1: %okd32.tpc, okd41.tpc, okdaa.tpc
2: \documentclass[12pt]{amsart}
3: \input definition.tex
4: \usepackage[dvips]{graphicx}
5: %\usepackage{e:/work/pentagons}
6: %\usepackage{faces}
7: %\usepackage{pentad}
8: \newcommand{\trace}{\hbox{\rm tr}}
9: %\newcommand{\Hom}{\hbox{\rm Hom}}
10: %\newcommand{\End}{\hbox{\rm End}}
11: %\newcommand{\K}{{\ym K}}
12: \newcommand{\Abar}{{\overline A}}
13: \newcommand{\Cbar}{{\overline C}}
14: \title[]
15: {Oriented Kauffman diagrams and universal quantum groups}
16: \author[Yamagami Shigeru]{Yamagami Shigeru}
17: \newcommand{\field}{\C}
18:
19: \begin{document}
20: \maketitle
21: \begin{center}
22: Department of Mathematics and Informatics
23: \end{center}
24: \begin{center}
25: Ibaraki University
26: \end{center}
27: \begin{center}
28: Mito, 310-8512, JAPAN
29: \end{center}
30:
31: \begin{abstract}
32: We study
33: the tensor category of oriented Kauffman diagrams
34: and determine fiber functors on them as well as
35: the associated Hopf algebras.
36: \end{abstract}
37: \bigskip
38: %\noindent{\bf Introduction}
39:
40: %\baselineskip=18pt
41:
42: % \rotatebox{180}{C}
43: % \rotatebox{60}{Q}
44: \section{Introduction}
45: In our previous paper \cite{CDA, FTL},
46: we determined fiber functors
47: on Temperley-Lieb categories with the help of universality
48: property on rigidity and classification results
49: on bilinear forms. The associated
50: Hopf algebra as a consequence of Tannaka-Krein duality
51: is identified with
52: the algebraic quantum group of Dubois-Violette and Launer.
53:
54: When C*-structure is entailed,
55: unitary fiber functors are also classified
56: with the associated compact quantum groups
57: isomorphic to the universal quantum groups of orthogonal type
58: due to Wang \cite{D-W} and Banica \cite{Ba}.
59: There is another related class
60: of compact quantum groups, called
61: the universal quantum group of unitary type,
62: investigated by the same authors.
63:
64: Here we shall study an oriented version of Temperley-Lieb
65: categories and obtain an analogous classification
66: of fiber functors on them.
67: Though we have failed in identifying the associated
68: algebraic quantum groups, when restricted to the unitary case,
69: they turn out to be the universal quantum groups of
70: unitary type, thus revealing
71: geometric structures behind them.
72:
73: \section{Oriented Kauffman Diagrams}
74: Let $D$ be a Kauffman diagram of type $(m,n)$
75: (\cite{Kau, Kau2}), i.e.,
76: $D$ is the isotopy class of planar strings in a rectangle
77: with the strings having $m$ terminal points on
78: the upper bounding line
79: and $n$ terminal points on the lower bounding line.
80: Thus $D$ contains $(m+n)/2$ strings in total and
81: there are $2^{(m+n)/2}$ possibilities in the choice of orientations
82: of strings in $D$.
83: A Kauffman diagram with a specific orientation is called
84: an \textbf{oriented Kauffman diagram}.
85: % Explicitly, we shall
86: For an explicit description of orientation,
87: % we
88: % specify the parity information on terminal points
89: % in the following way:
90: we assign one of symbols $X$ and $X^*$ to each point
91: so that $X$ (resp.~$X^*$) indicates starting
92: (resp.~ending) for upper bounding points and
93: ending (resp.~starting) for lower bounding points.
94: By this coding, an orientation for $D$ produces
95: two words $(X_1,X_2,\dots, X_m)$ and
96: $(X'_1,X'_2,\dots, X'_n)$ of two letters $\{ X, X^* \}$
97: corrsponding
98: to upper and lower sequences of vertices.
99: Clearly the set of such words is
100: identified with the free product monoid $\N * \N$
101: ($\N = \{ 1, 2, \dots\}$
102: being the additive monoid of natural numbers).
103: For an element $w \in \N * \N$, we use the notation $X^w$
104: to stand for the corresponding word.
105: Let $w, w' \in \N *\N$ be defined by
106: \[
107: X^w = (X_1, \dots, X_m),
108: \qquad
109: X^{w'} = (X'_1, \dots, X'_n).
110: \]
111: The pair $(w,w')$ is called the type of
112: an oriented Kauffman diagram.
113: Given words $w, w' \in \N * \N$, let $K_{w,w'}$ be the set of
114: oriented Kauffman diagrams of type $(w,w')$ and $\C[K_{w,w'}]$ be
115: the free complex vector space generated by the set $K_{w,w'}$
116: ($\C[\emptyset] = \{ 0\}$ by definition).
117: Let $|w|_+$ (resp.~$|w|_-$)
118: be the number of $X$'s (resp.~$X^*$'s) in $X^w$.
119: Then the set $K_{w,w'}$ is non-empty if and only if both of
120: $|w|_+ + |w'|_-$ and $|w|_- + |w'|_+$ are even numbers,
121: which can be seen by an easy induction argument.
122:
123: Just as in the unoriented case,
124: we introduce a (strict) tensor category
125: $\cO_{d_L,d_R}$ parametrized by $d_L, d_R \in \C^\times$:
126: \begin{enumerate}
127: \item
128: Objects are exactly the symobols $\{ X^w\}$ with $w \in \N * \N$.
129: \item
130: Hom-sets are set to be $\Hom(X^w,X^{w'}) = \C[K_{w,w'}]$
131: and the operation of composition
132: $\Hom(X^{w'},X^{w''})\times \Hom(X^w,X^{w'})
133: \to \Hom(X^w,X^{w''})$ is defined by
134: the concatenation of planar strings with clockwise
135: (resp.~anticlockwise) loops replaced by $d_R$ (resp.~$d_L$).
136: \item
137: The tensor product for morphisms is the linear extension of
138: the horizontal juxtaposition of diagrams in $K_{w,w'}$.
139: \end{enumerate}
140:
141: Notice that the unit object $I$ of $\cO_{d_L,d_R}$ is associated
142: to the empty word, i.e., $I = X^\emptyset$.
143:
144: The tensor category obtained in this way is rigid as in the case
145: of Temperley-Lieb categories and bears a kind of universality
146: on rigidity.
147:
148: Let $\epsilon_X: X\otimes X^* \to I$ and
149: $\epsilon_{X^*}:X^*\otimes X \to I$ be basic arcs in
150: $\cO_{d_L,d_R}$ with
151: $\delta_X: I \to X^*\otimes X$ and $\delta_{X^*}:I \to X\otimes X^*$
152: the associated copairings.
153:
154: \begin{Lemma}
155: Any (strictly) monoidal functor $F$ of
156: $\cO_{d_L,d_R}$ into a tensor category
157: $\cT$ is uniquely determined by the choice
158: \[
159: F(\epsilon_X): F(X)\otimes F(X^*) \to I,
160: \quad
161: F(\epsilon_{X^*}): F(X^*)\otimes F(X) \to I
162: \]
163: of rigidity pairings which satisfy the identities
164: \[
165: F(\epsilon_X) F(\delta_{X^*}) = d_L 1_I,
166: \quad
167: F(\epsilon_{X^*}) F(\delta_X) = d_R 1_I.
168: \]
169: Note here that $F(\delta_X)$ and $F(\delta_{X^*})$ are
170: characterized as the rigidity coparings associated with
171: $F(\epsilon_X)$ and $F(\epsilon_{X^*})$ respectively.
172:
173: Conversely, given rigidity pairings
174: $\epsilon_Y: Y\otimes Y^* \to I$ and $\epsilon_{Y^*}:
175: Y^*\otimes Y \to I$ in a tensor category $\cT$, satisfying
176: the relations
177: \[
178: \epsilon_Y\delta_{Y^*} = d_L 1_I,
179: \quad
180: \epsilon_{Y^*}\delta_Y = d_R 1_I,
181: \]
182: there exists a monoidal functor $F: \cO_{d_L,d_R} \to \cT$
183: such that $F(X) = Y$, $F(X^*) = Y^*$,
184: $F(\epsilon_X) = \epsilon_Y$ and
185: $F(\epsilon_{X^*}) = \epsilon_{Y^*}$.
186: \end{Lemma}
187:
188: Although the category $\cO_{d_L,d_R}$ has apparently
189: two parameters,
190: one freedom of them is superficial as we shall see below.
191:
192: \begin{Lemma}
193: Let $X^w$ ($w \in \N * \N$) be an object of $\cO_{d_L,d_R}$.
194: Then
195: $\dim \End(X^w) = 1$ if and only if $X^w = X^n$ or
196: $X^w = (X^*)^n$ for some $n \in \N$.
197: \end{Lemma}
198:
199: \begin{proof}
200: In fact, $\End(X\otimes X^*)$ and $\End(X^*\otimes X)$ are
201: two-dimensional, which are included in
202: $\End(X^w)$ unless $X^w = X^n$ or $X^w = (X^*)^n$.
203: \end{proof}
204:
205: Let $F: \cO_{d_L,d_R} \to \cO_{d'_L,d'_R}$ be an equivalence
206: of tensor categories, i.e., $F$ is an essentially surjective
207: and fully faithful tensor functor.
208: By the above lemma, we then have
209: $F(X) = X^n$ or $F(X) = (X^*)^n$ for some integer $n \geq 1$.
210: The case $n \geq 2$, however, contradicts with the essential
211: surjectivity of $F$ because it implies $X \not\cong F(X^w)$
212: for any $w \in \N *\N$.
213: Thus we have alternatives $F(X) = X$ or $F(X) = X^*$.
214:
215: The operation of $F$ on morphisms is then determined by
216: the effect on basic arcs $\epsilon_X$ and $\epsilon_{X^*}$:
217: \[
218: F(\epsilon_X) = \lambda \epsilon_X, \quad
219: F(\epsilon_{X^*}) = \mu \epsilon_{X^*}
220: \quad
221: (F(\delta_X) = \lambda^{-1}\delta_X, \quad
222: F(\delta_{X^*}) = \mu^{-1} \delta_{X^*})
223: \]
224: or
225: \[
226: F(\epsilon_X) = \lambda \epsilon_{X^*}, \quad
227: F(\epsilon_{X^*}) = \mu \epsilon_X
228: \quad
229: (F(\delta_X) = \lambda^{-1}\delta_{X^*}, \quad
230: F(\delta_{X^*}) = \mu^{-1} \delta_X).
231: \]
232: From the obvious equality
233: \[
234: F(\epsilon_X\delta_{X^*}) = \epsilon_X\delta_{X^*},
235: \quad
236: F(\epsilon_{X^*}\delta_X) = \epsilon_{X^*}\delta_X,
237: \]
238: we have
239: \[
240: d'_L = \lambda^{-1}\mu d_L, \quad
241: d'_R = \lambda\mu^{-1} d_R
242: \]
243: in the case $F(X) = X$ and
244: \[
245: d'_L = \lambda \mu^{-1} d_R, \quad
246: d'_R = \lambda^{-1}\mu d_L
247: \]
248: in the case $F(X) = X^*$.
249:
250: Conversely, these relations ensure the existence of
251: an equivalence functor $F$ by the generating property
252: of basic arcs (Lemma~2.1).
253: In this way, we have proved the following:
254:
255: \begin{Proposition}
256: We have
257: \[
258: \cO_{d_L,d_R} \cong \cO_{d'_L,d'_R}
259: \iff d_L d_R = d'_L d'_R.
260: \]
261:
262: In particular, if we write $\cO_d = \cO_{d,d}$,
263: $\cO_{d_L,d_R} \cong \cO_d$ for the choice
264: $d = \pm \sqrt{d_L d_R}$,
265: %if we set $\cO_d = \cO_{d,d}$ with $d \in \C^\times$,
266: with the condition
267: $\cO_d \cong \cO_{d'}$ equivalent to $d = \pm d'$.
268: Moreover self-equivalences of $\cO_d$ are given by
269: \[
270: F(\epsilon_X) = \lambda \epsilon_X,
271: \qquad
272: F(\epsilon_{X^*}) = \lambda \epsilon_{X^*}
273: \]
274: or
275: \[
276: F(\epsilon_X) = \lambda \epsilon_{X^*},
277: \qquad
278: F(\epsilon_{X^*}) = \lambda \epsilon_X
279: \]
280: with $\lambda \in \C^\times$.
281: \end{Proposition}
282:
283: In what follows, we shall concentrate on the category
284: $\cO_d = \cO_{d,d}$
285: withought loss of generality.
286: As in the case of Temperley-Lieb category, we then have
287: the natural operation of duality on the category $\cO_d$:
288: if we introduce the antimultiplicative involution $*$ on
289: the monoid $\N * \N$ by switching the role of two submonoids
290: $\N$, then $X^{w^*}$ gives a dual object of $X^w$ for
291: $w \in \N * \N$ with respect to the obvious pairings and
292: copairings by multiple arcs. The associated operation of transposed
293: maps is given by rotating diagrams by an angle of $\pi$.
294:
295: \section{Semisimplicity Analysis}
296: In the tensor category $\cO_d$, the semisimplicity analysis,
297: together with the Jones-Wenzl recursive formula,
298: works for objects of alternating tensor products
299: $X\otimes X^*\otimes X\otimes \dots$ exactly as
300: in the case of Temperley-Lieb categories (\cite{CDA}).
301: In particular, all of the alternating algebras
302: \[
303: \End(X\otimes X^*\otimes X\otimes \dots),
304: \quad
305: \End(X^*\otimes X\otimes X^*\otimes \dots)
306: \]
307: are semisimple
308: if and only if $d = q + q^{-1}$ with $q^2$ not a proper root of
309: unity.
310: Moreover,
311: under the assumption of semisimplicity on alternating algebras
312: $\End(X\otimes X^*\otimes X\otimes \dots)$,
313: we can inductively define simple objects
314: $\{ X_n, Y_n \}_{n \geq 1}$
315: in the idempotent-completion $\overline{\cO}_d$ of $\cO_d$
316: so that
317: $X_n$ and $Y_n$ are the new stuffs in
318: $\overbrace{X\otimes X^*\otimes \dots}^\text{$n$-factors}$ and
319: $\overbrace{X^*\otimes X\otimes \dots}^\text{$n$-factors}$
320: respectively, i.e.,
321: $X_n = f_n (X\otimes X^*\otimes \dots)$
322: with $f_n$ the $n$-th Jones-Wenzl idempotent and
323: similarly for $Y_n$.
324:
325: To a word $w \in \N * \N$, we associate a subobject
326: $X_w$ of $X^w$ in $\overline{\cO}_d$ by replacing maximal
327: alternating subwords in $w$ with the highest part just defined.
328: (Note that $X_w^* = X_{w^*}$ and the transposed ${}^tf_n$ is
329: again a Jones-Wenzl idempotent.)
330:
331: \begin{Example}
332: For the choice
333: \[
334: X^w = (X)(XX^*XX^*)(X^*XX^*XX^*)(X^*)(X^*),
335: \]
336: we have $X_w = X_1 \otimes X_4\otimes Y_5\otimes Y_1\otimes Y_1$.
337: \end{Example}
338:
339: \begin{Lemma}
340: For a non-empty word $w \in \N * \N$, %we have
341: $\Hom(X_w,I) = \{ 0\}$.
342: \end{Lemma}
343:
344: \begin{proof}
345: Let $w = w_1\dots w_l$ be the factorization into
346: maximal alternating parts and write
347: $X_{w_j} = P_j X^{w_j}$ with $P_j$ the Jones-Wenzl projection
348: to the highest part of the alternating tensor product $X^{w_j}$.
349: We shall show that any diagram $D$ in
350: $K_{w,\emptyset} \subset \Hom(X^w,I)$ annihilates
351: $P_1\otimes \dots \otimes P_l$ by an induction on the word length
352: $|w| = \sum_j |w_j|$ ($|w_j| = |w_j|_+ + |w_j|_-$).
353:
354: First observe that, if $D$ contains an arc connecting two vertices
355: in some $w_j$, then $P_j$ annihilates the arc (regarded as
356: a pairing morphism) by the highest assumption
357: (old stuffs being killed by $P_j$).
358: Thus, we need to deal with $D$ having no such arcs.
359: Then any vertex inside $w_1$ should be joined to a vertex in
360: $w_2\dots w_l$. The right-end vertex $v_1$ of $w_1$, however,
361: cannot be connected to any vertex in $w_2$
362: because of parity mismatch. So $v_1$ is joined to a vertex $v_k$
363: in $w_k$ with $k \geq 3$.
364: Now write $w_k = w'_k w''_k$, where $w''_k$ is the subword of $w_k$
365: starting at the vertex $v_k$ until the end of $w_k$ and
366: $w'_k$ is the complement to $w''_k$.
367: We then have the factorization
368: $D = D''
369: (1_{X^{w_1}}
370: \otimes D'\otimes
371: 1_{X^{w''}})$
372: with $w'' = w_k''w_{k+1}\dots w_l$ and
373: $D' \in K_{w_2\dots w_{k-1}w'_k}$ (see Fig.~1).
374:
375: On the other hand,
376: the Clebsh-Gordan fusion rule on alternating parts ensures
377: the relation $P_k = (P'_k\otimes P''_k)P_k$
378: ($P_k'$ and $P_k''$ being the Jones-Wenzl idempotents for
379: $w_k'$ and $w_k''$ respectively) and hence we see
380: \[
381: D(P_1\otimes \dots \otimes P_l) = D''
382: (1_{X^{w_1}}\otimes D'(P_2\otimes \dots \otimes P_{k-1}\otimes P'_k)
383: \otimes 1_{X^{w''}})
384: (P_1\otimes \dots \otimes P_l),
385: \]
386: which vanishes by the induction hypothesis
387: $D'(P_2\otimes \dots \otimes P_{k-1}\otimes P'_k) = 0$.
388: \end{proof}
389:
390: \begin{figure}[h]
391: %\includegraphics[width=12cm,clip]{three.eps}
392: \input okd32.tpc
393: \caption{\label{okd32}}
394: \end{figure}
395:
396:
397: \begin{Corollary}
398: We have $\Hom(X_w, X_{w'}) = \{ 0\}$ for $w \not= w' \in \N * \N$
399: and
400: $\Hom(X_w,X_w) = \C 1_{X_w}$.
401: \end{Corollary}
402:
403: \begin{proof}
404: Decompose $w = w_1\dots w_l$ and $w' = w'_1\dots w'_{l'}$ as
405: before. Then
406: $\Hom(X_w, X_{w'}) \cong \Hom(X_w\otimes X_{(w')^*},I)$ by rigidity.
407: If this vector space is non-trivial, the product
408: $w_l(w'_{l'})^*$ should interact at the contact point and
409: we have the decomposition of the form
410: \[
411: X_{w_l}\otimes X_{(w'_{l'})^*} \cong
412: X_{\widetilde w_1} \oplus \dots \oplus X_{\widetilde w_m}
413: \]
414: according to the Clebsh-Gordan rule.
415:
416: Thus, if the unit object $I$ does not appear inside
417: $X_{w_l}\otimes X_{(w'_{l'})^*}$, we know
418: \[
419: \Hom(X_w\otimes (X_{w'})^*,I) =
420: \bigoplus_{j=1}^m
421: \Hom(X_{w_1\dots w_{l-1} \widetilde w_j (w'_{l'-1})^*
422: \dots (w'_1)^*},I) = \{ 0\}.
423: \]
424: When $I$ is contained, $w_l = w'_{l'}$ and the unit object
425: appears exactly once in
426: $X_{w_l}\otimes X_{w'_{l'}}^*$ (say, $X_{\widetilde w_1} = I$
427: and $X_{\widetilde w_j} \not\cong I$ for $j \geq 2$) and
428: we get
429: \begin{align*}
430: \Hom(X_w\otimes (X_{w'})^*,I) &\cong
431: \bigoplus_{j=1}^m
432: \Hom(X_{w_1\dots w_{l-1}}\otimes X_{\widetilde w_j}
433: \otimes X_{w'_1\dots w'_{l'-1}}^*,I)\\
434: &= \Hom(X_{w_1\dots w_{l-1}}\otimes X_{w'_1\dots w'_{l'-1}}^*,I)\\
435: &\cong \Hom(X_{w_1\dots w_{l-1}}, X_{w'_1\dots w'_{l'-1}}).
436: \end{align*}
437: Now the induction argument is applied to see $w = w'$ and
438: \[
439: \End(X_{w_1\dots w_l}) \cong \End(X_{w_1\dots w_{l-1}})
440: \cong \C.
441: \]
442: \end{proof}
443:
444: \begin{Proposition}
445: The tensor category $\cO_d$ is semisimple if and only if
446: $d = q + q^{-1}$ with $q^2$ not a proper root of unit.
447: In this case, $\{ X_w \}_{w \in \N *\N}$ gives
448: a representative set of simple objects and the fusion rule is
449: given by the following recipe:
450: Let $w = w_1\dots w_l$ and $w' = w'_1\dots w'_{l'}$ be
451: the decompositions into maximally alternating parts.
452: \begin{enumerate}
453: \item
454: If $w_lw'_1$ matches in the parity,
455: the tensor product $X_{w_l}\otimes X_{w_1'}$ decomposes according to
456: the Clebsh-Gordan rule, otherwise $X_w\otimes X_{w'} = X_{ww'}$
457: remains simple.
458: \item
459: Let $X_{w_l}\otimes X_{w'_1} \cong
460: X_{u_1} \oplus \dots \oplus X_{u_k}$. If $I \not\cong U_j$ for
461: $1 \leq j \leq k$,
462: \[
463: X_w\otimes X_{w'} \cong
464: \bigoplus_{j=1}^k X_{\widetilde w_j}
465: \]
466: with $\widetilde w_j
467: = w_1\dots w_{l-1}u_jw'_2\dots w'_{l'} \in \N * \N$
468: gives an irreducible decompositionn.
469:
470: If $X_{u_1} \not\cong I$
471: (and hence $X_{u_j} \not\cong I$ for $j \geq 2$),
472: \[
473: X_w\otimes X_{w'} \cong
474: (X_{w_1\dots w_{l-1}}\otimes X_{w'_2\dots w'_{l'}})
475: \oplus X_{\widetilde w_2} \oplus \dots \oplus
476: X_{\widetilde w_k}
477: \]
478: and we are reduced to the decompotion of
479: $X_{w_1\dots w_{l-1}} \otimes X_{w'_2\dots w'_{l'}}$.
480: \end{enumerate}
481: \end{Proposition}
482:
483: \begin{Remark}
484: If we restrict ourselves to the unitary (i.e., C*-) case,
485: then the above formula produces the fusion rule in
486: \cite[Theorem~1]{Ba2} via Proposition~5.3 below.
487: \end{Remark}
488:
489: \section{Positivity Condition}
490: We shall here investigate possible *-structures on $\cO_d$.
491: Assume that there is a (compatible) *-structure on $\cO_d$.
492: Then we should have
493: \[
494: (\epsilon_X)^* = c_X \delta_{X^*},
495: \quad
496: (\epsilon_{X^*})^* = c_{X^*} \delta_X
497: \]
498: with $c_X, c_{X^*} \in \C^\times$.
499: % , i.e., see the figure below.
500:
501: %\input okd1.tpc
502:
503: To preserve the rigidity, we should have
504: \[
505: 1_X = 1_X^* =
506: \Bigl(
507: (1_X\otimes \epsilon_{X^*})(\delta_{X^*}\otimes 1_X)
508: \Bigr)^* = c_{X^*}(\delta_{X^*}^*\otimes 1_X)
509: (1_X\otimes \delta_X)
510: \]
511: and
512: \[
513: 1_{X^*} = (1_{X^*})^* =
514: \Bigl(
515: (1_{X^*}\otimes \epsilon_X)(\delta_X\otimes 1_{X^*})
516: \Bigr)^* = c_X(\delta_X^*\otimes 1_{X^*})
517: (1_{X^*}\otimes \delta_{X^*}),
518: \]
519: which are equivalent to
520: \[
521: (\delta_X)^* = c_X^{-1} \epsilon_{X^*},
522: \quad
523: (\delta_{X^*})^* = c_{X^*}^{-1} \epsilon_X.
524: \]
525: To preserve trace (loop) values, we should have
526: \begin{gather*}
527: \overline d = (\epsilon_X\delta_{X^*})^*
528: = c_X c_{X^*}^{-1} \epsilon_X \delta_{X^*} =
529: c_X c_{X^*}^{-1} d,\\
530: \overline d = (\epsilon_{X^*}\delta_X)^*
531: = c_X^{-1} c_{X^*} \epsilon_{X^*} \delta_X =
532: c_X^{-1} c_{X^*} d,
533: \end{gather*}
534: i.e.,
535: \[
536: c_X = \pm c_{X^*}\quad \text{and}\quad d = \pm \overline d.
537: \]
538: With these conditions satisfied, the universality on rigidity
539: (Lemma~2.1) allows us to extend the operation to the whole hom-sets
540: by antilinearity and antimultiplicativity. The obtained map is
541: then involutive if and only if
542: \[
543: \epsilon_X = (\epsilon_X)^{**} =
544: \overline{c_X} \delta_{X^*}^* = \frac{\overline{c_X}}{c_{X^*}}
545: \epsilon_X,\qquad
546: \epsilon_{X^*} = (\epsilon_{X^*})^{**} =
547: \overline{c_{X^*}} \delta_X^* = \frac{\overline{c_{X^*}}}{c_X}
548: \epsilon_{X^*},
549: \]
550: i.e., $\overline{c_X} = c_{X^*}$.
551:
552: \begin{Proposition}
553: The tensor category $\cO_d$ admits a compatible *-structure
554: if and only if $d^2 \in \R$, i.e., $d = \pm \overline{d}$.
555:
556: If this is the case, *-structures are parametrized by
557: $c \in \C^\times$ satisfying $\overline{cd} = cd$
558: with the associated *-operation given by
559: \[
560: D^* = c^{\sharp(D)}
561: \left(
562: \frac{\overline d}{d}
563: \right)^{l(D)} D'
564: \]
565: for an oriented diagram $D$, where $D'$ is the orientation reversion
566: of the reflection of $D$ (see Fig.~2),
567: \[
568: \sharp(D) = \sharp\{
569: \text{$\epsilon_X$'s and $\epsilon_{X^*}$'s inside $D$}\}
570: - \sharp\{
571: \text{$\delta_X$'s and $\delta_{X^*}$'s inside $D$}\},
572: \]
573: and
574: \[
575: l(D) = \sharp\{
576: \text{$\epsilon_{X^*}$'s inside $D$} \}
577: - \sharp\{
578: \text{$\delta_{X^*}$'s inside $D$} \}.
579: \]
580:
581: Moreover two *-structures are equivalent if and only if
582: $c = \lambda c'$ or $\overline c = \lambda c'$ for some
583: $\lambda >0$.
584: \end{Proposition}
585:
586: \begin{proof}
587: Let $c, c' \in \C^\times$ be parameters of *-structures
588: with the associated *-structures on $\cO_d$
589: denoted by $*$ and $\star$
590: respectively. Since monoidal automorphisms of $\cO_d$ are of the
591: form
592: \[
593: F(\epsilon_X) = \lambda \epsilon_X,
594: \quad
595: F(\epsilon_{X^*}) = \lambda \epsilon_{X^*}
596: \quad
597: \text{or}
598: \quad
599: G(\epsilon_X) = \lambda \epsilon_{X^*},
600: \quad
601: G(\epsilon_{X^*}) = \lambda \epsilon_X,
602: \]
603: the condition $F(D^*) = F(D)^\star$
604: (resp.~$G(D^*) = G(D)^\star$) is equivalent to $c = |\lambda|^2 c'$
605: (resp.~$\overline c = |\lambda|^2 c'$).
606: \end{proof}
607:
608: \begin{figure}[h]
609: \hspace{1cm}
610: \input okd41.tpc
611: \caption{\label{okd41}}
612: \vspace{5mm}
613: \end{figure}
614:
615: \begin{Theorem}
616: The tensor category $\cO_d$ admits a compatible C*-structure
617: if and only if $d \in \R^\times$ and $d^2 \geq 4$.
618:
619: If this is the case, C*-structure is unique up to
620: monoidal equivalences
621: and is given by
622: \[
623: D^* = \left(
624: \frac{d}{|d|}
625: \right)^{\sharp(D)} D'
626: \]
627: for a diagram $D$.
628: \end{Theorem}
629:
630: \begin{proof}
631: We argue as in the case of Temperley-Lieb category: Since
632: idempotents
633: \[
634: e_1 = \frac{1}{d} (\delta_{X^*}\epsilon_X)\otimes 1_X,
635: e_2 = \frac{1}{d} 1_X\otimes (\delta_X\epsilon_{X^*})
636: \]
637: are related to the central decomposition of
638: $\End(X\otimes X^*)$ and $\End(X^*\otimes X)$, we have
639: $e_1^* = e_1$ and $e_2^* = e_2$, which are equivalent to
640: $\overline c/c = \overline d/d$ (automatically satisfied).
641: From the relation
642: $e_1e_2e_1 = d^{-2} e_1$, we see $d^2 >0 \iff d \in \R$ and hence
643: $c \in \R^\times$ as well. Moreover $\epsilon_X\epsilon_X^* = cd>0$
644: implies the condition $c/|c| = d/|d|$.
645: Thus, up to equivalences, C*-structure (if any exists) is unique
646: and given by the above formula for the choice $|c|=1$.
647:
648: By a standard argument based on Jones-Wenzl formula
649: (cf.~\cite{CDA}), we can derive
650: the condition $d^2 \geq 4$ from the positivity of the *-structure.
651: On the other hand, if this is assumed, the Temperley-Lieb category
652: $\cK_d$ admits the (unique) C*-structure.
653: Since $\cO_d$ is a subcategory of $\cK_d$, the *-structure of
654: $\cO_d$ meets the positivity and we are done.
655: \end{proof}
656:
657: \section{Fiber Functors}
658: From the universality property,
659: any fiber functor $\Phi: \cO_d \to \cV ec$ is determined by
660: non-degenerate bilinear forms $\Phi(\epsilon_X)$ and
661: $\Phi(\epsilon_{X^*})$.
662: Set $V = \Phi(X)$, $W = \Phi(X^*)$,
663: $E = \Phi(\epsilon_X): V\otimes W \to \C$ and
664: $F = \Phi(\epsilon_{X^*}): W\otimes V \to \C$.
665: Let $\delta_E \in W\otimes V$ and
666: $\delta_F \in V\otimes W$ be the associated vector:
667: Given a basis $\{ v_i\}$ of $V$ and $\{ w_j\}$ of $W$,
668: \[
669: \delta_E = \sum c_{ji} w_j\otimes v_i,
670: \quad
671: \delta_F = \sum d_{ij} v_i\otimes w_j,
672: \]
673: with $\{ c_{ji}\}$ and $\{ d_{ij} \}$ defined by
674: \[
675: \sum_j E(v_k\otimes w_j)c_{ji} = \delta_{ki},
676: \quad
677: \sum_i F(w_k\otimes v_i) d_{ij} = \delta_{kj},
678: \]
679: give the covectors, whence they satisfy
680: \[
681: F(\delta_E) = \sum_{ij} c_{ji} F(w_j\otimes v_i) = d
682: = E(\delta_F) = \sum_{ij} d_{ij} E(v_i\otimes w_j).
683: \]
684:
685: Conversely, given invertible matrices
686: $A = (a_{ij})$ with $a_{ij} = E(v_i\otimes w_j)$ and
687: $B = (b_{ji})$ with $b_{ji} = F(w_j\otimes v_i)$ satisfying
688: \[
689: \text{trace}({}^tB A^{-1}) = d =
690: \text{trace}({}^tA B^{-1}),
691: \]
692: we obtain a fiber functor.
693:
694: Two fiber functors $\Phi$ and $\Phi'$ are naturally equivalent
695: if and only if we can find isomorphisms of vector spaces
696: \[
697: \Phi(X) \to \Phi'(X) = V',
698: \quad
699: \Phi(X^*) \to \Phi'(X^*) = W'
700: \]
701: such that the diagrams
702: \[
703: \begin{CD}
704: V \otimes W @>>> V'\otimes W'\\
705: @V{E}VV @VV{E'}V\\
706: \C @= \C
707: \end{CD}\ ,
708: \qquad
709: \begin{CD}
710: W \otimes V @>>> W'\otimes V'\\
711: @V{F}VV @VV{F'}V\\
712: \C @= \C
713: \end{CD}
714: \]
715: commute.
716: This means that $A$ and $B$ can be replaced by
717: \[
718: {}^tT A S,
719: \qquad
720: {}^tS BT
721: \]
722: with $S$ and $T$ invertible square matrices.
723: Thus we may choose $A = 1$ and there remains the gauge freedom
724: of ${}^tTS = 1$, i.e.,
725: \[
726: B \mapsto T^{-1} BT.
727: \]
728:
729: \begin{Theorem}
730: Fiber functors on $\cO_d$ is completely classified by similarity orbits in
731: \[
732: \{
733: B \in \text{GL}(n,\C); \text{trace}(B) = d =
734: \text{trace}(B^{-1})\}.
735: \]
736: \end{Theorem}
737:
738: For the description of the associated algebraic group (or
739: a Hopf algebra $H$),
740: we restore the matrix $A$ and choose bases $\{ v_i\}$ in $V$ and
741: $\{ w_j \}$ in $W$. Then $H$ is generated, as an algebra,
742: by $2n^2$ elements
743: $\{ v_{ij} \}$ and $\{ w_{kl}\}$ with
744: the defining relations
745: \begin{gather*}
746: \sum_{i,k} a_{ik} v_{ij} w_{kl} = a_{jl} 1,
747: \quad
748: \sum_{j,i} c_{ji} w_{kj} v_{li} = c_{kl} 1,\\
749: \sum_{i,k} b_{ik} w_{ij} v_{kl} = b_{jl} 1,
750: \quad
751: \sum_{j,i} d_{ji} v_{kj} w_{li} = d_{kl} 1
752: \end{gather*}
753: corresponding to basic morphisms
754: $\begin{CD} V\otimes W @>{E}>> \C @>{\delta_E}>> W\otimes V
755: \end{CD}$ and
756: $\begin{CD} W\otimes V @>{F}>> \C @>{\delta_F}>> V\otimes W
757: \end{CD}$.
758: The comultiplication is given by
759: \[
760: \Delta(v_{ij}) = \sum_k v_{ik}\otimes v_{kj},
761: \quad
762: \Delta(w_{kl}) = \sum_i w_{ki}\otimes w_{il}.
763: \]
764:
765: Next let $\Phi: \cO_d \to \cH ilb$ be a unitary fiber functor.
766: Here $\cO_d$ ($d^2 \geq 4$) is furnished with the *-structure
767: $D^* = (d/|d|)^{\sharp(D)} D'$ and $\cH ilb$ denotes
768: the C*-tensor category of finite-dimensional Hilbert spaces.
769: The condition $\Phi(D)^* = \Phi(D^*)$ for any $D$, i.e.,
770: \[
771: \Phi(\epsilon_X)^* = \frac{d}{|d|}\Phi(\delta_{X^*}),
772: \quad
773: \Phi(\epsilon_{X^*})^* = \frac{d}{|d|} \Phi(\delta_X)
774: \]
775: is equivalent to the relation
776: \[
777: % \overline A = \frac{d}{|d|} B^{-1},
778: % \qquad
779: \overline B = \frac{d}{|d|} A^{-1}
780: \]
781: with the trace-value condition given by
782: \[
783: \text{trace}(BB^*) = |d| =
784: \text{trace}((BB^*)^{-1}).
785: \]
786: As in the case of non-unitary fiber functors,
787: the gauge freedom is given by
788: \[
789: A \mapsto {}^tTAS,
790: \qquad
791: B \mapsto {}^tS BT
792: \]
793: with $S$, $T$ unitary matrices.
794: Note that the condition $\overline A = (d/|d|) B^{-1}$ as well as
795: the trace-value relation is preserved under this transformation.
796: Thus the orbit space for invertible matrices $B$ is identified with
797: \[
798: U(n)\backslash \text{GL}(n,\C)/U(n)
799: \cong \{ \text{positive definite matrices} \}/
800: \text{unitary similarity}.
801: \]
802:
803: \begin{Theorem}(cf.~\cite[Theorem~1.1]{Wa}, \cite[Theorem~6.2]{BDV})
804: Unitary fiber functors on the C*-tensor category $\cO_d$
805: are parametrized by positive eigenvalue lists
806: $\{ \mu_1, \dots, \mu_n\}$ satisfying
807: \[
808: \sum_j \mu_j^2 = |d| = \sum_j \mu_j^{-2}.
809: \]
810: \end{Theorem}
811:
812: We shall identify the associated compact quantum group
813: with the universal quantum group of unitary type
814: in \cite{Ba, D-W}.
815:
816: For this, we use the notation and the construction
817: in \cite{FTL}. Choose orthonormal bases $\xi = \{ v_i\}$ for
818: $V$ and $\eta = \{ w_k\}$ for $W$ with
819: the basis $\xi^* = \{ v^*_j\}$ of $W$ defined
820: by the relation
821: $E(v_j^*\otimes v_i) = \delta_{ji}$, i.e.,
822: $\sum_k (w_k|v^*_i) b_{kj} = \delta_{ij}$.
823: Let $\{ u^\xi_{ij} \}$ and $\{ u^\eta_{kl} \}$ be the associated
824: generators.
825: Then, from the covariance condition,
826: \[
827: \sum_l (w_l|v^*_i) u^\eta_{kl} =
828: \sum_j (w_k|v^*_j) u^{\xi^*}_{ji}
829: \]
830: and the *-relation $(u^\xi_{ij})^* = u^{\xi^*}_{ij}$, we see that
831: \[
832: u^\eta ({}^tB^{-1}) = ({}^tB^{-1}) u^{\xi^*}
833: \]
834: in the matrix notation for $u^\eta$ and $u^{\xi^*}$.
835: Thus the associated Hopf *-algebra $H$ is generated by
836: $u = (u^\xi_{ij})$, which constitutes a unitary matrix
837: by the orthonormality of the basis $\xi = \{ v_j\}$.
838:
839: From the universality on rigidity, the defining relations
840: are given by the covariance conditions for the morphisms
841: $\Phi(\epsilon_X)$, $\Phi(\epsilon_{X^*})$,
842: $\Phi(\delta_X)$ and $\Phi(\delta_{X^*})$:
843: \begin{align*}
844: \Phi(\epsilon_X)(v_i\otimes w_k) 1_H &=
845: \sum \Phi(\epsilon_X)(v_j\otimes w_l)
846: u^{\xi\otimes \eta}_{j,l;i,k},\\
847: \Phi(\epsilon_{X^*})(w_k\otimes v_i) 1_H &=
848: \sum \Phi(\epsilon_{X^*})(w_l\otimes v_j)
849: u^{\eta\otimes\xi}_{l,j;k,i},\\
850: \sum u^{\eta\otimes \xi}_{k,i;l,j}
851: (w_l\otimes v_j| \Phi(\delta_X)) &=
852: (w_k\otimes v_i| \Phi(\delta_X)) 1_H,\\
853: \sum u_{i,k;j,l}^{\xi\otimes \eta}
854: (v_j\otimes w_l|\Phi(\delta_{X^*})) &=
855: (v_i\otimes w_k| \Phi(\delta_{X^*})) 1_H.
856: \end{align*}
857: From the multiplication relations
858: $u^{\xi\otimes \eta}_{i,k;j,l} = u^\xi_{ij} u^\eta_{kl}$,
859: $u^{\eta\otimes \xi}_{k,i;l,j} = u^\eta_{kl} u^\xi_{ij}$,
860: the condition is equivalent to
861: \begin{align*}
862: A &= {}^t u^\xi A u^\eta,\\
863: B &= {}^t u^\eta B u^\xi,\\
864: u^\eta A^{-1} {}^t u^\xi &= A^{-1},\\
865: u^\xi B^{-1} {}^t u^\eta &= B^{-1}
866: \end{align*}
867: and therefore to the conditions $uu^* = 1 = u^*u$,
868: $\overline u ({}^tBA^{-1}) {}^t u = {}^tB A^{-1}$ and
869: ${}^t u A {}^tB^{-1} \overline u = A {}^tB^{-1}$.
870: Since $A^{-1} = (d/|d|) \overline B$, the conditions are further
871: equivalent to
872: \[
873: uu^* = 1 = u^* u,
874: \quad
875: \overline u (AA^*)^{-1} {}^t u = (AA^*)^{-1},
876: \quad
877: {}^t u AA^* \overline u = AA^*.
878: \]
879: Thus the associated compact quantum group is identified with
880: $A_u(AA^*)$ in \cite{D-W}.
881:
882: \begin{Proposition}
883: The compact quantum group associated to the unitary fiber functor
884: is naturally isomorphic to $A_u(AA^*)$ in \cite{D-W},
885: i.e., $A_u(A^*)$ in \cite{Ba}.
886: \end{Proposition}
887:
888: \begin{Remark}
889: In \cite{BDV}, unitary fiber functors are classified for
890: representation categories of the compact quantum group $A_u(F)$
891: based on representation theory of \cite{Ba2}.
892: Since these representation categories are exactly $\cO_d$ as tensor
893: categories by the above proposition, we get an access to some results
894: in \cite{BDV} in an elementary way.
895: \end{Remark}
896:
897: \appendix
898: \section{Faithfulness}
899: We shall here check the automatic faithfulness of relevant
900: functors.
901:
902: \begin{Proposition}
903: Fiber functors on the Temperley-Lieb categories
904: are faithful whenever the fundamental vector space $V$ has
905: dimension two or more.
906:
907: Fiber functors on $\cO_d$ are faithful if
908: the fundamental vector spaces $V$ and $W$ have dimension two or more.
909: \end{Proposition}
910:
911: \begin{proof}
912: Since Frobenius transforms are isomorphisms, we need to show that
913: the family $\{ \Phi(D); D \in K_{2n,0} \}$ is linearly indenpendent
914: for any $n \geq 1$, where
915: $\Phi: \cT\cL_d \to \cV ec$ denotes a fiber functor such that
916: the fundamental vector space $V = \Phi(X)$ has dimension
917: two or more.
918:
919: Given $1 \leq k \leq n$, let $\{ x_1, \dots x_k \}$ be
920: the first $k$-vertices (counting from the left end)
921: for diagrams in $K_{2n,0}$
922: and set
923: \[
924: \cD_{k,n} = \{
925: D \in K_{2n,0};
926: \text{there are no arcs connecting $x_i$ and $x_j$ in $D$}
927: \}.
928: \]
929: Note that $\cD_{n,n} \subset \cD_{n-1,n} \subset \dots
930: \subset \cD_{2,n} \subset \cD_{1,n} = K_{2n,0}$ and
931: $\cD_{n,n}$ consists of a single diagram.
932:
933: Since $\dim V \geq 2$,
934: we can find a vector $0 \not= v \in V$ such that
935: $F(\epsilon)(v\otimes v) = 0$
936: as a solution of a quadratic equation, for which
937: we shall show that the family
938: \[
939: \cE_{k,n} = \{ \Phi(D)(v^{\otimes k}\otimes \cdot)
940: V^{\otimes (2n-k)} \to \C; D \in \cD_{k,n} \}
941: \]
942: is linearly independent by an induction on $(k,n)$.
943: Observe that, if the family $\cD_{1,n}$ is linearly independent,
944: so is the family $\{ \Phi(D); D \in K_{2n,0} \}$.
945:
946: Given $1 \leq k \leq n$,
947: assume the linear independence of $\cE_{k',n'}$ for
948: $1 \leq k' \leq n' < n$ and for $k' > k, n' = n$.
949: We shall prove that the family $\cE_{k,n}$ is linearly independent.
950: Suppose that
951: \[
952: \sum_{D \in \cD_{k,n}} c_D \Phi(D)(v^{\otimes k}\otimes \cdot) = 0.
953: \]
954: Performing the evalution by
955: the vector $v$ one step further, we have
956: \[
957: \sum_{D \in \cD_{k,n}} c_D \Phi(D)(v^{\otimes (k+1)}\otimes \cdot)
958: = 0
959: \]
960: in $(V^{\otimes (2n-k-1)})^*$. Since $\Phi(D)$ is killed by
961: this evaluation for $D \in \cD_{k,n} \setminus \cD_{k+1,n}$,
962: the summation can be restricted to $\cD_{k+1,n}$
963: and the induction hypothesis ensures $c_D = 0$ for
964: $D \in \cD_{k+1,n}$.
965:
966: By the non-degeneracy of the bilinear form $\Phi(\epsilon)$,
967: we can find a vector $v' \in V$ satifying
968: $\Phi(\epsilon)(v\otimes v') = 1$. Evaluating by $v'$ in
969: the starting equation, we then have
970: \[
971: \sum_{D \in \cD_{k,n} \setminus \cD_{k+1,n}}
972: c_D \Phi(D)(v^{\otimes k}\otimes v'\otimes \cdot) = 0
973: \]
974:
975: Since $\cD_{k,n} \setminus \cD_{k+1,n}$ consists of diagrams
976: which contain the arc connecting $x_k$ and $x_{k+1}$
977: (Fig.~3),
978: we have the natural bijection
979: $\cD_{k,n} \setminus \cD_{k+1,n} \ni D \mapsto
980: D' \in \cD_{k-1,n-1}$ and then the last equation takes the form
981: \[
982: \sum_{D' \in \cD_{k-1,n-1}} c_D
983: \Phi(D')(v^{\otimes (k-1)}\otimes \cdot) = 0.
984: \]
985: Again by the induction hypothesis,
986: $c_D = 0 $ for $D \in \cD_{k,n} \setminus \cD_{k+1,n}$ and
987: we are done for the Temperley-Lieb case.
988:
989: For a fiber functor $\Phi: \cO_d \to \cV ec$, we normalize
990: the bilinear forms $\Phi(\epsilon_X): V\otimes W \to \C$ and
991: $\Phi(\epsilon_Y): W\otimes V \to \C$ so that the associated
992: matrices $A$ and $B$ are upper-triangular.
993: Then, for the choice $v = (1,0, \dots, 0)$ and
994: $w = (0,1,0,\dots,0)$, we see
995: $\Phi(\epsilon_X)(v\otimes w) = \Phi(\epsilon_Y)(w\otimes v) = 0$.
996:
997: Now, given an object $X^\omega$ of $\cO_d$, we can take the tensor
998: product of $v$ and $w$ according to the arrangement of $X$ and
999: $Y = X^*$ in $X^\omega$ up to the $k$-th factor.
1000: By using these as probing vectors, we can repeat the above argument
1001: to conclude the independence of the family
1002: $\{ \Phi(D); D \in K_{\omega, \emptyset}\}$.
1003:
1004: \begin{figure}[h]
1005: \hspace{1cm}
1006: \input okdaa.tpc
1007: \caption{\label{okdaa}}
1008: \vspace{5mm}
1009: \end{figure}
1010:
1011: \end{proof}
1012:
1013: \begin{thebibliography}{8}
1014: \bibitem{Ba}
1015: T.~Banica,
1016: Th\'eorie des repr\'esentations du groupe quantique compact libre
1017: $O(n)$,
1018: \textit{C.~R.~Acad.~Sci.~Paris},
1019: 322(1996), 241--244.
1020: \bibitem{Ba2}
1021: \underline{\phantom{T.~Banica}},
1022: Le groupe quantique compact libre $U(n)$,
1023: \textit{Commun.~Math.Phys.},
1024: 190(1997), 143--172.
1025: \bibitem{BDV}
1026: J.~Bichon, A.~De Rijdt and S.~Vaes,
1027: Ergodic coactions with large multiplicity and monoidal
1028: equivalence of quantum groups,
1029: \textit{Commun.~Math.Phys.},
1030: 262(2006), 703--728.
1031: \bibitem{D-W}
1032: A.~van Daele and S.Z.~Wang,
1033: Universal quantum groups,
1034: \textit{Intern.~J.~Math.}, 7(1996), 255--264.
1035: \bibitem{Kau}
1036: L.H.~Kauffman, State models and the Jones polynomial,
1037: {\it Topology},
1038: 26(1987), 395--407.
1039: \bibitem{Kau2}
1040: \underline{\phantom{Kauffman}},
1041: An invariant of regular isotopy,
1042: \textit{Trans.~Amer.~Math.~Soc.},
1043: 318(1990), 417--471.
1044: \bibitem{Wa}
1045: S.~Wang, Structure and isomorphism classification of
1046: compact quantum groups $A_u(Q)$ and $B_u(Q)$,
1047: \textit{J.~Operator Theory}, 48(2002), 573--583.
1048: \bibitem{CDA}
1049: S.~Yamagami,
1050: %\underline{\phantom{S.~Yamagami}},
1051: A categorical and diagrammatical approach to Temperley-Lieb
1052: algebras,
1053: math.QA/0405267.
1054: \bibitem{FTL}
1055: %S.~Yamagami,
1056: \underline{\phantom{S.~Yamagami}},
1057: Fiber functors on Temperley-Lieb categories,
1058: math.QA/0405517.
1059: \end{thebibliography}
1060:
1061: \end{document}
1062:
1063: %generating property of double arcs
1064: \begin{thebibliography}{8}
1065: \bibitem{Ba}
1066: T.~Banica, Quantum groups and Fuss-Catalan algebras,
1067: arXiv:math.QA/0010084.
1068: \bibitem{BW}
1069: J.W.~Barrett and B.W.~Westbury,
1070: %\underline{\phantom{J.W.~Barrett}},
1071: Spherical categories,
1072: {\it Adv.~in Math.},
1073: 143(1999), 357--375.
1074: \bibitem{BJ}
1075: D.~Bisch and V.~Jones,
1076: Algebras associated to intermediate subfactors,
1077: {\it Invent.~math.},
1078: 128(1997), 89--157.
1079: \bibitem{BD}
1080: M.~Borisavljevi\'c and K.~Dosen,
1081: Kauffman monoids,
1082: \textit{J.~Knot Theory and Its Ramifications},
1083: 11(2002), 127--143.
1084: \bibitem{CP}
1085: V.~Chari and A.~Pressley,
1086: {\it A Guide to Quantum Groups},
1087: Cambridge Univ.~Press,
1088: 1995.
1089: \bibitem{Da}
1090: A.A.~Davydov,
1091: Monoidal categories,
1092: {\it J.~Math.~Sci.~(New York)},
1093: 88(1998), 457--519.
1094: \bibitem{EK}
1095: D.~Evans and Y.~Kawahigashi,
1096: \textit{Quantum Symmetries on Operator Algebras}, Clarendon Press,
1097: Oxford, 1998.
1098: \bibitem{FK}
1099: J.~Fr\"ohlich and T.~Kerler,
1100: \textit{Quantum Groups, Quantum Categories and Quantum Field Theory},
1101: Lec.~Notes in Math.~1542, Springer-Verlag, 1993.
1102: \bibitem{GHJ}
1103: F.M.~Goodman, P.~de la Harpe and V.F.R.~Jones,
1104: {\it Coxeter Graphs and Towers of Algebras},
1105: Springer-Verlag, 1989.
1106: \bibitem{GW}
1107: F.M.~Goodman and H.~Wenzl,
1108: Ideals in the Temperley-Lieb category,
1109: arXiv:math.QA/0206301.
1110: \bibitem{Jo}
1111: V.F.R.~Jones, Index for subfactors,
1112: {\it Invent.~Math.},
1113: 72(1983), 1--25.
1114: \bibitem{Kas}
1115: C.~Kassel,
1116: {\it Quantum Groups},
1117: Springer-Verlag,
1118: Berlin-New York,
1119: 1995.
1120: \bibitem{Kau}
1121: L.H.~Kauffman, State models and the Jones polynomial,
1122: {\it Topology},
1123: 26(1987), 395--407.
1124: \bibitem{Kau2}
1125: \underline{\phantom{Kauffman}},
1126: An invariant of regular isotopy,
1127: \textit{Trans.~Amer.~Math.~Soc.},
1128: 318(1990), 417--471.
1129: \bibitem{Kau3}
1130: \underline{\phantom{Kauffman}},
1131: Knots and Physics, 1991, World Scientific Publishing.
1132: \bibitem{KW}
1133: D.~Kazhdan and H.~Wenzl,
1134: Reconstructing monoidal categories,
1135: {\it Adv.~Soviet Math.},
1136: 16(1993), 111--136.
1137: \bibitem{GKP}
1138: R.L.~Graham, D.E.~Knuth and O.~Patashnik,
1139: \textit{Concrete Mathematics},
1140: Addison-Wesley, 1994.
1141: \bibitem{La}
1142: Z.A.~Landau, Fuss-Catalan algebras and chains of
1143: intermediate subfactors,
1144: {\it Pacific J.~Math.},
1145: 197(2001), 325--367.
1146: \bibitem{Mac}
1147: S.~MacLane,
1148: {\it Categories for the Working Mathematician},
1149: Springer-Verlag,
1150: Berlin-New York,
1151: 1971.
1152: \bibitem{Mal}
1153: G.~Maltsiniotis,
1154: Traces dans les cat\'egories monoidales, dualit\'e et cat\'egories monoidales fibries,
1155: {\it Cahiers Topologie G\'eom. Diff\'erentielle Cat\'eg.},
1156: 36(1995), 195--288.
1157: \bibitem{PP}
1158: M.~Pimsner and S.~Popa,
1159: Iterating the basic constructions,
1160: {\it Trans.~Amer.~Math.~Soc.}, 310(1988), 127--134.
1161: \bibitem{TL}
1162: H.~Temperley and E.~Lieb, Relations between the
1163: `percolation' and `coloring' problem and
1164: other graph-theoretical problems associated with
1165: regular plane lattices,
1166: {\it Proc.~Roy.~Soc.~London A}, 322(1971), 251--280.
1167: \bibitem{Tu}
1168: V.G.~Turaev,
1169: \textit{Quantum Invariants of Knots and 3-Manifolds},
1170: Walter de Gruyter, Berlin-New York, 1994.
1171: \bibitem{W}
1172: B.W.~Westbury,
1173: The representation theory of the Temperley-Lieb algebras,
1174: \textit{Math.~Z.}, 219(1995), 539--565.
1175: \bibitem{NOA}
1176: S.~Yamagami,
1177: A note on Ocneanu's approach to Jones' index theory,
1178: {\it Internat.~J.~Math.},
1179: 4(1993), 859--871.
1180: \bibitem{FTC}
1181: \underline{\phantom{S.~Yamagami}},
1182: Free products of semisimple tensor categories, preprint.
1183: \end{thebibliography}
1184:
1185: \end{document}
1186:
1187: %\bibitem{BW1}
1188: %J.W.~Barrett and B.W.~Westbury,
1189: %Invariants of piecewise-linear $3$-manifolds,
1190: %{\it Trans.~Amer.~Math.~Soc.},
1191: %348(1996), 3997--4022.
1192: %\bibitem{BW2}
1193: %\underline{\phantom{J.W.~Barrett}},
1194: %Spherical categories,
1195: %{\it Adv.~in Math.},
1196: %143(1999), 357--375.
1197: %\bibitem{Da}
1198: %A.A.~Davydov,
1199: %Monoidal categories,
1200: %{\it J.~Math.~Sci.~(New York)},
1201: %88(1998), 457--519.
1202: % \bibitem{D}
1203: % J.~Dixmier,
1204: % {\it Les Alg\`ebres d'Op\'erateurs dans l'Espace Hilbertien},
1205: % Gauthier-Villars, Paris,
1206: % 1969.
1207: % \bibitem{DPR}
1208: % S.~Doplicher, C.~Pinzari and J.E.~Roberts,
1209: % An algebraic duality theory for multiplicative unitaries,
1210: % preprint.
1211: % \bibitem{GLR}
1212: % P.~Ghez, R.~Lima and J.E.~Roberts,
1213: % W*-categories,
1214: % {\it Pacific J.~Math.}, 120(1985), 79--109.
1215: % \bibitem{Ha}
1216: % U.~Haagerup, $L^p$-spaces associated with an arbitrary von Neumann algebra,
1217: % {\it Colloques internationaux du CNRS}, No.~274(1977).
1218: % \bibitem{Ki}
1219: % A.~A.~Kirillov,
1220: % {\it Elements of the Theory of Representations},
1221: % Springer-Verlag,
1222: % Berlin,
1223: % 1976.
1224: % \bibitem{L}
1225: % R.~Longo, Index of subfactors and statistics of quantum fields.~I,
1226: % {\it Commun.~Math.~Phys.}, 126(1989), 217--247.
1227: % \bibitem{LR}
1228: % R.~Longo and J.E.~Roberts,
1229: % A theory of dimension,
1230: % {\it K-Theory},
1231: % 11(1997), 103--159.
1232: %\bibitem{Mal}
1233: %G.~Maltsiniotis,
1234: %Traces dans les cat\'egories monoidales, dualit\'e et cat\'egories monoidales fibries,
1235: %{\it Cahiers Topologie G\'eom. Diff\'erentielle Cat\'eg.},
1236: %36(1995), 195--288.
1237: % \bibitem{O}
1238: % A.~Ocneanu,
1239: % Quantized group, string algebras and Galois theory for algebras.
1240: % In \textit{Operator algebras and applications}, vol.~2
1241: % (ed.~D.E.~Evans and M.~Takesaki) 119--172, Cambridge University Press,
1242: % Cambridge, 1988.
1243: % \bibitem{Pa}
1244: % B.~Pareigis, {\it Categories and Functors}, Academic Press, New York-London,
1245: % 1970.
1246: % \bibitem{Ri}
1247: % M.~Rieffel,
1248: % Morita equivalence for C*-algebras and W*-algebras,
1249: % {\it J.~Pure and Applied Algebras},
1250: % 5(1974), 51--96.
1251: % \bibitem{Ro}
1252: % J.E.~Roberts,
1253: % Cross products of von Neumann algebras by group duals,
1254: % {\it Symposia Mathematica}, 22(1976), 335--363.
1255: % \bibitem{S}
1256: % S.~Sakai,
1257: % {\it C*-algebras and W*-algebras},
1258: % Springer, Berlin,
1259: % 1971.
1260: % \bibitem{Sa}
1261: % J.-L.~Sauvageot, Sur le produit tensoriel relatif d'espaces de Hilbert,
1262: % {\it J.~Operator Theory}, 9(1983), 237--252.
1263: % \bibitem{T}
1264: % M.~Takesaki, {\it Theory of Operator Algebras}, I, Springer-Verlag, 1979.
1265: % \bibitem{Y1}
1266: % S.~Yamagami, Algebraic aspects in modular theory,
1267: % {\it Publ.~RIMS}, 28(1992), 1075--1106.
1268: % \bibitem{Y2}
1269: % \bysame, Modular theory for bimodules, {\it J.~Funct.~Anal.},
1270: % 125(1994), 327--357.
1271: