1: \documentclass{article}
2: \usepackage{amsmath}[1996/11/01]
3: \usepackage{amssymb,amsthm,amsxtra}
4: \usepackage{epsfig}
5:
6:
7: \headheight 0cm
8: \headsep 0cm
9: \newlength{\mytopmargin}
10: \newlength{\myleftmargin}
11: \setlength{\mytopmargin}{2.8cm}
12: \setlength{\myleftmargin}{2.5cm}
13: \setlength{\topmargin}{-1in}
14: \setlength{\oddsidemargin}{-1in}
15: \addtolength{\topmargin}{\mytopmargin}
16: \addtolength{\oddsidemargin}{\myleftmargin}
17: \textwidth 16cm
18: \textheight 24cm
19: \setlength{\parskip}{1ex}
20:
21: \def\zz{\rlx\hbox{\small \sf Z\kern-.4em Z}}
22:
23:
24: \newcommand{\ml}{\langle}
25: \newcommand{\mg}{\rangle}
26: \newtheorem{lemma}{Lemma}[section]
27:
28: \newtheorem{prop}[lemma]{Proposition}
29:
30:
31: \setlength{\parindent}{1.5em}
32: \renewcommand{\baselinestretch}{1.2}
33: \renewcommand{\theequation}{\thesection.\arabic{equation}}
34:
35: \begin{document}
36:
37: \vspace{1cm}
38: \noindent
39: \begin{center}{ \large \bf
40: Pieri-type formulas for the non-symmetric Jack polynomials}
41: \end{center}
42: \vspace{5mm}
43:
44: \noindent
45: \begin{center}
46: P.J.~Forrester and D.S.~McAnally \\
47:
48: \it Department of Mathematics and Statistics, \\
49: University of Melbourne, Parkville, Victoria
50: 3052, Australia
51: \end{center}
52: \vspace{.5cm}
53: \small
54: \begin{quote}
55: In the theory of symmetric Jack polynomials the coefficients in the
56: expansion of the $p$th elementary symmetric function $e_p(z)$ times
57: a Jack polynomial expressed as a series in Jack polynomials are known
58: explicitly. Here analogues of this result for the non-symmetric Jack
59: polynomials $E_\eta(z)$ are explored. Necessary conditions for non-zero
60: coefficients
61: in the expansion of $e_p(z) E_\eta(z)$ as a series in non-symmetric
62: Jack polynomials are given.
63: A known expansion formula for $z_i E_\eta(z)$
64: is rederived by an induction procedure, and this expansion is used to
65: deduce the corresponding result for the expansion of
66: $\prod_{j=1, \, j\ne i}^N z_j \, E_\eta(z)$, and consequently
67: the expansion of $e_{N-1}(z) E_\eta(z)$. In the general $p$ case
68: the coefficients for special terms in the expansion are presented.
69: \end{quote}
70:
71: \vspace{.5cm}
72: \noindent
73: \section{Introduction}
74: \setcounter{equation}{0}
75: Jack polynomials and Macdonald polynomials can be defined as homogeneous
76: multivariable orthogonal polynomials, or as eigenfunctions of a family of
77: commuting differential or difference operators respectively. From the latter
78: viewpoint these polynomials occur in the study of certain quantum many
79: body systems \cite{Fo94j,Ko96}. In their most basic form the polynomials are
80: non-symmetric, although eigenfunctions with a prescribed symmetry with
81: respect to interchange of coordinates are often required in application
82: \cite{BF97b}. The polynomials with a prescribed symmetry can be obtained
83: from the non-symmetric polynomials by an appropriate symmetry operation.
84: One consequence of this feature is that many properties of the symmetric
85: Jack and Macdonald polynomials can be obtained from the corresponding
86: properties of the non-symmetric polynomials \cite{BF99,Ma99}.
87:
88: There are, however, a number of properties of the symmetric Jack and
89: Macdonald polynomials which have no known relation to properties of the
90: non-symmetric polynomials. One example is the so-called Pieri formula
91: \cite{St89,Ma95,Ka97g}. To present this formula requires some notation. Let
92: $\kappa$ and $\lambda$ be partitions described by their diagrams and
93: suppose $\kappa \subset \lambda$. A skew diagram $\lambda/\kappa$ is said
94: to be a vertical $m$-strip if it consists of $m$ boxes, all of which
95: are in distinct rows. For $\lambda/\kappa$ a vertical $m$-strip define
96: $\chi_m$ by $\lambda = \kappa + \chi_m$, and put
97: $$
98: U^{(\alpha)}(\lambda/\kappa) := \frac{f_N^1(\alpha \kappa + \chi_m)
99: f_N^{1/\alpha}(\kappa)}{f_N^1(\alpha \kappa) f_N^{1/\alpha}
100: (\kappa + \chi_m)}
101: $$
102: where
103: $$
104: f_n^r(\kappa) = \prod_{1 \le i < j \le n}
105: {((j-i)r + \kappa_i - \kappa_j)_r \over ((j-i)r)_r}, \qquad
106: (u)_r := {\Gamma (u+r) \over \Gamma (u)}.
107: $$
108: With this notation the Pieri formula reads
109: \begin{equation}\label{pf1}
110: e_p(z) P_\kappa(z) = \sum_{\lambda \atop \lambda/\kappa \: {\rm
111: a \: vertical} \: m{\rm -strip}} U^{(\alpha)}(\lambda/\kappa)
112: P_\lambda(z)
113: \end{equation}
114: where
115: $$
116: e_p(z) := \sum_{1 \le i_1 < \cdots < i_p \le N}
117: z_{i_1} \cdots z_{i_p}
118: $$
119: denotes the $p$th elementary symmetric function, and $P_\kappa(x) :=
120: P_\kappa(x;\alpha)$ denotes the symmetric Jack polynomial indexed by
121: the partition $\kappa$ and normalized so that
122: when expanded in terms of monomial symmetric functions
123: the coefficient of the
124: monomial symmetric function $m_\kappa$ is unity.
125:
126: It is the objective of this paper to investigate non-symmetric
127: analogues of the Pieri formula (\ref{pf1}). Our original idea was to
128: adapt the method used by Knop and Sahi \cite{KS96} to derive (\ref{pf1}),
129: which involves the theory of the so-called shifted Jack polynomials.
130: This was passed on to D.~Marshall, who subsequently \cite{Ma00} obtained
131: the explicit form of the coefficients in the expansions
132: \begin{eqnarray}
133: z_i E_\eta(z) & = & \sum_{\nu: |\nu| = |\eta| + 1}
134: c_{\eta \nu}^{(i)} E_\nu(z) \label{m1}\\
135: \Big ( \sum_{i=1}^N z_i \Big ) E_\eta(z) & = & \sum_{\nu: |\nu| = |\eta| + 1}
136: C_{\eta \nu} E_\nu(z) . \label{m2}
137: \end{eqnarray}
138: In this work we will give an inductive proof of the evaluation of the
139: $c_{\eta \nu}^{(i)}$
140: which avoids all reference to the theory of the shifted Jack polynomials
141: (the evaluation of the $C_{\eta \nu}$ follows as a simple corollary
142: from knowledge of the $c_{\eta \nu}^{(i)}$).
143:
144: In Section 3 of the paper we present necessary conditions on $\nu$ for the
145: coefficients in the expansion
146: \begin{equation}\label{m3}
147: z_{i_1} \cdots z_{i_p} E_\eta(z) =
148: \sum_{\nu: |\nu| = |\eta| + p} c_{\eta \nu}^{(i_1, \dots, i_p)} E_\nu(z)
149: \end{equation}
150: to be non-zero. Here use is made of the theory of shifted Jack polynomials.
151: In Section 4 the result of Marshall for the explicit
152: value of $c_{\eta \nu}^{(i)}$ is revised, and in Section 5 we present our
153: inductive proof of this result. The expansion (\ref{m3}) in the case
154: $p = N-1$, where $N$ is the number of variables $z:= (z_1,\dots,z_N)$,
155: is given in Section 6. In the final section, Section 7, a coefficient
156: in the expansion of $e_p(z) E_\eta(z)$ as a series in
157: $\{E_\nu\}$ is evaluated for a special value of $\nu$ and the form
158: of the evaluation further explored for a larger class of $\nu$.
159:
160:
161: \section{The non-symmetric Jack polynomials}
162: \setcounter{equation}{0}
163: The non-symmetric Jack polynomials $E_\eta(z)$ can be specified as
164: the simultaneous polynomial eigenfunctions of the commuting operators
165: $$
166: \xi_i := \alpha z_i {\partial \over \partial z_i} +
167: \sum_{p<i} {z_i \over z_i - z_p} (1 - s_{ip}) +
168: \sum_{p > i} {z_p \over z_i - z_p} (1 - s_{ip}) + 1 - i,
169: $$
170: where $s_{ip}$ is the operator which permutes $z_i$ and $z_p$,
171: satisfying the eigenvalue equations
172: \begin{equation}\label{E1}
173: \xi_i E_\eta = \bar{\eta}_i E_\eta, \quad (i=1,\dots,N)
174: \end{equation}
175: and with coefficient of $z^\eta = z^{\eta_1} \cdots z^{\eta_N}$ unity.
176: For a given composition $\eta := (\eta_1,\dots, \eta_N)$, the eigenvalue
177: $\bar{\eta}_i$ in (\ref{E1}) is given by
178: \begin{equation}\label{ac1}
179: \bar{\eta}_i := \alpha \eta_i - \# \{ k < i | \eta_k \ge \eta_i \}
180: - \# \{ k > i| \eta_k > \eta_i \}.
181: \end{equation}
182:
183: An alternative characterization of the non-symmetric Jack polynomials is
184: as multivariable orthogonal polynomials.
185: With $z_j := e^{2 \pi i x_j}$, introduce the inner product
186: \begin{equation}\label{fg}
187: \ml f | g \mg := \int_0^1 dx_1 \cdots \int_0^1 dx_N \,
188: \prod_{1 \le j < k \le N} |z_k - z_j |^{2/\alpha}
189: f^*(z_1,\dots,z_N) g(z_1,\dots,z_N),
190: \end{equation}
191: where the $*$ denotes complex conjugation.
192: Suppose $|\eta| = |\nu|$ for compositions $\eta \ne \nu$.
193: Introduce the dominance partial ordering $<$ on
194: compositions by the statement that $\nu < \eta$ if
195: $\sum_{j=1}^p \nu_j < \sum_{j=1}^p \eta_j $
196: for each $p=1,\dots,N$.
197: Let $\eta^+$ denote the partition corresponding to the composition
198: $\eta$. Introduce a
199: further partial ordering $\triangleleft$ by the statement that
200: $\nu \triangleleft \eta$ if $\nu^+ < \eta^+$, or in the case $\nu^+ = \eta^+$,
201: if $\nu < \eta$. Then for a given value of $|\eta|$, the $E_\eta$ can
202: be constructed via a Gram-Schmidt procedure from the requirements that
203: \begin{equation}\label{fg1}
204: \ml E_\eta | E_\nu \mg = 0,
205: \end{equation}
206: for $\eta \ne \nu$, and that
207: \begin{equation}\label{2.8'}
208: E_\eta(z) = z^\eta + \sum_{\nu \triangleleft \eta} c_{\eta \nu } z^\nu.
209: \end{equation}
210:
211: We will have future use for the explicit value of
212: $$
213: {\cal N}_\eta := \ml E_\eta | E_\eta \mg.
214: $$
215: This requires the introduction of further quantities for its presentation.
216: Following \cite{Sa96}, define the arm and leg lengths at the node
217: $(i,j)$ of the diagram of a composition $\eta$ by
218: \begin{equation}\label{leg}
219: a(i,j) = \eta_i - j, \qquad
220: l(i,j) = \# \{ k < i | j \le \eta_k + 1 \le \eta_i \} +
221: \# \{ k > i | j \le \eta_k \le \eta_i \}
222: \end{equation}
223: and put
224: \begin{equation}\label{d1}
225: d_\eta' := \prod_{(i,j) \in \eta} \Big ( \alpha( a(i,j) + 1 ) +
226: l (i,j) \Big ), \qquad
227: d_\eta := \prod_{(i,j) \in \eta} \Big ( \alpha( a(i,j) + 1 ) +
228: l (i,j) +1 \Big ).
229: \end{equation}
230: Also, define the generalized factorial by
231: $$
232: [u]_{\eta^+}^{(\alpha)} =
233: \prod_{j=1}^N {\Gamma(u - (j-1)/\alpha + \eta_j^+) \over
234: \Gamma(u - (j-1)/\alpha)}
235: $$ and put
236: \begin{equation}\label{e1}
237: e_\eta = \alpha^{|\eta|} [1 + N/\alpha]_{\eta^+}^{(\alpha)}, \qquad
238: e_\eta' = \alpha^{|\eta|} [1 + (N-1)/\alpha]_{\eta^+}^{(\alpha)}.
239: \end{equation}
240: In terms of the quantities (\ref{d1}) and (\ref{e1}) we have
241: \cite{Op95,BF99}
242: \begin{equation}\label{en}
243: {{\cal N}_\eta \over {\cal N}_{(0^N)}} =
244: {d_\eta' e_\eta \over d_\eta e_\eta'}.
245: \end{equation}
246:
247: Starting with $E_{(0^N)}(z) = 1$, the non-symmetric Jack polynomials can be
248: recursively generated from the action of just two fundamental operators.
249: The first of these operators is the elementary permutation operator
250: $s_i := s_{i \, i+1}$, which permutes $z_i$ and $z_{i+1}$. It has the
251: action \cite{Op95}
252: \begin{equation}\label{u1}
253: s_i E_\eta(z) = \left\{ \begin{array}{ll}
254: {1 \over \bar{\delta}_{i,\eta}} E_\eta(z) + \Big ( 1 - {1 \over
255: \bar{\delta}_{i,\eta}^2} \Big ) E_{s_i \eta}(z), & \eta_i > \eta_{i+1} \\[.2cm]
256: E_\eta(z), & \eta_i = \eta_{i+1} \\[.2cm]
257: {1 \over \bar{\delta}_{i,\eta}} E_\eta(z) +
258: E_{s_i \eta}(z), & \eta_i < \eta_{i+1}
259: \end{array} \right.
260: \end{equation}
261: where
262: \begin{equation}\label{u1a}
263: \bar{\delta}_{i,\eta} := \bar{\eta}_i - \bar{\eta}_{i+1}.
264: \end{equation}
265: The second required operator is the raising type operator, defined when acting
266: on functions according to
267: $$
268: \Phi f(z_1,\dots, z_N) = z_N f(z_N,z_1,\dots, z_{N-1}),
269: $$
270: which has the property \cite{KS97}
271: \begin{equation}\label{u2}
272: \Phi E_\eta (z) = E_{\Phi \eta}(z), \qquad \Phi \eta := (\eta_2, \dots,
273: \eta_N, \eta_1 + 1).
274: \end{equation}
275: Starting from $\eta = (0^N)$, all compositions can be generated by the
276: action of $\Phi \eta$ and $s_i \eta$, so (\ref{u1}) and (\ref{u2})
277: provide the recursive generation of all the $E_\eta$.
278:
279: Future use will be made of the quantities (\ref{d1}) and (\ref{e1})
280: with $\eta$ replaced by $s_i \eta$ and $\Phi \eta$. In particular, we require
281: the formulas \cite{Sa96}
282: $$
283: e_{s_i \eta} = e_\eta, \quad e_{s_i \eta}' = e_\eta', \quad
284: {d_{s_i \eta} \over d_\eta} = \left \{
285: \begin{array}{ll}
286: {\bar{\delta}_{i,\eta} + 1 \over \bar{\delta}_{i,\eta}}, & \eta_i > \eta_{i+1}
287: \\[.2cm]
288: { \bar{\delta}_{i,\eta} \over
289: \bar{\delta}_{i,\eta} - 1 }, &
290: \eta_i < \eta_{i+1} \end{array} \quad
291: {d_{s_i \eta}' \over d_\eta'}\right. = \left \{
292: \begin{array}{ll}
293: {\bar{\delta}_{i,\eta} \over \bar{\delta}_{i,\eta}-1}, & \eta_i > \eta_{i+1}
294: \\[.2cm] { \bar{\delta}_{i,\eta}+1 \over
295: \bar{\delta}_{i,\eta}}, &
296: \eta_i < \eta_{i+1} \end{array} \right.
297: $$
298: \begin{equation}\label{15.red}
299: {d_{\Phi \eta} \over d_\eta} = {e_{\Phi \eta} \over e_\eta} =
300: \bar{\eta}_1 + \alpha + N, \qquad
301: {d_{\Phi \eta}' \over d_\eta'} = {e_{\Phi \eta}' \over e_\eta'}
302: = \bar{\eta}_1 + \alpha + N - 1.
303: \end{equation}
304:
305: Let us now revise some aspects of the theory of non-symmetric shifted
306: Jack polynomials $E_\eta^*$ \cite{Kn97}. The polynomial $E_\eta^*(z)$ is
307: the unique polynomial of degree $\le |\eta|$ with the property
308: $$
309: E_\eta^*(\bar{\rho}/\alpha) = 0, \qquad |\rho| \le |\eta|, \: \: \rho
310: \ne \eta
311: $$
312: and $E_\eta^*(\bar{\eta}/\alpha) \ne 0$ with coefficient of $z^\eta$
313: in its monomial expansion unity
314: ($\bar{\eta} := (\bar{\eta}_1,\dots,\bar{\eta}_N)$ where the $\bar{\eta}_j$
315: are specified by (\ref{ac1})). The non-symmetric Jack polynomial
316: $E_\eta$ is the leading homogeneous term of $E_\eta^*$ so that
317: \begin{equation}\label{lh}
318: E_\eta^*(z) = E_\eta(z) + {\rm lower \: degree \: terms}.
319: \end{equation}
320:
321: A fundamental property of the $E_\eta^*$ is the extra vanishing condition.
322: Introduce the partial ordering $\preceq$ on compositions by writing
323: $\nu \preceq \eta$ if there exisits a permutation $\pi$ such that
324: $\nu_i < \eta_{\pi(i)}$ for $i < \pi(i)$ and
325: $\nu_i \le \eta_{\pi(i)}$ for $i \ge \pi(i)$. Note that for $\nu$ and
326: $\eta$ partitions the statement $\nu \preceq \eta$ is equivalent to
327: $\nu \subseteq \eta$ (inclusion of diagrams) but for compositions,
328: although $\nu \subseteq \eta$ implies $\nu \preceq \eta$ (take $\pi$ to be the
329: identity), the converse is not true in general. The extra vanishing condition
330: states \cite{Kn97}
331: \begin{equation}\label{v0}
332: E_\eta^*(\bar{\nu}/\alpha) = 0 \quad {\rm for} \quad \eta \not\prec \eta.
333: \end{equation}
334:
335: \section{Structure of the Pieri type expansions for the non-symmetric
336: Jack polynomials}
337: \setcounter{equation}{0}
338: Our interest is in the coefficients $c_{\eta \nu}^{(i_1, \dots, i_p)}$
339: in the expansion (\ref{m3}). In this section we will use the theory of the
340: non-symmetric shifted Jack polynomials to present necessary conditions
341: for the coefficients to be non-zero.
342:
343: Now the extra vanishing condition (\ref{v0}) implies that any analytic
344: function vanishing on
345: $\{\bar{\rho}/\alpha: \eta \not\prec \rho \}$ can be written
346: in the form
347: \begin{equation}\label{v1}
348: f(z) = \sum_{\nu: \eta \preceq \nu} c_{\eta \nu} E_\nu^*(z).
349: \end{equation}
350: It follows from this that
351: \begin{equation}\label{v4}
352: z_{i_1} \cdots z_{i_p} E_\eta^*(z) = \sum_{\nu: \eta \preceq \nu
353: \atop |\nu| \le |\eta| + p} c_{\eta \nu}^{(i_1, \dots, i_p)}
354: E_\nu^*(z)
355: \end{equation}
356: for some coefficients $c_{\eta \nu}^{(i_1, \dots, i_p)}$. Taking the leading
357: homogeneous term on both sides using (\ref{lh}) gives
358: \begin{equation}\label{v4'}
359: z_{i_1} \cdots z_{i_p} E_\eta(z) = \sum_{\nu: \eta \preceq \nu
360: \atop |\nu| = |\eta| + p} c_{\eta \nu}^{(i_1, \dots, i_p)}
361: E_\nu(z)
362: \end{equation}
363: which is a refinement of (\ref{m3}).
364:
365: The statement (\ref{v4'}) can be further refined by making use of
366: the orthogonality (\ref{fg1}). Applying this orthogonality in (\ref{v4'})
367: shows that
368: \begin{equation}\label{f1}
369: c_{\eta \nu}^{(i_1, \dots, i_p)} =
370: {\ml E_\nu | z_{i_1} \cdots z_{i_p} E_\eta \mg \over \ml E_\nu |
371: E_\nu \mg}.
372: \end{equation}
373: Using the facts that
374: with $z^1 := z_1 \cdots z_N$ we have
375: $E_{\eta + (1^N)}(z) = z^1 E_\eta(z)$,
376: $\ml z^1 f| z^1 g \mg = \ml f|g \mg$ and $\ml f | g \mg = \ml g | f \mg$
377: (the latter provided $f$ and $g$ have real coefficients) it follows from
378: (\ref{f1}) that
379: \begin{equation}\label{f2}
380: c_{\eta \nu}^{(i_1, \dots, i_p)} =
381: {\ml z_{j_1} \cdots z_{j_{N-p}} E_\nu | E_{\eta+(1^N)} \mg \over
382: \ml E_\nu | E_\nu \mg} =
383: {\ml E_{\eta+(1^N)} | z_{j_1} \cdots z_{j_{N-p}} E_\nu \mg \over
384: \ml E_\nu | E_\nu \mg} =
385: c_{\nu \eta+(1^N)}^{(j_1,\dots, j_{N-p})} {\ml E_\eta | E_\eta \mg \over
386: \ml E_\nu | E_\nu \mg}
387: \end{equation}
388: where $j_1,\dots,j_{N-p}$ are such that $\{1,\dots,N\} =
389: \{i_1,\dots,i_p\} \cup \{j_1,\dots, j_{N-p} \}$. But
390: according to (\ref{v4'})
391: $c_{\nu \, \eta+(1^N)}^{(j_1,\dots, j_{N-p})} = 0$ for
392: $\nu \not\preceq \eta +(1^N)$
393: and thus (\ref{f2}) implies
394: \begin{equation}\label{f2'}
395: c_{\eta \nu}^{(i_1,\dots,i_p)} = 0 \quad {\rm for} \quad
396: \nu \not\preceq \eta + (1^N).
397: \end{equation}
398: Hence in (\ref{v4'}) we can make the additional restriction
399: $\nu \preceq \eta + (1^N)$, and so obtain
400: \begin{equation}\label{f4}
401: z_{i_1} \cdots z_{i_p} E_\eta(z) =
402: \sum_{\nu \in {\mathbb J}_{N,p}} c_{\eta \nu}^{(i_1,\dots,i_p)}
403: E_\nu(z)
404: \end{equation}
405: where
406: \begin{equation}\label{f3}
407: {\mathbb J}_{N,p} := \{ \nu: \eta \preceq \nu \preceq \eta + (1^N),
408: \: |\nu| = |\eta| + p \}.
409: \end{equation}
410: Note that by performing the sum $1 \le i_1 < \cdots < i_p \le N$ in
411: (\ref{f4}) we obtain
412: \begin{equation}\label{f4'}
413: e_p(z) E_\eta(z) = \sum_{\nu \in {\mathbb J}_{N,p}}
414: A_{\eta \nu}^{(p)} E_\nu(z)
415: \end{equation}
416: for some constants $A_{\eta \nu}^{(p)}$.
417:
418: Next we seek a more explicit description of the set ${\mathbb J}_{N,p}$.
419: Let $w_\eta$ be the shortest element of $S_N$ (the permutations of
420: $\{1,\dots,N\}$) such that $w_\eta^{-1}(\eta)$ is a partition and
421: similarly define $w_\nu$. It is straightforward to show \cite{Kn97} that
422: if $\nu \preceq \eta$ then the permutation $\pi$ in the definition
423: of the partial order can be represented $\pi = w_\nu \circ
424: w_\eta^{-1} =: \pi_{\nu,\eta}$. Now, members $\nu$ of the set
425: $\mathbb J_{N,p}$ require both $\eta \preceq
426: \nu$ and $\nu \preceq \eta + (1^N)$ with $|\nu| = |\eta| + p$.
427: For the former ordering constraint the relevant permutation is
428: $\pi_{\eta,\nu} = \pi_{\nu,\eta}^{-1}$. Replacing $\pi$ by
429: $\pi^{-1}$ in the definition of $\preceq$ shows we require
430: \begin{equation}\label{c1}
431: \eta_i < \nu_{\pi_{\nu,\eta}(i)} \quad {\rm for} \quad
432: i < \pi_{\nu,\eta}(i), \qquad
433: \eta_i \le \nu_{\pi_{\nu,\eta}(i)} \quad {\rm for} \quad
434: i \ge \pi_{\nu,\eta}(i).
435: \end{equation}
436: For the latter ordering constraint the relevant permutation is
437: $\pi_{\nu,\eta}$. Replacing $\pi$ by $\pi^{-1}$ and $i$ by $\pi(i)$ in the
438: definition of $\preceq$ shows we require
439: \begin{equation}\label{c2}
440: \nu_{\pi_{\nu,\eta}(i)}
441: < \eta_i + 1 \quad {\rm for} \quad
442: \pi_{\nu,\eta}(i) < i, \qquad
443: \nu_{\pi_{\nu,\eta}(i)} \le \eta_i + 1 \quad {\rm for} \quad
444: \pi_{\nu,\eta}(i) \ge i.
445: \end{equation}
446:
447: Combining (\ref{c1}) and (\ref{c2}) gives
448: $$
449: \eta_i < \nu_{\pi_{\nu,\eta}(i)} \le \eta_i + 1
450: \quad {\rm for} \quad i < \pi_{\nu,\eta}(i), \qquad
451: \eta_i \le \nu_{\pi_{\nu,\eta}(i)} < \eta_i + 1
452: \quad {\rm for} \quad i > \pi_{\nu,\eta}(i),
453: $$
454: and so
455: \begin{equation}\label{c3}
456: \nu_{\pi_{\nu,\eta}(i)} = \eta_i + 1 \quad {\rm for} \quad i < \pi_{\nu,\eta}(i), \qquad
457: \nu_{\pi_{\nu,\eta}(i)} = \eta_i \quad {\rm for} \quad i >
458: \pi_{\nu,\eta}(i).
459: \end{equation}
460: In the case $i = \pi_{\nu,\eta}(i)$ (\ref{c1}) and (\ref{c2}) give
461: $\eta_i \le \nu_{\pi_{\nu,\eta}(i)} \le \eta_i + 1$ and so
462: \begin{equation}\label{c4}
463: \nu_{\pi_{\nu,\eta}(i)} = \eta_i \quad {\rm or} \quad
464: \nu_{\pi_{\nu,\eta}(i)} = \eta_i + 1.
465: \end{equation}
466:
467: It remains to implement the requirement $|\nu| = |\eta| + p$. We see from
468: (\ref{c3}) and (\ref{c4}) that we must have
469: \begin{equation}\label{c5}
470: \nu_{\pi_{\nu,\eta}(i_r)} = \eta_{i_r} + 1 \qquad (r=1,\dots,p)
471: \end{equation}
472: for some $1 \le i_1 < \cdots < i_p \le N$ and
473: \begin{equation}\label{c6}
474: \nu_{\pi_{\nu,\eta}(j_r)} = \eta_{j_r} \qquad (r=1,\dots,N-p)
475: \end{equation}
476: where $\{i_1,\dots,i_p\} \cup \{j_1,\dots,j_{N-p} \} =
477: \{1,2,\dots, N\}$. Combining (\ref{c5}) and (\ref{c6}) with (\ref{c3})
478: and (\ref{c4}) shows compositions $\nu \in \mathbb J_{n,p}$ are
479: characterized by the properties
480: \begin{eqnarray}\label{c7}
481: \nu_{\pi(i_r)} = \eta_{i_r} + 1 \quad {\rm for} \quad
482: i_r \le \pi(i_r) \qquad r=1,\dots,p \nonumber \\
483: \nu_{\pi(j_r)} = \eta_{j_r} \quad {\rm for} \quad
484: j_r \ge \pi(j_r) \qquad r=1,\dots,N-p
485: \end{eqnarray}
486: for some permutation $\pi$ ($\pi = \pi_{\nu,\eta}$ suffices).
487: The characterization (\ref{c7}) can be interpreted in terms of the diagram
488: of $\eta$. We begin by adding one box to the rows $i_1,\dots,i_p$. Then
489: we consider all rearrangements of the rows such the rows with a box
490: added move downwards or stay stationary, while the rows with no box added
491: move upwards or stay stationary. An example is given in Figure
492: \ref{fig1.f1}.
493:
494: \vspace{.5cm}
495: \begin{figure}
496: \epsfxsize=10cm
497: \centerline{\epsfbox{david1.eps}}
498: \caption{\label{fig1.f1} Construction of the composition
499: $\nu: \, \eta \preceq \nu \preceq \eta + (1)^3$ with $\eta
500: = (2,0,1)$ and $|\nu| = |\eta| + 1$. The unshaded boxes originate from
501: the diagram of $\eta$. With reference to the original diagram of
502: $\eta$ the row with the additional box (shaded) must
503: move downwards or stay stationary, while the rows with no box added
504: move upwards or stay stationary. The labels in the description
505: (\ref{c8}) are also noted.}
506: \end{figure}
507:
508:
509: In the case $p=1$ the compositions $\nu$ defined by (\ref{c7}) and
510: thus belonging to the set $\mathbb J_{N,1}$ have the property of being
511: the minimal elements lying above $\eta$ \cite{Kn97}. Note that this
512: set can be indexed by subsets $I = \{t_1,\dots, t_s\}$ of
513: $\{1,\dots,N\}$ with $t_1 < \cdots < t_s$ which correspond to the element
514: \begin{equation}\label{c80}
515: \nu =: c_I(\eta) \in \mathbb J_{N,1}
516: \end{equation}
517: where
518: \begin{eqnarray}\label{c8}
519: \nu_{t_j} & = & \eta_{t_{j+1}} \qquad j=1,\dots, s-1 \nonumber \\
520: \nu_{t_s} & = & \eta_{t_1} + 1 \nonumber \\
521: \nu_i & = & \eta_i \qquad i \notin I
522: \end{eqnarray}
523: Furthermore, the subset $I$ is called maximal with respect to $\eta$ if
524: $I \neq \emptyset$ and
525: \begin{eqnarray}\label{c8a}
526: \eta_j & \ne & \eta_{t_u} \qquad j=t_{u-1} + 1, \dots, t_u - 1
527: \: \: (u=1,\dots,s; \, t_0 := 0) \nonumber \\
528: \eta_j & \ne & \eta_{t_1} + 1 \qquad j=t_s+1, \dots, N
529: \end{eqnarray}
530: It follows from (\ref{c8}) that an equivalent way to characterize the maximal
531: subsets is via the conditions
532: \begin{eqnarray}\label{c9}
533: \nu_j &\ne & \nu_{t_s} - 1, \qquad j=1,\dots, t_1 - 1 \nonumber \\
534: \nu_j & \ne & \nu_{t_u}, \qquad j=t_u+1, \dots, t_{u+1} - 1 \: \:
535: (u=1,\dots,s; \: t_{s+1} := N+1).
536: \end{eqnarray}
537: It is shown in \cite{Kn97} that it is only these maximal
538: subsets which give distinct compositions $\nu$ (we illustrate this point
539: in Figure \ref{fig1.f2}). Thus we can write
540: \begin{equation}\label{c8b}
541: \mathbb J_{N,1} :=
542: \mathbb J_{N,1}[\eta] =
543: \{ \nu: \, \nu = c_I(\eta), \: I \: {\rm maximal} \}.
544: \end{equation}
545: It is also convenient to introduce the set $\mathbb J_\eta$ of maximal
546: subsets
547: \begin{equation}\label{c8c}
548: \mathbb J_\eta = \{I: \, I \: {\rm is \: maximal \: w.r.t. \: } \eta \},
549: \end{equation}
550: so that $\mathbb J_{N,1} = \{c_I(\eta): \, I \in \mathbb J_\eta\}$.
551:
552: \vspace{.5cm}
553: \begin{figure}
554: \epsfxsize=6cm
555: \centerline{\epsfbox{david2.eps}}
556: \caption{\label{fig1.f2} In this example, starting with
557: $\eta = (0,2,2)$, two different choices of subsets
558: $I = \{t_1,\dots, t_s\}$ give the same composition, but only the second
559: subset is maximal (note that in the first diagram $\eta_2 = \eta_{t_1}$).
560: }
561: \end{figure}
562:
563:
564: \section{A Pieri type formula for the non-symmetric Jack polynomials
565: in the case $p=1$}
566: \setcounter{equation}{0}
567: So far the theory of shifted Jack polynomials \cite{Kn97} has been used
568: to deduce the structural formula (\ref{f4}), and also notions from
569: that theory are used to label the set $\mathbb J_{N,p}$ appearing
570: in (\ref{f4}) in terms of certain maximal subsets $I$. To now evaluate
571: the coefficients in (\ref{f4}), the most natural way to proceed is to
572: make further use of theory from \cite{Kn97}. In the case $p=1$
573: this part of the program
574: has recently been successfully undertaken by Marshall \cite{Ma00}.
575: The presentation of the result requires some notation.
576:
577: First write
578: \begin{equation}
579: a(x,y) := {1 \over \alpha (x-y)}, \qquad
580: b(x,y) := {x-y-1/\alpha \over x - y}.
581: \end{equation}
582: For $I = \{t_1,\dots,t_s\} \subseteq \{1,\dots,N\}$, $I \ne \emptyset$,
583: $t_1 < \cdots < t_s$ put
584: \begin{eqnarray}
585: A_I(x) &:=& \Big ( \prod_{u=1}^{s-1} a(x_{t_u}, x_{t_{u+1}}) \Big )
586: a(x_{t_s} - 1, x_{t_1}) \label{a2} \\
587: B_I(x) &:=& \Big ( \prod_{u=1}^{s} \prod_{j=t_u+1}^{t_{u+1}-1}
588: b(x_{t_u},x_j) \Big )
589: (x_{t_s} + (N-1)/\alpha) \prod_{j=1}^{t_1 - 1}
590: b(x_{t_s} -1, x_j), \quad t_{s+1} := N + 1 \label{a3} \\
591: \tilde{B}_I(x) &:=&
592: \Big ( \prod_{u=1}^{s} \prod_{j=t_{u-1}+1}^{t_{u}-1}
593: b(x_{t_u},x_j) \Big ) \Big ( \prod_{j=t_s +1}^N b(x_{t_1}+1,x_j) \Big )
594: (x_{t_1} +1 + (N-1)/\alpha) , \quad t_0 := 0
595: \label{a3'}
596: \end{eqnarray}
597: and for $i \in I$ write
598: \begin{eqnarray}
599: \chi_I^{(i)}(x) & = & \left \{ \begin{array}{ll}
600: \alpha(x_{t_{k-1}} - x_i), & i=t_k \: \: (k=2,\dots,s) \\
601: \alpha(x_{t_s} - x_i - 1), & i=t_1 \end{array} \right.
602: \nonumber \label{a3a}\\
603: \tilde{\chi}_I^{(i)}(x) &=& \left \{
604: \begin{array}{ll} \alpha (x_i - x_{t_{k+1}}), & i=t_k \: \:
605: (k=1,\dots,s-1) \\
606: \alpha (x_i - x_{t_1} - 1), & i = t_s. \end{array} \right. \label{a3b}
607: \end{eqnarray}
608:
609:
610: In terms of
611: these quantities,
612: and the quantity
613: $d_\eta'$ of (\ref{d1}), the result of Marshall \cite{Ma00} reads
614: \begin{equation}\label{fu1}
615: z_i E_\eta(z) = \alpha d_\eta' \sum_{I \in \mathbb J_{\eta} \atop
616: {\rm given} \, i \in I}
617: {\chi_I^{(i)}(\overline{c_I(\eta)}/\alpha)
618: A_I(\overline{c_I(\eta)}/\alpha)
619: B_I(\overline{c_I(\eta)}/\alpha)
620: \over d_{c_I(\eta)}' }
621: E_{c_I(\eta)}(z).
622: \end{equation}
623: Also, noting from (\ref{a3b}) that
624: \begin{equation}\label{4.47'}
625: \sum_{i \in I} \chi_I^{(i)}(x) = - \alpha
626: \end{equation}
627: it follows from (\ref{fu1}) that \cite{Ma00}
628: \begin{equation}\label{fu2}
629: \Big ( \sum_{i=1}^N z_i \Big ) E_\eta (z) = - \alpha^2
630: d_\eta'
631: \sum_{I \in \mathbb J_\eta}
632: {A_I((\overline{c_I(\eta)}/\alpha) B_I((\overline{c_I(\eta)}/\alpha) \over
633: d_{c_I(\eta)}' } E_{c_I(\eta)}(z).
634: \end{equation}
635: We remark that it follows from the definition (\ref{a3}) of $B_I(x)$ that
636: for $I$ not maximal (i.e.~cases for which the relations (\ref{c9}) are
637: not obeyed), $B_I(\overline{c_I(\eta)}/\alpha) = 0$. Thus the restriction to
638: maximal subsets in the summation of (\ref{fu1}) and (\ref{fu2}) is in
639: fact a feature of the analytic form of the coefficients.
640:
641: The dependence on $\overline{c_I(\eta)}$ in (\ref{fu1}) and (\ref{fu2}) can
642: be replaced by a dependence on $\bar{\eta}$. Thus we note from the
643: definitions (\ref{a2}), (\ref{a3'}) and (\ref{a3b}) that
644: $$
645: A_I(\overline{c_I(\eta)}/\alpha) = A_I(\bar{\eta}/\alpha), \quad
646: B_I(\overline{c_I(\eta)}/\alpha) = \tilde{B}_I(\bar{\eta}/\alpha), \quad
647: \chi_I^{(i)}(\overline{c_I(\eta)}/\alpha) =
648: \tilde{\chi}_I^{(i)}(\bar{\eta}/\alpha) \: \: (i \in I),
649: $$
650: which when substituted in (\ref{fu1}) and (\ref{fu2})
651: give
652: \begin{eqnarray}
653: z_i E_\eta(z) & = & \alpha d_\eta' \sum_{I \in \mathbb J_{\eta} \atop
654: {\rm given} \, i \in I}
655: {\tilde{\chi}_I^{(i)}(\bar{\eta}/\alpha)
656: A_I(\bar{\eta}/\alpha) \tilde{B}_I(\bar{\eta}/\alpha) \over d_{c_I(\eta)}' }
657: E_{c_I(\eta)}(z) \label{fz1} \\
658: \Big ( \sum_{i=1}^N z_i \Big ) E_\eta (z) & = & - \alpha^2
659: d_\eta'
660: \sum_{I \in \mathbb J_\eta}
661: {A_I(\bar{\eta}/\alpha) \tilde{B}_I(\bar{\eta}/\alpha) \over d_{c_I(\eta)}' }
662: E_{c_I(\eta)}(z). \label{fz2}
663: \end{eqnarray}
664: A still more useful form of (\ref{fz1}) results by introducing
665: \begin{equation}\label{a3'h}
666: \hat{B}_I(\bar{\eta}/\alpha) := \alpha {e_\eta' \over
667: e_{c_I(\eta)}'} \tilde{B}_I(\bar{\eta}/\alpha)
668: = \Big ( \prod_{u=1}^{s} \prod_{j=t_{u-1}+1}^{t_{u}-1}
669: b(x_{t_u},x_j) \Big ) \Big ( \prod_{j=t_s+1}^Nb(x_{t_s}+1,x_j) \Big ),
670: \quad t_0 := 0
671: \end{equation}
672: where the equality follows from (\ref{e1}) and (\ref{a3'}).
673: In terms of this quantity (\ref{fz1}) reads
674: \begin{equation}\label{obt}
675: z_i E_\eta(z) = {d_\eta' \over e_\eta'}
676: \sum_{I \in \mathbb J_{\eta} \atop
677: {\rm given} \, i \in I}
678: {e_{c_I(\eta)}'\tilde{\chi}_I^{(i)}(\bar{\eta}/\alpha)
679: A_I(\bar{\eta}/\alpha) \hat{B}_I(\bar{\eta}/\alpha) \over d_{c_I(\eta)}' }
680: E_{c_I(\eta)}(z).
681: \end{equation}
682:
683: \section{Inductive proof}
684: \setcounter{equation}{0}
685: In this section we will provide an inductive proof of (\ref{obt}). This
686: has the advantage of being independent of the theory of the shifted
687: Jack polynomials, relying only on the recurrence properties
688: (\ref{u1}) and (\ref{u2}) of the non-symmetric Jack polynomials
689: themselves.
690:
691: \subsection*{Strategy}
692: It has already been remarked that starting with $E_{(0^N)}(z) = 1$,
693: the non-symmetric Jack polynomials can be generated recursively from
694: the recurrence properties (\ref{u1}) and (\ref{u2}). To make use of
695: these properties, suppose for a given $\eta$ we know the coefficients
696: $c_{\eta, \nu}^{(j)}$ in the expansion
697: \begin{equation}\label{k1}
698: z_j E_\eta(z) = \sum_{\nu \in \mathbb J_{N,1}[\eta]}
699: c_{\eta, \nu}^{(j)} E_\nu(z)
700: \end{equation}
701: for each $j=1,\dots,N$. Then, with $z_{N+1} := z_1$ and
702: $c_{\eta, \nu}^{(N+1)} := c_{\eta, \nu}^{(1)}$, (\ref{u2}) gives
703: \begin{equation}\label{k2}
704: z_j E_{\Phi \eta}(z) = z_j \Phi E_\eta(z) = \Phi(z_{j+1}
705: E_\eta(z) ) =
706: \sum_{\nu \in \mathbb J_{N,1}[\eta]} c_{\eta, \nu}^{(j+1)}
707: E_{\Phi \nu}(z).
708: \end{equation}
709: This shows $z_j E_{\Phi \eta}(z)$ can be computed from knowledge of
710: the expansion (\ref{k1}) for the given $\eta$. Moreover, we can can
711: give an explicit relationship between coefficients.
712: To demonstrate this,
713: for $I \subseteq \{1,\dots,N\}$, $I \ne \emptyset$ put
714: \begin{equation}\label{thi}
715: \Phi (I) := \{ j-1 | j \in I \cap \{2,\dots,N\} \} \cup
716: \{N | 1 \in I \}.
717: \end{equation}
718: Then we can check that $\Phi(I)$ is maximal with respect to $\Phi \eta$ if
719: and only if $I$ is maximal with respect to $\eta$. This means that in
720: (\ref{k2}) we can replace the summation $\nu \in \mathbb J_{N,1}[\eta]$
721: by $\Phi \nu \in \mathbb J_{N,1}[\Phi \eta]$, which allows us
722: to change variables
723: $\Phi \nu \mapsto \nu$ to obtain
724: \begin{equation}\label{k3}
725: z_j E_{\Phi \eta}(z) =
726: \sum_{\nu \in \mathbb J_{N,1}[\Phi \eta]} c_{\eta, \Phi^{-1}\nu}^{(j+1)}
727: E_{\Phi \nu}(z).
728: \end{equation}
729: On the other hand (\ref{k1}) gives
730: \begin{equation}\label{k4}
731: z_j E_{\Phi \eta}(z) =
732: \sum_{\nu \in \mathbb J_{N,1}[\Phi \eta]} c_{\Phi \eta,\nu}^{(j)}
733: E_{\Phi \nu}(z).
734: \end{equation}
735: Comparing (\ref{k3}) and (\ref{k4}) shows we require
736: \begin{equation}\label{k5}
737: c_{\Phi \eta,\Phi\nu}^{(j)} = c_{\eta, \nu}^{(j+1)}, \quad
738: (j=1,\dots, N-1) \qquad
739: c_{\Phi \eta,\Phi\nu}^{(N)} = c_{\eta, \nu}^{(1)}.
740: \end{equation}
741:
742: Let us now consider the computation of $z_j E_{s_i \eta}(z)$ for
743: $\eta_i < \eta_{i+1}$ from knowledge of the expansion (\ref{k1}) for
744: the given $\eta$. For this purpose we rewrite (\ref{k1}) as
745: \begin{equation}\label{k1a}
746: z_j E_\eta(z) = \sum_\nu \alpha_{\eta,\nu}^{(j)} E_\nu(z)
747: \end{equation}
748: where $\alpha_{\eta,\nu}^{(j)} = c_{\eta,\nu}^{(j)}$,
749: $\nu \in \mathbb J_{N,1}[\eta]$ and $\alpha_{\eta,\nu}^{(j)} = 0$
750: otherwise. By doing this the sum over $\nu$ in (\ref{k1a}) is
751: unrestricted. Since from (\ref{u1}), with $\eta_i < \eta_{i+1}$,
752: \begin{eqnarray}\label{s1}
753: z_j E_{s_i \eta}(z)
754: & = & z_j \Big ( s_i E_\eta(z) - {1 \over \bar{\delta}_{i,
755: \eta}} E_\eta(z) \Big ) \nonumber \\
756: &=& \left \{ \begin{array}{ll}
757: s_i(z_j E_\eta(z)), & j \ne i,i+1 \\
758: s_i(z_i E_\eta(z)), & j = i+1 \\
759: s_i(z_{i+1} E_\eta(z)), & j = i \end{array} \right. -
760: {1 \over \bar{\delta}_{i,\eta}} z_j E_\eta(z)
761: \end{eqnarray}
762: we see that knowledge of $z_j E_\eta(z)$ for each $j=1,\dots,N$ implies
763: the value of $z_j E_{s_i \eta}(z)$. We want to exhibit this feature as
764: a recurrence for the coefficients $\alpha_{\eta,\nu}^{(j)}$. Now, from
765: (\ref{k1a}) and (\ref{u1})
766: \begin{eqnarray}
767: s_i(z_j E_\eta(z)) & = & \sum_{\nu_i < \nu_{i+1}}
768: \{ \alpha_{\eta,\nu}^{(j)} \bar{\delta}_{i,\nu}^{-1} +
769: \alpha_{\eta,s_i\nu}^{(j)}(1 - \bar{\delta}_{i,\nu}^{-2}) \} E_\nu(z) +
770: \sum_{\nu_i = \nu_{i+1}} \alpha_{\eta,\nu}^{(j)} E_\nu(z) \nonumber \\
771: && + \sum_{\nu_i > \nu_{i+1}}
772: \{ \alpha_{\eta,\nu}^{(j)} \bar{\delta}_{i,\nu}^{-1} +
773: \alpha_{\eta,s_i\nu}^{(j)} \} E_\nu(z) \label{s2}
774: \end{eqnarray}
775: while (\ref{k1a}) itself gives
776: \begin{equation}\label{s3}
777: z_j E_{s_i \eta}(z) = \sum_{\nu} \alpha_{s_i \eta, \nu}^{(j)}
778: E_\nu(z).
779: \end{equation}
780: Substituting (\ref{s3}), (\ref{s2}) and (\ref{k1a}) in (\ref{s1}) and
781: equating coefficients of $E_\nu(z)$ gives a recurrence allowing
782: $\alpha_{\eta,s_i\nu}^{(j)}$ to be computed. In the recurrence it is
783: necessary to distinguish the cases $\nu_i < \nu_{i+1}$ from
784: $\nu_i > \nu_{i+1}$. However this can be avoided if we write the
785: recurrence in terms of the
786: quantity
787: $$
788: \tilde{\alpha}_{\eta,\nu}^{(j)} := {d_\nu' e_\eta' \over d_\eta' e_\nu'}
789: \alpha_{\eta,\nu}^{(j)}
790: $$
791: and make use of (\ref{15.red}). We then find
792: for $\nu_i \ne \nu_{i+1}$
793: \begin{eqnarray}
794: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, \nu}^{(j)} & = &
795: (1 - \bar{\delta}_{i,\nu}^{-1} ) \tilde{\alpha}_{\eta,s_i\nu}^{(j)}
796: + (\bar{\delta}_{i,\nu}^{-1} - \bar{\delta}_{i,\eta}^{-1})
797: \tilde{\alpha}_{\eta,\nu}^{(j)} \quad (j \ne i,i+1 )
798: \nonumber \\
799: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, \nu}^{(i+1)} & = &
800: (1 - \bar{\delta}_{i,\nu}^{-1} ) \tilde{\alpha}_{\eta,s_i\nu}^{(i)} +
801: \bar{\delta}_{i,\nu}^{-1}\tilde{\alpha}_{\eta,\nu}^{(i)} -
802: \bar{\delta}_{i,\eta}^{-1} \tilde{\alpha}_{\eta,\nu}^{(i+1)} \nonumber \\
803: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, \nu}^{(i)} & = &
804: (1 - \bar{\delta}_{i,\nu}^{-1} ) \tilde{\alpha}_{\eta,s_i\nu}^{(i+1)} +
805: \bar{\delta}_{i,\nu}^{-1}\tilde{\alpha}_{\eta,\nu}^{(i+1)} -
806: \bar{\delta}_{i,\eta}^{-1} \tilde{\alpha}_{\eta,\nu}^{(i)} \label{s4}
807: \end{eqnarray}
808: while for $\nu_i = \nu_{i+1}$
809: \begin{eqnarray}
810: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, \nu}^{(j)} & = &
811: (1 - \bar{\delta}_{i,\eta}^{-1} ) \tilde{\alpha}_{\eta,\nu}^{(j)}
812: \quad (j \ne i,i+1 )
813: \nonumber \\
814: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, \nu}^{(i+1)} & = &
815: \tilde{\alpha}_{\eta,\nu}^{(i)} - \bar{\delta}_{i,\eta}^{-1}
816: \tilde{\alpha}_{\eta, \nu}^{(i+1)} \nonumber \\
817: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, \nu}^{(i)} & = &
818: \tilde{\alpha}_{\eta,\nu}^{(i+1)} - \bar{\delta}_{i,\eta}^{-1}
819: \tilde{\alpha}_{\eta, \nu}^{(i)}. \label{s5}
820: \end{eqnarray}
821: Noting that for $\nu_i = \nu_{i+1}$ we have $s_i \nu = \nu$, we see that
822: the equations (\ref{s4}) remain valid in that they reduce to
823: the equations (\ref{s5}),
824: so it suffices to consider (\ref{s4}) for all $\nu$.
825:
826: Starting from knowledge of $c_{(0^N),\nu}^{(j)}$ and $\tilde{\alpha}_{(0)^N,
827: \nu}^{(j)}$ the recurrences (\ref{k5}) and (\ref{s4}) can be
828: used to compute all the $c_{\eta,\nu}^{(j)}$ and $\tilde{\alpha}_{\eta,
829: \nu}^{(j)}$. Thus, after independently establishing their validity in
830: the case $\eta = (0^N)$, we want to show the functional forms
831: \begin{eqnarray}
832: \tilde{\alpha}_{\eta,\nu}^{(j)} & = & \left \{\begin{array}{ll}
833: \tilde{\chi}_I^{(j)}(\bar{\eta}/\alpha)
834: A_I(\bar{\eta}/\alpha) \hat{B}_I(\bar{\eta}/\alpha), &
835: I \in \mathbb J_\eta \: \: {\rm and} \: \: j \in I \: \: {\rm with}
836: \: \: \nu = c_I(\eta) \\
837: 0, & {\rm otherwise}
838: \end{array} \right. \label{s5a} \\
839: c_{\eta, \nu}^{(j)} & = & {d_\eta' e_\nu' \over d_\nu' e_\eta'}
840: \tilde{\chi}_I^{(j)}(\bar{\eta}/\alpha)
841: A_I(\bar{\eta}/\alpha) \hat{B}_I(\bar{\eta}/\alpha), \qquad
842: \nu = c_I(\eta), \label{s5b'}
843: \end{eqnarray}
844: with $\tilde{\chi}^{(j)}$, $A_I$ and $\hat{B}_I$ as specified by
845: (\ref{a3b}), (\ref{a2}) and (\ref{a3'h}) respectively, satisfy the
846: recurrences (\ref{k5}) and (\ref{s4}) as appropriate.
847:
848: \subsection*{Verification of the initial conditions}
849: We require the expansion of $z_i$ in terms of $\{E_\nu\}$. From the
850: recurrences (\ref{k5}) and (\ref{s4}) we can readily show
851: $$
852: E_{(0^k 1 0^{N-k-1})}(z) = z_{k+1} + {1 \over \alpha + k + 1}
853: (z_{k+2} + \cdots + z_N).
854: $$
855: Thus the expansion of $\{ E_{(0^k 1 0^{N-k-1})} \}$, $k=0,\dots, N-1$,
856: in terms of $\{z_i\}$ has a triangular structure. This makes the task of
857: inverting the formulas straightforward, provided we start with $z_N$
858: and then compute the expansion of $z_{N-1}$ etc.. We find
859: \begin{equation}\label{ag}
860: z_i = E_{(0^{i-1} 1 0^{N-i})}(z) - {1 \over \alpha + i}
861: E_{(0^i 1 0^{N-i-1})}
862: - {1 \over \alpha + i + 1} E_{(0^{i+1} 1 0^{N-i-2})}(z) - \cdots -
863: {1 \over \alpha + N - 1} E_{(0^{N-1} 1)}(z).
864: \end{equation}
865:
866: On the other hand for $\nu = c_I(0^N)$, $I \in \mathbb J_{(0^N)}$ and
867: $i \in I$ we see from (\ref{c8}) that the only possibilities are
868: $\nu = (0^{j-1} 1 0^{N-j})$ with $j \ge i$ so that (\ref{s5a}) gives
869: \begin{equation}\label{ag1}
870: z_i = \sum_{j=i}^N c_j^{(i)} E_{(0^{j-1} 1 0^{N-j})}(z)
871: \end{equation}
872: for some constants $c_j^{(i)}$. The general structure of (\ref{ag1})
873: is in agreement with (\ref{ag}). To check that the coefficients
874: agree we note from (\ref{c8}) that for $I \in
875: \mathbb J_{(0^N)}$, we must have $I = \{1,2,\dots,j\}$ and $t_l = l$
876: $(l=1,\dots,j)$. Noting also that with $\eta = (0^N)$,
877: $\bar{\eta}_i = - (i-1)$, we see that (\ref{a3b}), (\ref{a2}) and (\ref{obt})
878: give
879: \begin{eqnarray}\label{subs}
880: \tilde{\chi}_I^{(i)}(\overline{(0^N)}/\alpha) & = &
881: \left \{ \begin{array}{ll} 1, & i < j \\
882: - (N - 1 + \alpha), & i = j \end{array} \right.
883: \nonumber \\
884: A_I(\overline{(0^N)}/\alpha) & = & - {1 \over j - 1 + \alpha}, \nonumber \\
885: \hat{B}_I(\overline{(0^N)}/\alpha) & = & {j - 1 + \alpha \over N - 1 + \alpha}.
886: \end{eqnarray}
887: Substituting (\ref{subs}), together with the evaluations
888: $$
889: d_{(0^N)}' = e_{(0^N)}'=1, \quad
890: e_{(0^{j-1} 1 0^{N-j})}' = \alpha + N - 1, \quad
891: d_{(0^{j-1} 1 0^{N-j})}' = \alpha + j - 1
892: $$
893: from (\ref{e1}) and (\ref{d1}), in (\ref{s5a}) we see that the
894: coefficients in (\ref{ag1}) are as required by (\ref{ag}).
895:
896:
897: \subsection*{Verification of the recurrences}
898: Consider (\ref{k5}). From the definitions (\ref{ac1}) and (\ref{u2}) we
899: can check
900: \begin{equation}\label{ot}
901: (\overline{\Phi \eta})_i = (\bar{\eta})_{i+1}, \quad i \ne N \qquad
902: (\overline{\Phi \eta})_N = (\bar{\eta})_1 + \alpha.
903: \end{equation}
904: With $\Phi(I)$ defined by (\ref{thi}) and
905: $I = \{t_1,\dots,t_s\}$, making use of (\ref{ot}) it follows from the
906: definition (\ref{a2}) that
907: \begin{eqnarray}
908: \lefteqn{
909: A_{\Phi(I)}(\overline{\Phi \eta}/ \alpha)} \nonumber
910: \\&& = \left \{ \begin{array}{ll}
911: \Big ( \prod_{u=1}^{s-1}a((\overline{\Phi \eta})_{t_u - 1}/\alpha,
912: (\overline{\Phi \eta})_{t_{u+1} - 1}/\alpha) \Big )
913: a((\overline{\Phi \eta})_{t_s - 1}/\alpha -1,
914: (\overline{\Phi \eta})_{t_1 - 1}/\alpha), & t_1 \ne 1 \\[.2cm]
915: (a((\overline{\Phi \eta})_{t_{s-1}-1})/\alpha,
916: (\overline{\Phi \eta})_N/\alpha)
917: a((\overline{\Phi \eta})_N/\alpha - 1, (\overline{\Phi \eta})_{t_2-1}/
918: \alpha)
919: \prod_{u=2}^{s-1}a((\overline{\Phi \eta})_{t_{u} - 1}/\alpha,
920: (\overline{\Phi \eta})_{t_{u+1} - 1}/\alpha) , & t_1 =1 \end{array} \right.
921: \nonumber \\
922: && = \prod_{u=1}^{s-1}a(\bar{\eta}_{t_u}/\alpha,
923: \bar{\eta}_{t_{u+1}}/\alpha) \, a(\bar{\eta}_{t_s}/\alpha - 1,
924: \bar{\eta}_{t_1}/\alpha) \: = \:
925: A_I(\bar{\eta}/\alpha).
926: \end{eqnarray}
927: Similar calculations show
928: $$
929: \hat{B}_{\Phi(I)}(\overline{\Phi \eta}/ \alpha) =
930: \hat{B}_{I}(\bar{\eta}/\alpha), \quad
931: \tilde{\chi}_{\Phi({I})}^{(i)}(\overline{\Phi \eta}/ \alpha) =
932: \tilde{\chi}_{I}^{(i+1)}(\bar{\eta}/\alpha), \: (i=1,\dots,
933: N-1),
934: $$
935: $$
936: \tilde{\chi}_{\Phi({I})}^{(N)}(\overline{\Phi \eta}/ \alpha) =
937: \tilde{\chi}_{I}^{(1)}(\bar{\eta}/\alpha), \quad
938: c_{\Phi(I)}(\Phi \eta) =
939: \Phi (c_I(\eta)) .
940: $$
941: These formulas together with the appropriate
942: formula from (\ref{15.red})
943: immediately imply (\ref{k5}) is satisfied by (\ref{s5a}).
944:
945: The recurrences (\ref{s4}) are not so straightforward.
946: One complication is that the cases
947: \begin{equation}
948: {\rm (i) } \: i,i+1 \notin I \quad
949: {\rm (ii) } \: i \in I, \: i+1 \notin I \quad
950: {\rm (iii) } \: i \notin I, \: i+1 \in I \quad
951: {\rm (iv) } \: i, i+1 \in I
952: \label{Aiv}
953: \end{equation}
954: must be treated separately, in addition to the division of cases depending
955: on the value of $(j)$.
956: Independent of the division of cases (\ref{Aiv}), the fact that
957: $\tilde{\alpha}_{\eta,\nu}^{(j)} = 0$ for $\nu \notin
958: \mathbb J_{N,1}[\eta]$ used in (\ref{s4}) gives
959: \begin{equation}\label{s4m}
960: \tilde{\alpha}_{s_i \eta, \nu}^{(j)} = 0, \qquad
961: \nu \ne c_I(\eta), \: I \in \mathbb J_\eta \: \: {\rm and} \: \:
962: \nu \ne s_i c_I(\eta), \: I \in \mathbb J_\eta.
963: \end{equation}
964: For (\ref{s4m})
965: to be consistent with (\ref{s5a}) we must show
966: \begin{equation}\label{state}
967: c_{I'}(s_i \eta) = c_I(\eta) \quad
968: {\rm or} \quad c_{I'}(s_i \eta) = s_i c_I(\eta) \: {\rm for \: some
969: } \:
970: I \in \mathbb J_{\eta}.
971: \end{equation}
972: The validity of this statement will be verified for each of the
973: cases separately.
974:
975: First suppose $I' \in \mathbb J_{s_i \eta} \Big |_{i,i+1 \notin I'}$.
976: The definitions (\ref{c8}) and (\ref{c8a}) give that this is equivalent
977: to the statement that $I'=I$, $I \in \mathbb J_\eta \Big |_{i,i+1 \in I}$,
978: and
979: \begin{equation}\label{r1}
980: c_I(s_i \eta) = s_i c_I(\eta).
981: \end{equation}
982: Suppose next $I' \in \mathbb J_{s_i \eta} \Big |_{i+1 \in I', i \notin I'}$.
983: Then there are two possibilities. The first is $I' = (I \cup \{i+1\})
984: \backslash
985: \{i\}$, $I \in \mathbb J_\eta \Big |_{i \in I, i+1 \notin I}$, with
986: \begin{equation}\label{r2}
987: c_{(I \cup \{i+1\})\backslash\{i\}}(s_i \eta) = s_i c_I(\eta).
988: \end{equation}
989: The second is $I' = I \backslash\{i\}$, $I \in \mathbb J_\eta
990: \Big |_{i,i+1 \in I}$ with
991: \begin{equation}\label{r3}
992: c_{I\backslash\{i\}}(s_i \eta) = c_I(\eta).
993: \end{equation}
994: In the case $I' \in \mathbb J_{s_i \eta} \Big |_{i \in I, i+1 \notin I'}$
995: the only possibility is $I'=(I \cup \{i\}) \backslash\{i+1\}$,
996: $I \in \mathbb J_\eta \Big |_{i \notin I, i+1 \in I}$, with
997: \begin{equation}\label{r4}
998: c_{(I \cup \{i\}) \backslash\{i+1\}}(s_i \eta) = s_i c_I(\eta).
999: \end{equation}
1000: The remaining case is $I' \in \mathbb J_{s_i \eta} \Big |_{i,i+1 \in I'}$.
1001: Then we can have $I' = I \cup \{i\}$, $I \in \mathbb J_\eta
1002: \Big |_{i+1 \in I, i \notin I}$ with
1003: \begin{equation}\label{r5}
1004: c_{I \cup \{i\}}(s_i \eta) = c_I(\eta).
1005: \end{equation}
1006: These results together verify (\ref{state}). Thus we can restrict attention
1007: to the cases
1008: \begin{equation}\label{cases}
1009: \nu = c_I(\eta), \qquad \nu = s_i c_I(\eta), \quad (I \in \mathbb J_\eta).
1010: \end{equation}
1011:
1012: \noindent
1013: {\it The case $i, i+1 \notin I$}
1014:
1015: Because $\tilde{\alpha}_{\eta, \nu}^{(j)}$ requires $j \in I$ to be
1016: non-zero, while we are considering the case $i, i+1 \notin I$, the
1017: second and third equations in (\ref{s4}) give
1018: $\tilde{\alpha}_{s_i \eta, \nu}^{(i)} = 0$ and
1019: $\tilde{\alpha}_{s_i \eta, \nu}^{(i+1)} = 0$, which is consistent with
1020: (\ref{s5a}). Thus we can restrict attention to the first equation in
1021: (\ref{s4}). Also, if $\nu_i = \nu_{i+1}$ with $i, i+1 \notin I$ then
1022: we must have $\eta_i = \eta_{i+1}$. Since in the induction procedure
1023: it suffices to consider only the cases $\eta_{i+1} > \eta_i$ we can
1024: suppose $\nu_i \ne \nu_{i+1}$.
1025:
1026:
1027: Suppose $\nu = c_I(\eta)$. The assumptions that $i, i+1 \notin I$
1028: and $\eta_i \ne \eta_{i+1}$
1029: together with (\ref{r1}) imply there is no $I' \in \mathbb J_\eta$ such
1030: that $c_{I'}(\eta) = s_i c_I(\eta)$ and thus $\tilde{\alpha}_{\eta,
1031: s_i \nu}^{(j)} = 0$. Also in this case it follows from the definition
1032: (\ref{u1a}) that $\bar{\delta}_{i, c_I(\eta)} = \bar{\delta}_{i,\eta}$.
1033: Substituting these formulas in the first equation of (\ref{s4}) gives
1034: \begin{equation}\label{s4x}
1035: \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(j)} = 0, \qquad
1036: j \ne i, i+1, \quad I \in \mathbb J_\eta \Big |_{i,i+1 \notin I}.
1037: \end{equation}
1038: This is consistent with (\ref{s5a}) because (\ref{r1}) and the
1039: surrounding sentence implies there is no $I' \in \mathbb J_{s_i \eta}$
1040: such that $c_{I'}(s_i \eta) = c_I(\eta)$.
1041:
1042: According to (\ref{cases}) the remaining possibility for a non-zero value is
1043: $\nu = s_i c_I(\eta)$. From the above reasoning we know from this
1044: choice of $\nu$, $\tilde{\alpha}_{\eta,\nu}^{(j)} = 0$, while (\ref{u1a})
1045: gives $\bar{\delta}_{i,s_i c_I(\eta)} = - \bar{\delta}_{i,\eta}$.
1046: Thus the first equation in (\ref{s4}) reduces to
1047: \begin{equation}\label{s4y}
1048: \tilde{\alpha}_{s_i \eta, s_i c_I(\eta)}^{(j)} =
1049: \tilde{\alpha}_{\eta, c_I(\eta)}^{(j)} \qquad j \ne i, i+1, \quad
1050: I \in \mathbb J_\eta \Big |_{i,i+1 \notin I}.
1051: \end{equation}
1052: From (\ref{r1}) and the surrounding text we know that
1053: $s_i c_I(\eta) = c_I(s_i \eta)$ with $I \in \mathbb J_{s_i \eta}$.
1054: We note in general from (\ref{ac1}) that
1055: \begin{equation}\label{bb}
1056: (\overline{s_i \eta})_i =
1057: \left \{ \begin{array}{ll} \bar{\eta}_{i+1}, & \eta_i \ne \eta_{i+1} \\
1058: \bar{\eta}_{i}, & \eta_i = \eta_{i+1} \end{array} \right., \quad
1059: (\overline{s_i \eta})_{i+1} =
1060: \left \{ \begin{array}{ll} \bar{\eta}_{i}, & \eta_i \ne \eta_{i+1} \\
1061: \bar{\eta}_{i+1}, & \eta_i = \eta_{i+1} \end{array} \right., \quad
1062: (\overline{s_{i} \eta})_j = \bar{\eta}_j \quad (j \ne i,i+1).
1063: \end{equation}
1064: Using (\ref{bb}), we see from the definitions (\ref{a2}), (\ref{a3'h})
1065: that for $I \in \mathbb J_\eta \Big |_{i,i+1 \notin I} =
1066: \mathbb J_{s_i \eta} \Big |_{i,i+1 \notin I}$,
1067: $$
1068: A_I(\bar{\eta}/\alpha) = A_I(\overline{s_i \eta}/\alpha), \qquad
1069: \hat{B}_I(\bar{\eta}/\alpha) =
1070: \hat{B}_I(\overline{s_i \eta}/\alpha)
1071: $$
1072: while from (\ref{a3b}) we see that
1073: $$
1074: \tilde{\chi}_I^{(j)}(\bar{\eta}/\alpha) = \tilde{\chi}_I^{(j)}(
1075: \overline{s_i \eta}/\alpha).
1076: $$
1077: Hence (\ref{s5a}) satisfies (\ref{s4y}).
1078:
1079:
1080: \noindent
1081: {\it The case $i \in I$, $i+1 \notin I$}
1082:
1083: We note that in this case $s_i c_I(\eta) \neq c_I(\eta)$.
1084: Consider the first equation in (\ref{s4}) and suppose $\nu = c_I(\eta)$.
1085: We can check from the definitions (\ref{c8}) and (\ref{c8a}) that for
1086: $I \in \mathbb J_\eta \Big |_{i \in I, i+1 \notin I}$,
1087: \begin{equation}\label{mz1}
1088: s_i c_I(\eta) = c_{I \cup \{i+1\}}(\eta)
1089: \end{equation}
1090: so the value of $\tilde{\alpha}_{\eta, s_i \nu}^{(i)}$ on the
1091: right hand side of
1092: the first equation in (\ref{s4}) is non-zero. Noting from (\ref{u1a})
1093: and the definition (\ref{c8}) of $c_I(\eta)$ that
1094: $$
1095: \bar{\delta}_{i,c_I(\eta)} = (\overline{c_I(\eta)})_i - \bar{\eta}_{i+1},
1096: $$
1097: this equation reads
1098: \begin{equation}\label{mz2}
1099: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(j)}
1100: = {(\overline{c_I(\eta)})_i - \bar{\eta}_{i+1} -1 \over
1101: (\overline{c_I(\eta)})_i - \bar{\eta}_{i+1}}
1102: \tilde{\alpha}_{\eta, s_i c_I(\eta)}^{(j)} -
1103: \Big ( {1 \over \bar{\eta}_i - \bar{\eta}_{i+1}} -
1104: {1 \over (\overline{c_I(\eta)})_i - \bar{\eta}_{i+1}} \Big )
1105: \tilde{\alpha}_{\eta, c_I(\eta)}^{(j)}.
1106: \end{equation}
1107:
1108: The equation (\ref{mz1}) allows the quantities $A_{I'}$, $\hat{B}_{I'}$
1109: and $\tilde{\chi}^{(j)}_{I'}$, $I' = I \cup \{i+1\}$, making up
1110: $\tilde{\alpha}_{\eta, s_i c_I(\eta)}^{(j)}$ to be related to the
1111: corresponding quantities in $\tilde{\alpha}_{\eta, c_I(\eta)}^{(j)}$.
1112: Thus we can check from the definitions (\ref{a2}) and (\ref{a3'h}) that
1113: \begin{eqnarray}\label{mz2a}
1114: A_{I \cup \{i+1\}}( \bar{\eta}/\alpha) & = &
1115: {(\bar{\eta}_i - (\overline{c_I(\eta)})_i) \over
1116: (\bar{\eta}_i - \bar{\eta}_{i+1}) ( \bar{\eta}_{i+1} -
1117: (\overline{c_I(\eta)})_i) } A_I(\bar{\eta}/\alpha) \nonumber \\
1118: \hat{B}_{I \cup \{i+1\}}( \bar{\eta}/\alpha) & = &
1119: {(\bar{\eta}_{i+1} - (\overline{c_I(\eta)})_i)
1120: \over \bar{\eta}_{i+1} -
1121: (\overline{c_I(\eta)})_i + 1}
1122: \hat{B}_I( \bar{\eta}/\alpha)\nonumber \\
1123: \tilde{\chi}_{I \cup \{i+1\}}^{(j)} ( \bar{\eta}/\alpha) & = &
1124: \tilde{\chi}_{I}^{(j)} ( \bar{\eta}/\alpha), \qquad j \ne i,i+1.
1125: \end{eqnarray}
1126: When multiplied together according to (\ref{s5a}) to form
1127: $\tilde{\alpha}_{\eta, s_i c_I(\eta)}^{(j)}$ and substituted in
1128: (\ref{mz2}) we find all terms on the right hand side
1129: cancel giving the result
1130: \begin{equation}\label{mz2a'}
1131: \tilde{\alpha}^{(j)}_{s_i \eta, c_I(\eta)} = 0, \qquad j \ne i,i+1.
1132: \end{equation}
1133:
1134: Consider now the second and third equation equations in (\ref{s4}) in
1135: the case $\nu = c_I(\eta)$. The requirement in (\ref{s5a}) that
1136: $\tilde{\alpha}^{(j)}_{\eta, c_I(\eta)} \ne 0$ only if $j \in I$, while
1137: we are assuming $i \in I$, $i+1 \notin I$, means the equations read
1138: \begin{eqnarray}\label{mz3}
1139: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(i+1)}
1140: & = & (1 - \bar{\delta}_{i,\nu}^{-1})
1141: \tilde{\alpha}_{\eta, c_{I \cup \{i+1\}}(\eta)}^{(i)} +
1142: \bar{\delta}_{i,\nu}^{-1} \tilde{\alpha}_{\eta, c_{I}(\eta)}^{(i)}
1143: \nonumber \\
1144: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(i)}
1145: & = & (1 - \bar{\delta}_{i,\nu}^{-1})
1146: \tilde{\alpha}_{\eta, c_{I \cup \{i+1\}}(\eta)}^{(i+1)} -
1147: \bar{\delta}_{i,\eta}^{-1} \tilde{\alpha}_{\eta, c_{I}(\eta)}^{(i)}
1148: \end{eqnarray}
1149: where use has also been made of (\ref{mz1}). To simplify the
1150: right hand sides of
1151: these equations we note from (\ref{a3b}) that
1152: $$
1153: \tilde{\chi}_{I \cup \{i+1\}}^{(i)}(\bar{\eta}/\alpha) =
1154: {\bar{\eta}_i - \bar{\eta}_{i+1} \over \bar{\eta}_i -
1155: (\overline{c_I(\eta)})_i} \tilde{\chi}_{I}^{(i)}(\bar{\eta}/\alpha),
1156: \quad
1157: \tilde{\chi}_{I \cup \{i+1\}}^{(i+1)}(\bar{\eta}/\alpha) =
1158: {\bar{\eta}_{i+1} - (\overline{c_I(\eta)})_i \over \bar{\eta}_i -
1159: (\overline{c_I(\eta)})_i} \tilde{\chi}_{I}^{(i)}(\bar{\eta}/\alpha).
1160: $$
1161: Use of these equation, together with the first two equations of (\ref{mz2a})
1162: allows us to express $\tilde{\alpha}^{(i)}_{\eta, c_{I\cup
1163: \{i+1\}}(\eta)}$ and $\tilde{\alpha}^{(i+1)}_{\eta, c_{I\cup
1164: \{i+1\}}(\eta)}$ in terms of $\tilde{\alpha}^{(i)}_{\eta, c_{I}(\eta)}$.
1165: Doing this shows the right hand side
1166: is equal to zero in both cases and so for all
1167: $j=1,\dots,N$
1168: \begin{equation}\label{mz4}
1169: \tilde{\alpha}^{(j)}_{s_i\eta, c_I(\eta)} = 0.
1170: \end{equation}
1171: For $s_i c_I(\eta) \ne c_I(\eta)$
1172: the result (\ref{mz4}) is consistent with (\ref{s5a}) because the result
1173: (\ref{r2}) implies that for $I \in \mathbb
1174: J_\eta \Big |_{i+1 \notin I, i \in I}$
1175: there is no $I' \in \mathbb J_{s_i \eta}$ such that
1176: $c_{I'}(s_i \eta) = c_I(\eta)$.
1177:
1178: We now proceed to consider the equations (\ref{s4}) in the case
1179: $\nu = s_i c_I(\eta)$, $I \in \mathbb J_\eta
1180: \Big |_{i \in I, i+1 \notin I}$. Proceeding as in the derivation of
1181: (\ref{mz2a'}) we find that in this case the first equation of (\ref{s4})
1182: reads
1183: \begin{eqnarray}\label{su1}
1184: \tilde{\alpha}^{(j)}_{s_i\eta, s_i c_I(\eta)} & = &
1185: {(\bar{\eta}_{i} - \bar{\eta}_{i+1} -1)(\bar{\eta}_{i+1} -
1186: (\overline{c_I(\eta)})_i) \over
1187: (\bar{\eta}_{i} - \bar{\eta}_{i+1})(\bar{\eta}_{i+1} -
1188: (\overline{c_I(\eta)})_i + 1)} \tilde{\alpha}^{(j)}_{\eta,
1189: c_I(\eta)} \nonumber \\
1190: & = &
1191: \tilde{\chi}_{(I \cup \{i+1\})\backslash \{i\}}^{(j)}
1192: (\overline{s_i \eta}/\alpha)
1193: A_{(I \cup \{i+1\})\backslash \{i\}}(\overline{s_i \eta}/\alpha)
1194: \hat{B}_{(I \cup \{i+1\})\backslash \{i\}}(\overline{s_i \eta}/\alpha),
1195: \quad j \ne i,i+1
1196: \end{eqnarray}
1197: where the second equality follows after use of (\ref{s5a}) to
1198: substitute for $\tilde{\alpha}^{(j)}_{\eta,
1199: c_I(\eta)}$ and use of the definitions (\ref{a2}), (\ref{a3'h}) and
1200: (\ref{a3b}).
1201: An analogous calculation, involving the second and third equations of
1202: (\ref{s4}), gives
1203: \begin{equation}\label{su1a}
1204: \tilde{\alpha}^{(i)}_{s_i\eta, s_i c_I(\eta)} = 0
1205: \end{equation}
1206: as well as the equation (\ref{su1}) in the case $j=i+1$.
1207: Recalling (\ref{r2})
1208: we see the equations (\ref{su1}) and (\ref{su1a}) are consistent with
1209: (\ref{s5a}).
1210:
1211: \noindent
1212: {\it The case $i \notin I$, $i+1 \in I$}
1213:
1214: We distinguish the case
1215: \begin{equation}\label{cs1}
1216: s_i c_I(\eta) = c_I(\eta)
1217: \end{equation}
1218: from
1219: \begin{equation}\label{cs2}
1220: s_i c_I(\eta) \neq c_I(\eta)
1221: \end{equation}
1222: In the case (\ref{cs1}) we can check that
1223: \begin{equation}\label{cs2'}
1224: (\overline{ c_I(\eta)})_{i + 1} = \bar{\eta}_i - 1,
1225: \end{equation}
1226: while a feature of the case (\ref{cs2}) is that there is no $I' \in
1227: \mathbb J_\eta$ such that $s_i c_I(\eta) = c_{I'}(\eta)$ and
1228: therefore
1229: \begin{equation}\label{cs3}
1230: \tilde{\alpha}^{(j)}_{\eta, s_i c_I(\eta)} = 0.
1231: \end{equation}
1232:
1233: Consider first the equations (\ref{s4}) in the case (\ref{cs1}) (as already
1234: noted, the equations (\ref{s4}) are equivalent to the equations
1235: (\ref{s5}) for $\nu = s_i c_I(\eta) = c_I(\eta)$). The equations read
1236: \begin{eqnarray}\label{ww1}
1237: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(j)}
1238: &=& (1 - \bar{\delta}_{i,\eta}^{-1})
1239: \tilde{\alpha}_{\eta, c_I(\eta)}^{(j)},
1240: \quad j \ne i, i+1 \nonumber \\
1241: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(i+1)}
1242: &=& - \bar{\delta}_{i,\eta}^{-1}
1243: \tilde{\alpha}_{\eta, c_I(\eta)}^{(i+1)} \nonumber \\
1244: (1 + \bar{\delta}_{i,\eta}^{-1}) \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(i)}
1245: & = & \tilde{\alpha}_{ \eta, c_I(\eta)}^{(i+1)}.
1246: \end{eqnarray}
1247: To verify that (\ref{s5a}) satisfies these equations we note that in the
1248: case (\ref{cs1}) the equation (\ref{r5}) is valid, so we should seek
1249: to express (\ref{ww1}) in terms of $\tilde{\alpha}_{s_i \eta,
1250: c_{I \cup \{i\}}(s_i \eta)}$. Now (\ref{bb}), (\ref{a2}), (\ref{a3'h})
1251: and (\ref{a3b}) give
1252: \begin{eqnarray}\label{ww2a}
1253: A_{I \cup \{i\}}(\overline{s_i \eta}/\alpha) & = &
1254: {\bar{\eta}_i - \bar{\eta}_{i+1} - 1 \over
1255: \bar{\eta}_i - \bar{\eta}_{i+1}} A_I(\bar{\eta}/\alpha) \nonumber \\
1256: \hat{B}_{I \cup \{i\}}(\overline{s_i \eta}/\alpha) & = &
1257: {\bar{\eta}_i - \bar{\eta}_{i+1} \over
1258: \bar{\eta}_i - \bar{\eta}_{i+1} + 1}
1259: \hat{B}_{I}(\bar{\eta}/\alpha) \nonumber \\
1260: \tilde{\chi}^{(j)}_{I \cup \{i\}}(\overline{s_i \eta}/\alpha) & = &
1261: \tilde{\chi}^{(j)}_{I}(\bar{\eta}/\alpha), \quad j \ne i, i+1
1262: \nonumber \\
1263: \tilde{\chi}^{(i+1)}_{I \cup \{i\}}(\overline{s_i \eta}/\alpha) & = &
1264: {1 \over \bar{\eta}_{i+1} - \bar{\eta}_i + 1}
1265: \tilde{\chi}^{(i+1)}_{I}(\bar{\eta}/\alpha) \nonumber \\
1266: \tilde{\chi}^{(i)}_{I \cup \{i\}}(\overline{s_i \eta}/\alpha) & = &
1267: {\bar{\eta}_{i+1} - \bar{\eta}_i \over \bar{\eta}_{i+1} - \bar{\eta}_i + 1}
1268: \tilde{\chi}^{(i+1)}_{I}(\bar{\eta}/\alpha) .
1269: \end{eqnarray}
1270: Making use of these equations in the right hand side of (\ref{ww1}) we find
1271: that for each $j=1,2,\dots,N$
1272: \begin{equation}\label{ww3}
1273: \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(j)} =
1274: \tilde{\chi}_{I \cup \{i\}}^{(j)}(\overline{s_i \eta}/\alpha)
1275: A_{I \cup \{i\}}(\overline{s_i \eta}/\alpha)
1276: \hat{B}_{I \cup \{i\}}(\overline{s_i \eta}/\alpha)
1277: \end{equation}
1278: which by virtue of (\ref{r5}) is consistent with (\ref{s5a}).
1279:
1280: Consider now the equations (\ref{s4}) with $\nu = c_I(\eta)$ in the case
1281: (\ref{cs2}). Then (\ref{c3}) holds, so (\ref{s4}) can be appropriately
1282: simplified. Furthermore, we can check that the second and third members
1283: of (\ref{ww2a}) remain valid, while the remaining equations are to be
1284: replaced by
1285: \begin{eqnarray}\label{aov}
1286: A_{I \cup \{i\}}(\overline{s_i \eta}/\alpha) & = &
1287: - {(\bar{\eta}_{i+1} - (\overline{c_I(\eta)})_{i+1}) \over
1288: (\bar{\eta}_i - \bar{\eta}_{i+1})(\bar{\eta}_i -
1289: (\overline{c_I(\eta)})_{i+1})} A_I(\bar{\eta}/\alpha) \nonumber \\
1290: \tilde{\chi}_{I \cup \{i\}}^{(i+1)}(\overline{s_i \eta}/\alpha) & = &
1291: {\bar{\eta}_i - (\overline{c_I(\eta)})_{i+1} \over
1292: \bar{\eta}_{i+1} - (\overline{c_I(\eta)})_{i+1}}
1293: \tilde{\chi}_{I}^{(i+1)}((\bar{\eta}/\alpha) \nonumber \\
1294: \tilde{\chi}_{I \cup \{i\}}^{(i)}(\overline{s_i \eta}/\alpha) & = &
1295: {\bar{\eta}_{i+1} - \bar{\eta}_i \over
1296: \bar{\eta}_{i+1} - (\overline{c_I(\eta)})_{i+1}}
1297: \tilde{\chi}_{I}^{(i+1)}((\bar{\eta}/\alpha)
1298: \end{eqnarray}
1299: Using these equations to further simplify (\ref{s4}) again
1300: gives (\ref{ww3}), which we know is consistent with (\ref{s5a}).
1301: It remains to consider the case $\nu = s_i c_I(\eta)$,
1302: for which it suffices to restrict attention to the subcase (\ref{cs2})
1303: as the subcase (\ref{cs2}) is included in the above working.
1304: We first simplify the equations (\ref{s4}) according to (\ref{cs3})
1305: and then obtain the analogues of (\ref{ww2a}) for the quantities
1306: $A_{(I \cup \{i\})\backslash\{i+1\}}(\overline{s_i \eta}/\alpha)$
1307: etc.. We find, for $j \ne i+1$,
1308: $$
1309: \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(j)} =
1310: \tilde{\chi}_{I \cup \{i\}\backslash\{i+1\}}^{(j)}(\overline{s_i \eta}/\alpha)
1311: A_{(I \cup \{i\})\backslash\{i+1\}}(\overline{s_i \eta}/\alpha)
1312: \hat{B}_{I \cup \{i\}\backslash\{i+1\}}(\overline{s_i \eta}/\alpha)
1313: $$
1314: while
1315: $$
1316: \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(i+1)} = 0.
1317: $$
1318: We see from (\ref{r4}) that these equations are consistent with (\ref{s5a}).
1319:
1320: \noindent
1321: {\it The case $i,i+1 \in I$}
1322:
1323: Analogous to the case $i \notin I$, $i+1 \in I$ we distinguish the case
1324: $s_i c_I(\eta) = c_I(s_i \eta)$ from $s_i c_I(\eta) \notin c_I(s_i \eta)$.
1325: In the latter case
1326: \begin{equation}\label{f2a}
1327: s_i c_I(\eta) = c_{I\backslash \{i+1\}}(\eta).
1328: \end{equation}
1329: This tells us that this case is the same as that with $I' \in
1330: \mathbb J_\eta$, $i \in I'$, $i+1 \notin I'$, which has already been
1331: dealt with. Thus we can restrict attention to the case
1332: $s_i c_I(\eta) = c_I(s_i \eta)$, when the equations (\ref{s4}) reduce
1333: to the equations (\ref{s5}). Obtaining the analogue of
1334: (\ref{ww2a}) but for $A_{I\backslash \{i\}}(\overline{s_i \eta}/\alpha)$
1335: etc.~we find, for $j \neq i$,
1336: $$
1337: \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(j)} =
1338: \tilde{\chi}_{I \backslash\{i\}}^{(j)}(\overline{s_i \eta}/\alpha)
1339: A_{I \backslash\{i\}}(\overline{s_i \eta}/\alpha)
1340: \hat{B}_{I \backslash\{i\}}(\overline{s_i \eta}/\alpha)
1341: $$
1342: while
1343: $$
1344: \tilde{\alpha}_{s_i \eta, c_I(\eta)}^{(i)} = 0.
1345: $$
1346: By virtue of (\ref{r3}) these equations are consistent with
1347: (\ref{s5a}).
1348:
1349: \medskip
1350: This completes consideration of the choices of $\nu$ (\ref{cases}) in
1351: all four cases (\ref{Aiv}). In each case it was found (\ref{s5a})
1352: satisfies the recurrences (\ref{s4}), thereby completing the demonstration
1353: that for general $\nu$ (\ref{s5a}) satisfies (\ref{s4}). Since the other
1354: fundamental recurrence (\ref{k5}) has also been shown to be satisfied,
1355: as has the initial condition, our inductive proof is complete.
1356:
1357: \section{An equivalent expansion formula}
1358: \setcounter{equation}{0}
1359: The formula (\ref{fz2}) is the non-symmetric analogue of the Pieri
1360: formula (\ref{pf1}) in the case $p=1$. Here we will use this result
1361: and the formula (\ref{f2}) to derive the analogue of (\ref{pf1}) in the
1362: case $p = N-1$.
1363:
1364: First we note that in the case $p=N-1$, analogous to the case $p=1$, the
1365: set $\mathbb J_{N,p}$ appearing in (\ref{f4}) and (\ref{f4'}) can be
1366: indexed by subsets $I = \{t_1,\dots, t_s\}$ of $\{1,\dots,N\}$ with
1367: $t_1 < \cdots < t_s$. Each such subset corresponds to the element
1368: $$
1369: \nu =: \hat{c}_I(\eta)
1370: $$
1371: where
1372: \begin{eqnarray}\label{ptr1}
1373: (\hat{c}_I(\eta))_{t_1} & = & \eta_{t_s}, \nonumber \\
1374: (\hat{c}_I(\eta))_{t_u} & = & \eta_{t_{u-1}} + 1, \quad u=2,\dots,s
1375: \nonumber \\
1376: (\hat{c}_I(\eta))_k & = & \eta_k + 1, \quad k \notin I
1377: \end{eqnarray}
1378: (c.f.~(\ref{c8})). Furthermore, to avoid duplication within the set
1379: $\mathbb J_{N,N-1}$, as with the description (\ref{c8}) of
1380: $\mathbb J_{N,1}$, we must restrict $I$ to maximal subsets
1381: with respect to $\eta$,
1382: in this case specified by the requirements
1383: \begin{eqnarray}\label{ptr2}
1384: &&\eta_j \ne \eta_{t_s} - 1, \qquad j = 1,\dots, t_1 - 1 \nonumber \\
1385: &&\eta_j \ne \eta_{t_u}, \qquad
1386: j=t_u + 1, \dots, t_{u+1} - 1
1387: \end{eqnarray}
1388: for $u=1,\dots,s$ with $t_{s+1} := N + 1$. With this definition of maximal,
1389: analogous to (\ref{c8c}) we define
1390: $$
1391: \hat{\mathbb J}_\eta = \{I: I \: {\rm is \: maximal \: with \: respect
1392: \: to \:} \eta \}.
1393: $$
1394:
1395: According to (\ref{f2})
1396: \begin{equation}\label{pq}
1397: c_{\eta \nu}^{(i_1,\dots,i_{N-1})} = c_{\nu \,\eta+(1^N)}^{(j_1)}
1398: {\ml E_\eta | E_\eta \mg \over \ml E_\nu | E_\nu \mg}.
1399: \end{equation}
1400: Now $c_{\nu \, \eta + (1^N)}^{(j_1)}$ is non-zero only if
1401: $\eta + (1^N) = c_I(\nu)$, $I \in \mathbb J_{\nu}$,
1402: $j_1 \in I$. With $I = \{t_1,\dots, t_s\}$ we see from (\ref{c8}) that
1403: $\eta + (1^N) = c_I(\nu)$ gives
1404: \begin{eqnarray}
1405: \eta_{t_j} & = & \nu_{t_{j+1}} - 1, \qquad j=1,\dots, s-1 \nonumber \\
1406: \eta_{t_s} & = & \nu_{t_1} \nonumber \\
1407: \eta_i & = & \nu_i - 1, \qquad i \notin I
1408: \end{eqnarray}
1409: while from (\ref{c9}) the condition $I \in \mathbb J_{\nu}$
1410: gives
1411: \begin{eqnarray}
1412: \eta_j & \ne & \eta_{t_s} - 1 \qquad j=1,\dots, t_1 - 1 \\
1413: \eta_j & \ne & \eta_{t_u} \qquad j=t_u+1, \dots, t_{u+1} - 1
1414: \end{eqnarray}
1415: for $u=1,\dots,s$ with $t_{N+1} := N+1$. These are precisely the equations
1416: (\ref{ptr1}) with $\nu := \hat{c}_I(\eta)$ and the equations (\ref{ptr2})
1417: for $I \in \hat{\mathbb J}_{\eta}$, so we conclude
1418: $$
1419: \eta =
1420: c_I(\nu) \Big |_{I \in \mathbb J_{\nu}} \quad {\rm iff} \quad
1421: \nu + (1)\hat{c}_I(\eta) \Big |_{I \in \hat{{\mathbb J}}_\eta}.
1422: $$
1423:
1424: It remains to substitute for the explicit values in the right hand side
1425: of (\ref{pq}).
1426: With $\eta + (1^N) = c_I(\nu)$, $I \in \mathbb J_{\nu}$
1427: and thus $\nu = \hat{c}_I(\eta)$, $I \in
1428: \hat{\mathbb J}_\eta$ we read off from (\ref{obt}) and
1429: (\ref{en}) that
1430: \begin{equation}\label{fii}
1431: c_{\eta \nu}^{(i_1,\dots,i_{N-1})} =
1432: {e_\eta d_\nu \over d_\eta e_\nu}
1433: {d_\eta e_{\eta + (1^N)} \over d_{\eta + (1^N)} e_\eta}
1434: \tilde{\chi}^{(j_1)}_I(\overline{\eta}/\alpha)
1435: A_I(\overline{\eta}/\alpha)
1436: \hat{B}_I(\overline{\eta}/\alpha)
1437: \end{equation}
1438: where use has been made of the facts that ${\cal N}_\eta =
1439: {\cal N}_{\eta + (1^N)}$ and $\tilde{\chi}_I^{(j_1)}
1440: ((\overline{\eta} + c)/\alpha)
1441: = \tilde{\chi}^{(j_1)}_I(\overline{\eta}/\alpha)$ etc.~for any constant
1442: $c$.
1443: This further simplifies by noting from (\ref{d1}), (\ref{leg}) and (\ref{ac1})
1444: that
1445: $$
1446: {d_\eta \over d_{\eta + (1^N)}}
1447: = {1 \over \prod_{j=1}^N(\bar{\eta}_j + \alpha + N)}
1448: $$
1449: while (\ref{e1}) together with (\ref{bb}) and (\ref{ac1}) implies
1450: $$
1451: {e_{\eta + (1^N)} \over e_\eta} =
1452: \prod_{j=1}^N(\bar{\eta}_j + \alpha + N).
1453: $$
1454: Substituting these formulas in (\ref{fii}) gives
1455: \begin{equation}\label{56}
1456: c_{\eta \nu}^{(i_1,\dots,i_{N-1})} =
1457: {e_\eta d_\nu \over d_\eta e_\nu}
1458: \tilde{\chi}^{(j_1)}_I(\overline{\eta}/\alpha)
1459: A_I(\overline{\eta}/\alpha)
1460: \hat{B}_I(\overline{\eta}/\alpha), \quad
1461: \nu = \hat{c}_I(\eta), \: I \in \hat{\mathbb J}_\eta
1462: \end{equation}
1463: (c.f.~(\ref{s5b'})).
1464:
1465: As in the derivation of (\ref{fu2}) from (\ref{fu1}), if follows from
1466: (\ref{56}) that
1467: \begin{equation}\label{57}
1468: e_{N-1}(z) E_\eta(z) = - \alpha {e_\eta \over d_\eta}
1469: \sum_{I \in \hat{\mathbb J}_\eta}
1470: {e_{\hat{c}_I(\eta)} A_I(\bar{\eta}/\alpha) \hat{B}_I(\bar{\eta}/\alpha)
1471: \over d_{\hat{c}_I(\eta)} } E_{\hat{c}_I(\eta)}(z).
1472: \end{equation}
1473:
1474: \section{The coefficient $A_{\eta, \nu}^{(p)}$ of the Pieri type formula
1475: for general $p$}
1476: \setcounter{equation}{0}
1477: In this final section we will consider features of the coefficient
1478: $A_{\eta, \nu}^{(p)}$ in the expansion (\ref{f4'}) for general $p$.
1479: Our first result concerns the value of $A_{\eta, \nu}^{(p)}$ for a
1480: particular value of $\nu$. Now, according to (\ref{2.8'}) the coefficient
1481: of $z^\eta$ in $E_\eta(z)$ is unity and all other monomials are smaller with
1482: respect to the ordering $\triangleleft$. Consider all sets of the form
1483: $M = \{j_1,\dots,j_p | 1 \le j_1 < \cdots < j_p \le N \}$ and let
1484: $$
1485: (\chi_M)_i = \left \{ \begin{array}{ll}1, & i \in M \\
1486: 0, & {\rm otherwise} \end{array} \right.
1487: $$
1488: so that the $p$th monomial symmetric function can be written
1489: $$
1490: e_p(z) = \sum_M z^{\chi_M}.
1491: $$
1492: Let $M^*$ be the particular set $M$ such that
1493: $$
1494: \eta + \chi_M \triangleleft \eta + \chi_{M^*}
1495: $$
1496: for all $M \ne M^*$. Then we must have
1497: \begin{equation}\label{aw0}
1498: A_{\eta, \eta + \chi_{M^*}}^{(p)} = 1.
1499: \end{equation}
1500: Moreover, with
1501: $$
1502: l_\eta'(i) := \#\{k < i | \eta_k \ge \eta_i \} + \#\{k > i| \eta_k > \eta_i
1503: \}
1504: $$
1505: it follows from the definition of $\triangleleft$ that
1506: \begin{equation}\label{aw}
1507: ( \eta + \chi_{M^*})_i = \left \{
1508: \begin{array}{ll} \eta_i + 1, & l_\eta'(i) \le p-1 \\
1509: \eta_i, & l_\eta'(i) \ge p \end{array} \right.
1510: \end{equation}
1511:
1512: Associated with (\ref{aw}) are the sets
1513: \begin{equation}
1514: G_0 := \{i \in \{1,\dots,N\}: l_\eta'(i) \ge p \}, \qquad
1515: G_1 := \{i \in \{1,\dots,N\}: l_\eta'(i) \le p - 1 \}.
1516: \end{equation}
1517: An alternative characterization follows by noting that since $\eta
1518: \subseteq \eta + \chi_{M^*}$ we have $\eta \preceq \eta + \chi_{M^*}$
1519: and so from the definition of $\preceq$, for $\nu =
1520: \eta + \chi_{M^*}$,
1521: \begin{equation}
1522: G_0 := \{i \in \{1,\dots,N\}: \nu_{\pi(j)} = \eta_j \}, \quad
1523: G_1 := \{i \in \{1,\dots,N\}: \nu_{\pi(j)} = \eta_j + 1 \}.
1524: \end{equation}
1525: Let us now put
1526: \begin{equation}\label{siv}
1527: B_{\eta,\nu}^{(p)} := {d_\nu' e_\eta' \over e_\nu' d_\eta'}
1528: A_{\eta, \nu}^{(p)}.
1529: \end{equation}
1530: Then it follows from the definitions (\ref{d1}) and (\ref{e1}), together
1531: with (\ref{aw0}), that
1532: \begin{equation}\label{swa}
1533: B_{\eta, \eta + \chi_{M^*}}^{(p)} =
1534: \Big ( \prod_{j \in G_0, \, k \in G_1 \atop j<k}
1535: {\bar{\eta}_j - \bar{\eta}_k +1 \over \bar{\eta}_j - \bar{\eta}_k}
1536: \Big )
1537: \Big ( \prod_{j \in G_1, \, k \in G_0 \atop \pi(j)<\pi(k)}
1538: {\bar{\eta}_j - \bar{\eta}_k +\alpha - 1 \over
1539: \bar{\eta}_j - \bar{\eta}_k + \alpha}
1540: \Big )
1541: \end{equation}
1542: In the case $p=1$, comparison with the expression
1543: \begin{equation}\label{san}
1544: B_{\eta,\nu}^{(1)} = - \alpha A_I(\bar{\eta}/\alpha) \hat{B}_I
1545: (\bar{\eta}/\alpha), \quad I \in {\mathbb J}_\eta,
1546: \end{equation}
1547: which follows from (\ref{obt}), (\ref{4.47'}) and (\ref{siv}), we see
1548: that as written (\ref{swa}) is in fact valid for all $\nu = c_I(\eta)$
1549: with $I$ consisting of a single element $t_1$. More explicitly, we
1550: then have
1551: \begin{eqnarray}
1552: &&
1553: A_I(\bar{\eta}/\alpha) = a(\bar{\eta}_{t_1}/\alpha - 1,
1554: \bar{\eta}_{t_1}) = -
1555: {1 \over \alpha} \label{san1} \\
1556: &&\hat{B}_1(\bar{\eta}/\alpha) =
1557: \prod_{j=1}^{t_1 - 1} b(\bar{\eta}_{t_1}/\alpha,
1558: \bar{\eta}_j/\alpha)
1559: \prod_{j=t_1 + 1}^N b(\bar{\eta}_{t_1}/\alpha+1,
1560: \bar{\eta}_j/\alpha); \label{san2}
1561: \end{eqnarray}
1562: the factor (\ref{san1}) cancels with $-\alpha$ in (\ref{san}) while the
1563: two products (\ref{san2}) correspond with the two products in (\ref{swa})
1564: respectively.
1565:
1566: We can extend the form (\ref{swa}) so that for $p=1$ there is agreement
1567: with (\ref{san}). The extended form is
1568: \begin{eqnarray}
1569: B_{\eta,\nu}^{(p)} & = & \Big (
1570: \prod_{j \in G_0, k \in G_1 \atop j < k}
1571: {\bar{\eta}_j - \bar{\eta}_k + 1 \over
1572: \bar{\eta}_j - \bar{\eta}_k } \Big )
1573: \Big ( \prod_{j \in G_1, \, k \in G_0 \atop \pi(j)<\pi(k)}
1574: {\bar{\eta}_j - \bar{\eta}_k +\alpha - 1 \over
1575: \bar{\eta}_j - \bar{\eta}_k + \alpha}
1576: \Big ) \nonumber \\
1577: && \times
1578: \prod_{\pi^2(j) < \pi(j) < j}
1579: {1 \over \bar{\eta}_{\pi(j)} - \bar{\eta}_j}
1580: \prod_{j \le \pi^2(j) \le \pi(j)}
1581: {1 \over \bar{\eta}_{\pi(j)} - \bar{\eta}_j - \alpha},
1582: \end{eqnarray}
1583: valid for $p=1$ and $\nu = c_I(\eta)$, $I \in \mathbb J_\eta$.
1584: The significant feature of this form is that in the general $p$ case, with
1585: $\nu = c_I(\eta) \in \mathbb J_{N,p}$ and $I$ such that at most one part
1586: of $\eta$ in the formation of $\nu$ according to the prescription
1587: below (\ref{c7}) move downwards, explicit small $N$ calculations
1588: indicate it remains valid. However we have no proof of this empirical
1589: observation.
1590:
1591: \section*{Acknowledgements}
1592: We thank Dan Marshall for providing us with his notes on the derivation
1593: of (\ref{fu1}). This work was supported by the Australian Research
1594: Council.
1595:
1596: %\bibliographystyle{plain}
1597: %\bibliography{book}
1598:
1599:
1600: \begin{thebibliography}{10}
1601:
1602: \bibitem{BF97b}
1603: T.H. Baker and P.J. Forrester.
1604: \newblock The {Calogero-Sutherland} model and polynomials with prescribed
1605: symmetry.
1606: \newblock {\em Nucl. Phys. B}, 492:682--716, 1997.
1607:
1608: \bibitem{BF99}
1609: T.H. Baker and P.J. Forrester.
1610: \newblock Symmetric {Jack} polynomials from nonsymmetric theory.
1611: \newblock {\em Annals Comb.}, 3:159--170, 1999.
1612:
1613: \bibitem{Fo94j}
1614: P.J. Forrester.
1615: \newblock Addendum to {Selberg} correlation integrals and the $1/r^2$ quantum
1616: many body system.
1617: \newblock {\em Nucl. Phys. B}, 416:377--385, 1994.
1618:
1619: \bibitem{Ka97g}
1620: K.W.J. Kadell.
1621: \newblock The {Selberg-Jack} symmetric functions.
1622: \newblock {\em Adv. Math.}, 130:33--102, 1997.
1623:
1624: \bibitem{Kn97}
1625: F.~Knop.
1626: \newblock Symmetric and non-symmetric quantum {Capelli} polynomials.
1627: \newblock {\em Comm. Math. Helv.}, 72:84--100, 1997.
1628:
1629: \bibitem{KS96}
1630: F.~Knop and S.~Sahi.
1631: \newblock Difference equations and symmetric polynomials defined by their
1632: zeros.
1633: \newblock {\em Int. Math. Res. Not.}, 10:473--486, 1996.
1634:
1635: \bibitem{KS97}
1636: F.~Knop and S.~Sahi.
1637: \newblock A recursion and combinatorial formula for Jack polynomials
1638: \newblock {\em Inv. Math.}, 128:9--22, 1997.
1639:
1640: \bibitem{Ko96}
1641: H.~Konno.
1642: \newblock Relativistic {Calogero-Sutherland} model: spin generalization,
1643: quantum affine symmetry and dynamical correlation functions.
1644: \newblock {\em Nucl. Phys. B}, 473:579--601, 1996.
1645:
1646: \bibitem{Ma95}
1647: I.G. Macdonald.
1648: \newblock {\em Hall polynomials and symmetric functions}.
1649: \newblock Oxford University Press, Oxford, 2nd edition, 1995.
1650:
1651: \bibitem{Ma99}
1652: D.~Marshall.
1653: \newblock Symmetric and nonsymmetric {Macdonald} polynomials.
1654: \newblock {\em Annals Comb.}, 3:385--415, 1999.
1655:
1656: \bibitem{Ma00}
1657: D.~Marshall.
1658: \newblock In preparation, 2000.
1659:
1660: \bibitem{Op95}
1661: E.M. Opdam.
1662: \newblock Harmonic analysis for certain representations of graded {Hecke}
1663: algebras.
1664: \newblock {\em Acta Math.}, 175:75--121, 1995.
1665:
1666: \bibitem{Sa96}
1667: S.~Sahi.
1668: \newblock A new scalar product for nonsymmetric {Jack} polynomials.
1669: \newblock {\em Int. Math. Res. Not.}, 20:997--1004, 1996.
1670:
1671: \bibitem{St89}
1672: R.~P. Stanley.
1673: \newblock Some combinatorial properties of {Jack} symmetric functions.
1674: \newblock {\em Advances in Math.}, 77:76--115, 1989.
1675:
1676: \end{thebibliography}
1677: \end{document}
1678:
1679:
1680:
1681:
1682:
1683:
1684:
1685:
1686:
1687:
1688:
1689:
1690:
1691:
1692: