1: % Submitted to Journal of Physics A:Math. and Gene
2: % Authors: Accardi L, Pechen A N, Volovich I V
3: % This is written in Latex2
4:
5: \documentclass[12pt]{iopart}
6: \usepackage{amsfonts}
7: \newcommand{\ex}{\frac{\rme^{\rmi(t'-t)(E_1-E_2)/\la^2}}{\la^2}}
8: \newcommand{\ga}{\gamma}
9: \newcommand{\la}{\lambda}
10: \newcommand{\om}{\otimes}
11: \newcommand{\ep}{\varepsilon}
12: \newcommand{\dl}{\delta}
13: \newcommand{\ba}{\begin{array}{l}}
14: \newcommand{\ea}{\end{array}}
15: \newtheorem{theorem}{Theorem}
16: \newtheorem{lemma}{Lemma}
17: %\binoppenalty=10000
18: %\relpenalty=10000
19:
20:
21: \begin{document}
22:
23: \title[Quantum stochastic equation]
24: {Quantum stochastic equation for the low density limit}
25:
26: \author{L Accardi\dag, A N Pechen\dag\footnote[3]{permanent address:
27: Steklov Mathematical Institute of Russian Academy of
28: Sciences, Gubkin St.8, GSP-1, 117966, Moscow, Russia}
29: and I V Volovich\ddag}
30:
31: \address{\dag\ Centro Vito Volterra, Universita di Roma Tor Vergata 00133,Italia}
32:
33: \address{\ddag\ Steklov Mathematical Institute of Russian Academy of
34: Sciences, Gubkin St.8, GSP-1, 117966, Moscow, Russia}
35:
36: \eads{\mailto{accardi@volterra.mat.uniroma2.it}, \mailto{pechen@mi.ras.ru},
37: \mailto{volovich@mi.ras.ru}}
38:
39: \begin{abstract}
40: A new derivation of quantum stochastic differential equation
41: for the evolution operator in the low density limit is presented.
42: We use the distribution approach and derive a new algebra for
43: quadratic master fields in the low density limit by using
44: the energy representation. We formulate the stochastic golden rule
45: in the low density limit case
46: for a system coupling with Bose field via quadratic interaction.
47: In particular the vacuum expectation value of the evolution
48: operator is computed and its exponential decay is shown.
49:
50: \end{abstract}
51:
52: \submitto{Journal of Physics A: Mathematical and General}
53:
54: \pacs{02.50.EY, 02.50.FZ, 03.65.YZ, 05.10.Gg and 42.50.Lc}
55:
56: \maketitle
57:
58:
59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60:
61: \section{Introduction}
62:
63: There are many studies of the large time behavior in quantum
64: theory. One of the powerful methods is the stochastic limit. Many
65: important physical models have been investigated by using this
66: method (see \cite{AFL,AKV,ALV} for more discussions). This method,
67: however, is restricted to the studying of the long time behavior
68: of the models in the weak coupling case, i.e. when coupling
69: constant is a small parameter, and it can not be applied directly
70: to important class of models which contain terms in the
71: interaction without small coupling constant.
72:
73: This last class of
74: models includes models in which the small parameter is the
75: density. Such models naturally arise in the low density limit
76: (LDL). The LDL for a classical Lorentz gas is the Boltzmann-Grad limit.
77: For classical systems there has been a considerable progress in
78: the rigorous derivation of the Boltzmann equation.
79: Lanford~\cite{lanford} using ideas of Grad~\cite{grad} proved the
80: convergence of the hierarchy of correlation functions for a hard
81: sphere gas in the Boltzmann--Grad limit for sufficiently short
82: times. This proof was extended by King~\cite{king} to positive
83: potentials of finite range. The limiting evolution of the one
84: particle distribution is governed by the non-linear Boltzmann
85: equation.
86:
87: The test particle problem was studied by many authors(see
88: a review of Spohn~\cite{spohn2}). One considers
89: the motion of a single particle through an environment of randomly
90: placed, infinitely heavy scatterers (Lorentz gas). In the
91: Boltzmann--Grad limit successive collisions become independent and
92: the averaged over the positions of the scatterers the position and
93: velocity distribution of the particle converges to the solution of
94: the linear Boltzmann equation.
95:
96: To describe a quantum physical model to which LDL can be applied
97: let us consider an N-level atom immersed in a free gas whose
98: molecules can collide with the atom; the gas is supposed to be
99: very dilute. Then the reduced time evolution for the atom will be
100: Markovian, since the characteristic time $t_S$ for appreciable
101: action of the surroundings on the atom (time between collisions)
102: is much larger than the characteristic time $t_R$ for relaxation
103: of correlations in the surroundings. Rigorous results
104: substantiating this idea have been obtained in~\cite{dumcke}.
105:
106: It is known~\cite{AcLu} that the dynamics of the N-level atom
107: interacting with the free gas converges, in the low density limit,
108: to the solution of a quantum stochastic differential equation
109: driven by quantum Poisson noise. Indeed, from a semiclassical
110: point of view, collision times, being times of occurrence of rare
111: events, will tend to become Poisson distributed, whereas the
112: effect of each collision will be described by the
113: (quantum-mechanical) scattering operator of the atom with one gas
114: particle (see the description of the quantum Poisson process
115: in~\cite{kumerrer}).
116:
117: In this paper a derivation of the quantum stochastic differential
118: equation for the evolution operator in the low density limit is
119: presented. The equation obtained is equivalent to the stochastic
120: equation which has been derived in \cite{AcLu} but we use a new
121: method. We use the distribution approach~\cite{AKV,ALV}
122: and derive a new algebra for quadratic master
123: fields in the low density limit by using the energy
124: representation. An advantage of this method is the simplicity of
125: derivation of quantum stochastic equations and computation of
126: correlation functions. We formulate the stochastic golden rule in
127: the low density limit case for a system coupling with Bose field
128: via quadratic interaction. In particular the vacuum expectation
129: value of the evolution operator is computed and its exponential
130: decay is shown.
131:
132: Main results of the paper are quantum stochastic differential
133: equation~(\ref{normordeq}),
134: the new algebra of commutation relations for the master
135: field (Theorem~\ref{mastfi}) and the expression for the expectation value
136: of the evolution operator~(\ref{expvalue}).
137:
138: In this paper we obtain unitary evolution which is given by
139: the solution of the quantum stochastic differential equation.
140: Using this equation one can obtain corresponding quantum
141: Langevin and master equations.
142:
143: An important problem in theory of open quantum systems is the
144: rigorous derivation of quantum Boltzmann equation from microscopic
145: dynamics. Aa approach to the derivation of classical and quantum
146: Boltzmann equations, based on BBGKY-hierarchy, has been presented
147: in the work of Bogoliubov~\cite{bogol}. The low density limit for the model
148: under consideration, with completely different
149: methods, based on quantum BBGKY hierarchy has been investigated by
150: D\"umcke~\cite{dumcke}.
151:
152: Let us explain our notations. We consider a quantum model of
153: a test particle interacting with a reservoir (heat bath). We
154: shall restrict ourselves to the case of a Boson reservoir in this
155: paper. Let $ {\cal H}_{\rm S} $ be the Hilbert space of the system
156: (test particle) with the system Hamiltonian $ H_{\rm
157: S} $. The system Hilbert space for the $N$-level atom ${\cal
158: H}_{\rm S} = \mathbb C^N$. The reservoir is described by the
159: Boson Fock space $\Gamma({\cal H}_1)$ over the one particle
160: Hilbert space ${\cal H}_1 = L^2(\mathbb R^d)$, where $d=3$ in
161: physical case. Moreover, the Hamiltonian of the reservoir is
162: given by $H_{\rm R}:=\rmd\Gamma(H_1)$ (the second quantization of
163: the one particle Hamiltonian $ H_1 $) and the total Hamiltonian
164: of the compound system is given by a self--adjoint operator
165: on the total Hilbert space ${\cal H}_{\rm S}\otimes \Gamma({\cal
166: H}_1)$, which has the form
167: \[
168: H_{\rm tot}:=H_{\rm S}\otimes1+1\otimes H_{\rm R}+H_{\rm int}=:H_{\rm free} +
169: H_{\rm int}.
170: \]
171: Here $H_{\rm int}$ is the interaction Hamiltonian between the
172: system and the reservoir.
173: The evolution operator at time $t$ is given by:
174: \[
175: U_t:=\rme^{\rmi tH_{\rm free}}\cdot \rme^{-\rmi tH_{\rm tot}}.
176: \]
177: Obviously, it satisfies the differential equation
178: \[
179: \partial_tU_t=-\rmi H_{\rm int}(t)U_t
180: \]
181: where the quantity $H_{\rm int}(t)$ will be
182: called the evolved interaction and defined as
183: \[
184: H_{\rm int}(t)=\rme^{\rmi tH_{\rm free}} H_{\rm int}\rme^{-\rmi tH_{\rm free}}.
185: \]
186:
187: The interaction Hamiltonian will be assumed to have the following
188: form:
189: \[
190: H_{\rm int}:= D\otimes A^+(g_0)A(g_1)+D^+\otimes A^+(g_1)A(g_0)
191: \]
192: where $ D $ is a bounded operator in $ {\cal H}_S $, $ D\in {\bf
193: B}({\cal H}_S) $, $ A $ and $ A^+ $ are annihilation and creation
194: operators and $ g_0, g_1\in {\cal H}_1$ are form-factors
195: describing the interaction of the system with the reservoir.
196: This Hamiltonian preserves the particle number of the reservoir,
197: and therefore the particles of the reservoir are only scattered
198: and not created or destroyed. This model was considered by
199: Davies~\cite{davies} in the analysis of the weak coupling limit.
200: The development of the method to the Bose gas is a subject
201: of further works.
202:
203: With the notion
204: \[
205: S^0_t:=\rme^{\rmi tH_1} \ ;\qquad D(t):=\rme^{\rmi tH_{\rm S}}D\rme^{-\rmi tH_{\rm S}}
206: \]
207: the evolved interaction can be written in the form
208: \begin{equation}\fl
209: H_{\rm int}(t):=D(t)\otimes A^+(S^0_tg_0)A(S^0_tg_1)
210: +D^+(t)\otimes A^+(S^0_tg_1)A(S^0_tg_0)\label{Hint}
211: \end{equation}
212:
213: The initial state of the compound system is supposed to be of
214: the form
215: \[
216: \rho = \rho_{\rm S}\otimes\varphi^{(\xi)}.
217: \]
218: Here $\rho_{\rm S}$ is arbitrary density matrix of the system and
219: the initial state of the reservoir is the Gibbs state, at inverse
220: temperature $\beta$, of the free evolution, i.e. the gauge
221: invariant quasi-free state $\varphi^{(\xi)}$, characterized by
222: \begin{equation}\label{state}
223: \varphi^{(\xi)}(W(f))=\exp\Bigl(-{1\over 2}<f,(1+\xi\rme^{-\beta H_1})
224: (1-\xi\rme^{-\beta H_1})^{-1}f>\Bigr)
225: \end{equation}
226: for each $f\in {\cal H}_1$. Here $ W(f) $ is the Weyl operator,
227: $\beta$ the inverse temperature of the reservoir,
228: $\xi=\rme^{\beta\mu}$ the fugacity, $\mu$ the chemical potential.
229: We suppose that the temperature $\beta^{-1}>0$. Therefore for
230: sufficiently low density one is above the transition temperature,
231: and no condensate is present. The generalization to the case then
232: the condensate is present, is a subject of further investigations.
233:
234: We will study the dynamics, generated by the
235: Hamiltonian~(\ref{Hint}) and the initial state of the
236: reservoir~(\ref{state}) in the low density regime: $n\to 0$,
237: $t\sim 1/n$ ($n$ is the density of particles of the reservoir). In
238: the low density limit the fugacity $\xi$ and the density of
239: particles of the reservoir $n$ have the same asymptotic, i.e.
240: \[
241: \lim\limits_{n\to 0}\frac{\xi(n)}{n}=1
242: \]
243: Therefore the limit $n\to 0$ is equivalent to the limit $\xi\to
244: 0$.
245:
246: The low density limit for this model, with completely different
247: methods, based on quantum BBGKY hierarchy has been considered by
248: D\"umcke~\cite{dumcke}.
249:
250: Throughout the paper, for simplicity, the following technical
251: condition is assumed: the two test functions in the interaction
252: Hamiltonian have disjoint supports in the energy representation.
253: Thus the disjointness is invariant under the action of any
254: function of $H$. This assumption means that the two test function
255: $g_0, g_1$ in the interaction Hamiltonian satisfy:
256: \[
257: <g_0,S^0_t\rme^{-\beta H}g_1>=0\qquad \forall t\in{\mathbb R}.
258: \]
259: This means that, even if the particles of the reservoir have
260: generically a continuous energy spectrum, they behave like a
261: 2--level system as far as their interaction with the system is
262: concerned: if $P_0$ and $P_1$ project onto disjoint intervals
263: (energy bands) $I_0$ and $I_1$, these energy bands act as the
264: counterpart of the energy levels $\epsilon_0,\epsilon_1$ of the
265: system.
266:
267: Also the rotating wave approximation condition will be assumed.
268: This condition means that
269: \[
270: \rme^{\rmi tH_{\rm S}}D\rme^{-\rmi tH_{\rm S}}=\rme^{-\rmi t\omega_0}D
271: \]
272: where $ \omega_0 $ is a real number. This is a familiar
273: assumption, satisfied by all the Hamiltonians commonly used in
274: quantum optics. This assumption is satisfied if
275: $D=|\epsilon_0><\epsilon_1|$, where $|\epsilon_0>$ and $
276: |\epsilon_1>$ are eigenvectors of the free system Hamiltonian with
277: eigenvalues $\epsilon_0$ and $\epsilon_1$ so that $\omega =
278: \epsilon_1-\epsilon_0$.
279:
280: We will fix a projection operator $ P_0 $ in $ {\cal H}_1 $
281: commuting with $ H_1 $ and $ H $ and such that
282: \[
283: P_0g_0=g_0\quad {\rm and}\quad P_0g_1=0.
284: \]
285: Using this projection let us define the group $ \{S_t;\, t\in {\mathbb R}\} $
286: of unitary operators on $ {\cal H}_1 $ by
287: \[
288: S_t=S^0_t\rme^{-\rmi t\omega_0P_0}=\rme^{\rmi t(H_1-\omega_0P_0)}.
289: \]
290: The infinitesimal generator $ H_1' $ of $ S_t $ is given by
291: \[
292: H_1'=H_1-\omega_0P_0.
293: \]
294:
295: Following Palmer~\cite{Pl} we realize the representation space
296: as the tensor product of a Fock and anti-Fock representations.
297: Then the expectation values with respect to the state
298: $ \varphi^{(\xi)} $ for the
299: model with the interaction Hamiltonian~(\ref{Hint}) can be
300: conveniently represented as the vacuum expectation values in the
301: Fock-anti-Fock representation for the modified Hamiltonian.
302:
303: Denote by ${\cal H}_1^\iota$ the conjugate of ${\cal H}_1$, i.e.
304: \begin{eqnarray}
305: \iota :\ {\cal H}_1\longrightarrow {\cal H}_1\ ,\ \ \iota
306: (\la f):=\bar \la \iota (f) \nonumber \\
307: <\iota(f),\iota(g)>_\iota \ :=\ <g,f> \nonumber
308: \end{eqnarray}
309: then, ${\cal H}_1^\iota $ is a Hilbert space.
310: The corresponding Fock space $ \Gamma({\cal H}_1^\iota) $ is called
311: in this context the anti-Fock space.
312:
313: It was shown in~\cite{AcLu}
314: that with notations $ D_0=D,\, D_1=D^+ $ the modified Hamiltonian acting in
315: $ \Gamma({\cal H}_1)\otimes \Gamma({\cal H}_1^\iota) $ has the form
316: \begin{eqnarray*}\fl
317: H_\la(t)=\sum\limits_{\ep=0,1}D_\ep\otimes\bigl(A^+(S_tg_\ep)A(S_tg_{1-\ep})
318: \otimes 1\\
319: +\la(A(S_tg_{1-\ep})\otimes
320: A(S_tLg_\ep)+A^+(S_tg_\ep)\otimes A^+(S_tLg_{1-\ep})\bigr).
321: \end{eqnarray*}
322: Here $ A $ and $ A^+ $ are Bose annihilation and creation operators
323: acting in the Fock spaces $ \Gamma({\cal H}_1) $ and
324: $ \Gamma({\cal H}_1^\iota) $ and $ L:=\rme^{-\beta H/2} $.
325:
326: Moreover, it will be assumed that there exists a subset ${\cal K}$
327: (which includes $g_0, g_1$) of the one particle Hilbert space ${\cal
328: H}_1$, such that
329: \[
330: \int_{\mathbb R}|<f,S_tg>|dt<\infty\qquad \forall f,g\in {\cal K}.
331: \]
332:
333: The interaction Hamiltonian determines the evolution operator $ U_t^{(\la)} $
334: which is the solution of the Schr\"odinger equation
335: in interaction representation:
336: \[
337: \partial_tU_t^{(\la)}=-\rmi H_\la(t)U_t^{(\la)}
338: \]
339: with initial condition
340: \[
341: U_0^{(\la)}=1.
342: \]
343: One has the following integral equation for the evolution operator.
344: \[
345: U_t^{(\la)}=1-\rmi\int\limits_0^t\rmd t'H_\la(t')U_{t'}^{(\la)}.
346: \]
347:
348:
349: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
350:
351: \section{Energy representation}
352:
353: We will investigate the limit of the evolution operator when $ \xi\to +0 $
354: after the time rescaling $ t\to t/\xi $, where $ \xi=\la^2 $.
355: After this time rescaling
356: the equation for the evolution operator becomes
357: \[
358: \partial_tU_{t/\la^2}^{(\la)}=-\rmi\sum\limits_{\ep=0,1}D_\ep\om
359: \bigl(N_{\ep,1-\ep,\la}(t)+B_{1-\ep,\ep,\la}(t)+B^+_{\ep,1-\ep,\la}(t)\bigr)
360: U^{(\la)}_{t/\la^2}
361: \]
362: where we introduced the notations:
363: \begin{eqnarray*}
364: N_{\ep_1,\ep_2,\la}(t)=\frac{1}{\la^2}A^+(S_{t/\la^2}g_{\ep_1})
365: A(S_{t/\la^2}g_{\ep_2})\om 1 \\
366: B^+_{\ep_1,\ep_2,\la}(t)=\frac{1}{\la}
367: A^+(S_{t/\la^2}g_{\ep_1})\om A^+(S_{t/\la^2}Lg_{\ep_2}).
368: \end{eqnarray*}
369:
370: Let us introduce the energy representation for the creation
371: and annihilation operators by the formulae
372: \begin{equation}\label{AEdef}\fl
373: A^+_{\rm E}(g)=A^+(P_{\rm E}g)=\int\rmd k(P_Eg)(k)a^+(k)=\int\rmd k\dl(H_1'-E)g(k)a^+(k)
374: \end{equation}
375: \[
376: A_E(g)=A(P_Eg)
377: \]
378: Here
379: \[
380: P_E=\frac{1}{2\pi}\int\limits_{-\infty}^{\infty}\rmd tS_t\rme^{-\rmi tE}=
381: \dl(H_1'-E).
382: \]
383: It has the properties
384: \begin{eqnarray*}
385: S_t=\int\rmd EP_E\rme^{\rmi tE} \\
386: P_EP_{E'}=\dl(E-E')P_E \\
387: P^*_E=P_E
388: \end{eqnarray*}
389:
390: In the case when $ {\cal H}_1=L^2({\mathbb R}^d) $ the one-particle
391: Hamiltonian is the multiplication operator to the function
392: $ \omega(k) $ and acts on any element $ f\in L^2({\mathbb R}^d) $ as
393: $ H_1f(k)=\omega(k)f(k) $.
394:
395: It is easy to check that
396: \[
397: [A_E(f),A_{E'}^+(g)]=\dl(E-E')<f,P_Eg>.
398: \]
399: Here $ <\cdot ,\cdot > $ means the scalar product in $ {\cal H}_1 $.
400:
401: Using the energy representation one gets
402: \begin{eqnarray*}
403: N_{\ep_1,\ep_2,\la}(t)=\int\rmd E_1\rmd E_2N_{\ep_1,\ep_2,\la}(E_1,E_2,t) \\
404: B_{\ep_1,\ep_2,\la}(t)=\int\rmd E_1\rmd E_2B_{\ep_1,\ep_2,\la}(E_1,E_2,t)
405: \end{eqnarray*}
406: where
407: \begin{eqnarray*}
408: N_{\ep_1,\ep_2,\la}(E_1,E_2,t):=\frac{\rme^{\rmi t(E_1-E_2)/\la^2}}{\la^2}
409: A^+_{E_1}(g_{\ep_1})A_{E_2}(g_{\ep_2})\om 1 \\
410: B_{\ep_1,\ep_2,\la}(E_1,E_2,t):=
411: \frac{\rme^{\rmi t(E_2-E_1)/\la^2}}{\la}A_{E_1}(g_{\ep_1})\om A_{E_2}(Lg_{\ep_2}).
412: \end{eqnarray*}
413:
414: Let us also denote
415: \[
416: \ga_\ep(E):=\int\limits^0_{-\infty}\rmd t<g_\ep,S_tg_\ep>\rme^{-\rmi tE}.
417: \]
418:
419: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
420:
421: \section{The limiting commutation relations}
422:
423: Besides the operators $ B_{\ep_1,\ep_2,\la}(E_1,E_2,t) $ and $
424: N_{\ep_1,\ep_2,\la}(E_1,E_2,t) $ defined above let us consider the
425: following operators:
426: \[
427: N^{\iota}_{\ep_1,\ep_2,\la}(E_1,E_2,t)=
428: \frac{\rme^{\rmi t(E_2-E_1)/\la^2}}{\la^2}
429: \phantom{i}1\om A^+_{E_1}(Lg_{\ep_1})A_{E_2}(Lg_{\ep_2})
430: \]
431: with $ A^+_E(Lg_\ep) $ has been defined in (\ref{AEdef}).
432: The commutators of these operators are:
433:
434: \begin{eqnarray}
435: \fl [B_{\ep_1,\ep_2,\la}(E_1,E_2,t),B^+_{\ep_3,\ep_4,\la}(E_3,E_4,t')]\nonumber\\
436: \fl = \ex\Bigl(\dl_{\ep_1,\ep_3}\dl_{\ep_2,\ep_4}\dl(E_1-E_3)\dl(E_2-E_4)
437: <g_{\ep_1},P_{E_1}g_{\ep_1}><g_{\ep_2},P_{E_2}L^2g_{\ep_2}>\nonumber \\
438: \vphantom{\ex}
439: +\la^2\dl_{\ep_1,\ep_3}\dl(E_1-E_3)<g_{\ep_1},P_{E_1}g_{\ep_1}>
440: N^{\iota}_{\ep_4,\ep_2,\la}(E_4,E_2,t')& \nonumber\\
441: \vphantom{\ex}
442: +\la^2\dl_{\ep_2,\ep_4}\dl(E_2-E_4)<g_{\ep_2},P_{E_2}L^2g_{\ep_2}>
443: N_{\ep_3,\ep_1,\la}(E_3,E_1,t')\Bigr) \label{cr} \\
444: \fl [B_{\ep_1,\ep_2,\la}(E_1,E_2,t),N_{\ep_3,\ep_4,\la}(E_3,E_4,t')]\nonumber\\
445: =\dl_{\ep_1,\ep_3}\ex\dl(E_1-E_3)\nonumber\\
446: \times <g_{\ep_1},P_{E_1}g_{\ep_1}>
447: B_{\ep_4,\ep_2,\la}(E_4,E_2,t') \label{cr2} \\
448: \fl [N_{\ep_1,\ep_2,\la}(E_1,E_2,t),N_{\ep_3,\ep_4,\la}(E_3,E_4,t')]\nonumber\\
449: =\frac{\rme^{\rmi (t'-t)(E_3-E_1)/\la^2}}{\la^2}
450: \Bigl(\dl_{\ep_2,\ep_3}\dl(E_2-E_3)<g_{\ep_2},P_{E_2}g_{\ep_2}>
451: N_{\ep_1,\ep_4,\la}(E_1,E_4,t') \nonumber \\
452: \vphantom{\ex}
453: -\dl_{\ep_1,\ep_4}\dl(E_1-E_4)<g_{\ep_1},P_{E_1}g_{\ep_1}>{}
454: N_{\ep_3,\ep_2,\la}(E_3,E_2,t)\Bigr).\label{cr3}
455: \end{eqnarray}
456:
457: Notice that in the sense of distributions one has the limit
458: \begin{equation}\label{dllimit}
459: \lim\limits_{\la\to 0}\ex = 2\pi\dl(t'-t)\dl(E_1-E_2)
460: \end{equation}
461: and, in the sense of distributions over the standard simplex
462: (cf.~\cite{ALV}) one has the limit
463: \begin{equation}\label{dlplimit}
464: \lim\limits_{\la\to 0}\ex =\dl_+(t'-t)\frac{1}{\rmi (E_1-E_2-\rmi 0)}.
465: \end{equation}
466:
467: The following theorem describes the algebra of commutation
468: relations for the master field in the LDL.
469: \begin{theorem}\label{mastfi}
470: The limits
471: \[
472: X_{\ep_1,\ep_2}(E_1,E_2,t):=
473: \lim\limits_{\la\to 0}X_{\ep_1,\ep_2,\la}(E_1,E_2,t)\qquad (X=B,N)
474: \]
475: exist in the sense of convergence of correlators
476: and satisfy the (causal) commutation relations
477: \begin{eqnarray}
478: \fl [B_{\ep_1,\ep_2}(E_1,E_2,t),B^+_{\ep_3,\ep_4}(E_3,E_4,t')] \nonumber\\
479: \vphantom{\frac{P}{E_1}}
480: = \dl_{\ep_1,\ep_3}\dl_{\ep_2,\ep_4}\dl_+(t'-t)\dl(E_1-E_3)\dl(E_2-E_4)
481: \nonumber\\
482: \times\frac{<g_{\ep_1},P_{E_1}g_{\ep_1}>}{\rmi (E_1-E_2-\rmi 0)}
483: <g_{\ep_2},P_{E_2}L^2g_{\ep_2}>\label{crmf1} \\
484: \fl\vphantom{\frac{P}{E_1}}
485: [B_{\ep_1,\ep_2}(E_1,E_2,t),N_{\ep_3,\ep_4}(E_3,E_4,t')] \nonumber\\
486: = \dl_{\ep_1,\ep_3}\dl_+(t'-t)\dl(E_1-E_3)\frac{<g_{\ep_1},P_{E_1}g_{\ep_1}>}
487: {\rmi (E_1-E_2-\rmi 0)}B_{\ep_4,\ep_2}(E_4,E_2,t') \label{crmf2} \\
488: \fl [N_{\ep_1,\ep_2}(E_1,E_2,t),N_{\ep_3,\ep_4}(E_3,E_4,t')]=
489: \dl_+(t'-t)\frac{1}{\rmi (E_3-E_1-\rmi 0)} \nonumber \\
490: \times\Bigl(\dl_{\ep_2,\ep_3}\dl(E_2-E_3)<g_{\ep_2},P_{E_2}g_{\ep_2}>
491: N_{\ep_1,\ep_4}(E_1,E_4,t') \nonumber \\
492: - \dl_{\ep_1,\ep_4}\dl(E_1-E_4)<g_{\ep_1},P_{E_1}g_{\ep_1}>{}
493: N_{\ep_3,\ep_2}(E_3,E_2,t)\Bigr) \label{crmf3}
494: \end{eqnarray}
495: The commutation relations of the master field are obtained
496: by (\ref{crmf1}),(\ref{crmf2}),(\ref{crmf3}) replacing the factor
497: $ \dl_+(t'-t) $ by $ \dl(t'-t) $ and
498: $ (\rmi (E_1-E_2-\rmi 0))^{-1} $ by $ 2\pi\dl(E_1-E_2) $.
499: \end{theorem}
500: {\bf Proof.} The proof of the theorem follows by induction from
501: the commutation relations~(\ref{cr})-(\ref{cr3}) using the
502: limits~(\ref{dllimit}) and~(\ref{dlplimit}) and standard methods
503: of the stochastic limit.
504:
505: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
506:
507:
508: \section{The master space and the associated white noise}
509:
510: Let $ {\cal H}_{0,1} $ denote the closed subspace of
511: $ {\cal H}_1=L^2({\mathbb R}^d) $ spanned by the vectors
512: \[
513: S_tg_\ep,\quad \ep\in\{0,1\},\quad t\in{\mathbb R}.
514: \]
515: Let $ K $ be a non zero subspace of $ {\cal H}_{0,1} $
516: such that $ g_\ep\in K $ ($ \ep=0,1 $) and
517: \[
518: \int\limits_{-\infty}^\infty|<f,S_tg>|\rmd t<\infty\qquad\forall f,g\in K.
519: \]
520: This assumption implies that the sesquilinear form $ (\cdot|\cdot):K\times K
521: \longrightarrow C $ defined by
522: \[
523: (f|g)=\int\limits_{-\infty}^\infty<f,S_tg>\rmd t, \qquad f,g\in K
524: \]
525: is well defined. Moreover it defines
526: a pre-scalar product on $ K $. We denote $ \{K,(\cdot|\cdot)\} $ or
527: simply $ K $, the completion of the quotient of $ K $ by the zero
528: $ (\cdot|\cdot) $-norm elements.
529:
530: Define then Hilbert space $ K_{0,1}$ which will be denoted also as
531: $K\om_\beta K $ as the completion of the algebraic tensor product
532: $ K\om K^* $ with respect to the scalar product
533: \begin{eqnarray*}\fl
534: (f_0\om_\beta f_1|f'_0\om_\beta f'_1)&:=\int\limits_{-\infty}^\infty
535: <f_0,S_tf'_0>\overline{<f_1,S_tL^2f'_1>}\rmd t\\
536: & = \int\limits_{-\infty}^\infty <f_0,S_tf'_0><f'_1,S_{-t}L^2f_1>\rmd t.
537: \end{eqnarray*}
538:
539: Bounded operators acts naturally on $ K_{0,1} $ by
540: \[
541: (A\om_\beta B)(f_0\om_\beta f_1)=Af_0\om_\beta Bf_1
542: \qquad\forall A,B\in {\bf B}(K).
543: \]
544:
545: The limit reservoir (or master) space is the space
546: \[
547: {\cal F}(L^2({\mathbb R})\om K_{0,1}).
548: \]
549:
550: The (non-causal) commutation relations
551: (\ref{crmf1}),...,(\ref{crmf3}) mean that operators
552: $ B_{\ep_1,\ep_2}(E_1,E_2,t) $
553: are the white noise operators $ b_t(\cdot) $ in
554: $ {\cal F}(L^2({\mathbb R})\om K_{0,1}) $:
555: \[
556: B_{\ep_1,\ep_2}(E_1,E_2,t)=:b_t(P_{E_1}g_{\ep_1}\om_\beta P_{E_2}g_{\ep_2}).
557: \]
558:
559: The number operator is
560: \begin{eqnarray}\fl
561: N_{\ep_1,\ep_2}(E_1,E_2,t)=\sum\limits_{\ep'=0,1}n_{\ep'}(E_1)
562: b^+_t(g_{\ep_1}\om_\beta P_{E_1}g_{\ep'})
563: b_t(P_{E_2}g_{\ep_2}\om_\beta P_{E_1}g_{\ep'})\nonumber\\
564: =\sum\limits_{\ep'=0,1}\int\rmd En_{\ep'}(E_1)
565: B^+_{\ep_1,\ep'}(E,E_1,t)B_{\ep_2,\ep'}(E_2,E_1,t)\label{noperator}
566: \end{eqnarray}
567: where we denoted
568: \[
569: n_\ep(E):=\frac{1}{<g_\ep,P_EL^2g_\ep>}.
570: \]
571:
572: By identifying the element of the algebraic tensor product
573: $ f\om g\in K\om K^* $ with the operator
574: \[
575: |f><g|:\xi\in{\cal H}_1\to<g,\xi>f\in{\cal H}_1
576: \]
577: so that
578: \[
579: \bigl(|f><g|\bigr)^*=|g><f|
580: \]
581: and introducing the scalar product of such operators $ X,Y $
582: (notice that $ L^2=e^{-\beta H} $) by
583: \[
584: <Y,X>:=Tr\int\rme^{-\beta H}Y^*S_tXS^*_t\rmd t=
585: 2\pi Tr\int\rmd E\rme^{-\beta H}Y^*P_EXP_E
586: \]
587: one can rewrite white noise $ b_t(g\om_\beta f) $ as $ b_t(|g><f|) $
588: with commutation relations defined by
589: \[
590: [b_t(Y),b_{t'}^+(X)]=\dl(t'-t)<Y,X>.
591: \]
592:
593: Let us introduce for simplicity
594: \begin{eqnarray*}
595: B_{\ep_1,\ep_2}(E,t):=\int\rmd E'B_{\ep_1,\ep_2}(E',E,t)\\
596: N_{\ep_1,\ep_2}(E,t):=\int\rmd E'N_{\ep_1,\ep_2}(E,E',t)
597: \end{eqnarray*}
598: with (causal) commutation relations:
599: \begin{eqnarray*}\fl
600: [B_{\ep_1,\ep_2}(E,t),B^+_{\ep_3,\ep_4}(E',t')]=
601: \dl_{\ep_1,\ep_3}\dl_{\ep_2,\ep_4}\dl_+(t'-t)\dl(E-E')\ga_{\ep_1}(E)
602: <g_{\ep_2},P_EL^2g_{\ep_2}>\\
603: \fl [B_{\ep_1,\ep_2}(E,t),N_{\ep_3,\ep_4}(E',t')]=
604: \dl_{\ep_1,\ep_3}\dl_+(t'-t)\frac{<g_{\ep_1},P_{E'}g_{\ep_1}>}
605: {\rmi (E'-E-\rmi 0)}B_{\ep_4,\ep_2}(E,t').
606: \end{eqnarray*}
607:
608: In these notations the limiting Hamiltonian
609: acts on $ {\cal H}_S\om{\cal F}(L^2({\mathbb R})\om K_{0,1}) $ as
610: \[
611: H(t)=\int\rmd E\sum\limits_{\ep=0,1}D_\ep\om\bigl(
612: N_{\ep,1-\ep}(E,t)+B_{1-\ep,\ep}(E,t)+B_{\ep,1-\ep}^+(E,t)\bigr).
613: \]
614:
615: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
616:
617:
618: \section{Emergence of the drift term and annihilation process}
619:
620: The results of the preceding section allow us to write the
621: equation for the evolution operator in the stochastic limit
622: \begin{equation}\label{equ}\fl
623: \partial_tU_t=-\rmi H(t)U_t=-\rmi\sum\limits_{\ep=0,1}D_\ep\om(N_{\ep,1-\ep}(t)+
624: B_{1-\ep,\ep}(t)+B_{\ep,1-\ep}^+(t))U_t
625: \end{equation}
626: In order to bring it to the normally ordered form one needs to compute
627: the commutator
628: \[
629: -\rmi D_\ep[B_{1-\ep,\ep}(t),U_t]=-\rmi D_\ep\int\rmd E[B_{1-\ep,\ep}(E,t),U_t].
630: \]
631:
632: Notice that $ D_\ep D_{1-\ep} $ is a positive self-adjoint operator.
633: Therefore one can assume that for each $ E\in {\mathbb R} $,
634: the inverse operator
635: \[
636: T_\ep(E):=\bigl(1+(\ga_\ep\ga_{1-\ep})(E)D_\ep D_{1-\ep}\bigr)^{-1}
637: \]
638: always exists. Notice also that, since $ D_\ep D_{1-\ep} $
639: commutes with $ T_\ep(E) $, one has
640: \[
641: 1-D_\ep D_{1-\ep}(\ga_\ep\ga_{1-\ep})(E)T_\ep(E)=T_\ep(E).
642: \]
643: Therefore
644: \begin{equation}\label{TE}
645: \rmi D_\ep(1-D_{1-\ep}(\ga_\ep\ga_{1-\ep})(E)T_\ep(E)D_\ep)=\rmi T_\ep(E)D_\ep
646: \end{equation}
647: \begin{theorem}\label{normordu}
648: For the model described above one has
649: \begin{eqnarray}\fl
650: -\rmi D_\ep[B_{1-\ep,\ep}(t),U_t]=-\int\rmd E\ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)\nonumber\\
651: \times\Bigl(<g_\ep,P_EL^2g_\ep>U_t-
652: \rmi D_\ep\ga_\ep(E)U_tB_{1-\ep,\ep}(E,t)+U_tB_{\ep,\ep}(E,t)\Bigr)\label{commbu}
653: \end{eqnarray}
654: \end{theorem}
655: {\bf Proof.} Using the integral equation for the evolution operator
656: and the commutation relations~(\ref{crmf1}),(\ref{crmf2}), one gets
657: \begin{eqnarray}\fl
658: -\rmi D_\ep[B_{1-\ep,\ep}(E,t),U_t]\nonumber\\
659: \fl =-\sum\limits_{\ep'=0,1}D_\ep D_{\ep'}
660: \int\rmd E'\int\limits_0^t\rmd t_1[B_{1-\ep,\ep}(E,t),
661: N_{\ep',1-\ep'}(E',t_1)+B^+_{\ep',1-\ep'}(E',t_1)]U_{t_1}\nonumber\\
662: =-D_{\ep}D_{1-\ep}\ga_{1-\ep}(E)\Bigl(
663: <g_\ep,P_{E}L^2g_\ep>+B_{\ep,\ep}(E,t)\Bigr)U_t\label{commbu1}
664: \end{eqnarray}
665: Notice that the first equality in~(\ref{commbu1}) holds
666: because, due to the time consecutive principle
667: \[
668: [B_{\ep,\ep'}(E,t),U_{t_1}]=0.
669: \]
670: Similarly one computes the commutator
671: \begin{eqnarray}\fl
672: [B_{\ep,\ep}(E,t),U_t]=-\rmi\sum\limits_{\ep'}D_{\ep'}\int\rmd E'
673: \int\limits_0^t\rmd t_1[B_{\ep,\ep}(E,t),N_{\ep',1-\ep'}(E',t_1)]U_{t_1}\nonumber\\
674: = -\rmi D_\ep\ga_\ep(E)B_{1-\ep,\ep}(E,t)U_t\label{commbu1a}
675: \end{eqnarray}
676: After substitution of this commutator into (\ref{commbu1}) one gets
677: \begin{eqnarray*}\fl
678: -\rmi D_\ep[B_{1-\ep,\ep}(E,t),U_t]=-D_{\ep}D_{1-\ep}\ga_{1-\ep}(E)
679: \bigl(<g_\ep,P_EL^2g_\ep>U_t-\rmi D_\ep\ga_\ep(E)\\
680: \times ([B_{1-\ep,\ep}(E,t),U_t]+U_tB_{1-\ep,\ep}(E,t))+U_tB_{\ep,\ep}(E,t)\bigr)
681: \end{eqnarray*}
682: Then for
683: \[
684: f_\ep(E,t):=-\rmi D_\ep[B_{1-\ep,\ep}(E,t),U_t]
685: \]
686: one has
687: \begin{eqnarray*}
688: \fl\bigl(1+(\ga_\ep\ga_{1-\ep})(E)D_\ep D_{1-\ep}\bigr)f_\ep(E,t)
689: =-\ga_{1-\ep}(E)D_\ep D_{1-\ep}\bigl(<g_\ep,P_EL^2g_\ep>U_t \\
690: -\rmi D_\ep\ga_\ep(E)U_tB_{1-\ep,\ep}(E,t)+U_tB_{\ep,\ep}(E,t)\bigr).
691: \end{eqnarray*}
692:
693: Since the inverse operator
694: $ \bigl(1+(\ga_\ep\ga_{1-\ep})(E)D_\ep D_{1-\ep}\bigr)^{-1} $ exists
695: we can solve the equation above for $ f_\ep(E,t) $. Using this
696: solution we find
697: \begin{eqnarray*}\fl
698: -\rmi D_\ep[B_{1-\ep,\ep}(t),U_t]=\int\rmd Ef_\ep(E,t)=
699: -\int\rmd E\ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)\\
700: \times\Bigl(<g_\ep,P_EL^2g_\ep>U_t
701: -\rmi D_\ep\ga_\ep(E)U_tB_{1-\ep,\ep}(E,t)+U_tB_{\ep,\ep}(E,t)\Bigr).
702: \end{eqnarray*}
703:
704: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
705:
706:
707: \section{Emergence of the number and creation processes}
708:
709: In order to bring equation (\ref{equ}) to the normally ordered form
710: one needs also to move the annihilation operators in $ N_{\ep,1-\ep}(t) $ to the
711: right of the evolution operator. Using~(\ref{noperator}) this leads to
712: \begin{eqnarray}\fl
713: -\rmi D_\ep N_{\ep,1-\ep}(t)U_t=-\rmi D_\ep\int\rmd E\sum\limits_{\ep'=0,1}
714: n_{\ep'}(E)B^+_{\ep,\ep'}(E,t)B_{1-\ep,\ep'}(E,t)U_t\nonumber\\
715: \fl = -\rmi D_\ep\int\rmd E\sum\limits_{\ep'=0,1}n_{\ep'}(E)B^+_{\ep,\ep'}(E,t)
716: \Bigl([B_{1-\ep,\ep'}(E,t),U_t]+U_tB_{1-\ep,\ep'}(E,t)\Bigr)\label{normordu1}
717: \end{eqnarray}
718: and the commutator is evaluated using~(\ref{commbu1}) and (\ref{commbu1a}).
719: \begin{theorem}\label{normordu2} For the model described above one has
720: \begin{eqnarray*}\fl
721: -\rmi D_\ep N_{\ep,1-\ep}(t)U_t=-\int\rmd E\Bigl(\ga_{1-\ep}(E)D_\ep
722: D_{1-\ep}T_\ep(E)\\
723: \times\Bigl(-\rmi D_\ep\ga_\ep(E)B^+_{\ep,1-\ep}(E,t)+B^+_{\ep,\ep}(E,t)\Bigr)U_t\\
724: +\sum\limits_{\ep'=0,1}n_{\ep'}(E)
725: \Bigl(\rmi T_\ep(E)D_\ep B^+_{\ep,\ep'}(E,t)U_tB_{1-\ep,\ep'}(E,t)\\
726: +\ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)B^+_{\ep,\ep'}(E,t)U_tB_{\ep,\ep'}(E,t)\Bigr)\Bigr)
727: \end{eqnarray*}
728: \end{theorem}
729: {\bf Proof.} From~(\ref{commbu1}) and (\ref{commbu1a}) it follows that
730: \begin{eqnarray*}\fl
731: -\rmi D_\ep[B_{1-\ep,\ep'}(E,t),U_t]=-\ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)
732: \Bigl(n^{-1}_{\ep'}(E)(\dl_{\ep,\ep'}-\dl_{1-\ep,\ep'}\rmi D_\ep\ga_\ep)U_t \\
733: -\rmi D_\ep\ga_\ep(E)U_tB_{1-\ep,\ep'}(E,t)+U_tB_{\ep,\ep'}(E,t)\Bigr).
734: \end{eqnarray*}
735: After substitution of these commutators in~(\ref{normordu1})
736: and using~(\ref{TE}) one finishes the proof of the theorem.
737:
738:
739: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
740:
741:
742: \section{The normally ordered equation}
743:
744: Theorems~(\ref{normordu}) and~(\ref{normordu2})
745: allow us to obtain immediately the normally ordered equation for the
746: evolution operator in the LDL. This procedure of deduction of
747: quantum stochastic differential equation is being called a stochastic golden
748: rule. The normally ordered equation has the form
749: \begin{eqnarray}\fl
750: \partial_tU_t=-\sum\limits_{\ep=0,1}\int\rmd E\Bigl\{
751: \rmi T_\ep(E)D_\ep\sum\limits_{\ep'=0,1}
752: n_{\ep'}(E)B^+_{\ep,\ep'}(E,t)U_tB_{1-\ep,\ep'}(E,t)\nonumber\\
753: +\vphantom{\sum\limits_0\int}
754: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)\sum\limits_{\ep'=0,1}
755: n_{\ep'}(E)B^+_{\ep,\ep'}(E,t)U_tB_{\ep,\ep'}(E,t)\nonumber\\
756: +\vphantom{\sum\limits_0\int}
757: \rmi T_\ep(E)D_\ep B^+_{\ep,1-\ep}(E,t)U_t+
758: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)B^+_{\ep,\ep}(E,t)U_t\nonumber\\
759: +\vphantom{\sum\limits_0\int}
760: \rmi T_\ep(E)D_\ep U_tB_{1-\ep,\ep}(E,t)+
761: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)U_tB_{\ep,\ep}(E,t)\nonumber\\
762: +\vphantom{\sum\limits_0\int}
763: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)<g_\ep,P_EL^2g_\ep>U_t\Bigr\}\label{normordeq}
764: \end{eqnarray}
765: We will represent equation~(\ref{normordeq}) in the form of
766: quantum stochastic differential equation~\cite{HP}.
767: \begin{theorem}
768: Equation~(\ref{normordeq}) is equivalent to the
769: quantum stochastic differential equation
770: \begin{eqnarray}\fl
771: \rmd U_t=-\sum\limits_{\ep=0,1}\int\rmd E\Bigl\{
772: \rmi T_\ep(E)D_\ep\rmd N_t(2\pi |g_\ep><g_{1-\ep}|P_E\om_\beta P_E)\nonumber\\
773: +\vphantom{\sum\limits_0\int}
774: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)
775: \rmd N_t(2\pi |g_\ep><g_\ep|P_E\om_\beta P_E)\nonumber\\
776: +\vphantom{\sum\limits_0\int}
777: \rmi T_\ep(E)D_\ep\rmd B^+_t(g_\ep\om_\beta P_Eg_{1-\ep})+
778: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)\rmd B^+_t(g_\ep\om_\beta P_Eg_\ep)\nonumber\\
779: +\vphantom{\sum\limits_0\int}
780: \rmi T_\ep(E)D_\ep\rmd B_t(g_{1-\ep}\om_\beta P_Eg_\ep)+
781: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)\rmd B_t(g_\ep\om_\beta P_Eg_\ep)\nonumber\\
782: +\vphantom{\sum\limits_0\int}
783: \ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)<g_\ep,P_EL^2g_\ep>
784: \rmd t\Bigr\}U_t\label{normordeqa}
785: \end{eqnarray}
786: \end{theorem}
787: {\bf Proof.} Let us consider the following term in~(\ref{normordeq})
788: \begin{eqnarray}\fl
789: \sum\limits_{\ep'=0,1}n_{\ep'}(E)B^+_{\ep,\ep'}(E,t)U_tB_{1-\ep,\ep'}(E,t)=
790: n_{1-\ep}(E)b^+_t(g_\ep\om_\beta P_Eg_{1-\ep})U_tb_t(g_{1-\ep}\om_\beta
791: P_Eg_{1-\ep})\nonumber\\
792: +n_\ep(E)b^+_t(g_\ep\om_\beta P_Eg_\ep)U_tb_t(g_{1-\ep}\om_\beta
793: P_Eg_\ep)\label{1N}
794: \end{eqnarray}
795: The matrix element of this expression on the exponential vectors $ \psi(f) $,
796: $ \psi(f') $ is (we use Dirac's notations also for bra- and ket-vectors
797: from $ K_{0,1} $)
798: \begin{eqnarray}\fl
799: \bigl(n_{1-\ep}(E)<f|g_\ep\om_\beta P_Eg_{1-\ep}><g_{1-\ep}\om_\beta
800: P_Eg_{1-\ep}|f'>\nonumber\\
801: +n_\ep(E)<f|g_\ep\om_\beta P_Eg_\ep><g_{1-\ep}\om_\beta P_Eg_\ep|f'>\bigr)
802: <\psi(f)|U_t|\psi(f')>\nonumber\\
803: =<f|T_{\ep,1-\ep}(E)|f'><\psi(f)|U_t|\psi(f')>\label{1Na}
804: \end{eqnarray}
805: which is the time derivative of the matrix element
806: $ <\psi(f)|\rmd N_t(T_{\ep,1-\ep}(E))|\psi(f')> $
807: where $ N_t(T_{\ep,1-\ep}(E)) $ is the number process with intensity
808: \begin{eqnarray*}\fl
809: T_{\ep,1-\ep}(E)=n_{1-\ep}(E)|g_\ep\om_\beta P_Eg_{1-\ep}><g_{1-\ep}\om_\beta
810: P_Eg_{1-\ep}|\\
811: +n_\ep(E)|g_\ep\om_\beta P_Eg_\ep><g_{1-\ep}\om_\beta P_Eg_\ep|
812: \end{eqnarray*}
813: Let us now prove that
814: \begin{equation}
815: T_{\ep,1-\ep}(E)=2\pi |g_\ep><g_{1-\ep}|P_E\om_\beta P_E\label{1Nb}
816: \end{equation}
817: To this goal let us consider the action of the
818: $ T_{\ep,1-\ep}(E) $ on vectors of the form
819: \[
820: |f>=|P_{E_1}g_{\ep_1}\om_\beta P_{E_2}g_{\ep_2}>.
821: \]
822: One has
823: \begin{eqnarray}\fl
824: T_{\ep,1-\ep}(E)|f>\nonumber\\
825: =2\pi\dl_{\ep,1-\ep_1}\dl(E_1-E)\dl(E_2-E)<g_{1-\ep},P_Eg_{1-\ep}>\nonumber\\
826: (\dl_{\ep,1-\ep_2}|g_\ep\om_\beta P_Eg_{1-\ep}>+
827: \dl_{\ep,\ep_2}|g_\ep\om_\beta P_Eg_\ep>)\nonumber\\
828: =2\pi\dl_{\ep,1-\ep_1}<g_{1-\ep},P_Eg_{1-\ep}>\dl(E_1-E)\dl(E_2-E)\nonumber\\
829: (\dl_{\ep,1-\ep_2}|g_\ep\om_\beta P_Eg_{\ep_2}>+
830: \dl_{\ep,\ep_2}|g_\ep\om_\beta P_Eg_{\ep_2}>)\nonumber\\
831: =2\pi\dl_{\ep,1-\ep_1}<g_{1-\ep},P_Eg_{1-\ep}>\dl(E_1-E)\dl(E_2-E)
832: |g_\ep\om_\beta P_Eg_{\ep_2}>\nonumber\\
833: =2\pi |g_\ep><g_{1-\ep}|P_E\om_\beta P_E|f>\label{1Nc}
834: \end{eqnarray}
835: Therefore (\ref{1Nb}) holds and
836: the term~(\ref{1N}) corresponds to the number process
837: \[
838: \rmd N_t(2\pi |g_\ep><g_{1-\ep}|P_E\om_\beta P_E)
839: \]
840:
841: Computing the same matrix element for the term
842: \begin{eqnarray}\fl
843: \sum\limits_{\ep'=0,1}n_{\ep'}(E)B^+_{\ep,\ep'}(E,t)U_tB_{\ep,\ep'}(E,t)=
844: n_{1-\ep}(E)b^+_t(g_\ep\om_\beta P_Eg_{1-\ep})U_tb_t(g_\ep\om_\beta
845: P_Eg_{1-\ep})\nonumber\\
846: +n_\ep(E)b^+_t(g_\ep\om_\beta P_Eg_\ep)U_tb_t(g_\ep\om_\beta P_Eg_\ep)\label{2N}
847: \end{eqnarray}
848: one finds an expression like~(\ref{1Na}) with $ T_{\ep,1-\ep} $
849: replaced by the operator
850: \[\fl
851: T_{\ep,\ep}(E)=n_{1-\ep}(E)|g_\ep\om_\beta P_Eg_{1-\ep}><g_\ep\om_\beta
852: P_Eg_{1-\ep}|+n_\ep(E)|g_\ep\om_\beta P_Eg_\ep><g_\ep\om_\beta P_Eg_\ep|
853: \]
854: and a calculation similar to the one done in~(\ref{1Nc}) leads to the
855: conclusion that
856: \[
857: T_{\ep,\ep}(E)=2\pi |g_\ep><g_\ep|P_E\om_\beta P_E.
858: \]
859: Therefore the term~(\ref{2N}) corresponds to the number process
860: \[
861: \rmd N_t(2\pi |g_\ep><g_\ep|P_E\om_\beta P_E).
862: \]
863: This finishes the proof of the theorem.
864:
865: Let us introduce the notations:
866: \begin{eqnarray*}
867: R_{\ep,\ep}(E)=-\ga_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep(E)\\
868: R_{\ep,1-\ep}(E)=-\rmi T_\ep(E)D_\ep .
869: \end{eqnarray*}
870: In these notations the quantum stochastic differential equation for the
871: evolution operator can be rewritten as
872: \begin{eqnarray*}\fl
873: \rmd U_t=\sum\limits_{\ep=0,1}\int\rmd E\Bigl\{
874: \vphantom{\sum\limits_0\int}
875: R_{\ep,1-\ep}(E)\rmd N_t(2\pi|g_\ep><g_{1-\ep}|P_E\om_\beta P_E)\\
876: +\vphantom{\sum\limits_0\int}
877: R_{\ep,\ep}(E)\rmd N_t(2\pi|g_\ep><g_\ep|P_E\om_\beta P_E)\\
878: +\vphantom{\sum\limits_0\int}
879: R_{\ep,1-\ep}(E)\rmd B^+_t(g_\ep\om_\beta P_Eg_{1-\ep})+
880: R_{\ep,\ep}(E)\rmd B^+_t(g_\ep\om_\beta P_Eg_\ep)\\
881: +\vphantom{\sum\limits_0\int}
882: R_{\ep,1-\ep}(E)\rmd B_t(g_{1-\ep}\om_\beta P_Eg_\ep)+
883: R_{\ep,\ep}(E)\rmd B_t(g_\ep\om_\beta P_Eg_\ep)\\
884: +\vphantom{\sum\limits_0\int}
885: R_{\ep,\ep}(E)<g_\ep,P_EL^2g_\ep>\rmd t\Bigr\}U_t.
886: \end{eqnarray*}
887:
888: Notice that the quantum stochastic differential
889: equation~(\ref{normordeqa}) can be written also in the
890: Frigerio-Maassen form~\cite{FrMa}. In order to prove this recall
891: that for any pair of Hilbert spaces $ {\cal X}_0,{\cal X}_1 $ if $ N,A $
892: denote the number and annihilation processes on the Fock space
893: $ {\cal F}({\cal X}_1) $ then for $ X_0\in B({\cal X}_0) $,
894: $ X_1\in B({\cal X}_1) $, $ x\in{\cal X}_1 $, Frigerio and Maassen~\cite{FrMa}
895: introduced the notation:
896: \begin{eqnarray*}
897: N(X_0\om X_1):=X_0\om N(X_1)\\
898: A(X_0\om X_1x):=X_0\om A(X_1x)\\
899: <x,X_0\om X_1x>:=X_0\om 1<x,X_1x>
900: \end{eqnarray*}
901: Let us also introduce an operator $ T_3(E) $ acting on the triple
902: $ {\cal H}_S\om K\om_\beta K $ (this is the reason for introducing index $3$)
903: as
904: \[
905: T_3(E):=2\pi\sum\limits_{\ep,\ep'=0,1}R_{\ep,\ep'}(E)
906: \om|g_\ep><g_{\ep'}|P_E\om_\beta P_E
907: \]
908: and the vector $ \xi(E)\in K\om_\beta K $
909: \[
910: \xi(E):=\frac{1}{2\pi}\sum\limits_{\ep=0,1}
911: \frac{1}{<g_\ep,P_Eg_\ep>}g_\ep\om_\beta g_\ep .
912: \]
913:
914: In these notations equation~(\ref{normordeqa}) can be written as
915: \begin{eqnarray*}\fl
916: \rmd U_t=\int\rmd E\Bigl(\rmd N_t\bigl(T_3(E)\bigr)+\rmd B^+_t\bigl(T_3(E)\xi(E)\bigr)\\
917: +\rmd B_t\bigl(T^*_3(E)\xi(E)\bigr)+<\xi(E),T_3(E)\xi(E)>\rmd t\Bigr)U_t.
918: \end{eqnarray*}
919:
920:
921: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
922:
923:
924: \section{Connection with scattering theory}
925:
926: Here we consider relation between the evolution operator and scattering
927: theory.
928: Because of the number conservation,
929: the closed subspace of $ {\cal H}_S\om {\cal F} $ generated by vectors of the
930: form $ u\om A^+(f)\Phi $\enskip ($ u\in {\cal H}_{\rm S} $,
931: $ f\in {\cal H}_1=L^2({\mathbb R}^d) $) which is naturally isomorphic to
932: $ {\cal H}_{\rm S}\om {\cal H}_1 $, is globally invariant under the time evolution
933: operator $ \exp[\rmi (H_{\rm S}\om 1+1\om H_{\rm R}+V)t] $. Therefore the restriction of the time
934: evolution operator to this subspace corresponds to an evolution operator on
935: $ {\cal H}_{\rm S}\om {\cal H}_1 $ given by
936: \[
937: \exp[\rmi(H_{\rm S}\om 1+1\om H_1+V_1)t]
938: \]
939: where
940: \begin{equation}\label{defV}
941: V_1=\sum\limits_{\ep=0,1}D_\ep\om|g_\ep><g_{1-\ep}|
942: \end{equation}
943:
944: The 1-particle M\o ller wave operators are defined as
945: \[\fl
946: \Omega_{\pm}=s-\lim\limits_{t\to\pm\infty}\exp[\rmi(H_{\rm S}\om 1+1\om H+V_1)t]
947: \exp[-\rmi(H_{\rm S}\om 1+1\om H)t]
948: \]
949: and the 1-particle $ T $-operator is defined as
950: \begin{equation}\label{defT}
951: T=V_1\Omega_+
952: \end{equation}
953: From (\ref{defV}) it follows that
954: \[
955: \Omega_{\pm}=s-\lim\limits_{t\to\pm\infty}U^{(1)}_t
956: \]
957: where $ U_t^{(1)} $ is the solution of
958: \[
959: \partial_tU_t^{(1)}=-\rmi\left(\sum\limits_{\ep=0,1}
960: D_\ep\om|S_tg_\ep><S_tg_{1-\ep}|\right)U_t^{(1)}\qquad U_0^{(1)}=1.
961: \]
962:
963: In order to make a connection between the stochastic process $ U_t $
964: and scattering theory notice that the operator $ T_3(E) $ can be written as
965: \[
966: T_3(E)=2\pi T(E)\om_\beta P_E
967: \]
968: where operator $ T(E) $ acts on $ {\cal H}_{\rm S}\om K $ as
969: \[
970: T(E)=\sum\limits_{\ep,\ep'=0,1}R_{\ep,\ep'}(E)\om |g_\ep><g_{\ep'}|P_E.
971: \]
972: In~\cite{AcLu} it was proved that $ T $-operator defined by (\ref{defT})
973: connected with $ T(E) $ by the following formula
974: \[
975: T=\int\rmd ET(E).
976: \]
977:
978: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
979:
980: \section{Vacuum expectation value}
981:
982: For the vacuum matrix element of the evolution
983: operator
984: from~(\ref{normordeq}) one immediately gets
985: \begin{equation}\label{expvalue}
986: <U(t)>_{vac}=\rme^{-\Gamma t}.
987: \end{equation}
988: The operator $ \Gamma $ acts in $ {\cal H}_{\rm S} $ as
989: \[
990: \Gamma =\sum\limits_{\ep=0,1}\int
991: \rmd E\ga_{1-\ep}(E)D_{\ep}D_{1-\ep}T_{\ep}(E)<g_\ep,P_EL^2g_\ep>.
992: \]
993: \begin{theorem}
994: The operator $ \Gamma $ has a
995: non-negative real part (i.e. this operator describes
996: the damping).
997: \end{theorem}
998: {\bf Proof.} From the definition of $T_\ep(E)$ we know that
999: \[
1000: \gamma_{1-\ep}(E)D_\ep D_{1-\ep}T_\ep (E)={D_\ep
1001: D_{1-\ep}\over{\gamma^{-1}_{1-\ep}(E)}+\gamma_\ep (E)
1002: D_\ep D_{1-\ep}}
1003: \]
1004: But $\gamma_\ep$ and $\gamma_{1-\ep}$
1005: (hence also $1/\gamma_{1-\ep}$) have positive real part
1006: and $D_\ep D_{1-\ep}$ is positive self--adjoint.
1007:
1008: Hence the above expression has a positive real part
1009: because it is of the form:
1010: \[
1011: {H\over z_1+z_2H}\,={H(\hbox{Re }z_1+H\hbox{ Re
1012: }z_2)\over|z_1+
1013: z_2H|^2}\,-\rmi{H(\hbox{Im }z_1+H\hbox{ Im
1014: }z_2)\over|z_1+z_2H|^2}
1015: \]
1016: where $H$ is positive self--adjoint and $z_1$, $z_2$
1017: have a positive
1018: real part. Since $\langle g_\ep,P_EL^2g_\ep\rangle\geq0$ the thesis follows.
1019:
1020: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1021:
1022: \ack{
1023: One of the authors (A.P.) is grateful to the Centro Vito Volterra for kind
1024: hospitality, and to Y.G.Lu for useful discussions. This work is partially
1025: supported by the RFFI 02-01-01084 and INTAS 99-00545 for L.A. and I.V. and by the
1026: Vito Volterra Fellowship and INTAS 01/1-200 for A.P.}
1027:
1028: \section*{References}
1029: \begin{thebibliography}{99}
1030:
1031: \bibitem{AFL} Accardi L, Frigerio A and Lu Y G 1987
1032: {\it Quantum Probability and Applications. IV. Lect. Notes Math.}
1033: {\bf 1396} 20--58
1034:
1035: \bibitem{AKV} Accardi L, Kozyrev S V and Volovich I V 1997
1036: {\it Phys. Rev. A} {\bf 56}(4) 2557--62\\
1037: (Accardi L, Kozyrev S V and Volovich I V 1997 {\it Preprint} quant-ph/9706021)
1038:
1039: \bibitem{ALV} Accardi L, Lu Y G and Volovich I V 2001
1040: {\it Quantum Theory and Its Stochastic Limit}, Springer
1041:
1042: \bibitem{lanford} Lanford O E 1976 {\it Ast\'erisque}
1043: {\bf 40} 117--37
1044:
1045: \bibitem{grad} Grad H 1958 {\it Principles of the kinetic theory
1046: of gases. In: Handbuch der physik} {\bf 12} Fl\"ugge S (ed.),
1047: Berlin, Heidelberg, New York: Springer
1048:
1049: \bibitem{king} King F 1975 {\it PhD Thesis} University of
1050: California, Berkeley
1051:
1052: \bibitem{spohn2} Spohn H 1980 {\it Rewievs in Modern Physics}
1053: {\bf 52} 569--615
1054:
1055: \bibitem{dumcke} D\"umcke R 1984 {\it Lect. Notes in Math.}
1056: {\bf 1136} 151--61
1057:
1058: \bibitem{AcLu} Accardi L and Lu Y G 1991
1059: {\it J. Phys. A: Math. Gen.} {\bf 24} 3483--512
1060:
1061: \bibitem{kumerrer} Kummerer B 1986 {\it Markov dilations and non-commutative Poisson
1062: processes} Preprint Tubingen
1063:
1064: \bibitem{bogol} Bogoliubov N N 1946 {\it Porblems of Dynamical
1065: Theory in Statistical Physics}, (Moscow: Gostechizdat)
1066:
1067: \bibitem{davies} Davies E B 1974 {\it Comm. Math. Phys.}
1068: {\bf 39} 91--110
1069:
1070: \bibitem{Pl} Palmer P F 1977 {\it PhD Thesis} Oxford University
1071:
1072: \bibitem{HP} Hudson R and Parthasarathy K R 1984 {\it Comm. Math. Phys.}
1073: {\bf 93} 301--23
1074:
1075: \bibitem{FrMa} Frigerio A and Maassen H 1989 {\it Prob. Th. Rel. Fields}
1076: {\bf 83} 489--508
1077:
1078:
1079: \end{thebibliography}
1080:
1081:
1082:
1083:
1084: \end{document}
1085: