pair the result
3167: $
3168: \rho^* ( \tilde\rho (d)) = d
3169: $
3170: is clear from the definitions.
3171:
3172: For $d \in B^{xe}_{m}$ let $\{1,i_1 \},\ldots, \{i_l,1'\}$
3173: be the $L$-decorated pairs.
3174: By Corollary~\ref{ok} at most one of these is $R$-decorated
3175: --- $\{ i_r, i_{r+1} \}$ say.
3176: Then $\tilde\rho(d)$ has the same
3177: (suitably relabeled) non-$L$-decorated pairs, and the $L$-decorated
3178: pairs $\{i_1,i_2 \},\ldots, \{i_{l-1},i_l\}$;
3179: with the further $R$-decoration (if any) on
3180: $\{ i_{r-1}, i_{r} \}$ or $\{ i_{r+1}, i_{r+2} \}$
3181: as appropriate:
3182:
3183: \eql(move4)
3184: \includegraphics{xfig/move4.eps}
3185: \eq
3186: Note that this establishes that
3187: $\tilde\rho( B^{xe}_{m} ) \subseteq B^{x}_{m-1}$.
3188:
3189: For $d \in B^{x}_{m-1}$ let $\{i_1,i_2 \},\ldots, \{i_{l-1},i_l\}$
3190: be the list of $L$-decorated pairs. As before there is at most one
3191: $R$-decoration in this list, on $\{ i_{r-1}, i_{r} \}$ say.
3192: Then $\rho^*(d)$ has the same non-$L$-decorated pairs as $d$, and
3193: $L$-decorated pairs $\{1,i_1 \},\ldots, \{i_l,1'\}$.
3194: It has at most one further $R$, on
3195: $\{ i_{r-2}, i_{r-1} \}$ or $\{ i_{r}, i_{r+1} \}$
3196: as appropriate.
3197: Note that it follows that
3198: $\rho^*( B^{x}_{m-1} ) \subseteq B^{xe}_{m}$.
3199:
3200: Considering $\rho^* ( \tilde\rho (d))$ then, the lists of $L$-decorated
3201: (and undecorated) pairs are manifestly restored.
3202: The location of the $R$-decoration (if any) in the $L$-decorated list
3203: may be seen to be restored by considering the four cases
3204: in Proposition~\ref{LR} (illustrated above).
3205:
3206: A similar argument shows the right inverse property.
3207: \Qed
3208:
3209: %}}}
3210: %{{{ footnote: claim -> \stuffa
3211: \newcommand{\stuffa}{{
3212: \parker{I THINK WE CAN DROP THIS NOW?:}
3213:
3214: A pictorial version of prop~\ref{p*p} is as follows.
3215: \prl(claim big)
3216: The map $\tilde\rho$ restricts to a bijection:
3217: \[
3218: \tilde\rho : B^{xe}_{m} \rightarrow B^{x}_{m-1}
3219: \]
3220: \end{pr}
3221: \footnote{OLD VERSION:
3222: \begin{claim}
3223: This requirement is satisfied in general by imposing
3224: the `topological' quotient (\ref{top quot}).
3225: \end{claim}
3226: }
3227: {\em Proof:}
3228: First we shall show that this is a map --- i.e. that the range may be
3229: restricted as indicated.
3230: To begin, consider the range of $\tilde\rho$ as originally defined.
3231: We shall argue that no `reducible component'
3232: (e.g. closed loop, or LL on the same line segment) appears in any
3233: pseudodiagram in the range, except possibly RLR
3234: (whose appearance is then excluded by the restriction we have now imposed).
3235:
3236: The restriction to $B^e_{n}$ is $\rho_1$, with range $B_{n-1}$ which,
3237: note, has no closed loops.
3238: Every diagram in $B^{x'e}_{n}$ is obtained from one in $B^e_{n}$ by
3239: adding some R decorations, but R decorations do not affect the
3240: underlying shape change produced by $\tilde\rho$, so:
3241: \\
3242: 1. the range of $\tilde\rho$ includes no diagram with loops.
3243: \\
3244: In order for a component of form LL to appear as an image,
3245: the diagram in its
3246: preimage would have to contain a loop, by the definition of $\rho_1$
3247: (which is straightforwardly reversible on appropriate
3248: components of pseudodiagrams):
3249: \[
3250: {\tilde\rho}^{-1}: \raisebox{-0.542in}{\includegraphics{xfig/rhoinv1.eps}}
3251: \]
3252: and indeed for RR to appear the preimage would also have to contain
3253: RR.
3254: For LRL to appear would similarly require a preimage with a
3255: (decorated) loop,
3256: so:
3257: \\
3258: 2. the range of $\tilde\rho$ includes no diagram with LL or RR or LRL.
3259: \\
3260: Components with RLR can occur, as we have already demonstrated, but an
3261: inverse image analysis similar to the above
3262: shows that a preimage precisely of the form
3263: eliminated by the topological quotient
3264: (i.e. not present in the domain in the proposition) is required.
3265:
3266: Next we shall show that with the domain in the proposition no diagram
3267: of the form eliminated by the topological quotient can appear as an
3268: image. The component inverse image picture is of the form:
3269: \[
3270: {\tilde\rho}^{-1}: \raisebox{-0.542in}{\includegraphics{xfig/rhoinv2.eps}}
3271: \]
3272: The connection shown as a dashed line is forced since their can be no
3273: lines decorated with L to the right of the line(s) decorated with R,
3274: by the definition of LR-pseudodiagram.
3275: The dashed line gives a component which does not occur in any diagram
3276: in the domain, so we are done here.
3277:
3278:
3279: NOW Think about showing surjectivity. (And
3280: construct the inverse?)
3281:
3282: \parker{PROOF end NEEDED.}
3283: }}
3284: %}}}
3285: %{{{ remark/footnote: choice of final extension
3286:
3287: \medskip
3288: \noindent {\em Remark:}
3289: Map $\tilde\rho$ extends in the obvious way
3290: to a map from $\pseud(m,m)^e$
3291: to $\pseud(m-1,m-1)$ ---
3292: it is only necessary further to specify that
3293: every blob deformation
3294: (in the sense of the illustration to the definition of $\rho_1$)
3295: that is not on a closed loop,
3296: except the lowest such, passes
3297: to the north of every blob deformation that is
3298: --- see the figure below for an illustration.
3299:
3300: Map $\rho^*$ extends in the obvious way to the domain $\pseud(m-1,m-1)$,
3301: provided again that closed loops are unambiguously treated
3302: in the deformation process --- let us say that they are deemed to
3303: congregate in the north-west.
3304:
3305:
3306: The extended map $\tilde\rho$ is illustrated by the following example:
3307: \[
3308: \includegraphics{xfig/eTLB15pseud2.eps}
3309: \]
3310: Note that we have
3311: \[
3312: \pseud(m,m)^e \stackrel{\tilde\rho}{\rightarrow} \pseud(m-1,m-1)
3313: \stackrel{\rho^*}{\rightarrow} \pseud(m,m)^e
3314: \]
3315: but the two maps are not inverse with this domain and range,
3316: because of the necessarily
3317: arbitrary choice of treatment of closed loops.
3318: With the choice specified, $\rho^*(\tilde\rho(d))$ is similar to $d$,
3319: except that all `extra' blobs are gathered on the top left line,
3320: regardless of where they where in $d$.
3321:
3322: %}}}
3323: \subsection{Algebra localisation and the \achiralb\ algebra}
3324: %{{{ the obvious candidate
3325:
3326: The obvious candidate to construct our algebra isomorphism is
3327: $\tilde\rho$ on $B^{x'e}_{m}$.
3328: But note that the range is not $B^{x'}_{m-1}$ here,
3329: and this is not a bijection.
3330: Consider the examples in (\ref{top q1})
3331: (neglecting the $1/\delta_e$ factor for the moment):
3332: Note that the right hand side of the bottom example
3333: in (\ref{top q1}) is $k_L$
3334: times the right hand side of the upper example (by the $RLR$ relation).
3335: Thus if we want to have an isomorphism we must require the same of the
3336: left hand sides, that is, we must impose a further quotient relation.
3337: This means in particular
3338: that we eliminate the diagram with two double-decorated
3339: lines from the basis.
3340:
3341: %}}}
3342: %{{{ pre-claim defn
3343:
3344: \del(top quot)
3345: The `topological' quotient
3346: of $b^{x'}_{m}(\abparams)$
3347: is defined by
3348: \eql(topquot)
3349: \kappa_{LR} \;\; \raisebox{-0.2in}{\includegraphics{xfig/eTLB8x.eps}}
3350: \eq
3351: Here each labelled shaded area is shorthand for a certain subdiagram.
3352: Thus each diagram restricts to the same subdiagram in
3353: the shaded region marked A (resp. B, C, D),
3354: but (\ref{topquot}) represents an identity for each such
3355: arrangement.
3356: (Note that in C and D there must be a route for the adjacent blob to
3357: the edge, hence no propagating lines.
3358: Indeed there are no propagating lines at all on the right hand side)
3359: \end{de}
3360:
3361: %}}}
3362: %{{{ affine blob
3363:
3364: \del(affine blob)
3365: Define the \achiralb\ algebra $b^{x}_{m}(\abparams)$
3366: to be the quotient of $b^{x'}_{m}(\abparams)$ by the
3367: additional set of relations (\ref{topquot}).
3368: \end{de}
3369:
3370: \prl(a b basis)
3371: The set $B^x_m$ is a basis for $b^{x}_{m}(\abparams)$.
3372: \end{pr}
3373: {\em Proof:}
3374: We use Bergman's diamond lemma \cite{Bergman}.
3375: %{{{ ver 2
3376:
3377:
3378: Let us define the {\it height} of a
3379: (pseudo)diagram to be the sum of the number of
3380: loops and the number of decorations. We observe that all the nontrivial
3381: relations alter the height of a diagram. This is easily checked except in
3382: the case of the topological relation, in which case two of the decorations
3383: are removed. The vertical edges shown on the left hand diagram $D_0$ in
3384: Definition~\ref{top quot}
3385: are the only vertical edges in $D_0$, and thus these edges can
3386: only participate in a total of one loop in a product $D_1 D_0 D_2$
3387: containing $D_0$. This means that at most one loop can be deleted when
3388: applying the topological relation to reduce the number of decorations, and
3389: thus that the height decreases by at least $2-1 = 1$.
3390:
3391: We may now choose a semigroup order, $<$, on the corresponding free
3392: diagram algebra in such a way that smaller diagrams have strictly smaller
3393: height. We aim to conclude that a minimal diagram in this sense is a basis
3394: element. To do so, we need to worry about inclusion ambiguities (of which
3395: there are none) and overlap ambiguities. (We are assuming familiarity
3396: with Bergman's set-up.) Imagine that the diagrams shown in
3397: Definition~\ref{top quot} are
3398: sandwiched between other diagrams to the top and bottom in a triple
3399: product.
3400:
3401: There is nothing to check for $k_L$-type relations, because they cannot
3402: occur in ``even index" algebras, which is where this problem arises.
3403:
3404: The $\kappa_L$ and $\kappa_R$ relations cause no problem because they commute
3405: with the topological relation.
3406:
3407: The $\delta_L$ and $\delta_R$ relations also commute with the topological
3408: relation, because they never remove the last $L$ (respectively, $R$)
3409: decoration.
3410:
3411: The $\delta$ relation is easy to deal with because it cannot interact with
3412: the topological relation, and thus the relations commute.
3413:
3414: The only nontrivial case is the $\kappa_{LR}$ relation:
3415:
3416: Suppose the top of the right hand side of
3417: Definition~\ref{top quot} is part of a
3418: $\kappa_{LR}$-type loop.
3419: Then we have a choice: we can contract the $\kappa_{LR}$
3420: loop first and then apply the topological relation, or vice versa. The
3421: ambiguity is resolvable here, however, because the region A (plus whatever
3422: is just above it) can only be a disjoint collection of undecorated loops,
3423: so they can be deleted and then recreated anywhere in the diagram where
3424: they will not cause an intersection.
3425:
3426: A similar case deals with the region B and a $\kappa_{LR}$ loop at the bottom.
3427:
3428: If there is a $\kappa_{LR}$ feature on the left hand side, then there must be
3429: two such features on the right hand side, and a similar argument again
3430: applies.
3431:
3432: According to Bergman's diamond lemma, we can conclude that the minimal
3433: diagrams in this ``height" sense are a basis, as desired.
3434: %}}}
3435: \Qed
3436:
3437: %}}}
3438: %{{{ Prop chufster: tilde\rho isomorphism
3439:
3440: \prl(chufster)
3441: Provided that $\delta_L = \delta_e$ is invertible,
3442: the map $\tilde\rho$ extends to an isomorphism
3443: \[
3444: e \; b^{x}_{m} (\abparams) \; e \; \;
3445: \cong \;
3446: b^{x}_{m-1}(\delta,\kappa_L,\delta_R,\delta_L,\kappa_R,\kappa_{LR})
3447: \]
3448: \end{pr}
3449: {\em Proof:} We have to check
3450: $\tilde\rho (d_1 d_2) = \tilde\rho(d_1) \tilde\rho(d_2)$.
3451: Composition on both sides begins with pseudodiagram concatenation
3452: --- so it remains to check that pseudodiagram reduction is consistent.
3453: This is a routine `diagram chase' similar to the blob case.
3454: The difference is that there are $R$s present
3455: --- these largely play no role,
3456: except that the $\kappa_{LR}$ reduction on the left becomes a $k_L$
3457: reduction on the right,
3458: \[
3459: \includegraphics{xfig/kLRkL.eps}
3460: \]
3461: and vice versa
3462: (however note that these matters are already
3463: resolved by our identification of these parameters in (\ref{KLR=KL})).
3464: \Qed
3465:
3466: See (\ref{pc1}) and (\ref{pc2}) for diagrams exemplifying the
3467: parameter change.
3468: Also:
3469: \[
3470: \includegraphics{xfig/eTLB19.eps}
3471: \]
3472: illustrating that $\delta_L$ and $\kappa_L$ are swapped.
3473:
3474:
3475: Note that $b^{x}_{m}$ is a radical departure from the original blob algebra,
3476: in that the topological quotient
3477: mixes between diagrams with different numbers of
3478: `propagating lines'. This appears to deny us a powerful tool in
3479: representation theory (cf. \cite{MartinSaleur94a}).
3480: However,
3481: we will determine the structure of this algebra by appealing to a
3482: slightly different realisation.
3483:
3484: %}}}
3485: %{{{ alternate
3486: {\em To reiterate:}
3487: Just as the blob algebra is isomorphic to the 0-cover $m=2$ contour algebra, so
3488: the idea of the east-west composite $m=2$ contour algebra
3489: (that is the variation in which lines which are 0-covered to the east or
3490: the west may be decorated with a left (respectively right) blob)
3491: also turns out to include a rather interesting case.
3492: There are a couple of ways in which such an algebra can be defined.
3493: Firstly note that some lines, in some diagrams,
3494: are 0-covered both to east and to west.
3495: Then left and right-blobs can meet on such a line. In general we will
3496: consider them to be distinct, and even noncommuting on the line.
3497: Consider the case in which we
3498: disallow multiple decorations with the property that it is not
3499: possible to deform both the east leaning blob to the eastern edge and
3500: the west leaning blob to the western edge simultaneously.
3501: %}}}
3502: %}}}
3503: \section{Affine symmetric TL algebra} \label{ASTLA'}
3504: %{{{ section ASTLA
3505: %{{{ para
3506: One reason why this algebra $b_m^x$ is interesting is
3507: the existence of a doubled version of the {\em unfolding} map $\mu$.
3508: As with $\mu$ on $B_m$,
3509: the western blobs in an element of $B_m^x$
3510: may again be used to map the diagram into left-right symmetric
3511: versions (so far still with eastern blobs,
3512: now with mirror
3513: images) about a reflection wall corresponding to the western edge.
3514: If we play the same game with eastern blobs we have another
3515: reflection wall corresponding to the eastern edge, which is thus
3516: {\em affine} in the affine reflection group sense
3517: \cite{Humphreys90}!
3518: Altogether we have a fundamental domain (as it were) between these two
3519: reflection walls, with a mirror image on each side (and then
3520: repeated reflections beyond).
3521: Obviously then, the mirror image on the right is a translate of that
3522: on the left and we have {\em periodicity}.
3523: Here is an example (the embossed letter `R' added to the fundamental
3524: rectangle in this figure
3525: is only intended to emphasise the mirror images):
3526: \[
3527: \includegraphics{xfig/eTLB6.eps}
3528: \]
3529: (Note as before that this is a well defined construct on isotopy classes.)
3530: We observe that the resultant diagram is an element of
3531: $\Dppc{2m}$,
3532: the set of periodic TL diagrams with $2m$ vertices along each edge
3533: in the fundamental period (see \S\ref{deform}).
3534: We denote by $\mu^x$ this unfolding map:
3535: \[
3536: \mu^x : B^x_m \longrightarrow \Dppc{2m} .
3537: \]
3538: %}}}
3539: \subsection{On properties of the unfolding map $\mu^x$}
3540: %{{{ we can extend
3541: We can extend $\mu^x$ $\Ring$-linearly into a map from the algebra $b_m^x$
3542: to $\Ring \Dppc{2m}$.
3543: It takes the generator $U_i$ in that algebra to a product
3544: $U_{i}U_{-i}$ (in a suitable labelling) in $\Dppc{2m}$,
3545: and so on.
3546: In other words we have a left-right symmetric
3547: subquotient-algebra of a {\em periodic} TL diagram algebra.
3548:
3549: Here is an example, of $\mu^x$ mapping to the cylinder realisation,
3550: viewing
3551: the cylinder along the axis, so that it appears as an annulus:
3552: \[
3553: \includegraphics{xfig/eTLB7.eps}
3554: \]
3555: \label{Dph}
3556: \del(sym period diagrams)
3557: Let
3558: $\Dph_{2m}$ denote the set of left-right
3559: symmetric periodic diagrams contained in $\Dppc{2m}$.
3560: (NB, noncontractible loops are still possible in $\Dph_{2m}$.)
3561: \end{de}
3562:
3563: %}}}
3564: %{{{ CC
3565: There is a subalgebra of the periodic algebra spanned by $D^{\phi}_{2m}$.
3566: It will be evident from the illustration above
3567: that $\mu^x(B_m^x)$ lies in this set.
3568:
3569: Note further
3570: that these diagrams can be two-coloured like ordinary TL diagrams.
3571: (Periodic diagrams with odd numbers of vertices cannot be
3572: two-coloured on the cylinder or annulus
3573: without a cohomology seam, but this need not concern us here.)
3574: For definiteness we fix that
3575: the region touching the interval of the northern edge
3576: (which becomes the {\em inner} edge in the annular realisation)
3577: astride the 0-reflection line is coloured {\em white}.
3578: For example
3579: \[
3580: \includegraphics{xfig/annulBWhite01.eps}
3581: \]
3582: Note
3583: that there is a subset of these left-right symmetric periodic diagrams
3584: with the property that the induced
3585: colouring of the intervals of the southern edge coincides with
3586: that on the northern edge.
3587: (Note that this is a proper subset in general, since it is
3588: possible to draw a symmetric periodic diagram in which precisely one
3589: line crosses the reflection line.)
3590: Diagrams with this property are called {\em colouring composable} (CC)
3591: diagrams because, when they are concatenated in the usual way
3592: (top edge to bottom edge, or inner edge to outer edge in the annular
3593: realisation) the colouring we have specified
3594: gives colours that agree across the join.
3595: It follows that the set
3596: of colouring composable
3597: diagrams again spans a subalgebra.
3598: It will be evident that $\mu^x(B_m^x)$ lies in this set
3599: (since $\mu^x$ drags finite decorated segments of decorated lines
3600: out of the diagram, creating {\em pairs} of crossings of the
3601: reflection line).
3602:
3603: %}}}
3604: %{{{ para
3605:
3606: To see that the image $\mu^x(B_m^x)$ generates a quotient
3607: of the corresponding subalgebra of the
3608: periodic algebra, we should explicitly consider the image
3609: in the set $D^{\phi}_{2m}$ of
3610: symmetric periodic diagrams described above.
3611: \prl(goophy)
3612: The map $\mu^x: B^x_{m} \longrightarrow D^{\phi}_{2m}$ is injective.
3613: \end{pr}
3614: {\em Proof: }
3615: The map is reversible at the point of
3616: deforming a blob out of the frame, so the issue is if isotopy on the
3617: target side can equivalence two diagrams.
3618: However no contractible loops are
3619: produced from diagrams in
3620: $B^x_{m}$, so no such isotopy can arise.
3621: \Qed
3622:
3623: Note that this map is not surjective on arbitrary
3624: CC symmetric periodic diagrams,
3625: since the maximum
3626: number of noncontractible loops is 1 in the image.
3627: The quotients associated to the parameters $k_L$ and $\kappa_{LR}$ on
3628: the blob side both have the effect of replacing a pair of
3629: noncontractible loops with the factor $\kappa_{LR}$.
3630:
3631: %}}}
3632: \subsection{Periodic pseudodiagram reduction}\label{reduc}
3633: %{{{ features: periodic pseudodiagrams
3634:
3635: Recall
3636: from section~\ref{deform0}
3637: that
3638: a periodic pseudodiagram is the generalisation of an ordinary
3639: pseudodiagram from a rectangular to a cylindrical geometry
3640: (or unboundedly wide rectangles with finite periodic repetition).
3641:
3642: Here we will restrict to even period left-right symmetric colouring
3643: composable pseudodiagrams.
3644: (Periodic left-right symmetric is the same as
3645: affine symmetric, as already noted.)
3646: Write $\CC_{2m}$ for the set of these pseudodiagrams with period $2m$.
3647: We will colour such that the central northern interval is white.
3648:
3649: As usual these diagrams are isotopy classes of concrete diagrams.
3650: But as in section~\ref{deform2}, having taken a subset we have the option of
3651: correspondingly strengthening the notion of isotopy.
3652: Here we consider isotopies that preserve the symmetry.
3653: (Note, then, that $\CC_{2m}$ is not the same thing
3654: as the subset of general periodic pseudodiagrams with symmetric
3655: representative elements, where a diagram with two contractible loops
3656: on either side of the reflection line is isotopic to one with two
3657: contractible loops both astride the reflection line.)
3658:
3659: Composition on $\CC_{2m}$ is defined as before.
3660:
3661: %}}}
3662: %{{{ features: periodic pseudodiagrams II
3663:
3664: \prl(features)
3665: The following list of features
3666: of concrete pseudodiagrams
3667: are preserved by the
3668: $\CC_{2m}$
3669: isotopy in this
3670: setting,
3671: and hence can be considered to appear (with well defined multiplicities)
3672: in these pseudodiagrams:
3673: \newline
3674: ($\delta$) symmetric pair of loops (one each side of the symmetry line
3675: --- the 0-reflection line);
3676: \newline
3677: ($\delta_L'$) white loop astride the 0-reflection line;
3678: \newline
3679: ($\kappa_L'$) black loop astride the 0-reflection line;
3680: \newline
3681: ($\delta_R'$) white ($m$ even) (\resp\ black ($m$ odd))
3682: loop astride the 1-reflection line;
3683: \newline
3684: ($\kappa_R'$) black ($m$ even) (\resp\ white ($m$ odd))
3685: loop astride the 1-reflection line;
3686: \newline
3687: ($\kappa_{}'$) pair of noncontractible loops.
3688: (Two such loops are `adjacent' if they may be deformed to touch, and the pair
3689: is called black (\resp\ white) if one is on the black (\resp\ white)
3690: side of the partition formed by the other.)
3691: \Qed
3692: \end{pr}
3693:
3694: %}}}
3695: %{{{ $B^{\phi}_{2m}$
3696:
3697: Let us write $B^{\phi}_{2m}$ for the subset of pseudodiagrams in
3698: $\CC_{2m}$ with
3699: {\em none} of these features.
3700: For given period $n$ there are finitely many such pseudodiagrams.
3701:
3702: Define a map $$
3703: \nu :
3704: \CC_{2m} \rightarrow
3705: D^o(V) |_{S=\{ b,w\}}
3706: $$
3707: (that is, $D^o(V)$ with two types of decoration) as follows.
3708: Given a diagram in $CC_{2m}$ consider the ur-diagram which is the
3709: strip between the 0-reflection line on the left and the 1-reflection
3710: line on the right (so the affine reflection group orbit of this strip is the
3711: whole diagram). By the CC condition there are an even number of lines
3712: leaving the strip through the 0-reflection line (and similalrly for
3713: the 1-reflection line). Thus the lines
3714: leaving the strip through the 0-reflection line may be collected into
3715: pairs,
3716: such that the two lines in a pair are consecutive on the reflection
3717: line.
3718: This means that they can be brought arbitrarily close together
3719: at the reflection line.
3720: Joining each such pair with a blob (and similarly with a white-blob
3721: at the 1-reflection line) we get an element of $D^o(V)$.
3722:
3723: Note that $\nu$ is not injective on $\CC_{2m}$.
3724: A diagram with two non-contractible loops and a (mirror) pair of
3725: contractible loops above them is mapped to the same element of
3726: $D^o(V)$ as a diagram with two non-contractible loops and a (mirror) pair of
3727: contractible loops below them.
3728:
3729: %}}}
3730: %{{{ lemma
3731:
3732: \begin{lem}
3733: The maps $\nu$ and $\mu^x$ induce a bijection between
3734: $B^x_m$ and $B^{\phi}_{2m}$.
3735: \end{lem}
3736: {\em Proof:}
3737: Firstly note that $\nu \circ \mu^x$ is the identity map on $B^x_m$
3738: ($\nu$ is the reverse of $\mu^x$, which is injective).
3739:
3740: Secondly, consider $d \in \CC_{2m} \setminus B^{\phi}_{2m}$.
3741: It has at least one of the listed features.
3742: It is routine to check
3743: that each of these produces at least one contractible
3744: loop in $\nu(d)$.
3745: Thus $\nu ( \mu^x(B^x_m) \setminus B^{\phi}_{2m})$ does not
3746: intersect $B^x_m$.
3747: But by the previous paragraph
3748: $\nu ( \mu^x(B^x_m) \setminus B^{\phi}_{2m})$
3749: is contained in
3750: $B^x_m$, so it is empty.
3751: Thus $\mu^x(B^x_m) \subseteq B^{\phi}_{2m}$.
3752:
3753: Next we show that $\nu(B^{\phi}_{2m}) \subset B^x_m$.
3754: First consider $d \in \nu(B_{2m}^\phi)$ that does not have any
3755: non-contractible loops. Any line in $d$
3756: starting at the northern edge, say, and
3757: crossing the $0$ or $1$ line
3758: cannot be propagating, and will have a
3759: corresponding
3760: line
3761: starting at the northern or southern edge
3762: paired to it by the CC'ness of $d$. Thus
3763: no string in $\nu(d)$ has more than one blob on it,
3764: and thus $\nu(d) \in B_m^x$.
3765: If $d \in \nu(B_{2m}^\phi)$ does have a
3766: (necessarily unique)
3767: non-contractible loop, then
3768: this line under $\nu$ becomes
3769: part of
3770: the unique propagating line and is
3771: decorated by exactly one black and one white blob. All other blobs come
3772: from non-propagating lines
3773: combining in pairs
3774: as before.
3775:
3776: It is easy to see
3777: that $\mu^x \circ \nu |_{B^{\phi}_{2m}}$ is the identity map.
3778: Thus $\nu$ is injective when restricted to $B^{\phi}_{2m}$.
3779: Thus finally the two sets have the same cardinality.
3780: \Qed
3781:
3782: %}}}
3783: %{{{ define b phi
3784:
3785: Denote by
3786: $b_{2m}^{\phi}(\delta,\delta_{L}',\delta_{R}',\kappa_L',\kappa_R',\kappa_{}') $
3787: the quotient of the $\Ring$-algebra spanned by
3788: $\CC_{2m}$
3789: by the relations that each feature
3790: itemised in Proposition~\ref{features}
3791: may be removed at the
3792: cost of introducing a scalar factor as indicated
3793: in Proposition~\ref{features}
3794: in brackets (each such factor then appearing, note,
3795: as an argument to $b_{2m}^{\phi}$).
3796:
3797: Since
3798: all of the features of pseudodiagrams in Proposition~\ref{features}
3799: have multiplicity weakly increasing in composition
3800: we have
3801: \prl(bbbbbasis)
3802: The \aSTL\ algebra
3803: $b_{2m}^{\phi}(\delta,\delta_{L}',\delta_{R}',\kappa_L',\kappa_R',\kappa_{}')$
3804: has basis $B^{\phi}_{2m}$.
3805: \end{pr}
3806: {\em Proof:}
3807: One uses Bergman's diamond lemma much as in Proposition~\ref{a b basis}.
3808: \Qed
3809:
3810:
3811:
3812: %}}}
3813: %{{{ hom when apply topo quotient
3814:
3815: Again this is not the only way to produce a finite rank quotient.
3816: For example $\kappa_{}'$ could be for the excision of black pairs
3817: only
3818: (see also \cite{MartinSaleur93}).
3819: However
3820: \prl(x-hom)
3821: The map $\mu^x$ extends to an algebra
3822: isomorphism
3823: \[
3824: \mu^x: b^x_{m} \longrightarrow b^{\phi}_{2m}
3825: \]
3826: with the obvious identification of parameters.
3827: \end{pr}
3828: {\em Proof:}
3829: Note from the construction that $\nu$ commutes with diagram
3830: composition, considered as a map from $\CC_{2m}$ to $D^o(V)$.
3831: It remains to show that the two different kinds of
3832: pseudodiagram reduction yield the same
3833: factors on each side.
3834: Applying $\nu$ to a diagram with two non-contractible loops will give
3835: a diagram with a loop with both types of decoration on it
3836: --- thus we set $\kappa' = \kappa_{LR}$. \\
3837: Applying $\nu$ to a diagram with a pair of contractible loops will give
3838: a diagram with an undecorated loop
3839: --- reduction on either side gives a factor $\delta$. \\
3840: Applying $\nu$ to a diagram with a white loop astride the 0-line will give
3841: a diagram with a line with two left-blobs on it
3842: --- thus we set $\delta_L' = \delta_{L}$. \\
3843: Applying $\nu$ to a diagram with a black loop astride the 0-line will give
3844: a diagram with a loop with a left-blob on it
3845: --- thus we set $\kappa_L' = \kappa_{L}$. \\
3846: The loops astride the 1-line pass across similarly.
3847: \Qed
3848:
3849:
3850:
3851: {\em Remark:}
3852: Note that $b^{x'}_{m}$ does not map injectively into $b^{\phi}_{2m}$
3853: without the `topological' quotient, since
3854: \be \non
3855: \includegraphics{xfig/eTLB9a.eps}
3856: \\
3857: \includegraphics{xfig/eTLB9b.eps}
3858: \eeq
3859:
3860: %}}}
3861: %{{{ paras
3862: We will see that
3863: even before the quotient the affine symmetric subalgebra
3864: is not complicated by as many
3865: `infinities' as the ordinary periodic TLA.
3866: We will also see that it is amenable to the same recollement treatment
3867: as the non-affine case above.
3868:
3869: The claim is that the set of diagrams that contain a cup and cap
3870: astride the 0-reflection line is a subset that spans an (idempotent)
3871: subalgebra (a similar statement holds for the 1-reflection line).
3872: There is a bijective map into the set of all diagrams with one fewer vertex on
3873: each side obtained by simply removing this cup and cap.
3874: In order to elevate this to the status of an algebra homomorphism we
3875: will again have to take care with the parameters.
3876:
3877: %}}}
3878: %}}}
3879: \section{Representation theory of ASTLA} \label{rep ASTLA}
3880: %{{{ repn thy
3881: %{{{ f
3882: In what follows $f$ corresponds to the right blob, in the way that $e$
3883: (or $\abe'$)
3884: corresponds to the left blob:
3885: \[
3886: f \mapsto \;\;\;
3887: \raisebox{-2.cm}{
3888: \includegraphics{xfig/fannulBWhite01.eps}
3889: }
3890: \]
3891: %}}}
3892: \subsection{General and generic results}
3893: %{{{ cup cap insert
3894: Assume that $\delta_L$ is invertible in $\Ring$.
3895: Let
3896: $\rho': D^{\phi}_{2m-2}
3897: \longrightarrow \frac{1}{\delta_L} D^{\phi}_{2m}
3898: \; \subset K D^{\phi}_{2m}$
3899: denote the map that inserts a cup and cap astride the
3900: 0-reflection line and then rescales by $ \frac{1}{\delta_L} $.
3901:
3902: Note that the map $\rho'$ is injective, with image
3903: the set of {\em all} (rescaled) diagrams in $D^{\phi}_{2m}$
3904: with a cup and cap astride the
3905: 0-reflection line.
3906: Thus
3907: $\rho'( B^{\phi}_{2m-2})$ spans the
3908: subalgebra
3909: $\frac{\abe'}{\delta_L} b_{2m}^{\phi}(\delta,...) \frac{\abe'}{\delta_L}$
3910: of $b_{2m}^{\phi}$
3911: in a similar way to the ordinary blob case.
3912:
3913: Let $\rho$ denote the map on
3914: $ \rho'( D^{\phi}_{2m-2}) $
3915: that removes this cup and cap and normalisation (inverse to $\rho'$).
3916:
3917: %}}}
3918: %{{{ constructing hom
3919: \prl(rho hom)
3920: The map $\rho$ extends to an algebra isomorphism
3921: \eql(ebe)
3922: \abe' \;
3923: b_{2m}^{\phi}(\delta,\delta_{L},\delta_{R},\kappa_L,\kappa_R,\kappa_{LR})
3924: \; \abe' \;\;
3925: \stackrel{\rho}{\longrightarrow} \;
3926: b_{2m-2}^{\phi}(
3927: \delta,\kappa_L,\delta_{R},\delta_{L},\kappa_R,\kappa_{LR} ) .
3928: \eq
3929: \end{pr}
3930: {\em Proof:}
3931: This follows from Propositions~\ref{chufster} and~\ref{x-hom}.
3932: However it is useful to sketch a direct proof analogous to
3933: Proposition~\ref{recool}.
3934:
3935: In order to readily distinguish $\delta_L$ and $\kappa_L$ in the
3936: periodic realisation it is again useful to two-colour the diagrams. As
3937: before we set the interior of the region whose closure includes the
3938: northern interval astride the 0-reflection line to {\em white}.
3939: Then $\delta_L$ loops of
3940: $b_{2m}^{\phi}(\delta,\delta_{L},\delta_{R},\kappa_L,\kappa_R,\kappa_{LR})$
3941: are white loops astride this line.
3942: The colour of the corresponding interval (and hence loops) astride the
3943: 1-reflection line depends on whether $m$ is odd or even.
3944: (Thus the image of $f.f=\delta_R f$ is a black loop if $m$ is odd.)
3945:
3946: As in the ordinary blob case, comparing $\rho(a)\rho(b)$ to $\rho(ab)$
3947: the underlying diagrams agree, and there is a correspondence between
3948: the loops produced on each side, but the cup and cap removal means
3949: that they all change colour.
3950: Thus, on applying $\rho$ to a pseudodiagram
3951: (a diagram with loops which reduce to scalars),
3952: the roles of $\delta_L$ and $\kappa_L$ are interchanged.
3953: As for $\delta_R$ and $\kappa_R$, there is a colour change due to
3954: $\rho$, which nominally interchanges them, but $\rho$ also changes $m$
3955: between odd and even, changing back.
3956: Altogether, we have $\rho(a)\rho(b)=\rho(ab)$ with $\rho$ as in (\ref{ebe}).
3957: \Qed
3958:
3959: %}}}
3960: %{{{ meta-category
3961: We could bring the two sides of (\ref{ebe}) closer together by making
3962: further constrained parameter choices.
3963: However, here we will concentrate on the generic case, i.e.
3964: {\em working with field $k=\C$, so that parameter space
3965: can be endowed with the Zariski topology, }
3966: we assume that Zariski open subsets of points in parameter space
3967: all have basically the same representation theory
3968: (in the sense that the basis for simple module $M_{\lambda}$ (say)
3969: is in each case the image of the {\em same} module basis over ground ring $K$
3970: under $M^K_{\lambda} \mapsto k \otimes_K M^K_{\lambda}$
3971: by specialisation).
3972: We will verify this assumption shortly.
3973: Under this assumption we may consider a single meta-category
3974: $b_{2m}^{\phi}$-mod of left modules for $b_{2m}^{\phi}$.
3975: Then by (\ref{glob})
3976: \prl(meta-cat)
3977: Map (\ref{ebe}) provides a full embedding $G$ of $b_{2m-2}^{\phi}$-mod in
3978: $b_{2m}^{\phi}$-mod.
3979: \Qed
3980: \end{pr}
3981: This means in particular that we can construct prestandard modules by
3982: a recursive procedure.
3983: (NB, much representation theory
3984: can still be done with the restrictions on the field imposed here removed,
3985: but at considerable cost in brevity.)
3986:
3987: %}}}
3988: %{{{ simples theo
3989:
3990: \begin{theo}\label{index wangy}
3991: An index set for equivalence classes of simple $b_{2m}^{\phi}$-modules
3992: for generic parameters is $\{ 0 \}$ for $m=0$; and
3993: $\Lambda^{\phi}_{m} = \{ -m,-m+1,..,0,..,m-1 \}$ for $m>0$.
3994: \end{theo}
3995: {\em Proof:}
3996: By Proposition~\ref{grebo} %%the standard machinery
3997: there is for each simple module (irreducible representation)
3998: $S_{\lambda}(2m-2)$
3999: in $b_{2m-2}^{\phi}$-mod a simple module in $b_{2m}^{\phi}$-mod
4000: which is the head of prestandard module $\glob(S_{\lambda})$
4001: (using the $\glob$ corresponding to (\ref{ebe})).
4002:
4003: The simple modules not constructed in this way
4004: are those $S$ obeying $eS=0$. Since $U_1eU_1 \sim U_1$
4005: the element $U_1$ acts as zero in such representations,
4006: and hence so do all the $U_i$s.
4007: Thus only 1 and (possibly) $f$ are (represented by) non-zero.
4008: Since $f$ is (pre) idempotent there are precisely two such simple
4009: modules in general, both one-dimensional, one where $f$ is zero
4010: (which module we will give the label $\lambda=-m$) and
4011: the other not (which module we will give the label $\lambda=m-1$).
4012: (NB
4013: the `bootstrap' case at $m=1$
4014: is the exception, since there, uniquely,
4015: $fef\sim f$ and $f \cong 0$ is also forced.)
4016: \Qed
4017:
4018: %}}}
4019: \subsection{Combinatorics of the basis $B^{\phi}_{2m}$}
4020: %{{{ ss combi
4021: %{{{ preamble
4022:
4023: \newcommand{\BD}{B}
4024:
4025: Note that if a diagram in $B^{\phi}_{2m}$
4026: has any propagating lines then it has at least
4027: two (by the symmetry), and that there is a unique mirror image pair
4028: that can be deformed to touch the 0-reflection line at some point
4029: --- the pair `closest' to the 0-reflection line.
4030: There is thus a unique region of the diagram that touches both
4031: elements of this pair {\em and} contains a segment of the 0-reflection line.
4032: This region can be black or white. We will call it the {\em inner}
4033: region
4034: (and call the pair of lines the inner lines).
4035:
4036: One way of organising diagrams into subsets is by the number of
4037: propagating lines. Within this, we may subdivide the set of those with
4038: $2l>0$ propagating lines into those in which the
4039: inner region is black or white.
4040: Let us denote these subsets $\BD^{\phi}_{2m}[\pm l]$ respectively (note
4041: that this works at $l=+0=-0$ since there is no inner region there).
4042:
4043: For example:
4044: \eql(bwex)
4045: \BD^{\phi}_{2m}[-m] = \{ 1 \} ; \qquad
4046: \BD^{\phi}_{2m}[-(m-1)] = \{ f \} ; \qquad
4047: \BD^{\phi}_{2m}[m-1] = \{ e \} .
4048: \eq
4049: Remark: It is necessary to have such a labelling scheme for these
4050: subsets, and this scheme will serve our purposes. However, it is not
4051: canonical.
4052:
4053: %}}}
4054: %{{{ ideals
4055:
4056: Following \cite{MartinSaleur94a} let us write $\hash(d)$ for the
4057: number of propagating lines in diagram $d$.
4058: Similarly we extend this to apply to any scalar multiple of $d$,
4059: so that $\hash(dd')$ is defined for any two diagrams $d,d'$.
4060: We write $c_i(d) \in \{ b,w \}$ for the inner region colour of $d$
4061: (and again similarly for $dd'$).
4062:
4063: \begin{lem} \label{filtre}
4064: For all $d,d'$ \\
4065: (1) We have $\hash(dd') \leq \hash(d)$.
4066: \\
4067: (2) If $\hash(dd') = \hash(d)$ then $c_i(dd')=c_i(d)$.
4068: \end{lem}
4069: {\em Proof:}
4070: (1) is straightforward.
4071: (2) Suppose that a certain pair of lines are inner in $d$.
4072: These lines are still identifiable beginning at the
4073: northern edge of $dd'$ (which is inherited from $d$).
4074: If they
4075: remain inner in the extension of $d$ by $d'$ then obviously the colour is
4076: unchanged. On the other hand if they are not inner in $dd'$
4077: then they are no longer propagating and $\hash(dd') < \hash(d)$.
4078: \Qed
4079:
4080: For $l \geq 0$ define
4081: \newcommand{\BBD}{BB^{\phi}_{2m}}
4082: \[
4083: \BBD(l) \; = \;
4084: \bigcup_{0 \leq j\leq l}
4085: \left( \BD^{\phi}_{2m}[j] \cup \BD^{\phi}_{2m}[-j] \right)
4086: \]
4087: and $\BD^{\phi}_{2m}(0)= \BD^{\phi}_{2m}[0]$
4088: and for $l \geq 1$
4089: \[
4090: \BD^{\phi}_{2m}(\pm l) \; = \;
4091: \BD^{\phi}_{2m}[ \pm l] \cup \BBD(l-1) .
4092: \]
4093:
4094: %}}}
4095: %{{{ prl ideallypoo
4096:
4097: \prl(ideallypoo)
4098: For $l \in \{ -m,-m+1,...,m-1 \}$ the set $\BD^{\phi}_{2m}(l)$
4099: is a basis for an ideal of $b^{\phi}_{2m}$, and the subset structure
4100: \[
4101: \UseComputerModernTips
4102: \xymatrix{
4103: & \BD^{\phi}_{2m}(m-1) \ar @{_(->}[dl] <1ex>
4104: & \BD^{\phi}_{2m}(m-2) \ar@{_{(}->}[l]
4105: \ar@{_(->}[ddl] <1ex>
4106: &&
4107: \\
4108: \BD^{\phi}_{2m}(-m) & & &
4109: \cdots & & \BD^{\phi}_{2m}(0) \ar@{_{(}->}[ul]
4110: \ar@{_{(}->}[dl] <1ex>
4111: \\
4112: & \BD^{\phi}_{2m}(-(m-1)) \ar@{_{(}->}[ul]
4113: & \BD^{\phi}_{2m}(-(m-2)) \ar@{_{(}->}[l]
4114: \ar@{_{(}->}[uul]
4115: &&
4116: }
4117: \]
4118: passes to a subideal structure (over any ring).
4119: \end{pr}
4120: {\em Proof:} This follows directly from Lemma~\ref{filtre}. \Qed
4121:
4122:
4123: %}}}
4124: %{{{ I ideal
4125:
4126: Let us name the subideals
4127: $
4128: I^{\phi}_{2m}(l) \; = \; k \BD^{\phi}_{2m}(l) .
4129: $
4130: Noting this structure,
4131: associate a partial order $\lhd$ to $\Lambda^{\phi}_{m}$ by
4132: $l \;\lhd\; l'$ if and only if $|l| < |l'|$
4133: (NB, this order is not total).
4134: We then define $I^{\phi}_{2m}[l]$ as
4135: the $l$-th section of this ideal structure, that is
4136: \[
4137: I^{\phi}_{2m}[ l] \;
4138: = \left\{ \begin{array}{ll}
4139: \; k \BD^{\phi}_{2m}(l) / k \BBD(-l-1) & l<0 \cr
4140: \; k \BD^{\phi}_{2m}(l) & l=0 \cr
4141: \; k \BD^{\phi}_{2m}(l) / k \BBD(l-1) & l>0 . \cr
4142: \end{array} \right.
4143: \]
4144: Note that $I^{\phi}_{2m}[l]$ has basis $\BD^{\phi}_{2m}[l]$
4145: (where the action is the algebra multiplication, but taking account of
4146: the quotient).
4147:
4148:
4149: %}}}
4150:
4151: Next we want to decompose these sections as left-modules,
4152: and equip their component modules with an inner product.
4153:
4154: %{{{ half-diagrams
4155: \pdef{Half-diagrams:}
4156: Note that it is always possible to cut a diagram
4157: $d \in B^{\phi}_{2m}$
4158: from the eastern to
4159: the western edge in such a way that only propagating lines are cut
4160: (and these exactly once each).
4161: Note, however, that this process is not always unique (even up to isotopy),
4162: since some diagrams
4163: with { no} propagating lines
4164: contain a noncontractible loop, which could lie
4165: above or below the cut.
4166: However, considering the set of {\em half-diagrams} produced in this
4167: way
4168: { ignoring} any noncontractible line,
4169: any top-bottom pair of half-diagrams
4170: with the same number of propagating lines, and inner region colour if defined,
4171: can always be recombined to produce a full CC diagram,
4172: and in exactly one way, with the caveat
4173: that the CC requirement will determine if a noncontractible loop must
4174: be inserted.
4175: We will call any such loop a {\em belt}.
4176:
4177: \newcommand{\ket}[1]{| #1 \rangle}%
4178: \newcommand{\bra}[1]{\langle #1 |}%
4179:
4180: Let us denote by $\ket{\BD^{\phi}_{2m}[l]}$ the set of upper half
4181: diagrams associated to $\BD^{\phi}_{2m}[l]$ (any $l$),
4182: and by $\ket{d}$ (the ``ket'')
4183: the upper half diagram obtained from diagram $d$
4184: (write $\bra{d}$ (the ``bra'') for the corresponding lower half diagram).
4185: There is an obvious isomorphism of the set
4186: $\ket{\BD^{\phi}_{2m}[l]}$ with the set of bottom
4187: halves $\bra{\BD^{\phi}_{2m}[l]}$, obtained by reflecting in an east-west
4188: line,
4189: that is, $\ket{a} \stackrel{\sim}{\mapsto} \bra{a^o}$.
4190: It follows that
4191: \begin{eqnarray}
4192: \BD^{\phi}_{2m}[l] & \cong &
4193: \ket{\BD^{\phi}_{2m}[l]} \times \bra{\BD^{\phi}_{2m}[l]}
4194: \\
4195: d & \mapsto & ( \; \ket{d} \; , \;\; \bra{d} \;)
4196: \end{eqnarray}
4197: where the map is the cut map.
4198: (Note that elements of $\ket{\BD^{\phi}_{2m}[+l]}$ and
4199: $\bra{\BD^{\phi}_{2m}[-l]}$
4200: can be concatenated with $l>0$, but they will not produce a CC
4201: diagram.)
4202: We may write the inverse map as a multiplication:
4203: \[
4204: ( \; \ket{d} \; , \;\; \bra{d} \;) \mapsto \ket{d} \bra{d} .
4205: \]
4206:
4207: %}}}
4208: %{{{ lemma
4209:
4210: \newcommand{\Belt}{{\mathrm{Belt}}}%
4211: For $\ket{a} \in \ket{\BD^{\phi}_{2m}[0]}$ write $\Belt_a$ for the
4212: subset of elements $\bra{b} \in \bra{\BD^{\phi}_{2m}[0]}$
4213: such that $\ket{a}\bra{b}$ has a belt.
4214: The partition of $\bra{\BD^{\phi}_{2m}[0]}$ into two parts defined by
4215: $\Belt_a$ (as one of the parts) is independent of $a$, and written
4216: simply as $\Belt$.
4217: Note that
4218: \begin{lem} \label{(x)-}
4219: If $d,d' \in \BD^{\phi}_{2m}[l]$ and
4220: $\hash(dd') = |l|$
4221: then
4222: \\
4223: (i) there is a monomial in the parameters
4224: $k_{dd'}$ such that
4225: \eql(dd')
4226: dd'
4227: = k_{dd'} \ket{d} \bra{d'} .
4228: \eq
4229: \\
4230: (ii) This $k_{dd'}$ depends on $ \bra{d}$ and $ \ket{d'}$ but
4231: does not depend on
4232: $ \bra{d'}$ and $ \ket{d}$, except in case $l=0$ through the
4233: noncontractible loop caveat.
4234: \\
4235: (iii) No top-bottom symmetric diagram $\ket{a} \bra{a^o}$ has a belt.
4236: If $l=0$ and $dd' = \ket{a} \bra{b} \ket{c} \bra{a^o}$
4237: write $ \bra{b} \ket{c}_a$ for $k_{dd'}$.
4238: Let $M_a$ be the matrix $(\bra{b} \ket{c}_a)_{b,c}$.
4239: Let $M_a'$ be
4240: the matrix obtained from $M_a$ by dividing every row with $b \in
4241: \Belt_a$ by $\kappa_{LR}$;
4242: and let $M_a''$ be
4243: the matrix obtained from $M_a$ by dividing every column with $c \in
4244: \Belt_a$ by $\kappa_{LR}$.
4245: If $a$, $a'$ are in the same part of $\Belt$, then $M_a=M_{a'}$.
4246: If $a$, $a'$ are in different parts of $\Belt$, then $M_a'=M_{a'}''$.
4247: \\
4248: (iv) Now let $d' \in B_{2m}^\phi[l]$ and $d'' \in B_{2m}^\phi[l']$ be such
4249: that $\#(d'' d') = |l|$
4250: ($d''$ could have more than $|l|$ propagating lines).
4251: Then
4252: $$
4253: d'' d' = \sum_{d \in |B_{2m}^\phi[l]\rangle} k'_d |d \rangle \langle d'|
4254: ,$$
4255: where the $k'_d$ depend on $d''$ and $\langle d'|$, but not on
4256: $|d' \rangle$.
4257: \end{lem}
4258: {\em Proof:} (i) The product
4259: $dd'$ is some scalar times a diagram. Under the given conditions it is
4260: clear that this diagram must have the given ket-bra form.
4261: \\
4262: (ii) This follows from the definition of the cut map.
4263: \\
4264: (iii) The first part is obvious. If $a,a'$ are in the same part of
4265: $\Belt$, then the calculations for each $ \bra{b} \ket{c}_a$ and
4266: $ \bra{b} \ket{c}_{a'}$ are identical.
4267: Otherwise there are four types of case in comparing $ \bra{b} \ket{c}_a$ and
4268: $ \bra{b} \ket{c}_{a'}$:
4269: \\
4270: 1) if $\{b,c\} \cap \Belt_a = \{b,c\}$ then the matrix elements differ
4271: precisely by a factor $\kappa_{LR}$ on the $a$ side, which is adjusted by the
4272: division on the $a$ side;
4273: \\
4274: 2) if $\{b,c\} \cap \Belt_a = \{b\}$ then both sides have at least one
4275: factor $\kappa_{LR}$, and are the same, and both have this factor
4276: divided out;
4277: \\
4278: 3) if $\{b,c\} \cap \Belt_a = \{c\}$ similarly;
4279: \\
4280: 4) if $\{b,c\} \cap \Belt_a = \emptyset$ then the matrix elements differ
4281: precisely by a factor $\kappa_{LR}$ on the $a'$ side, which is adjusted by the
4282: division on the $a'$ side.
4283: \\
4284: (iv) As for (i).
4285: (NB,
4286: there can only be one term in the sum, because the diagram
4287: basis is a monomial basis, i.e. the product of any two diagrams
4288: is a scalar multiple of another.)
4289: \Qed
4290: \\
4291: Indeed, since
4292: \eql(dd'xx)
4293: dd' = \ket{d}\bra{d} \; \ket{d'}\bra{d'}
4294: = k_{dd'} \ket{d} \bra{d'}
4295: \eq
4296: we will sometimes write $\bra{d} \ket{d'}$ for $k_{dd'} $
4297: when no ambiguity arises.
4298:
4299:
4300: %}}}
4301: %{{{ structure prop
4302:
4303: \prl(struc1)
4304: Let $d,d' \in \BD^{\phi}_{2m}[l]$.
4305: \\
4306: (i)
4307: There exist diagrams $a,b$, and a
4308: nonzero monomial in the
4309: parameters $k$, such that
4310: $adb=kd'$.
4311: That is, provided all the parameters are units then every diagram in
4312: $\BD^{\phi}_{2m}[l]$
4313: generates
4314: $ \BD^{\phi}_{2m}(l)$.
4315: \\
4316: (ii) For $a$ any diagram, if $\hash(dad')=l$ then
4317: $dad'=k_a \ket{d}\bra{d'}$ where $k_a$ is a nonzero monomial in the
4318: parameters.
4319: \end{pr}
4320: {\em Proof:} (i) Note that if $d^o$ is the `opposite' diagram of $d$
4321: (the same diagram
4322: drawn upside-down) then $\hash(d^o d)=\hash(d d^o)=l$.
4323: Consider $a = \ket{d'}\bra{d^o}$ and $b=\ket{d^o}\bra{d'}$, then
4324: \[
4325: adb = \ket{d'}\bra{d^o} \ket{d}\bra{d} \ket{d^o}\bra{d'}
4326: = k \ket{d'}\bra{d'} =kd'
4327: \]
4328: For the second part of (i) it is now enough to show that we can get from some
4329: diagram in $\BD^{\phi}_{2m}[l]$ to (some appropriate scalar multiple of)
4330: some diagram in $\BD^{\phi}_{2m}[\pm(l-1)]$.
4331: This is routine for a suitable choice of diagram in each case.
4332: For example, for $2j \leq m-2$ then $w=e U_2 U_4 \ldots U_{2j} f
4333: \in \BD^{\phi}_{2m}[m-2j-2]$; while for $2j < m-2$ then
4334: $U_{m-1} w U_{m-1} = \kappa_R w'$ with
4335: $w' = e U_2 U_4 \ldots U_{2j} U_{m-1} \in \BD^{\phi}_{2m}[m-2j-3]$.
4336: \\
4337: (ii) Similarly we have:
4338: \[
4339: dad' = \ket{d}\bra{d} \ket{a}\bra{a} \ket{d'}\bra{d'}
4340: = ( \bra{d} \ket{a} \; \bra{a} \ket{d'} ) \; \ket{d}\bra{d'}
4341: \]
4342: and $ \bra{d} \ket{a}$ and $\bra{a} \ket{d'}$ are
4343: nonzero by construction.
4344: \Qed
4345:
4346: An immediate corollary to \ref{struc1}(i) is
4347: {\co{ \label{imco}
4348: Subject to the same parameter restriction as in \ref{struc1}(i), no
4349: unit multiple of any diagram is in the radical.
4350: \Qed
4351: }}
4352:
4353: %}}}
4354: %{{{ parker woz ere
4355: %{{{ preamble
4356:
4357:
4358:
4359: \newcommand{\rad}{\mbox{rad}}
4360: %}}}
4361:
4362: For $i=0,1,2,...,m-1$ define $S_i = I_{2m}(-i)$ and
4363: $T_i = I_{2m}(i) + I_{2m}(-i)$.
4364: Note by Proposition~\ref{ideallypoo}
4365: that the following is a chain of ideals in $b_{2m}^\phi$
4366: \eql(hch)
4367: S_0
4368: \subseteq
4369: S_1
4370: \subseteq
4371: T_1
4372: \subseteq
4373: S_2
4374: \subseteq
4375: T_2
4376: \subseteq
4377: \cdots \subseteq
4378: T_{m-1}
4379: \subseteq
4380: S_{m} = I_{2m}(-m) = b_{2m}^\phi .
4381: \eq
4382:
4383: %}}}
4384: %{{{ some modules
4385:
4386: For each $l \in \Lambda^{\phi}_m$ and $d \in \BD_{2m}^\phi[l]$
4387: define $b_{2m}^\phi$ modules by
4388: $\mathcal{S}^d_{2m}(0) := b_{2m}^\phi d $
4389: if $l=0$ and otherwise
4390: \eql(standby)
4391: \mathcal{S}^d_{2m}(l) := \frac{b_{2m}^\phi d \; + \; S}{S}
4392: \eq
4393: where
4394: $S=T_{|l|-1} $.
4395:
4396: Note that right multiplication by $d'$ gives a map
4397: $\gamma_{d'}: \mathcal{S}^d_{2m}(l) \mapsto \mathcal{S}^{d'}_{2m}(l)$.
4398: Thus by Proposition~\ref{struc1}(i):
4399: \begin{lem}
4400: If $\delta, \delta_L, \delta_R, \kappa_L, \kappa_R, \kappa_{LR} $
4401: are units
4402: then
4403: the precise choice of $d$ is irrelevant,
4404: up to isomorphism, in $\mathcal{S}^d_{2m}(l)$.
4405: In this case define
4406: $ \mathcal{S}^{}_{2m}(l) = \mathcal{S}^{d}_{2m}(l)$.
4407: \end{lem}
4408:
4409: Note that $d \in \BD_{2m}^\phi[l]$ can always be chosen to have no
4410: non-contractible lines. In this case
4411: the module
4412: $ \mathcal{S}_{2m}(l) $
4413: has basis
4414: $b_{2m}^\phi d \cap \BD_{2m}^{\phi}[l] $,
4415: where the action is the algebra multiplication, but taking account
4416: of the quotient. Further
4417: $b_{2m}^\phi d \cap \BD_{2m}^{\phi}[l] \cong \ket{\BD^{\phi}_{2m}[l]}$.
4418:
4419: %}}}
4420: %{{{ inner prod
4421:
4422: We may extend the notation $ \bra{d} \ket{d'}$
4423: to a map
4424: $( \bra{ \BD^{\phi}_{2m}[l] } , \ket{ \BD^{\phi}_{2m}[l] } ) \rightarrow K$
4425: by
4426: $ \bra{d} \ket{d'}=0$ if $\#(dd')<l$.
4427: Via the involution $\ket{} \stackrel{\sim}{\rightarrow} \bra{}$ the map
4428: $\bra{-} \ket{-}$ extends (bi)linearly to an inner product on
4429: $\mathcal{S}^d_{2m}(l)$.
4430:
4431: %}}}
4432: %{{{ inner product
4433:
4434: \newcommand{\brac}{\langle}
4435: \newcommand{\cket}{\rangle}
4436:
4437: Define
4438: \eql(innuit)
4439: \Deltag^{d''}_{2m}(l) = \mathrm{det}\left( (\langle d | | d'
4440: \rangle)_{n\times n}\right)
4441: \eq
4442: where $d^o,d' \in b^{\phi}_{2m} d ''\cap B^{\phi}_{2m}[l]$,
4443: for a fixed $d'' \in
4444: B^{\phi}_{2m}$ (the basis of $\sS_{2m}(l)$)
4445: and $n = \dim \sS_{2m}(l)$.
4446: Note
4447: from Lemma~\ref{(x)-}
4448: that $\Deltag^{d''}_{2m}(l)$ does not depend on $d''$ if
4449: $l \neq 0$;
4450: and depends on $d''$ only through at most an overall factor of
4451: a power of $\kappa_{LR}$
4452: if $l=0$.
4453: Choosing the lowest power in this overall factor, we will write
4454: simply $\Deltag^{}_{2m}(l)$ in all cases.
4455:
4456: %}}}
4457: %{{{ prl(gen semisimp)
4458:
4459: \newcommand{\lambdal}{l}
4460:
4461: We will address the specific computation of the Gram
4462: determinant later, but note that the
4463: matrix entries
4464: are monomial in the parameters.
4465: We have
4466: \prl(gen semisimp)
4467: For each $\lambdal \in \Lambda^{\phi}_{m}$ there is a
4468: polynomial $P$ in the parameters such that
4469: the prestandard module of $b_{2m}^{\phi}$ associated to $\lambdal$
4470: has Gram determinant given by evaluation of $P$ at the
4471: appropriate specialisation.
4472: Every prestandard is generically simple.
4473: \end{pr}
4474: {\em Proof:}
4475: By computing $\Deltag^{}_{2m}(\lambdal)$
4476: with the parameters treated as
4477: indeterminates we obtain the polynomial $P$.
4478: By Proposition~\ref{orestes} the inner product we have
4479: defined is unique up to scalars,
4480: and the Gram determinant is nonvanishing
4481: in any given parameter choice if and only if the module is
4482: simple there.
4483: `Generically' has the meaning of Zariski open here, so
4484: it is only necessary to show that no such polynomial $P$ is
4485: identically the zero polynomial.
4486: This can be done by considering asymptotic cases
4487: of the parameters.
4488: (The power of $\delta$, say, is maximal on the
4489: diagonal for all rows of the matrix, and uniquely maximal there for
4490: some rows, so the determinant is not zero.)
4491: \Qed
4492:
4493: %}}}
4494: %{{{ cellular
4495:
4496: \begin{theo}
4497: Let $*$ be the $\Ring$-linear involutory antiautomorphism
4498: on $b^{\phi}_{2m}$ defined by flipping each diagram upside-down,
4499: i.e. by reflection in a horizontal line.
4500: For each $l \in \Lambda^{\phi}_{m}$ let $M(l)$ be one of the
4501: diagram bases of $ \mathcal{S}^{}_{2m}(l) $;
4502: and let $C$ be the ket-bra combination of basis elements.
4503: Then
4504: $b^{\phi}_{2m}$ is cellular over $\Ring$ with cell datum
4505: $( \Lambda^{\phi}_{m},M,C,*)$.
4506: \end{theo}
4507: {\em Proof:}
4508: Note that
4509: $*$ is an algebra antiautomorphism
4510: by top-bottom symmetry of the reduction rules.
4511: Proposition~\ref{ideallypoo}, Lemma~\ref{(x)-} and
4512: (\ref{standby}) verify the axioms given
4513: in \cite{GrahamLehrer96}.
4514: \Qed
4515:
4516: %}}}
4517: %{{{ parker II
4518:
4519: \begin{theo}
4520: If
4521: $\delta, \delta_L, \delta_R, \kappa_L, \kappa_R, \kappa_{LR} $
4522: are units
4523: then
4524: $b_{2m}^{\phi}$ is quasihereditary,
4525: and the chain (\ref{hch}) is a heredity chain.
4526: \end{theo}
4527: {\em Proof:}
4528: It is enough to show that the chain is heredity
4529: (one might also see \cite{KoenigXi99b}).
4530: We need to show for $S= S_i$ or $S= T_i$
4531: where
4532: $A=b_{2m}^\phi/T_{i-1}$ if $S=S_i$ and
4533: $A=b_{2m}^\phi/S_{i}$ if $S=T_i$:
4534: \\ \noindent (1) that $S^2 = S$
4535: \\ \noindent (2) that $S J S = 0$ for $J = \rad(A)$
4536: \\ \noindent (3) that each section defined by the proposed hereditary chain
4537: of ideals is projective in the quotient.
4538: I.e. that $S/T_{i-1}$ (resp. $S/S_i$ is projective in $A$).
4539:
4540: %{{{ ignore
4541: %}}}
4542:
4543: Suppose $S=S_i$ for some $i$,
4544: so $S_i$ contains all diagrams with $i$ propagating lines and white
4545: inner region and all diagrams with fewer than $i$ propagating lines.
4546:
4547: Take $d \in \BD_{2m}[-i]$. Then $\# (d d^{\circ}) =i$.
4548: Thus $S_i^2$
4549: contains a diagram with white inner region and $i$ propagating
4550: lines. Proposition~\ref{struc1}(i) then says that $S_i \subset S_i^2$ and so
4551: $S_i^2 = S_i$.
4552:
4553: We now show that $SJS =0$ where $J=\rad b^{\phi}_{2m}/T_{i-1} $.
4554: Consider $djd'$, with $d$, $d' \in
4555: \BD^{\phi}_{2m}(-i)$ and $j \in J$.
4556: Now write $j = ( \sum m_{\alpha} j_{\alpha}) +T_{i-1}$
4557: where $m_\alpha$ are in the
4558: base ring $K$ and $j_{\alpha} \in B_{2m}^{\phi}$.
4559: If $\# (d j_{\alpha} d')\le i-1$ then
4560: $ d j_{\alpha} d' $
4561: is zero in the quotient
4562: $b_{2m}^{\phi} /T_{i-1}$.
4563:
4564: If $\# (d j_{\alpha} d')= i$ then
4565: Proposition~\ref{struc1}(ii) says that
4566: $d j_{\alpha} d' = k_{\alpha} dd'$ for some
4567: $k_{\alpha}$ in the ring and so
4568: $d j d' = \left(\sum m_{\alpha} k_{\alpha}\right) dd' +T_{i-1}$
4569: (where the sum is over those $j_{\alpha}$ such that
4570: $\# dj_{\alpha}d' = i$).
4571:
4572: If $\# (d d') <i$ then $djd'=0$ in the quotient.
4573: If $\# (d d') =i$ then $(dd')^r = k^r dd' \not \in T_{i-1}$
4574: for some $k$ a non-zero
4575: monomial in the parameters and for all $r$ and so $dd' + T_{i-1}
4576: \not\in J$.
4577: Thus for $djd'$ to be nilpotent in the quotient we need
4578: $\left(\sum m_{\alpha} k_{\alpha}\right) dd'$ to be zero and thus
4579: $d j d' =0$.
4580: Thus $SJS=0$.
4581:
4582: Finally we need $S_i/T_{i-1}$ to be projective as a
4583: $b_{2m}^{\phi}/T_{i-1}$-module.
4584:
4585:
4586: The section $S_i/T_{i-1}$ splits up into a direct sum of
4587: modules
4588: $b_{2m}^{\phi} d /T_{i-1}$ for $d \in \BD_{2m}^\phi[-i]$.
4589: (Note that $b_{2m}^{\phi} d /T_{i-1} = b_{2m}^{\phi} d' /T_{i-1} $
4590: if and only if $\bra{d}=\bra{d'}$.)
4591: The action of the algebra on the {\em right} gives maps between these
4592: modules for different choices of $d$. These maps are invertible by
4593: Proposition~\ref{struc1}, so the summands are isomorphic.
4594: Note that there exists at least one $d$ in each
4595: $\BD_{2m}^\phi[-i]$ that is idempotent up to a unit,
4596: thus each summand is projective.
4597:
4598: The argument for $S=T_i$ is very similar.
4599: \Qed
4600:
4601: %}}}
4602: %{{{ corollary
4603: \begin{co}
4604: If
4605: $\delta, \delta_L, \delta_R, \kappa_L, \kappa_R, \kappa_{LR} $
4606: are units
4607: then the modules $ \mathcal{S}_{2m}(l) $
4608: are the standard modules of $b_{2m}^\phi$.
4609: \end{co}
4610:
4611: %}}}
4612: %{{{ aep later pt 1
4613:
4614: By Proposition~\ref{rho hom} the globalisation functor $G$ is
4615: $$G: b_{2m-2}^{\phi}
4616: (\delta, \delta_L,\delta_R, \kappa_L, \kappa_R, \kappa_{LR})
4617: \to b_{2m}^{\phi}
4618: (\delta, \kappa_L,\delta_R, \delta_L, \kappa_R, \kappa_{LR})
4619: $$
4620: We may `dually' define another globalisation functor using the right
4621: hand blob
4622: (in Proposition~\ref{rho hom})
4623: rather than the left hand one. We get a functor
4624: $$G': b_{2m-2}^{\phi}
4625: (\delta, \delta_L,\delta_R, \kappa_L, \kappa_R, \kappa_{LR})
4626: \to b_{2m}^{\phi}
4627: (\delta, \delta_L,\kappa_R, \kappa_L, \delta_R, \kappa_{LR}).
4628: $$
4629:
4630: It is clear
4631: that $G \circ G' = G' \circ G$.
4632: We thus get three functors from
4633: $b_{2m-4}^{\phi}$ to
4634: $b_{2m}^{\phi}$
4635: (as meta-categories, i.e ignoring the swapping of parameters).
4636:
4637: \prl(old 824)
4638: If $\delta, \delta_L, \delta_R, \kappa_L, \kappa_R, \kappa_{LR} $
4639: are units
4640: then
4641: $$ G(\mathcal{S}_{2m-2} (l)) \cong \mathcal{S}_{2m} (-l)$$
4642: $$ G'(\mathcal{S}_{2m-2} (l)) \cong \mathcal{S}_{2m} (l)$$
4643: \end{pr}
4644: {\em Proof:}
4645: Since $b_{2m}^\phi$ is quasi-hereditary under the assumption on the
4646: parameters, globalising takes standard modules to standard ones
4647: \cite[Proposition 4]{MartinRyom02}),
4648: and we
4649: need only determine which one. Globalising does not change the number
4650: of propagating lines. The colour of the inner region changes for
4651: $G$ but stays the same for $G'$, hence the change in sign for $G$ but
4652: not for $G'$.
4653: \Qed
4654:
4655: Note that for these algebras
4656: $b^{\phi}_{2m}$
4657: we have shown that there is a set of
4658: prestandard modules which are in fact standard.
4659:
4660: %}}}
4661:
4662: %}}}
4663: \subsection{Prestandard modules by $G_e$: Low rank examples}
4664: %{{{ low rank
4665: %{{{ begin recursion
4666: \newcommand{\Si}{S}%% SIMPLE MODULE BODY
4667: \newcommand{\Sit}[2]{S_{#2}(#1)}%% SIMPLE MODULE BODY
4668:
4669: Let us begin the recursion,
4670: implicit in the proof of Theorem~\ref{index wangy},
4671: to construct prestandard modules using $G$:
4672: \newline
4673: We can apply $G$ equally to simple modules directly, or to the regular
4674: representation (and hence to diagrams), since the regular
4675: representation is a direct sum of projective representations, each
4676: with an appropriate unique simple module in its head.
4677: Firstly,
4678: $b^{\phi}_0$ is spanned by the empty diagram, which is thus also a
4679: basis for the unique simple left module, $\Sit{0}{0}$.
4680: Applying $G$ to this will give $\Sit{0}{2}$.
4681: Now by virtue of Proposition~\ref{basis thang}
4682: the image of this under $\glob$
4683: is given by the image under $\rho'$,
4684: which is
4685: \[
4686: \raisebox{-.542cm}{\includegraphics{xfig/eTLB10.eps}}
4687: \qquad \in b^{\phi}_2
4688: \]
4689: In this case the multiplication map is a bijection, and the above element
4690: generates the left prestandard module $\Sit{0}{2}$ with basis
4691: consisting of this and one other diagram:
4692: \eql(bas1)
4693: \includegraphics{xfig/eTLB11.eps}
4694: \eq
4695: (note that this set is spanning by the $\kappa'$ relation).
4696: Observe that this basis has no intrinsic dependence on the parameters.
4697:
4698: %}}}
4699: %{{{ back to m=1
4700:
4701: Consider the module morphism between ideals given by $m \mapsto mf$ here.
4702: We are considering the generic case
4703: (so that $\kappa'$ is invertible) so
4704: \eql(bas2)
4705: \includegraphics{xfig/eTLB12.eps}
4706: \eq
4707: span the isomorphic image module to that above
4708: (we will touch on the chiral case $RLR \not\propto R$ elsewhere).
4709: Note that the
4710: elements in (\ref{bas1}),(\ref{bas2}) span $ b^{\phi}_2 e b^{\phi}_2$.
4711: By proposition~\ref{grebo}
4712: the simple module missing from this construction ($\Sit{-1}{2}$)
4713: may be constructed as $b^{\phi}_2 / b^{\phi}_2 e b^{\phi}_2$.
4714: Thus $\Sit{-1}{2}$ has basis
4715: \eql(bas3)
4716: \includegraphics{xfig/eTLB101.eps}
4717: \eq
4718: where the action of the algebra is algebra multiplication modulo the
4719: elements in (\ref{bas1}),(\ref{bas2}).
4720: Note that $| B^{\phi 0}_2| = 5$ so this is a complete decomposition of
4721: the regular module.
4722:
4723: %}}}
4724: %{{{ again
4725:
4726: Applying $\glob$ again we determine the structure of $b^{\phi}_4$.
4727: The image of $b^{\phi}_2$ is as follows.
4728: The image of the basis elements for $\Sit{0}{2}$ in (\ref{bas1}) is the
4729: first two elements of:
4730: \eql(bas4)
4731: \includegraphics{xfig/eTLB141.eps}
4732: \eq
4733: (the other two are generated from these by the algebra action);
4734: thus these objects are a basis for $\Sit{0}{4}$.
4735: Another basis for this module is obtained as the image of the elements
4736: in (\ref{bas2}) (and two further elements generated from these):
4737: \[
4738: \includegraphics{xfig/eTLB142.eps}
4739: \]
4740: It is left as an exercise to write down two more sets of four elements
4741: giving bases for isomorphic modules.
4742: The object in (\ref{bas3}) is not strictly an element of the algebra
4743: (because of the quotient).
4744: Thus we cannot apply $\rho'$ to it directly.
4745: However we can consider $\Sit{-1}{2}$ as
4746: $b^{\phi}_2 / b^{\phi}_2 e b^{\phi}_2$ and apply $\rho'$ to
4747: $b^{\phi}_2$.
4748: Applying $\rho'$ to (\ref{bas3})
4749: and to the quotienting module spanned by (\ref{bas1}),(\ref{bas2})
4750: we have a basis for $\Sit{-1}{4}$:
4751: \[
4752: \includegraphics{xfig/eTLB143.eps}
4753: \]
4754:
4755: The two missing simple modules are in
4756: $b^{\phi}_4 / b^{\phi}_4 e b^{\phi}_4$.
4757: Discarding all the diagrams in $ b^{\phi}_4 e b^{\phi}_4$
4758: (constructed above) these are
4759: given by
4760: \[ \Sit{-2}{4} = k \; %%\sim
4761: \raisebox{-.542cm}{\includegraphics{xfig/eTLB144.eps}}
4762: \]
4763: and
4764: \[ \Sit{+1}{4} = k \; %%\sim
4765: \raisebox{-.542cm}{\includegraphics{xfig/eTLB14-4.eps}}
4766: \]
4767: This completes the arrangement of the basis elements for the
4768: left regular representation.
4769: We have total rank $4^2 +1+1+1$.
4770:
4771: The $\rho'$-image of the basis elements for $\Sit{0}{4}$ in (\ref{bas4}) is the
4772: first four elements of:
4773: \eql(bas44)
4774: \includegraphics{xfig/eTLB170.eps}
4775: \eq
4776: (the other four are generated from these by the algebra action);
4777: thus these objects are a basis for $\Sit{0}{6}$.
4778:
4779: %}}}
4780: %{{{ The general pattern
4781:
4782: The general pattern
4783: can now be given.
4784:
4785: %}}}
4786: %}}}
4787: \subsection{Top-half combinatorics}
4788: %{{{ top-half combinatorics 0 prop
4789:
4790: Suppose that we view the right half of the fundamental region of a
4791: diagram,
4792: and concentrate for the moment on the $m$ vertices in the northern
4793: edge in this interval
4794: --- as it were, the upper right-hand {\em quarter} of the diagram.
4795: In basis elements with no propagating lines, the line from each of
4796: these vertices descends (initially) and turns either to Left or to
4797: Right. Considering each such line in turn, starting from the left,
4798: say, we may construct (the upper half of) a diagram by choosing the
4799: direction of these lines. Each direction may be chosen freely, in turn,
4800: irrespective of the direction of previously chosen lines: we can
4801: always choose Right since the direction of lines {\em to the right}
4802: have yet to be chosen; we can always choose Left since either there is
4803: a path to the left edge (the existing choices make Right-Left pairs,
4804: i.e. cups, possibly nested, possibly together with some additional
4805: Left directed lines); or there is a preceding Right not in a
4806: Right-Left pair, which can then form a Right-Left pair with this new
4807: Left.
4808: Some examples of these quarter diagrams
4809: are as follows (the shorthand on the left in each
4810: example is L for line-turn-left; R for line-turn-right):
4811: \eql(table key 0)
4812: \includegraphics{xfig/LLRL.eps}
4813: \eq
4814: In consequence the rank of the prestandard module $\Sit{0}{2m}$ is
4815: \[
4816: | \ket{\BD^{\phi}_{2m}[0] } | \; = \; 2^{m}
4817: \]
4818: This is because the composites of these elements with any element of
4819: $\bra{\BD^{\phi}_{2m}[0]}$ produces a basis for $\Sit{0}{2m}$.
4820:
4821: %}}}
4822: %{{{ the fig
4823:
4824: Altogether the tower of bases starts as shown in figure~\ref{notbrat},
4825: where $o$ denotes a propagating line.
4826: The key for the shorthand used in this table is indicated in
4827: (\ref{table key 0}) above, and by
4828: \eql(table key)
4829: \includegraphics{xfig/LLLLRL.eps}
4830: \eq
4831: \newcommand{\vect}[1]%
4832: {\! \left\{ \begin{array}{c} #1 \end{array} \right\} \!}
4833: \newcommand{\psign}{}
4834: \newcommand{\msign}{-}
4835: \begin{figure}
4836: \begin{turn}{90} \vbox{
4837: $$ \begin{array}{c|ccccccccc}
4838: m_= &l=\psign 3 &\psign 2 &\psign 1 & {0} & \msign 1 &\msign 2 &\msign
4839: 3
4840: &\msign 4\\
4841: \hline
4842: 0&&&&\vect{\emptyset} \\
4843: 1&&&&\vect{L\\ R}& \vect{o} \\
4844: 2&&&\vect{oR}&\vect{LL\\ LR\\ RL\\ RR}&\vect{Lo}&\vect{oo} \\
4845: 3&&\vect{ooR}&\vect{LoR}&
4846: \vect{LLL\\ LLR\\ LRL\\ LRR\\ RLL\\ RLR\\ RRL\\ RRR}&
4847: \vect{LLo\\ RLo\\ oRL\\ oRR}&\vect{Loo}&\vect{ooo} \\
4848: 4&\vect{oooR}&\vect{LooR}&
4849: \vect{oRRR\\ oRLR\\ oRRL\\ RLoR\\ LLoR} &
4850: \vect{LLLL\\ LLLR\\ LLRL\\ LLRR\\ LRLL\\ \ldots\\ RRRR} &
4851: \vect{LLLo\\ LRLo\\ RLLo\\ LoRL\\ LoRR} &
4852: \vect{LLoo\\ RLoo\\ oRLo\\ ooRL\\ ooRR} &
4853: \vect{Looo} &
4854: \vect{oooo}
4855: \end{array} $$}
4856: \end{turn}
4857: \caption{\label{notbrat} Table of standard $\Sit{l}{2m}$ bases
4858: up to $m=4$.
4859: NB, this $LR$ shorthand should NOT be confused with the nonabelian
4860: ring elements which live on strings
4861: --- see main text for key.}
4862: \end{figure}
4863: %}}}
4864: %{{{ ur_1
4865: \newcommand{\ur}{\mbox{ur}_1}%
4866: \newcommand{\uro}{\mbox{ur}_0}%
4867:
4868: Let us write
4869: $\ur(d)$ for the number of lines passing out of the fundamental region
4870: of a (half-)diagram through the 1-wall on the right,
4871: not counting those lines which also pass out on the left
4872: (`equatorial' lines, as it were).
4873: Thus for example the 4-th diagram in (\ref{bas44}) has
4874: $\ur(d) = 2$.
4875: (We will also later use $\uro(d)$ for the number of lines crossing the
4876: 0-wall of a (half-)diagram:
4877: the left hand edge, then, of a quarter diagram.)
4878:
4879: Note that the set of half-diagrams with given $m$ and $l=+x$
4880: ($x >0$) is that set with $x$ propagating lines and
4881: $\ur(d) \equiv 1$ mod~2,
4882: while the set of half-diagrams with given $m$ and $l=-x$
4883: ($x >0$) is that set with $x$ propagating lines and
4884: $\ur(d) \equiv 0$ mod~2.
4885:
4886: %}}}
4887: \subsection{Restriction of prestandards to blob algebra standards}\label{wa}
4888: %{{{ restriction
4889: \newcommand{\Deltab}[2]{\Delta^{b}_{#2}(#1)}
4890: The representation of $b_3$ induced on the basis for $\Sit{0}{6}$ in
4891: (\ref{bas44}) is as follows (we use the isomorphism with $b_6'$
4892: --- and by virtue of proposition~\ref{blob iso} we use $b_m$ and $b_{2m}'$
4893: interchangeably in this section):
4894: %{{{ example
4895: \[
4896: R_0( \; \raisebox{-.1542cm}{\includegraphics{xfig/e_in_6.eps}} \; )
4897: =\mat{llllllll}
4898: \delta_L \\
4899: & \delta_L \\
4900: & & \delta_L \\
4901: & & & \delta_L \\
4902: 1 &&&& 0 \\
4903: &1&&& &0 \\
4904: &&& \kappa &&& 0 \\
4905: && \kappa &&&&& 0 \tam
4906: \]
4907: \[
4908: R_0( \; \raisebox{-.1542cm}{\includegraphics{xfig/U1_in_6.eps}} \; )
4909: =\mat{llllllll}
4910: 0 &&&& \kappa_L \\
4911: & 0 &&&& \kappa_L \\
4912: & & 0 && 1\\
4913: & & & 0 && 1\\
4914: &&&& \delta \\
4915: &&&& &\delta \\
4916: &&& &&\kappa_R& 0 \\
4917: && &&&1&& 0 \tam
4918: \]
4919: \[
4920: R_0( \; \raisebox{-.1542cm}{\includegraphics{xfig/U2_in_6.eps}} \; )
4921: =\mat{llllllll}
4922: 0 && \delta_L \\
4923: & 0 & \kappa \\
4924: & & \delta && \\
4925: & & \delta_R& 0 && \\
4926: &&1&& 0 \\
4927: &&&& &0&&1 \\
4928: &&& &&& 0 & \delta_R \\
4929: && &&&&& \delta \tam
4930: \]
4931: (all unmarked entries zero).
4932: %}}}
4933: Note that basis elements in positions 1,3,5 span a $b_3$-submodule
4934: isomorphic to $\Deltab{1}{3}$
4935: (where $\Deltab{l}{m} = \Delta_{m}(l)$
4936: of \cite{MartinWoodcock2000}).
4937: The quotient by this has
4938: $b_3$-submodule spanned by elements 2,6,8,
4939: isomorphic to $\Deltab{-1}{3}$. The quotient by this has
4940: $b_3$-submodule spanned by element 4,
4941: isomorphic to $\Deltab{-3}{3}$.
4942: The remaining quotient is isomorphic to $\Deltab{3}{3}$.
4943:
4944: %}}}
4945: %{{{ general
4946:
4947: We can see this decomposition directly by looking at (\ref{bas44}).
4948: Note for general $\Sit{l}{2m}$ that the number
4949: $\ur(d)$
4950: cannot be increased by acting on a half-diagram
4951: by an element of $b_m$ (i.e. of $b_{2m}'$). Thus
4952: \prl(ur1)
4953: The restriction $\Res(b_{2m}',b_{2m}^{\phi},) \Sit{l}{2m}$
4954: has submodule structure filtered
4955: by $\ur$. In particular
4956: $\Res(b_{2m}',b_{2m}^{\phi},) \Sit{0}{2m}$
4957: has $m+1$ sections, since all the $\ur$ values are realised from 0 to
4958: $m$.
4959: Meanwhile $\Res(b_{2m}',b_{2m}^{\phi},) \Sit{l}{2m}$
4960: with $l=\pm x$ ($x>0$)
4961: has $\ur$ values from, and hence
4962: sections indexed by: \\
4963: $\{ m-x-1, m-x-3, \ldots, 0 \}$ if $m-x$ odd and $l<0$; \\
4964: $\{ m-x, m-x-2, \ldots, 1 \}$ if $m-x$ odd and $l>0$; \\
4965: $\{ m-x-1, m-x-3, \ldots, 1 \}$ if $m-x$ even and $l>0$; \\
4966: $\{ m-x, m-x-2, \ldots, 0 \}$ if $m-x$ even and $l<0$.
4967: \end{pr}
4968: {\em Proof:} Only the $l \neq 0$ cases require further explanation.
4969: Here, since there are propagating lines, there can be no equatorial line
4970: (passing from one reflection wall to the other), so the
4971: lines contributing to $\ur$ all start from the northern edge of the diagram.
4972: Consider the lines passing out of the $m$ vertices on the northern
4973: edge of the diagram.
4974: The number of propagating lines is fixed at $x$, and the total number
4975: $\ur + \uro$ of lines passing to the reflection walls
4976: is of definite parity, since all other lines return to the north edge
4977: and hence contribute to $m$ in pairs. But the parity of $\uro$ is
4978: fixed by the sign of $l$, so the parity of $\ur$ is also fixed.
4979: It is routine to check the extremal numbers.
4980: \Qed
4981:
4982: %}}}
4983: %{{{ prop
4984:
4985: Consider the $r$-th $\ur$-section of
4986: $\Res(b_{2m}',b_{2m}^{\phi},) \Sit{l}{2m}$ ($l=\pm x$ ($x>0$)).
4987: As already noted,
4988: $\uro$ is of definite parity in this section.
4989: If $\uro$ is even,
4990: there is an injective map from the basis elements in this section
4991: into
4992: the basis elements of a $b_{2m}'$ standard module
4993: $\Deltab{(x+r)}{}$,
4994: obtained by
4995: deforming the ends of the $r$ lines that pass out through the 1-wall
4996: until they pass out through the bottom of the diagram (i.e. become
4997: propagating lines).
4998: It is easy to see that this extends to a module morphism.
4999: Since the section is a blob module the map must be
5000: onto and hence a bijection.
5001: There is a similar morphism for $\uro$ odd.
5002: \prl(ur1 2)
5003: In the $\ur$-sections of
5004: $\Res(b_{2m}',b_{2m}^{\phi},) \Sit{l}{2m}$ ($l=\pm x$ ($x>0$))
5005: an $\ur$-line acts like a propagating line.
5006: If $m-x$ odd and $l<0$, or $m-x$ even and $l>0$, then the inner
5007: region is black so the first $\uro$ line also
5008: acts as a propagating (and blobbed) line.
5009: Taking into account the $x$ lines that are already propagating,
5010: the section with $\ur = r$ is thus isomorphic to a
5011: $b_{2m}'$ standard module of form $\Deltab{-(x+r+1)}{}$
5012: if one of the `black' conditions above is satisfied;
5013: and of form $\Deltab{x+r}{}$ otherwise.
5014: (NB, the sign on the weight here does not affect the dimension of the
5015: module.)
5016: A similar statement holds for $l=0$, so that
5017: $\Res(b_{2m}',b_{2m}^{\phi},) \Sit{0}{2m}$
5018: is a sum of one copy of each blob standard.
5019: \Qed
5020: \end{pr}
5021:
5022: %}}}
5023: %{{{ co
5024:
5025: Recall that the dimension of $\Sit{0}{2m}$ is $2^m$.
5026:
5027: \begin{co} \label{wangy}
5028: Consider the prestandard $\Sit{l}{2m}$, where $l = \pm x$ and $x > 0$.
5029: Define the integer $\epsilon$ by
5030:
5031: $ \epsilon = 1 $ if $m-x$ odd,
5032:
5033: $ \epsilon = 2 $ if $m-x$ even and $l > 0$,
5034:
5035: $ \epsilon = 0 $ if $m-x$ even and $l < 0$,
5036: \\
5037: and let $k = (m - (x+\epsilon))/2$.
5038: Then the dimension of $S_l(2m)$ is given by
5039: $\sum_{i = 0}^k {m \choose i}$.
5040:
5041: \end{co}
5042: {\em Proof:}
5043: From \cite{MartinSaleur94a}, the dimension of the $b'_{2m}$-standard module
5044: $\Deltab{\pm c}{}$ is given by ${m \choose {(m - c)/2}}$. The result now
5045: follows from Proposition~\ref{ur1 2}, summing over the $r$-values specified in
5046: Proposition~\ref{ur1}.
5047: \Qed
5048: \\
5049: Thus we have determined the complete generic representation theory.
5050:
5051: Note that the globalisation and localisation functors act
5052: in a natural way on blob
5053: categories as well as \achiralb\ categories.
5054: We did not need this fact here, but it is useful in computing
5055: non-generic representation theory.
5056:
5057:
5058: %}}}
5059:
5060: %{{{ discussion
5061: \section{Discussion}
5062: Having determined the generic representation theory,
5063: and set up the homological machinery for analysing the
5064: exceptional (non-semisimple) cases,
5065: in our next paper we will turn to computing the
5066: representation theory of the exceptional cases.
5067: We conclude here with a brief
5068: introduction to
5069: this problem.
5070:
5071:
5072: %{{{ 9
5073:
5074: \newtheorem{proof}{Proof}
5075: %{{{ 1
5076:
5077: With the Temperley--Lieb and blob algebras, the symplectic blob
5078: algebra (or isomorphically, $b^{\phi}_{2m}$)
5079: belongs to an intriguing class of Hecke algebra quotients.
5080: The first two
5081: have representation theories beautifully and efficiently described in
5082: alcove geometrical language, where the precise geometry is determined,
5083: in the non-semisimple cases,
5084: by the parameters of the algebra.
5085: In these first two algebras the parameterisation appropriate to
5086: reveal this structure is not that in which the algebras were first
5087: described. Rather, it was discovered during efforts to put the low
5088: rank
5089: data on non-semisimple manifolds in parameter space in a coherent
5090: format \cite{MartinSaleur94a}.
5091: The determination of the representation theory of $b^{\phi}_{2m}$
5092: in the non-semisimple cases is the next important problem in the
5093: programme initiated in this paper.
5094: We therefore conclude the paper with one result on the
5095: characterisation of non-semisimple manifolds.
5096: The crucial point is that because of the globalisation map,
5097: this is derived from a
5098: low rank result, which then globalises to all levels in the tower.
5099:
5100: %}}}
5101: %{{{ qh and the params and polynomials
5102:
5103: Throughout this section we will assume that all the parameters are
5104: units, so $b_{2m}^{\phi}$ is quasihereditary.
5105:
5106:
5107: %}}}
5108: %{{{ Set
5109:
5110: Set $L_{2m}(l)$ to be the irreducible head of the standard module
5111: $ \sS_{2m}(l)$ .
5112: At any point in parameter space for which the
5113: Gram determinant $\Deltag_{2m}(l) $
5114: evaluates to zero,
5115: there is a proper submodule of $\sS_{2m}(l)$
5116: and so $\sS_{2m}(l)$ is not simple. In this case using the fact that
5117: $b_{2m}^\phi$ is quasihereditary we can find a non-zero map
5118: $\sS_{2m}(j) \to \sS_{2m}(l)$ for some $j \ne l$.
5119: Once we have found a non-zero map we can then globalise it to larger
5120: $m$, using functor $G$
5121: and
5122: Proposition~\ref{old 824}.
5123:
5124: %}}}
5125: %{{{ low rank
5126: Define polynomials
5127: in the six parameters $\{\delta,
5128: \delta_L, \delta_R, \kappa_L,
5129: \kappa_R, \kappa_{LR}\}$:
5130: \begin{eqnarray*}
5131: K_{0}\! \!\! &=& \!\! \kappa_{LR} \\
5132: K_{1}\! \!\! &=& \!\! \delta_L \delta_R - \kappa_{LR} \\
5133: K_{2}\! \!\! &=& \!\! \kappa_{LR} - \delta_L \kappa_R - \kappa_L \delta_R
5134: +\delta\delta_L\delta_R \\
5135: K_{3}\! \!\! &=& \!\! \delta^2\delta_L\delta_R-\delta\delta_L\kappa_R
5136: -\delta\delta_R\kappa_L-\delta_l\delta_R+\kappa_L\kappa_R \\
5137: K_{1,3}\! \!\! &=&
5138: \delta^2\delta_L\delta_R-\delta\delta_L\kappa_R
5139: -\delta\delta_R\kappa_L+\kappa_L\kappa_R-\kappa_{LR}
5140: \end{eqnarray*}
5141: and commuting operators $\Phi$ and $\Psi$ on the space of six-parameter
5142: polynomials which swap the second and fourth, \resp\ third and fifth,
5143: parameters.
5144: The non-trivial Gram determinants for $ b_6^\phi$ are:
5145: %{{{ ignore
5146: %}}}
5147: $$
5148: \Deltag_6(-1) =
5149: \kappa_L \kappa_R K_3
5150: $$
5151: \begin{eqnarray*}
5152: \Deltag_6(0) \!\!\!&=&\!\!
5153: \kappa_{LR}^4 K_1^4 \Psi\Phi(K_1)
5154: \Psi(K_2) \Phi(K_2) K_{1,3}
5155: \end{eqnarray*}
5156:
5157: %}}}
5158: %{{{ We set...
5159:
5160:
5161: We may deduce maps:
5162: $ S_6 (-1) \hookrightarrow S_6(0)$ for
5163: $K_1=0$,
5164: $ S_6 (1) \hookrightarrow S_6(0)$ for
5165: $\Psi\Phi(K_1) =0$,
5166: $ S_6 (2) \hookrightarrow S_6(0)$
5167: for
5168: $\Phi(K_2)=0$
5169: and
5170: $ S_6 (-2) \hookrightarrow S_6(0)$ for
5171: $\Psi(K_2)=0$.
5172: We may deduce that the only possible
5173: non-zero map to $S_6(-1)$ is
5174: $S_6(-3) \hookrightarrow S_6(-1)$ and this therefore must occur when
5175: $K_3=0$.
5176:
5177: We also get a non-zero map $S_6(-3) \hookrightarrow S_6(0)$ when
5178: $K_{1,3} =0$.
5179: Thus when both
5180: $K_1=0$
5181: and $K_3=0$
5182: we get two copies of $S_{6}(-3)$ in the socle of $S_6(0)$.
5183:
5184:
5185: %}}}
5186: %{{{ prop
5187: \begin{pr}
5188: $b_{2m}^\phi(\delta, \delta_L,\delta_R, \kappa_L, \kappa_R,
5189: \kappa_{LR}) $
5190: is not semisimple when
5191:
5192: \begin{tabular}{l} \hspace{-.3in}(1)
5193: $K_3=0$
5194: and $m$ is odd and $m \ge 3$ \\
5195: $\Phi (K_3)=0$
5196: and $m$ is even and $m \ge 4$ \\
5197: $\Psi (K_3)=0$
5198: and $m$ is even and $m \ge 4$ \\
5199: $\Phi\Psi (K_3)=0$
5200: and $m$ is odd and $m \ge 5$. \\
5201: \end{tabular}
5202:
5203: \begin{tabular}{l} \hspace{-.3in}(2)
5204: $K_{1,3}=0$
5205: and $m$ is odd and $m \ge 3$ \\
5206: $\Phi(K_{1,3})=0$
5207: and $m$ is even and $m \ge 4$ \\
5208: $\Psi(K_{1,3})=0$
5209: and $m$ is even and $m \ge 4$ \\
5210: $\Psi\Phi(K_{1,3})=0$
5211: and $m$ is odd and $m \ge 5$. \\
5212:
5213: \end{tabular}
5214: \end{pr}
5215: Cases (1) are proved by globalising the map $ S_6 (-3) \to S_6(-1)$
5216: for
5217: $b_{6}^{\phi}(\delta, \delta_L,\delta_R, \kappa_L, \kappa_R,
5218: \kappa_{LR}) $ with
5219: $K_3=0$;
5220: cases (2) are proved by globalising the map $S_6(-3) \to S_6(0)$
5221: for
5222: $b_{6}^{\phi}(\delta, \delta_L,\delta_R, \kappa_L, \kappa_R,
5223: \kappa_{LR}) $ with
5224: $K_{1,3} =0$.
5225:
5226: %}}}
5227: %}}}
5228:
5229:
5230:
5231:
5232: %}}}
5233:
5234:
5235: %}}}
5236: %{{{ APPENDICES
5237: \vspace{1.398in}
5238:
5239: \appendix
5240: \medskip\noindent
5241: {\LARGE\bf Appendix} \vspace{-.1in}
5242: \stuffb
5243:
5244: \stuffbb
5245:
5246:
5247:
5248: %}}}
5249: \bibliographystyle{amsplain}
5250: \bibliography{new31}
5251: \end{document}
5252: %{{{ Local Variables
5253: % Local Variables:
5254: % eval: (standard-display-european 1)
5255: % folded-file: t
5256: % End:
5257:
5258: %}}}
5259:
5260: