1: \documentclass{icsm}
2: \usepackage{graphicx}% Include figure files
3: \usepackage{bm}% bold math
4: \usepackage{amsmath}
5: \usepackage{amsfonts}
6: \usepackage{amssymb}
7: \usepackage{amsthm}
8:
9: \usepackage[dvips]{epsfig}
10:
11: \slacs{.4ex}
12:
13: \begin{document}
14: \title{Partition functions and graphs: A combinatorial approach}
15: %
16: \authori{A. I. Solomon$^{a,b}$, P. Bl{}asiak$^{b,c}$,}
17: \addressi{$^{a}$The Open University, Physics and Astronomy Department\\
18: Milton Keynes MK7 6AA, United Kingdom\\
19: $^{b}$Laboratoire de Physique Th\'eorique des Liquides,
20: Universit\'e Pierre et Marie Curie\\
21: Tour 24 -- 2e \'et., 4 Pl.Jussieu, F 75252 Paris Cedex 05, France\\
22: e-mail: a.i.solomon@open.ac.uk, blasiak@lptl.jussieu.fr
23: \\
24: $^{c}$H.Niewodnicza\'nski Institute of Nuclear Physics, Polish Academy of Sciences\\
25: ul. Eliasza-Radzikowskiego 152, PL 31342 Krakow, Poland\\}
26: \authorii{G. E. H. Duchamp,}
27: \addressii{Universit\'{e} de Rouen, LIFAR, F 76821 Mont-Saint Aignan Cedex, France\\
28: e-mail: gduchamp2@free.fr}
29: \authoriii{A. Horzela,}
30: \addressiii{H.Niewodnicza\'nski Institute of Nuclear Physics, Polish Academy of Sciences\\
31: ul. Eliasza-Radzikowskiego 152, PL 31342 Krakow, Poland\\
32: e-mail: andrzej.horzela@ifj.edu.pl}
33: \authoriv{K. A. Penson}
34: \addressiv{Laboratoire de Physique Th\'eorique des Liquides,
35: Universit\'e Pierre \& Marie Curie\\
36: Tour 24 -- 2e \'et., 4 Pl.Jussieu, F 75252 Paris Cedex 05, France\\
37: e-mail: penson@lptl.jussieu.fr}
38: %
39: \authorv{} \addressv{}
40: \authorvi{} \addressvi{}
41: %
42: \headauthor{A.I. Solomon et al.}
43: %
44: \headtitle{Partition functions and graphs: A combinatorial approach}
45: %
46: \lastevenhead{A.I. Solomon et al.: Partition functions and graphs:
47: A combinatorial approach}
48: %
49: \pacs{03.65.Fd, 05.30.Jp,}
50: \keywords{boson normal ordering, combinatorics}
51: \maketitle
52:
53: \begin{abstract}
54: Although symmetry methods and analysis are a necessary ingredient
55: in every physicist's toolkit, rather less use has been made of
56: combinatorial methods. One exception is in the realm of
57: Statistical Physics, where the calculation of the partition
58: function, for example, is essentially a combinatorial problem. In
59: this talk we shall show that one approach is via the normal
60: ordering of the second quantized operators appearing in the
61: partition function. This in turn leads to a combinatorial
62: graphical description, giving essentially ‘Feynman--type’ graphs
63: associated with the theory. We illustrate this methodology by the
64: explicit calculation of two model examples, the free boson gas and
65: a superfluid boson model. We show how the calculation of partition
66: functions can be facilitated by knowledge of the combinatorics of
67: the boson normal ordering problem; this naturally gives rise to
68: the {\em Bell} numbers of combinatorics. The associated graphical
69: representation of these numbers gives a perturbation expansion in
70: terms of a sequence of graphs analogous to zero--dimensional
71: Feynman diagrams.
72: \end{abstract}
73:
74: \section{Introduction}
75:
76: According to the axiomatic approach, it is possible to reconstruct
77: a quantum field theory from the vacuum expectation values. When we
78: {\em normally order} a set of operators so that all the
79: annihilation operators are to the right, then in such a vacuum
80: expectation value only the constant term remains.\footnote{Note
81: that this process of normally ordering, denoted by $\mathcal{N}$,
82: does {\em not} change the value of the operator. Thus
83: $\mathcal{N}(f(a,a^{\dagger}))=f(a,a^{\dagger})$.} Similarly, when
84: we normally order a string of boson operators $a$,$a ^{\dagger}$
85: satisfying $[a,a ^{\dagger}]=1$ , then the vacuum expectation
86: reduces to the constant term; and likewise we may recover the
87: expectation in the coherent state $|z\rangle$ by replacing $a$ by
88: the $c$--number $z$. Therefore the ability to normally order a
89: string of operators is a powerful calculational tool in physics.
90: It was recognized early on that the procedure of normally ordering
91: a string of bosons leads to classical combinatorial numbers, the
92: Stirling and Bell numbers \cite{[1]}. Apart from the generation of
93: interesting combinatorial sequences, and their extensions
94: \cite{[2]}, the associated graph theory representations of these
95: numbers have analogies to the Feynman diagrams of a
96: zero--dimensional quantum field theory.
97:
98: In this note we shall present a standard graphical
99: representation of the Bell numbers which we show is in some
100: sense {\em generic}. That is, we use these graphs to provide a
101: representation of a perturbation expansion for the partition
102: function of a free boson gas, and show that the same sequence of
103: graphs can be used to give a perturbation expansion for a general
104: partition function.
105:
106: We illustrate this approach by applying it to a superfluid boson
107: system.
108:
109: \section{Combinatorial sequences, generating functions and graphs}
110:
111: \subsection{Some combinatorial sequences}
112:
113: Sequences of numbers of combinatorial interest are common in
114: mathematics; for example, some sequences depending on one
115: parameter are
116: \begin{enumerate}
117: \item $\{n\}$ The integers.
118: \item $\{n!\}$ The number of ways of putting $n$ different objects into $n$ different
119: containers (leaving none empty).
120: \item $\{2^n\}$ The number of maps from an $n$--element set to $\{0,1\}$.
121: \item $\{B(n)\}$ The Bell numbers; the number of ways of putting $n$
122: different objects into $n$ identical containers (some may be left
123: empty).
124: \end{enumerate}
125: These are all familiar except perhaps the last, the Bell numbers.
126: These have the values
127: $$
128: B(n)=1,\,2,\,5,\,15,\,52,\,203,\,\ldots,\quad n=1,\,2,\,\ldots\,.
129: $$
130: and $B(n)\leq n!$.
131:
132: We may also define sequences which depend on, say, two parameters,
133: for exam\-ple the 'choose' symbols
134: $^nC_k\equiv\dfrac{n!}{(n-k)!\;k!}$. Related to the Bell numbers
135: are the {\em Stirling numbers of the second kind} $S(n,k)$, which
136: are defined as the number of ways of putting $n$ different objects
137: into $k$ identical containers, leaving none empty. From the
138: definition we have
139: \begin{equation}\label{sti}
140: B(n)=\sum_{k=1}^n S(n,k)\,.
141: \end{equation}
142:
143: \subsection{Normal order}
144:
145: Although somewhat unfamiliar to physicists, the Bell and Stirling
146: numbers are fundamental in quantum theory. This is because they
147: arise naturally in the {\em normal ordering problem}. For
148: canonical bosons $[a,a^\dag]=1$ the Stirling numbers of the
149: second kind $S(n,k)$ intervene through \cite{[3]}
150: \begin{eqnarray}
151: (a^\dag a)^n=\sum_{k=1}^nS(n,k) (a^\dag)^k a^k\,.
152: \end{eqnarray}
153: The corresponding Bell numbers $B(n)=\sum_{k=1}^n S(n,k)$ are
154: simply the expectation values
155: \begin{equation}\label{Bell}
156: B(n)=\langle z|(a^{\dagger}a)^n|z\rangle_{z=1}
157: \end{equation}
158: taken in the coherent state defined by
159: \begin{equation}\label{cs}
160: a|z\rangle=z|z\rangle
161: \end{equation}
162: for $z=1$. In fact, for physicists, these equations may be taken
163: as the {\em definitions} of the Stirling and Bell numbers.
164:
165: \subsection{Generating functions}
166:
167: Corresponding to a sequence of combinatorial numbers $\{a_n\}$ we
168: may define the {\em exponential generating function (egf)}
169: \begin{equation}\label{egf}
170: A(x)=\sum_{n=0}^\infty a_n\frac{x^n}{n!}\,.
171: \end{equation}
172: This provides a compact formula for the combinatorial sequence.
173: For example for the sequences above
174: \begin{enumerate}
175: \item $a_n=n$,\quad $A(x)=x\E^x$,
176: \item $a_n=n!$,\quad $A(x)=1/(1-x)$,
177: \item $a_n=2^n$,\quad $A(x)=\E^{2x}$,
178: \item $a_n=B(n)$,\quad $A(x)=\exp\left(\E^{x}-1\right)$.
179: \end{enumerate}
180: Except for the last, which we shall prove in what follows, these
181: exponential generating functions are immediate. Note that we
182: consider the expansion in a {\em formal} sense; we are not
183: concerned here about the convergence of the series defined by the
184: egf's. We may similarly define egf's corresponding to 2--parameter
185: sequences, thus:
186: \begin{equation}\label{egf1}
187: A(x,y)=\sum_{n=0}^\infty\biggl(\;\sum_{k=0}^\infty
188: a_{n,k}y^k\biggr) \frac{x^n}{n!}\,.
189: \end{equation}
190: For the 2--parameter sequences above we obtain:
191: \begin{enumerate}
192: \item $a_{n,k}\equiv ^n\!C_k$,\quad $A(x,y)=\exp((1+y)x)$,
193: \item $a_{n,k}\equiv S(n,k)$,\quad $A(x,y)=\exp\bigl(y(\E^{x}-1)\bigr)$.
194: \end{enumerate}
195: In the latter function, the coefficient of the $n$--th power of
196: $x$ is an $n$--degree polynomial in $y$, a Bell polynomial,
197: $B_n(y)$.
198:
199: Therefore we have the generating function for the Bell numbers and
200: Bell polynomials
201: \begin{eqnarray}\label{bgf}
202: A_{B}(x)&=&\sum_{n=0}^\infty B(n)\,\frac{x^n}{n!}= \nonumber \\
203: &=&\exp\left(x+\frac{x^2}{2!}+\frac{x^3}{3!}+\ldots\right)
204: \end{eqnarray}
205: and
206: \begin{eqnarray}\label{bpgf}
207: A_{B}(x,y)&=&\sum_{n=0}^\infty\biggl(\;\sum_{k=0}^\infty
208: S(n,k)y^k\biggr)\frac{x^n}{n!}= \nonumber \\
209: &=&\sum_{n=0}^\infty B_n(y)\,\frac{x^n}{n!}= \nonumber \\
210: &=&\exp\bigl(y(\E^{x}-1)\bigr).
211: \end{eqnarray}
212:
213: \subsection{Graphs}
214:
215: We now give a graphical representation of the Bell numbers.
216: Consider labelled lines which emanate from a white dot, the
217: origin, and finish on a black dot, the vertex. We shall allow
218: only one line from each white dot but impose no limit on the
219: number of lines ending on a black dot. Clearly this simulates the
220: definition of $S(n,k)$ and $B(n)$, with the white dots playing the
221: role of the distinguishable objects, whence the lines are
222: labelled, and the black dots that of the indistinguishable
223: containers. The identification of the graphs for 1,2 and 3 lines
224: is given in the Figure 1.
225:
226: \bfg[ht] \bc \resizebox{8.5cm}{!}{\includegraphics{sol.eps}} \ec
227: \vspace{-7mm} \caption{Graphs for $B(n)$, $n=1$, 2, 3.}
228: \label{inter} \efg
229:
230: We have concentrated on the Bell number sequence and its
231: associated graphs since, as we shall show, there is a sense in
232: which this sequence of graphs is {\em generic}. That is, we can
233: represent {\em any} combinatorial sequence by the same sequence of
234: graphs as in the Figure 1, with suitable vertex multipliers
235: (denoted by the $V$ terms in the same figure). One reason the
236: graphical representation is useful is that there exist some rather
237: powerful results which apply to graphs and their associated
238: generating functions. We give one such result now.
239:
240: \subsection{Connected graph theorem}
241:
242: This states that if $C(x)=\sum_{n=1}^{\infty} c(n)x^n/n!$ is the
243: exponential generating function of {\em labelled connected}
244: graphs, {\em viz.} $c(n)$ counts the number of connected graphs of
245: order $n$, that is with $n$ lines, then
246: \begin{equation}
247: A(x) = \exp\bigl(C(x)\bigr)
248: \end{equation}
249: is the exponential generating function for {\em all}
250: graphs \cite{[4]}.
251:
252: We may apply this very simply to the case of the $B(n)$ graphs
253: in Figure 1. For each order $n$, the {\em connected} graphs
254: clearly consist of a single graph. Therefore for each $n$ we have
255: $c(n)=1$; whence, $C(x)=exp(x)-1.$ It follows that the generating
256: function for {\em all} the graphs $A(x)$ is given by
257: \begin{equation}
258: A(x)=\exp\bigl(\exp(x) - 1\bigr)\,, \label{bell}
259: \end{equation}
260: which is therefore the generating function for the Bell numbers.
261:
262: Given a general generating function
263: $C(x)=\sum_{n=1}^\infty{V_nx^{n}/n!}$ we may represent this by the
264: same sequence of graphs as in Figure 1, but now with vertex
265: factors ${V_n}= c(n)$, as shown in Figure 1.
266:
267: This is of course a trivial observation, since any sequence of
268: graphs each term of which consists of a single connected graph
269: will, with the appropriate multipliers $V_{n}$, give by definition
270: the exponential generating function of many combinatorial
271: sequences. This however shows the generic nature of the $B(n)$
272: sequence, as previously noted in \cite{[4a]}; and will prove of
273: value in our discussion of partition functions.
274:
275: \section{Partition functions}
276:
277: We first illustrate the use of normal ordering in the context of
278: {\em partition functions} by the most simple example. We define
279: as usual the partition function $Z$ associated with a Hamiltonian
280: $H$ by $Z\equiv {\rm Tr}\exp(-\beta H)$.
281:
282: \subsection{Free boson gas and Bell polynomials}
283: \label{sub;ho}
284:
285: We take as our example the Hamiltonian for the single--mode free
286: boson gas $H=\epsilon a^{\dagger}a$ (ignoring an additive
287: constant), $\epsilon>0$. The usual computation of the partition
288: function, exploiting the completeness property
289: $\sum_{n=0}^{\infty}|n\rangle\langle n|=I$, is immediate:
290: \begin{eqnarray}\label{pf1}
291: Z&=& {\rm Tr}\exp(-\beta \epsilon a^{\dagger}a)=\\
292: &=& \sum_{n=0}^\infty{\langle n|\E^{-\beta \epsilon \hat{n}}|n\rangle}= \\
293: &=& \sum _{n=0}^\infty{\E^{-\beta \epsilon n}}=\\
294: &=&\left(1-\E^{-\beta \epsilon}\right)^{-1}\,.
295: \end{eqnarray}
296: However, we may, somewhat perversely in this simple case, use
297: {\em any} complete set to perform the trace. Choosing coherent
298: states, for which the completeness or {\em resolution of unity}
299: property is
300: \begin{equation}\label{cs1}
301: \frac{1}{\pi}\int \D^{2}z |z\rangle\langle z|=I\equiv
302: \int \D\mu(z)|z\rangle\langle z|
303: \end{equation}
304: the appropriate trace calculation is
305: \begin{eqnarray}\label{tr1}
306: Z&=&\frac{1}{\pi}\int \D^{2}z \langle z|\exp\bigl(-\beta \epsilon a^{\dagger}a\bigr)|z\rangle=\\
307: &=&\frac{1}{\pi}\int \D^{2}z \langle
308: z|:\exp\bigl(a^{\dagger}a(\E^{-\beta\epsilon}-1)\bigr):|z\rangle\,,
309: \end{eqnarray}
310: where we have used the following well-known relation
311: \cite{[5],[6]} for the {\em forgetful} normal ordering operator
312: $:\!f(a, a^{\dagger})\!:$ which means ``normally order the
313: creation and annihilation operators in $f$ {\em forgetting} the
314: commutation relation $[a,a^{\dagger}]=1$''\footnote{Of course,
315: this procedure may alter the value of the operator to which it is
316: applied.}:
317: \begin{equation}\label{no1}
318: \mathcal{N}\bigl(\exp(x
319: a^{\dagger}a)\bigr)=:\exp\bigl(a^{\dagger}a(\E^{x}-1)\bigr):.
320: \end{equation}
321: We therefore obtain, integrating over the angle variable $\theta$
322: and the radial variable $r=|z|$,
323: \begin{equation}
324: Z=\frac{1}{\pi}\int_0^{2\pi}\D\theta\int_0^\infty r\,\D r
325: \exp\left(r^2\bigl(\E^{-\beta \epsilon}-1\bigr)\right)\,,
326: \label{z1}
327: \end{equation}
328: which gives us $Z=\bigl(1-\E^{-\beta\epsilon}\bigr)^{-1}$ as
329: before.
330:
331: We rewrite the above equation to show the connection with our
332: previously--defined combinatorial numbers. Writing $y=r^2$ and
333: $x=-\beta \epsilon$, Eq.(\ref{z1}) becomes
334: \begin{equation}\label{z2}
335: Z=\int\limits_0^\infty \D y\exp\bigl(y(\E^{x}-1)\bigr)\,.
336: \end{equation}
337: This is an integral over the classical generating function for
338: the Bell polynomials
339: \begin{equation}\label{bpgf1}
340: \exp\left(y\bigl(\E^{x}-1\bigr)\right)=\sum_{n=0}^\infty B_n(y)\,\frac{x^n}{n!}\,,
341: \end{equation}
342: where $B_{n}(1)=B(n)$, as defined in Eq.(\ref{Bell}). This leads
343: to the combinatorial form for the partition function
344: \begin{equation}\label{z3}
345: Z=\int_0^\infty\D y\sum_{n=0}^\infty B_n(y)\,\frac{x^n}{n!}\,.
346: \end{equation}
347:
348: Although Eq.(\ref{z3}) is remarkably simple in form, it is often
349: by no means a straightforward matter to evaluate the analogous
350: integral for other than the free boson system considered here.
351: Further, it is also clear that we may not interchange the integral
352: and the summation, as each individual $y$ integral diverges. We
353: shall therefore concentrate in what follows on the {\em partition
354: function integrand} (PFI) $F(z)=\langle z|\exp(-\beta
355: H)|z\rangle$, whence $Z=\int F(z)\,\D\mu(z)$, to give a graphical
356: description of a perturbation approach. The function $F$ maps
357: coherent states $|z\rangle$ to (real) numbers, and is related to,
358: but not identical with, the previously-introduced {\em free energy
359: functional} \cite{[7]} associated with the Hamiltonian system.
360:
361: \subsection{General partition functions}
362:
363: We now apply this graphical approach to the general partition
364: function in second quantized form. With the usual definition for
365: the partition function
366: \begin{equation}\label{pf}
367: Z={\rm Tr}\exp(-\beta H)\,.
368: \end{equation}
369: In general the Hamiltonian is given by $H=\epsilon
370: w(a,a^{\dagger})$, where $\epsilon$ is the energy scale, and $w$
371: is a string (= sum of products of positive powers) of boson
372: creation and annihilation operators. The partition function
373: integrand $F$ for which we seek to give a graphical expansion, is
374: \begin{equation}\label{pfz}
375: Z(x)=\int{F(x,z)\,\D\mu(z)}\,,
376: \end{equation}
377: where
378: \begin{eqnarray}
379: F(x,z)&=&\langle z|\exp(xw)|z\rangle= \hskip20mm(x=-\beta \epsilon)\nonumber \\
380: &=&\sum_{n=0}^{\infty}\langle z|w^n|z\rangle\,\frac{x^n}{n!}=\nonumber \\
381: &=&\sum_{n=0}^{\infty}W_n(z)\,\frac{x^n}{n!}=\nonumber \\
382: &=&\exp\biggl(\;\sum_{n=1}^{\infty}V_n(z)\,\frac{x^n}{n!}\biggr),
383: \end{eqnarray}
384: with obvious definitions of $W_n$ and $V_n$. The sequences
385: $\left\{W_n\right\}$ and $\left\{V_n\right\}$ may each be
386: recursively obtained from the other \cite{[7a]}. This relates the
387: sequence of multipliers $\{V_n\}$ of Figure 1 to the Hamiltonian
388: of Eq.(\ref{pf}). The lower limit $1$ in the $V_n$ summation is a
389: consequence of the normalization of the coherent state
390: $|z\rangle$.
391:
392: We conclude this discussion with an example.
393:
394: \subsection{Superfluid boson system}
395:
396: We may write the Hamiltonian for a superfluid boson system as
397: \cite{[8]}
398: \begin{equation}\label{sbs}
399: H=\epsilon\left\{\sfrac12\,c_1a^2+\sfrac12\,\bar{c}_1(a^{\dagger})^2+
400: c_2\bigl(a^{\dagger}a+\sfrac12\bigr)\right\}.
401: \end{equation}
402: As in the non-interacting boson gas example, this represents a
403: single mode of the system; the complete Hamiltonian is a direct
404: sum of non-interacting terms. The constants $c_1$ and $c_2$ are
405: complex and real respectively, but otherwise arbitrary. Noting
406: that the Hamiltonian Eq.(\ref{sbs}) is an element of the
407: non-compact Lie algebra $su(1,1)$ enables an expression for the
408: partition function integrand to be readily obtained (see the
409: Appendix). To illustrate the graphical series for this PFI, we
410: shall choose real arguments; that is we put $z$ and the constant
411: $c_1$ real in Eq.(\ref{pf8}) below to get
412: \begin{equation}\label{pf4}
413: F(x,y)=\sqrt{\frac1{\mu}}\,\exp\left[y\left(y_1+\frac{1}{\mu}-1\right)\right],
414: \end{equation}
415: where $y=z^2$ and $y_1$, $\mu$ are functions of $x$. The PFI
416: $F(x,y)$ of Eq.(\ref{pf4}) generates an integer sequence for even
417: $c_1$, $c_2$ as may be easily verified using algebraic software.
418: For example, choosing $c_1=2$, $c_2=4$ we have
419: \begin{equation}\label{pf5}
420: F(x,y)=\exp\left(\frac{V_1(y)x}{1!}+\frac{V_2(y)x^2}{2!}+
421: \frac{V_3(y)x^3}{3!}+\ldots\right),
422: \end{equation}
423: with
424: \begin{equation}
425: \begin{array}{l}
426: V_1(y)=2+6y\,,\\
427: V_2(y)=2+36y\,,\\
428: V_3(y)=16+288y\,,\\
429: V_4(y)=144+3024y\,,\\
430: \ldots\,.
431: \end{array}
432: \end{equation}
433: leading to a corresponding Bell--type graphical expansion.
434:
435: \section{Discussion}
436:
437: The normal ordering of creation and annihilation operators leads
438: naturally to Stirling and Bell numbers. Therefore these
439: combinatorial numbers, and their graphical representations, are
440: ubiquitous in second--quantized quantum physics. In this talk we
441: emphasized this fact by application to a perturbative expansion
442: for the partition function of statistical physics. We did this
443: through the introduction of the {\em partition function integrand}
444: or PFI, a function which when integrated over the complete set of
445: coherent states gives the partition function proper. The
446: evaluation of the PFI depends on the normal ordering of the the
447: exponentiated Hamiltonian given in terms of second--quantized
448: operators. In general this is a difficult if not intractable
449: problem, although approaches based on the {\em Product Theorem} of
450: combinatorial graph theory lead to some
451: simplification\footnote{See the talk by A. Horzela {\em et al} in
452: this conference.}. However, even with the straightforward methods
453: illustrated here, solutions are obtainable in many cases of
454: physical interest. In particular, we illustrated this approach in
455: the case of the free boson gas, and a superfluid boson model. We
456: also gave a graphical representation for the perturbation
457: expansion of the PFI, an expansion in terms of $x=-\beta\epsilon$,
458: essentially energy scale times inverse temperature. The graphical
459: series is that of the Bell numbers, together with appropriate
460: multipliers $V_n(z)$, where $z$ is the coherent state parameter
461: which must be integrated over to obtain the partition function.
462: We showed how to determine these multipliers in terms of the
463: Hamiltonian. We conclude that there is a sense in which the Bell
464: graph series is generic for the evaluation of PFI's.
465:
466: The operators of this theory are quantized, but have no explicit
467: dependence on space or time, which justifies considering the
468: theory as a zero--dimensional field theory, with the graphs as
469: being analogous to zero-dimensional Feynman diagrams.
470:
471: \section*{Appendix: Partition function integrand for a superfluid boson system}
472:
473: The partition function integrand $F(x,z)$ corresponding to the
474: hamiltonian Eq.(\ref{sbs}) is the expectation of the following
475: operator
476: \begin{eqnarray}
477: \exp(-\beta H)&=&\exp\left\{\sfrac12\,xc_1a^2+\sfrac12\,x\bar{c}_1(a^{\dagger})^2+
478: xc_2(a^{\dagger}a+\sfrac12)\right\}= \nonumber \\
479: &=&\exp\left\{\sfrac12\,x_1
480: a^2+\sfrac12\,\bar{x}_1(a^{\dagger})^2+x_2(a^{\dagger}a+\sfrac12)\right\}=
481: \qquad\qquad\left(\mathrm{for~}\ba{c}x_1=xc_1\,,\\x_2=xc_2\,,\ea\right)\nonumber \\
482: &=&\exp\{x_1 K^{-}+\bar{x}_1K^{+}+2x_2 K^{0}\}\,, \label{pf6}
483: \end{eqnarray}
484: where we have identified the generators $\{K^{+},K^{-},K^{0}\}$ of
485: $su(1,1)$ with the appropriate operators
486: $\bigl\{\frac12(a^{\dagger})^2,\frac12a^2,\frac{1}{2}(a^{\dagger}a+\frac12)\bigr\}$.
487: The normally ordered form of Eq.(\ref{pf6}) is
488: \begin{equation}\label{pf7}
489: \exp(-\beta H)=\E^{\bar{y}_1K^+}\E^{2y_2K^0}\E^{y_1K^-}\,.
490: \end{equation}
491: Since the equality of Eq.(\ref{pf6}) and Eq.(\ref{pf7}) is a
492: group--theoretical result, we may use the $2\times2$
493: representation
494: \begin{equation}\label{2x2}
495: {\slacs{1ex} \{K^{+},K^{-},K^{0}\}=\left\{\begin{pmatrix}
496: 0&1\\
497: 0&0 \end{pmatrix},
498: \begin{pmatrix}
499: 0&0\\
500: -1&0 \end{pmatrix},
501: \begin{pmatrix}
502: \frac{1}{2}&0\\
503: 0&\frac{1}{2} \end{pmatrix}
504: \right\},}
505: \end{equation}
506: in which to implement the equality, thus evaluating $y_1$ and
507: $y_2$. The result is
508: $$
509: y_1=xc_1\sinh\left(\frac{\delta}{\mu\delta}\right),\quad
510: y_2=-\ln\mu\,,
511: $$
512: where
513: $$
514: \delta=x\sqrt{{c_2}^2-|c_1|^2}\,,\quad
515: \mu=\cosh\delta-xc_2\frac{\sinh\delta}{\delta}\,.
516: $$
517: We therefore have for the Partition Function Integrand
518: \begin{eqnarray}\label{pf8}
519: F(x,z)&=&\langle z|\exp\bigl\{\sfrac12\,x_1 a^2+\sfrac12\,\bar{x}_1(a^{\dagger})^2+
520: x_2 (a^{\dagger}a+\sfrac12)\bigr\}|z\rangle= \nonumber \\
521: &=&\bigl<z\bigr|\E^{\bar{y}_1 (a^{\dagger})^2/2}\E^{y_2(a^{\dagger}a+1/2)}
522: \E^{y_1 a^2/2}\bigl|z\bigr>= \nonumber \\
523: &=&\E^{\frac12\bar{y}_1\bar{z}^2}\E^{|z|^2(e^{y_2}-1)}\E^{y_1z^2/2}e^{\frac{1}{2}y_2}\,.
524: \end{eqnarray}
525:
526: \bigskip\noindent
527: {\small P.B. wishes to thank the Polish Ministry of Scientific
528: Research and Information Technology for support under Grant no: 1
529: P03B 051 26}
530:
531: \bbib{99}
532: \bibitem{[1]}
533: J. Katriel: Lett. Nuovo Cimento \textbf{10} (1974) 565.
534: \bibitem{[2]}
535: P. Blasiak, K.A. Penson, and A.I. Solomon: Ann. Comb. \textbf{7}
536: (2003) 127.
537: \bibitem{[3]}
538: J. Katriel: Phys. Lett. A. \textbf{273} (2000) 159.
539: \bibitem{[4]}
540: G.W. Ford and G.E. Uhlenbeck: Proc. Nat. Acad. \textbf{42} (1956)
541: 122.
542: \bibitem{[4a]}
543: C.M. Bender, D.C. Brody, and B.K. Meister: J. Math. Phys.
544: \textbf{40} (1999) 3229.
545: \bibitem{[5]}
546: J.R. Klauder and E.C.G. Sudarshan: \textit{Fundamentals of Quantum
547: Optics}. Benjamin, New York, 1968.
548: \bibitem{[6]}
549: W.H. Louisell: \textit{Quantum Statistical Properties of
550: Radiation}. J. Wiley, New York, 1990.
551: \bibitem{[7]}
552: M. Rasetti: Int. J. Theor. Phys. \textbf{14} (1975) 1.
553: \bibitem{[7a]}
554: M. Pourahmadi: Amer. Math. Monthly \textbf{91} (1984) 303.
555: \bibitem{[8]}
556: A.I. Solomon: J. Math. Phys. \textbf{12} (1971) 390.
557: \ebib
558: \end{document}
559: