Introduction
One of the principal challenges in kinetic theory is understanding how to describe the evolution of a system composed of many particles through the behavior of a typical particle. The conventional philosophy of describing kinetic quantum evolution states that if a separable initial state is specified by a one-particle (reduced) density operator, then at an arbitrary time, the evolution of this state in the appropriate scaling limit can be effectively described by a one-particle density operator governed by a nonlinear kinetic equation. Recently, there has been a resurgence of interest in deriving kinetic equations from the dynamics of many quantum particles. One key challenge was largely due to unresolved issues in the quantum theory of information transfer processes [1 ] , which are linked to the description of the evolution of correlations within complex systems [2 , 3 , 4 ] .
The problem of deriving kinetic equations has deeper mathematical implications. These lie in the relationships between the fundamental evolution equations that describe many-particle systems and the heuristically formulated equations that explain the evolution of systems at various scales. The origins of this problem can be traced back to H. Poincaré in 1905 and D. Hilbert in 1912, when it was framed in the context of deriving hydrodynamic equations from the Boltzmann kinetic equation. This concept was later developed by M. Bogolyubov in 1945, who derived kinetic equations from the hierarchy of fundamental evolution equations known as the BBGKY (Bogolyubov–Born–Green–Kirkwood–Yvon) hierarchy, which describes the evolution of states in many-particle systems. A rigorous theory for describing equilibrium states was developed in the 1960s, while the development of a theory of nonequilibrium states began in the 1980s. We refer to review [5 ] for a synopsis of the chronology of kinetic theory. The mathematical theory of the quantum BBGKY hierarchy originates from D. Petrina’s article [6 ] . At the end of the XXth century, the results of the development of the mathematical theory of the BBGKY hierarchy of evolution equations and its relationship to kinetic equations for classical systems were summed up in books [7 , 8 , 9 ] .
The approaches to deriving kinetic equations rely on our existing capabilities to construct solutions for the BBGKY hierarchy [7 ] , [10 , 11 ] . Until now, such solutions and their scaling asymptotics have been constructed using perturbation theory methods. The greatest progress has been associated with the derivation of the Boltzmann equation for the collision dynamics of hard spheres [5 ] , [12 , 13 , 14 , 15 ] . It should be noted that the problem of deriving the Boltzmann equation is, in a sense, an evergreen issue. This year, a recent contribution to this problem, the rigorous solution of which was first published in the paper by D. Petrina and the author of this article [16 ] , was awarded a Fields Medal [17 ] . The last two decades have also seen significant progress in deriving quantum kinetic equations [18 , 19 , 20 ] in scaling limits of the perturbative solution of the BBGKY hierarchy [21 , 22 , 23 ] , in particular, the quantum Boltzmann equation in the weak coupling limit [24 , 25 , 26 ] , the nonlinear Schrödinger equation and the Gross–Pitaevskii equation in the mean field limit [27 , 28 , 29 , 30 , 31 ] .
This article formulates methods for deriving kinetic equations based on non-perturbative solutions of hierarchies of evolution equations for quantum systems [32 ] . One of the achievements of these methods is the construction of kinetic equations for non-separable initial states, i.e., in the presence of correlations between initial states, as well as the description of the propagation processes of correlations that are described by their solution [3 , 4 ] , [33 , 34 ] . The hope is that such approaches will eventually yield a mathematically rigorous derivation of kinetic equations that aligns with physical intuition.
The article is structured around the key themes related to the challenge of the relationship of the evolution equations that describe many-particle systems. The main issues being considered are:
Hierarchies of fundamental evolution equations of quantum systems and non-perturbative methods for constructing their solutions (§ 2 );
Mathematical origins of quantum kinetic equations and the evolution of correlations governed by kinetic equations (§ 3 );
Novel methods for deriving kinetic equations from hierarchies of evolution equations of quantum systems, specifically focusing on kinetics within the framework of the evolution of observables and the generalized quantum kinetic equation with initial correlations (§ 4 ).
In Section 5, we provide a brief overview of perspectives for future research.
Hierarchies of evolution equations for quantum systems
Many-particle systems are described in terms of concepts of observables and the state. The mean-value functional of observables defines the duality between observables and a state. As a consequence, there are two equivalent approaches to describing the evolution of systems: either as the evolution of observables or as the evolution of a state.
In the paper, we will explore quantum systems of a non-fixed number of identical spinless particles, i.e., an arbitrary but finite average number of particles that obey the Maxwell–Boltzmann statistics, and use a system of units where h = 2 π ℏ = 1 h = 2\pi\hbar= 1 h = 2 π ℏ = 1 is Planck’s constant and m = 1 m = 1 m = 1 is the particle mass. The results can be extended to systems of many bosons or fermions, following [34 ] .
We first describe the traditional approach to describing observables and the state of quantum many-particle systems and introduce the notation needed for the subsequent presentation in terms of the nonequilibrium grand canonical ensemble.
Let H ⊗ n \mathcal{H}^{\otimes n} H ⊗ n be an n n n -particle Hilbert space. The Fock space over the space H \mathcal{H} H is denoted by F H = ⨁ n = 0 ∞ H ⊗ n \mathcal{F}_{\mathcal{H}} = \bigoplus_{n=0}^{\infty}\mathcal{H}^{\otimes n} F H = ⨁ n = 0 ∞ H ⊗ n . Let L ( H ⊗ n ) \mathcal{L}(\mathcal{H}^{\otimes n}) L ( H ⊗ n ) be the space of bounded self-adjoint operators b n ∈ L ( H ⊗ n ) b_n \in\mathcal{L}(\mathcal{H}^{\otimes n}) b n ∈ L ( H ⊗ n ) with operator norm ∥ ⋅ ∥ L ( H ⊗ n ) \lVert\cdot\rVert_{\mathcal{L}(\mathcal{H}^{\otimes n})} ∥ ⋅ ∥ L ( H ⊗ n ) . The self-adjoint operator b n b_n b n is defined on the space H ⊗ n \mathcal{H}^{\otimes n} H ⊗ n , hereinafter also denoted by the symbol b n ( 1 , … , n ) b_n(1,\ldots,n) b n ( 1 , … , n ) , which for arbitrary values of the indices ( i 1 , … , i n ) ∈ ( 1 , … , n ) (i_1,\ldots,i_n) \in(1,\ldots,n) ( i 1 , … , i n ) ∈ ( 1 , … , n ) satisfies the symmetry condition: b n ( 1 , … , n ) = b n ( i 1 , … , i n ) b_n(1,\ldots,n) = b_n(i_1,\ldots,i_n) b n ( 1 , … , n ) = b n ( i 1 , … , i n ) . Let us introduce the space L γ ( F H ) \mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}}) L γ ( F H ) of sequences of bounded operators b = ( b 0 , b 1 ( 1 ) , … , b n ( 1 , … , n ) , … ) b = (b_0,b_1(1),\ldots,b_n(1,\ldots,n),\ldots) b = ( b 0 , b 1 ( 1 ) , … , b n ( 1 , … , n ) , … ) with norm
∥ b ∥ L γ ( F H ) = max n ≥ 0 γ n n ! ∥ b n ∥ L ( H ⊗ n ) , \lVert b\rVert_{\mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}})} = \max_{n\geq0}\frac{\gamma^n}{n!}\lVert b_n\rVert_{\mathcal{L}(\mathcal{H}^{\otimes n})}, ∥ b ∥ L γ ( F H ) = n ≥ 0 max n ! γ n ∥ b n ∥ L ( H ⊗ n ) ,
where 0 < γ < 1 0 < \gamma< 1 0 < γ < 1 is a real number. The subspace of finite sequences of degenerate operators whose kernels are infinitely differentiable functions with compact supports is denoted by L 0 ( F H ) \mathcal{L}_0(\mathcal{F}_{\mathcal{H}}) L 0 ( F H ) .
Accordingly, let L 1 ( H ⊗ n ) \mathcal{L}^1(\mathcal{H}^{\otimes n}) L 1 ( H ⊗ n ) be the space of trace class operators f n ≡ f n ( 1 , … , n ) ∈ L 1 ( H ⊗ n ) f_n \equiv f_n(1,\ldots,n) \in\mathcal{L}^1(\mathcal{H}^{\otimes n}) f n ≡ f n ( 1 , … , n ) ∈ L 1 ( H ⊗ n ) , which for arbitrary values of the indices ( i 1 , … , i n ) ∈ ( 1 , … , n ) (i_1,\ldots,i_n) \in(1,\ldots,n) ( i 1 , … , i n ) ∈ ( 1 , … , n ) satisfy the symmetry condition: f n ( 1 , … , n ) = f n ( i 1 , … , i n ) f_n(1,\ldots,n) = f_n(i_1,\ldots,i_n) f n ( 1 , … , n ) = f n ( i 1 , … , i n ) , with norm: ∥ f n ∥ L 1 ( H ⊗ n ) = Tr 1 , … , n ∣ f n ( 1 , … , n ) ∣ \lVert f_n\rVert_{\mathcal{L}^1(\mathcal{H}^{\otimes n})} = \operatorname{Tr}_{1,\ldots,n}\lvert f_n(1,\ldots,n)\rvert ∥ f n ∥ L 1 ( H ⊗ n ) = Tr 1 , … , n ∣ f n ( 1 , … , n )∣ , where the symbol Tr 1 , … , n \operatorname{Tr}_{1,\ldots,n} Tr 1 , … , n denote the partial traces of the operator f n f_n f n . Let us introduce the space L α 1 ( F H ) \mathcal{L}^1_{\alpha}(\mathcal{F}_{\mathcal{H}}) L α 1 ( F H ) of sequences of trace class operators f = ( f 0 , f 1 ( 1 ) , … , f n ( 1 , … , n ) , … ) f = (f_0,f_1(1),\ldots,f_n(1,\ldots,n),\ldots) f = ( f 0 , f 1 ( 1 ) , … , f n ( 1 , … , n ) , … ) with norm
∥ f ∥ L α 1 ( F H ) = ∑ n = 0 ∞ α n Tr 1 , … , n ∣ f n ( 1 , … , n ) ∣ , \lVert f\rVert_{\mathcal{L}^1_{\alpha}(\mathcal{F}_{\mathcal{H}})} = \sum_{n=0}^{\infty}\alpha^n\operatorname{Tr}_{1,\ldots,n}\lvert f_n(1,\ldots,n)\rvert, ∥ f ∥ L α 1 ( F H ) = n = 0 ∑ ∞ α n Tr 1 , … , n ∣ f n ( 1 , … , n )∣ ,
where α > 1 \alpha> 1 α > 1 is a real number. The subspace of finite sequences of degenerate operators whose kernels are infinitely differentiable functions with compact supports will be denoted by L 0 1 ( F H ) \mathcal{L}^1_0(\mathcal{F}_{\mathcal{H}}) L 0 1 ( F H ) .
Evolution of observables of a finite number of quantum particles
The evolution of observables of a non-fixed but finite average number of quantum particles is described by sequences of self-adjoint operators defined on the Fock space F H = ⨁ n = 0 ∞ H ⊗ n \mathcal{F}_{\mathcal{H}} = \bigoplus_{n=0}^{\infty}\mathcal{H}^{\otimes n} F H = ⨁ n = 0 ∞ H ⊗ n
O n ( t , 1 , … , n ) = G n ( t ) O n 0 ( 1 , … , n ) , n ≥ 1 , (1) O_n(t,1,\ldots,n) = \mathcal{G}_n(t)O_n^0(1,\ldots,n),\qquad n \geq1, \tag*{(1)} O n ( t , 1 , … , n ) = G n ( t ) O n 0 ( 1 , … , n ) , n ≥ 1 , (1)
which, for an n n n -particle quantum subsystem, are determined by the following one-parameter mapping defined on the space L ( H ⊗ n ) \mathcal{L}(\mathcal{H}^{\otimes n}) L ( H ⊗ n ) :
G n ( t ) b n ≔ e i t H n b n e − i t H n , (2) \mathcal{G}_{n}(t)b_{n} \coloneqq e^{itH_{n}}b_{n}e^{-itH_{n}}, \tag*{(2)} G n ( t ) b n : = e i t H n b n e − i t H n , (2)
where the self-adjoint operator H n H_{n} H n is the Hamiltonian of n n n quantum particles. This operator has the form H n = ∑ j = 1 n K ( j ) + ∑ 1 ≤ j 1 < j 2 ≤ n Φ ( j 1 , j 2 ) H_{n} = \sum_{j=1}^{n} K(j) + \sum_{1 \leq j_{1}<j_{2}\leq n} \Phi(j_{1},j_{2}) H n = ∑ j = 1 n K ( j ) + ∑ 1 ≤ j 1 < j 2 ≤ n Φ ( j 1 , j 2 ) , where K ( j ) K(j) K ( j ) is the kinetic energy operator for the j j j -th particle, and Φ \Phi Φ is the pair-interaction potential, which satisfies the Kato conditions [35 ] . From now on, we take the interaction potential to be bounded.
In the space L ( H ⊗ n ) \mathcal{L}(\mathcal{H}^{\otimes n}) L ( H ⊗ n ) the one-parameter mapping (2) forms a ∗ * ∗ -weakly continuous group of operators, the infinitesimal generator of which, on the domain of its definition, coincides with the Heisenberg operator [35 ]
N n = ∑ j = 1 n N ( j ) + ∑ j 1 < j 2 = 1 n N i n t ( j 1 , j 2 ) , (3) \mathcal{N}_{n} = \sum_{j=1}^{n} \mathcal{N}(j) + \sum_{j_{1}<j_{2}=1}^{n} \mathcal{N}_{\mathrm{int}}(j_{1},j_{2}), \tag*{(3)} N n = j = 1 ∑ n N ( j ) + j 1 < j 2 = 1 ∑ n N int ( j 1 , j 2 ) , (3)
where, for b n ∈ L 0 ( H ⊗ n ) b_{n} \in\mathcal{L}_{0}(\mathcal{H}^{\otimes n}) b n ∈ L 0 ( H ⊗ n ) , we introduced the notations:
N ( j ) b n ≔ − i ( b n K ( j ) − K ( j ) b n ) , \mathcal{N}(j)b_{n} \coloneqq-i\bigl(b_{n}K(j)-K(j)b_{n}\bigr), N ( j ) b n : = − i ( b n K ( j ) − K ( j ) b n ) ,
N i n t ( j 1 , j 2 ) b n ≔ − i ( b n Φ ( j 1 , j 2 ) − Φ ( j 1 , j 2 ) b n ) . \mathcal{N}_{\mathrm{int}}(j_{1},j_{2})b_{n} \coloneqq-i\bigl(b_{n}\Phi(j_{1},j_{2})-\Phi(j_{1},j_{2})b_{n}\bigr). N int ( j 1 , j 2 ) b n : = − i ( b n Φ ( j 1 , j 2 ) − Φ ( j 1 , j 2 ) b n ) .
The sequence of operators (1) is determined by the Cauchy problem for the sequence of Heisenberg equations:
∂ ∂ t O n ( t ) = N n O n ( t ) , (4) \frac{\partial}{\partial t}O_{n}(t)=\mathcal{N}_{n}O_{n}(t), \tag*{(4)} ∂ t ∂ O n ( t ) = N n O n ( t ) , (4)
O n ( t ) ∣ t = 0 = O n 0 , n ≥ 1 , (5) O_{n}(t)\big|_{t=0}=O_{n}^{0},\qquad n\geq1, \tag*{(5)} O n ( t ) t = 0 = O n 0 , n ≥ 1 , (5)
where the generator of these evolution equations is given by formula (3) . In view of the properties of operator groups (2) , an existence theorem holds [35 ] .
Theorem 1. For initial data O ( 0 ) ∈ L 0 ( F H ) O(0) \in\mathcal{L}_{0}(\mathcal{F}_{\mathcal{H}}) O ( 0 ) ∈ L 0 ( F H ) , the sequence (1) is the unique global in time classical solution to the Cauchy problem for the Heisenberg equations (4) , (5) . For arbitrary initial data in the space L γ ( F H ) \mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}}) L γ ( F H ) , it is the unique global in time generalized solution.
A similar result also holds for sequences of unbounded operators [35 ] .
Evolution of a state of a finite number of quantum particles
The mean value (mathematical expectation) at a time t ∈ R t \in\mathbb{R} t ∈ R of an observable O ( t ) = ( O 0 , O 1 ( t , 1 ) , … , O n ( t , 1 , … , n ) , … ) O(t)=(O_{0},O_{1}(t,1),\ldots,O_{n}(t,1,\ldots,n),\ldots) O ( t ) = ( O 0 , O 1 ( t , 1 ) , … , O n ( t , 1 , … , n ) , … ) for quantum systems with a non-fixed number of identical particles is defined by a continuous linear functional, which can be expressed in the form of the following series expansion: [21 ] :
⟨ O ⟩ ( t ) ≔ ( I , D ( 0 ) ) − 1 ∑ n = 0 ∞ 1 n ! Tr 1 , … , n O n ( t ) D n 0 ≡ (6) \langle O\rangle(t) \coloneqq(I,D(0))^{-1}\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{1,\ldots,n}O_{n}(t)D_{n}^{0} \equiv \tag*{(6)} ⟨ O ⟩ ( t ) : = ( I , D ( 0 ) ) − 1 n = 0 ∑ ∞ n ! 1 Tr 1 , … , n O n ( t ) D n 0 ≡ (6)
( I , D ( 0 ) ) − 1 ( O ( t ) , D ( 0 ) ) , (I,D(0))^{-1}(O(t),D(0)), ( I , D ( 0 ) ) − 1 ( O ( t ) , D ( 0 )) ,
where the abbreviated notation was introduced. The sequence D ( 0 ) = ( I , D 1 0 , … , D n 0 , … ) D(0) = (I, D_1^0, \ldots, D_n^0, \ldots) D ( 0 ) = ( I , D 1 0 , … , D n 0 , … ) of positive self-adjoint operators D n 0 ∈ L 1 ( H ⊗ n ) D_n^0 \in\mathcal{L}^1(\mathcal{H}^{\otimes n}) D n 0 ∈ L 1 ( H ⊗ n ) , whose kernels are known as density matrices, is the sequence of density operators that describe all possible states of the quantum system with a non-fixed number of particles at the initial instant. The coefficient ( I , D ( 0 ) ) = ∑ n = 0 ∞ 1 n ! Tr 1 , … , n D n 0 (I,D(0)) = \sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{1,\ldots,n}D_n^0 ( I , D ( 0 )) = ∑ n = 0 ∞ n ! 1 Tr 1 , … , n D n 0 is the normalizing factor known as the grand canonical partition function.
The mean-value functional (6) defines a duality between observables and the state. For D n 0 ∈ L 1 ( H ⊗ n ) D_n^0 \in\mathcal{L}^1(\mathcal{H}^{\otimes n}) D n 0 ∈ L 1 ( H ⊗ n ) and O n ( t ) ∈ L ( H ⊗ n ) O_n(t) \in\mathcal{L}(\mathcal{H}^{\otimes n}) O n ( t ) ∈ L ( H ⊗ n ) , it exists.
The sequence of density operators at instant t ∈ R t \in\mathbb{R} t ∈ R we denote by D ( t ) = ( I , D 1 ( t ) , … , D n ( t ) , … ) D(t) = (I,D_1(t),\ldots,D_n(t),\ldots) D ( t ) = ( I , D 1 ( t ) , … , D n ( t ) , … ) . Given the validity of the fundamental relation for representations of the mean-value functional for the observables
⟨ O ⟩ ( t ) = ( I , D ( 0 ) ) − 1 ( O ( t ) , D ( 0 ) ) = ( I , D ( 0 ) ) − 1 ∑ n = 0 ∞ 1 n ! Tr 1 , … , n G n ( t ) O n 0 D n 0 = \langle O\rangle(t) = (I,D(0))^{-1}(O(t),D(0)) = (I,D(0))^{-1}\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{1,\ldots,n}\mathcal{G}_n(t)O_n^0D_n^0 = ⟨ O ⟩ ( t ) = ( I , D ( 0 ) ) − 1 ( O ( t ) , D ( 0 )) = ( I , D ( 0 ) ) − 1 n = 0 ∑ ∞ n ! 1 Tr 1 , … , n G n ( t ) O n 0 D n 0 =
( I , D ( t ) ) − 1 ∑ n = 0 ∞ 1 n ! Tr 1 , … , n O n 0 G n ∗ ( t ) D n 0 = ( I , D ( t ) ) − 1 ( O ( 0 ) , D ( t ) ) , (7) (I,D(t))^{-1}\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{1,\ldots,n}O_n^0\mathcal{G}_n^*(t)D_n^0 = (I,D(t))^{-1}(O(0),D(t)), \tag*{(7)} ( I , D ( t ) ) − 1 n = 0 ∑ ∞ n ! 1 Tr 1 , … , n O n 0 G n ∗ ( t ) D n 0 = ( I , D ( t ) ) − 1 ( O ( 0 ) , D ( t )) , (7)
the density operator of an n n n -particle quantum system
D n ( t , 1 , … , n ) = G n ∗ ( t ) D n 0 ( 1 , … , n ) , n ≥ 1 , (8) D_n(t,1,\ldots,n) = \mathcal{G}_n^*(t)D_n^0(1,\ldots,n), \qquad n \ge1, \tag*{(8)} D n ( t , 1 , … , n ) = G n ∗ ( t ) D n 0 ( 1 , … , n ) , n ≥ 1 , (8)
is determined by the group of operators adjoint to operators (2) for $f_n \in \mathcal{L}^1(\mathcal{H}^{\otimes n})
G n ∗ ( t ) f n = e − i t H n f n e i t H n . (9) \mathcal{G}_n^*(t)f_n = e^{-itH_n}f_ne^{itH_n}. \tag*{(9)} G n ∗ ( t ) f n = e − i t H n f n e i t H n . (9)
In the space L 1 ( H ⊗ n ) \mathcal{L}^1(\mathcal{H}^{\otimes n}) L 1 ( H ⊗ n ) , the dual one-parameter mapping (9) in the sense of functional (7) forms a strongly continuous isometric group of operators that preserves positivity and self-adjointness of operators. The infinitesimal generator of the operator group (9) on the subspace L 0 1 ( H ⊗ n ) \mathcal{L}_0^1(\mathcal{H}^{\otimes n}) L 0 1 ( H ⊗ n ) coincides with the following von Neumann operator [21 ] :
N n ∗ = ∑ j = 1 n N ∗ ( j ) + ∑ j 1 < j 2 = 1 n N i n t ∗ ( j 1 , j 2 ) . (10) \mathcal{N}_n^* = \sum_{j=1}^{n}\mathcal{N}^*(j) + \sum_{j_1<j_2=1}^{n}\mathcal{N}_{\mathrm{int}}^*(j_1,j_2). \tag*{(10)} N n ∗ = j = 1 ∑ n N ∗ ( j ) + j 1 < j 2 = 1 ∑ n N int ∗ ( j 1 , j 2 ) . (10)
For f n ∈ L 0 1 ( H ⊗ n ) f_n \in\mathcal{L}_0^1(\mathcal{H}^{\otimes n}) f n ∈ L 0 1 ( H ⊗ n ) , we introduced the notations:
N ∗ ( j ) f n ≔ − i ( K ( j ) f n − f n K ( j ) ) , \mathcal{N}^*(j)f_n \coloneqq-i(K(j)f_n-f_nK(j)), N ∗ ( j ) f n : = − i ( K ( j ) f n − f n K ( j )) ,
N i n t ∗ ( j 1 , j 2 ) f n ≔ − i ( Φ ( j 1 , j 2 ) f n − f n Φ ( j 1 , j 2 ) ) , \mathcal{N}_{\mathrm{int}}^*(j_1,j_2)f_n \coloneqq-i(\Phi(j_1,j_2)f_n-f_n\Phi(j_1,j_2)), N int ∗ ( j 1 , j 2 ) f n : = − i ( Φ ( j 1 , j 2 ) f n − f n Φ ( j 1 , j 2 )) ,
where the operator K ( j ) K(j) K ( j ) is the kinetic energy of the j j j -th particle, and Φ \Phi Φ is the pair-interaction potential.
For further discussion, we introduce the key concept of cumulants (semi-invariants) of operator groups [36 ] , that is, the connected part of operator groups.
The cumulant of the n n n th-order of operator groups (9) is defined by the following expansion [21 ] :
A n ( t , 1 , … , n ) = ∑ P : ( 1 , … , n ) = ⋃ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! ∏ X i ⊂ P G ∣ X i ∣ ∗ ( t , X i ) . (11) \mathfrak{A}_n(t,1,\ldots,n) = \sum_{\mathrm{P}:(1,\ldots,n)=\bigcup_i X_i}(-1)^{|\mathrm{P}|-1}(|\mathrm{P}|-1)!\prod_{X_i\subset\mathrm{P}}\mathcal{G}_{|X_i|}^*(t,X_i). \tag*{(11)} A n ( t , 1 , … , n ) = P : ( 1 , … , n ) = ⋃ i X i ∑ ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 )! X i ⊂ P ∏ G ∣ X i ∣ ∗ ( t , X i ) . (11)
where the symbol ∑ P : ( 1 , … , n ) = ⋃ i X i \sum_{\mathcal{P}:(1,\ldots,n)=\bigcup_i X_i} ∑ P : ( 1 , … , n ) = ⋃ i X i denotes the sum over all partitions P \mathcal{P} P of the index set ( 1 , … , n ) (1,\ldots,n) ( 1 , … , n ) into ∣ P ∣ |\mathcal{P}| ∣ P ∣ non-empty subsets X i ⊂ ( 1 , … , n ) X_i \subset(1,\ldots,n) X i ⊂ ( 1 , … , n ) that are mutually disjoint. The simplest examples of cumulants of operator groups (11) are presented by expansions:
A 1 ( t , 1 ) = G 1 ∗ ( t , 1 ) , A 2 ( t , 1 , 2 ) = G 2 ∗ ( t , 1 , 2 ) − G 1 ∗ ( t , 1 ) G 1 ∗ ( t , 2 ) , A 3 ( t , 1 , 2 , 3 ) = G 3 ∗ ( t , 1 , 2 , 3 ) − G 1 ∗ ( t , 1 ) G 2 ∗ ( t , 2 , 3 ) − G 1 ∗ ( t , 2 ) G 2 ∗ ( t , 1 , 3 ) − G 1 ∗ ( t , 3 ) G 2 ∗ ( t , 1 , 2 ) + 2 ! G 1 ∗ ( t , 1 ) G 1 ∗ ( t , 2 ) G 1 ∗ ( t , 3 ) , ⋮ \begin{aligned}
\mathfrak{A}_1(t,1) &= \mathfrak{G}_1^*(t,1),\\
\mathfrak{A}_2(t,1,2) &= \mathfrak{G}_2^*(t,1,2)-\mathfrak{G}_1^*(t,1)\mathfrak{G}_1^*(t,2),\\
\mathfrak{A}_3(t,1,2,3) &= \mathfrak{G}_3^*(t,1,2,3)-\mathfrak{G}_1^*(t,1)\mathfrak{G}_2^*(t,2,3)-\mathfrak{G}_1^*(t,2)\mathfrak{G}_2^*(t,1,3)\\
&\quad-\mathfrak{G}_1^*(t,3)\mathfrak{G}_2^*(t,1,2)+2!\mathfrak{G}_1^*(t,1)\mathfrak{G}_1^*(t,2)\mathfrak{G}_1^*(t,3),\\
&\vdots
\end{aligned} A 1 ( t , 1 ) A 2 ( t , 1 , 2 ) A 3 ( t , 1 , 2 , 3 ) = G 1 ∗ ( t , 1 ) , = G 2 ∗ ( t , 1 , 2 ) − G 1 ∗ ( t , 1 ) G 1 ∗ ( t , 2 ) , = G 3 ∗ ( t , 1 , 2 , 3 ) − G 1 ∗ ( t , 1 ) G 2 ∗ ( t , 2 , 3 ) − G 1 ∗ ( t , 2 ) G 2 ∗ ( t , 1 , 3 ) − G 1 ∗ ( t , 3 ) G 2 ∗ ( t , 1 , 2 ) + 2 ! G 1 ∗ ( t , 1 ) G 1 ∗ ( t , 2 ) G 1 ∗ ( t , 3 ) , ⋮
Cumulants of operator groups (11) are solutions of recursive equations known as cluster expansions of operator groups (9)
G n ∗ ( t , 1 , … , n ) = ∑ P : ( 1 , … , n ) = ⋃ i X i ∏ X i ∈ P A ∣ X i ∣ ( t , X i ) , n ≥ 1 , (12) \mathfrak{G}_n^*(t,1,\ldots,n)=\sum_{\mathcal{P}:(1,\ldots,n)=\bigcup_i X_i}\prod_{X_i\in\mathcal{P}}\mathfrak{A}_{|X_i|}(t,X_i),\qquad n\geq1, \tag*{(12)} G n ∗ ( t , 1 , … , n ) = P : ( 1 , … , n ) = ⋃ i X i ∑ X i ∈ P ∏ A ∣ X i ∣ ( t , X i ) , n ≥ 1 , (12)
where the notation of the formula (11) is used. It should be noted that the uniqueness of the structure of cluster expansions is determined by the symmetry of the infinitesimal generator (10) of the operator groups (9) .
For further consideration, we will also generalize the concepts of cluster and cumulant expansions of operator groups for systems consisting of clusters of quantum particles. We will use such abridged notations. The symbol { { i 1 , … , i s } } \{\{i_1,\ldots,i_s\}\} {{ i 1 , … , i s }} denotes a connected subset of indices ( i 1 , … , i s ) (i_1,\ldots,i_s) ( i 1 , … , i s ) of the set of indices ( 1 , … , n ) (1,\ldots,n) ( 1 , … , n ) , i.e., a set consisting of a single index element. Then the set { { X 1 } , … , { X ∣ P ∣ } } ∈ ( 1 , … , n ) \{\{X_1\},\ldots,\{X_{|\mathcal{P}|}\}\}\in(1,\ldots,n) {{ X 1 } , … , { X ∣ P ∣ }} ∈ ( 1 , … , n ) is the set of the ∣ P ∣ |\mathcal{P}| ∣ P ∣ elements that are mutually disjoint subsets X i ⊂ ( 1 , … , n ) X_i\subset(1,\ldots,n) X i ⊂ ( 1 , … , n ) of the partition P \mathcal{P} P of the set of indices ( 1 , … , n ) = ⋃ i = 1 ∣ P ∣ X i (1,\ldots,n)=\bigcup_{i=1}^{|\mathcal{P}|}X_i ( 1 , … , n ) = ⋃ i = 1 ∣ P ∣ X i .
The cumulant of the ∣ P ∣ |\mathcal{P}| ∣ P ∣ th-order of operator groups (9) for ∣ P ∣ |\mathcal{P}| ∣ P ∣ clusters of n n n quantum particles is denoted by A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ } ) \mathfrak{A}_{|\mathcal{P}|}(t,\{X_1\},\ldots,\{X_{|\mathcal{P}|}\}) A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ }) , and it is defined by the following expansion:
A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ } ) ≔ ∑ P ′ : { { X 1 } , … , { X ∣ P ∣ } } = ⋃ k Y k ( − 1 ) ∣ P ′ ∣ − 1 ( ∣ P ′ ∣ − 1 ) ! ∏ Y k ∈ P ′ G ∣ θ ( Y k ) ∣ ∗ ( t , θ ( Y k ) ) , (13) \mathfrak{A}_{|\mathcal{P}|}(t,\{X_1\},\ldots,\{X_{|\mathcal{P}|}\})\coloneqq\sum_{\mathcal{P}':\{\{X_1\},\ldots,\{X_{|\mathcal{P}|}\}\}=\bigcup_k Y_k}(-1)^{|\mathcal{P}'|-1}(|\mathcal{P}'|-1)!\prod_{Y_k\in\mathcal{P}'}\mathfrak{G}_{|\theta(Y_k)|}^*(t,\theta(Y_k)), \tag*{(13)} A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ }) : = P ′ : {{ X 1 } , … , { X ∣ P ∣ }} = ⋃ k Y k ∑ ( − 1 ) ∣ P ′ ∣ − 1 ( ∣ P ′ ∣ − 1 )! Y k ∈ P ′ ∏ G ∣ θ ( Y k ) ∣ ∗ ( t , θ ( Y k )) , (13)
where the notations are similar to that in formula (11) and the operator θ \theta θ is the declusterization mapping defined by θ ( { X 1 } , … , { X ∣ P ∣ } ) ≔ ( 1 , … , n ) \theta(\{X_1\},\ldots,\{X_{|\mathcal{P}|}\})\coloneqq(1,\ldots,n) θ ({ X 1 } , … , { X ∣ P ∣ }) : = ( 1 , … , n ) .
Note that the cumulant (13) of the ∣ P ∣ |\mathcal{P}| ∣ P ∣ th-order of operator groups for the ∣ P ∣ |\mathcal{P}| ∣ P ∣ clusters of n n n quantum particles is expressed in terms of the cumulants (11) of operator groups for n n n particles. A typical example of the cumulant of a cluster of s s s particles and n n n particles is determined by the expansions:
A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) = ∑ P : ( { 1 , … , s } , s + 1 , … , s + n ) = ⋃ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! × ∏ X i ∈ P ∑ P ′ : θ ( X i ) = ⋃ j i Y j i ∏ Y j i ∈ P ′ A ∣ Y j i ∣ ( t , Y j i ) , n ≥ 0 , (14) \begin{aligned}
\mathfrak{A}_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)
&=\sum_{\mathcal{P}:(\{1,\ldots,s\},s+1,\ldots,s+n)=\bigcup_i X_i}(-1)^{|\mathcal{P}|-1}(|\mathcal{P}|-1)!\\
&\quad\times\prod_{X_i\in\mathcal{P}}\sum_{\mathcal{P}':\theta(X_i)=\bigcup_{j_i}Y_{j_i}}\prod_{Y_{j_i}\in\mathcal{P}'}\mathfrak{A}_{|Y_{j_i}|}(t,Y_{j_i}),\qquad n\geq0, \tag*{(14)}
\end{aligned} A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) = P : ({ 1 , … , s } , s + 1 , … , s + n ) = ⋃ i X i ∑ ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 )! × X i ∈ P ∏ P ′ : θ ( X i ) = ⋃ j i Y j i ∑ Y j i ∈ P ′ ∏ A ∣ Y j i ∣ ( t , Y j i ) , n ≥ 0 , (14)
for example,
A 1 + 0 ( t , { 1 , 2 } ) = A 2 ( t , 1 , 2 ) + A 1 ( t , 1 ) A 1 ( t , 2 ) , \mathcal{A}_{1+0}(t,\{1,2\})=\mathcal{A}_{2}(t,1,2)+\mathcal{A}_{1}(t,1)\mathcal{A}_{1}(t,2), A 1 + 0 ( t , { 1 , 2 }) = A 2 ( t , 1 , 2 ) + A 1 ( t , 1 ) A 1 ( t , 2 ) ,
A 1 + 1 ( t , { 1 , 2 } , 3 ) = A 3 ( t , 1 , 2 , 3 ) + A 2 ( t , 1 , 3 ) A 1 ( t , 2 ) + A 2 ( t , 2 , 3 ) A 1 ( t , 1 ) , \mathcal{A}_{1+1}(t,\{1,2\},3)=\mathcal{A}_{3}(t,1,2,3)+\mathcal{A}_{2}(t,1,3)\mathcal{A}_{1}(t,2)+\mathcal{A}_{2}(t,2,3)\mathcal{A}_{1}(t,1), A 1 + 1 ( t , { 1 , 2 } , 3 ) = A 3 ( t , 1 , 2 , 3 ) + A 2 ( t , 1 , 3 ) A 1 ( t , 2 ) + A 2 ( t , 2 , 3 ) A 1 ( t , 1 ) ,
If f n ∈ L 1 ( H ⊗ n ) f_n\in\mathcal{L}^{1}(\mathcal{H}^{\otimes n}) f n ∈ L 1 ( H ⊗ n ) , then for the cumulant of n n n th-order (11) , the following estimate is valid
∥ A n ( t , 1 , … , n ) f n ∥ L 1 ( H ⊗ n ) ≤ ∑ P : ( 1 , … , n ) = ⋃ i X i ( ∣ P ∣ − 1 ) ! ∥ f n ∥ L 1 ( H ⊗ n ) ≤ ∑ k = 1 n s ( n , k ) ( k − 1 ) ! ∥ f n ∥ L 1 ( H ⊗ n ) ≤ n ! e n ∥ f n ∥ L 1 ( H ⊗ n ) , (15) \left\|\mathcal{A}_{n}(t,1,\ldots,n)f_n\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes n})}
\leq
\sum_{\mathrm{P}:(1,\ldots,n)=\bigcup_i X_i} (|\mathrm{P}|-1)!\left\|f_n\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes n})}
\leq
\sum_{k=1}^{n} s(n,k)(k-1)!\left\|f_n\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes n})}
\leq n!e^n\left\|f_n\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes n})}, \tag*{(15)} ∥ A n ( t , 1 , … , n ) f n ∥ L 1 ( H ⊗ n ) ≤ P : ( 1 , … , n ) = ⋃ i X i ∑ ( ∣ P ∣ − 1 )! ∥ f n ∥ L 1 ( H ⊗ n ) ≤ k = 1 ∑ n s ( n , k ) ( k − 1 )! ∥ f n ∥ L 1 ( H ⊗ n ) ≤ n ! e n ∥ f n ∥ L 1 ( H ⊗ n ) , (15)
where s ( n , k ) s(n,k) s ( n , k ) are the Stirling numbers of the second kind.
Evolution of states in terms of density operators
The evolution of a state of non-fixed, but finite average number of quantum particles, i.e., the sequence of density operators (8) , is determined by the Cauchy problem for the sequence of the von Neumann equations:
∂ ∂ t D n ( t ) = N n ∗ D n ( t ) , (16) \frac{\partial}{\partial t}D_n(t)=\mathcal{N}_n^*D_n(t), \tag*{(16)} ∂ t ∂ D n ( t ) = N n ∗ D n ( t ) , (16)
D n ( t ) ∣ t = 0 = D n 0 , n ≥ 1 , (17) D_n(t)\big|_{t=0}=D_n^0,\qquad n\geq1, \tag*{(17)} D n ( t ) t = 0 = D n 0 , n ≥ 1 , (17)
where the generator of these evolution equations is given by formula (10) .
As a consequence of the properties of operator groups (9) , an existence theorem holds [21 ] .
Theorem 2. For initial data D ( 0 ) ∈ L 0 1 ( F H ) D(0)\in\mathcal{L}^{1}_{0}(\mathcal{F}_{\mathcal{H}}) D ( 0 ) ∈ L 0 1 ( F H ) , the sequence (8) is the unique global-in-time strong solution to the Cauchy problem for the von Neumann equations (16) ,(17) , whereas for arbitrary initial data from the space L α 1 ( F H ) \mathcal{L}^{1}_{\alpha}(\mathcal{F}_{\mathcal{H}}) L α 1 ( F H ) , it is a unique global-in-time weak solution.
This statement also holds for interaction potentials that satisfy the Kato conditions [35 ] .
Evolution of states in terms of correlation operators
An alternative way to describe the states of quantum particle systems involves using operators defined as cumulant expansions of density operators, which are interpreted as correlations of particle states.
We introduce sequence g ( t ) = ( I , g 1 ( t , 1 ) , … , g s ( t , 1 , … , s ) , … ) g(t)=(I,g_1(t,1),\ldots,g_s(t,1,\ldots,s),\ldots) g ( t ) = ( I , g 1 ( t , 1 ) , … , g s ( t , 1 , … , s ) , … ) of correlation operators by means of cluster expansions of the density operators D ( t ) = ( I , D 1 ( t , 1 ) , … , D n ( t , 1 , … , n ) , … ) D(t)=(I,D_1(t,1),\ldots,D_n(t,1,\ldots,n),\ldots) D ( t ) = ( I , D 1 ( t , 1 ) , … , D n ( t , 1 , … , n ) , … ) :
D n ( t , 1 , … , n ) = ∑ P : ( 1 , … , n ) = ⋃ i X i ⨂ X i ⊂ P g ∣ X i ∣ ( t , X i ) , n ≥ 1 , (18) D_n(t,1,\ldots,n)=\sum_{\mathrm{P}:(1,\ldots,n)=\bigcup_i X_i}\bigotimes_{X_i\subset\mathrm{P}}g_{|X_i|}(t,X_i),\qquad n\geq1, \tag*{(18)} D n ( t , 1 , … , n ) = P : ( 1 , … , n ) = ⋃ i X i ∑ X i ⊂ P ⨂ g ∣ X i ∣ ( t , X i ) , n ≥ 1 , (18)
where the symbol ∑ P : ( 1 , … , n ) = ⋃ i X i \sum_{\mathrm{P}:(1,\ldots,n)=\bigcup_i X_i} ∑ P : ( 1 , … , n ) = ⋃ i X i denotes the sum over all partitions P \mathrm{P} P of the index set ( 1 , … , n ) (1,\ldots,n) ( 1 , … , n ) into ∣ P ∣ |\mathrm{P}| ∣ P ∣ non-empty subsets X i ⊂ ( 1 , … , n ) X_i\subset(1,\ldots,n) X i ⊂ ( 1 , … , n ) that are mutually disjoint.
The solutions of recurrent relations (18) are determined by the following expansions:
g n ( t , 1 , … , n ) = ∑ P : ( 1 , … , n ) = ⋃ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! ⨂ X i ⊂ P D ∣ X i ∣ ( t , X i ) , n ≥ 1. (19) g_n(t,1,\ldots,n)=\sum_{\substack{\mathrm{P}:(1,\ldots,n)=\bigcup_i X_i}}(-1)^{|\mathrm{P}|-1}(|\mathrm{P}|-1)!\bigotimes_{X_i\subset\mathrm{P}}D_{|X_i|}(t,X_i),\qquad n\geq1. \tag*{(19)} g n ( t , 1 , … , n ) = P : ( 1 , … , n ) = ⋃ i X i ∑ ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 )! X i ⊂ P ⨂ D ∣ X i ∣ ( t , X i ) , n ≥ 1. (19)
The structure of expansions (19) is such that correlation operators can be interpreted as cumulants (semi-invariants) of density operators (8) . The origin of the structure of cluster expansions for density operators will be argued further in subsection 2.5. The structure of cumulant expansions for states (19) is determined by explicitly incorporating the normalizing factor in mean-value functional (7) , which is represented through reduced density operators.
We will also expand the concept of cumulant expansions for correlation operators in systems with many quantum particles, including clusters of particles. The correlation operator (19) for ∣ P ∣ |\mathrm{P}| ∣ P ∣ clusters of n n n -particle system, i.e., ( 1 , … , n ) = ⋃ P θ ( { X ∣ P ∣ } ) (1,\ldots,n)=\bigcup_{\mathrm{P}}\theta(\{X_{|\mathrm{P}|}\}) ( 1 , … , n ) = ⋃ P θ ({ X ∣ P ∣ }) , we will denote by g ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ } ) g_{|\mathrm{P}|}(t,\{X_1\},\ldots,\{X_{|\mathrm{P}|}\}) g ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ }) , and it is defined by the following expansion:
g ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ } ) = ∑ P ′ : ( { X 1 } , … , { X ∣ P ∣ } ) = ⋃ k Y k ( − 1 ) ∣ P ′ ∣ − 1 ( ∣ P ′ ∣ − 1 ) ! ⨂ Y k ⊂ P ′ D ∣ θ ( Y k ) ∣ ( t , θ ( Y k ) ) , (20) g_{|\mathrm{P}|}(t,\{X_1\},\ldots,\{X_{|\mathrm{P}|}\})=
\sum_{\substack{\mathrm{P}':(\{X_1\},\ldots,\{X_{|\mathrm{P}|}\})=\bigcup_k Y_k}}(-1)^{|\mathrm{P}'|-1}(|\mathrm{P}'|-1)!\bigotimes_{Y_k\subset\mathrm{P}'}D_{|\theta(Y_k)|}(t,\theta(Y_k)), \tag*{(20)} g ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ }) = P ′ : ({ X 1 } , … , { X ∣ P ∣ }) = ⋃ k Y k ∑ ( − 1 ) ∣ P ′ ∣ − 1 ( ∣ P ′ ∣ − 1 )! Y k ⊂ P ′ ⨂ D ∣ θ ( Y k ) ∣ ( t , θ ( Y k )) , (20)
where we used the abridged notation similar to the formula (13) . In the particular case of a cluster consisting of n n n particles, this expansion (20) takes the form
g 1 ( t , { 1 , … , n } ) = D n ( t , 1 , … , n ) . g_1(t,\{1,\ldots,n\})=D_n(t,1,\ldots,n). g 1 ( t , { 1 , … , n }) = D n ( t , 1 , … , n ) .
According to the cluster expansions (18) , we derive the relation between the correlation operator of a cluster of particles and the correlation operators of individual particles
g 1 ( t , { 1 , … , n } ) = ∑ P : θ ( { 1 , … , n } ) = ⋃ i X i ⨂ X i ⊂ P g ∣ X i ∣ ( X i ) . g_1(t,\{1,\ldots,n\})=\sum_{\substack{\mathrm{P}:\theta(\{1,\ldots,n\})=\bigcup_i X_i}}\bigotimes_{X_i\subset\mathrm{P}}g_{|X_i|}(X_i). g 1 ( t , { 1 , … , n }) = P : θ ({ 1 , … , n }) = ⋃ i X i ∑ X i ⊂ P ⨂ g ∣ X i ∣ ( X i ) .
By applying the cluster expansions (12) of operator groups (9) to operator groups in the definition (20) of cumulants of density operators (8) and considering the definition of the correlation operators at the initial instant, we derive that the evolution of all possible states of a quantum system of non-fixed, i.e., an arbitrary but finite number of particles, is described by means of the sequence of the following correlation operators:
g n ( t , 1 , … , n ) = ∑ P : ( 1 , … , n ) = ⋃ j X j ∏ X j ⊂ P A ∣ X j ∣ ( t , X j ) g ∣ P ∣ 0 ( { X 1 } , … , { X ∣ P ∣ } ) , n ≥ 1 , (21) g_n(t,1,\ldots,n)=\sum_{\substack{\mathrm{P}:(1,\ldots,n)=\bigcup_j X_j}}\prod_{X_j\subset\mathrm{P}}\mathcal{A}_{|X_j|}(t,X_j)g^0_{|\mathrm{P}|}(\{X_1\},\ldots,\{X_{|\mathrm{P}|}\}),\qquad n\geq1, \tag*{(21)} g n ( t , 1 , … , n ) = P : ( 1 , … , n ) = ⋃ j X j ∑ X j ⊂ P ∏ A ∣ X j ∣ ( t , X j ) g ∣ P ∣ 0 ({ X 1 } , … , { X ∣ P ∣ }) , n ≥ 1 , (21)
where the abridged notation, similar to that used in formula (13) , is applied here. For arbitrary initial states described in terms of individual particle correlation operators, the evolution of correlation operators can also be represented by similar expansions, which are equivalent to the previous expansions (21) :
g n ( t , 1 , … , n ) = ∑ P : ( 1 , … , n ) = ⋃ j X j A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ } ) ⨂ X i ⊂ P g ∣ X i ∣ 0 ( X i ) , n ≥ 1. (22) g_n(t,1,\ldots,n)=\sum_{\substack{\mathrm{P}:(1,\ldots,n)=\bigcup_j X_j}}\mathcal{A}_{|\mathrm{P}|}(t,\{X_1\},\ldots,\{X_{|\mathrm{P}|}\})\bigotimes_{X_i\subset\mathrm{P}}g^0_{|X_i|}(X_i),\qquad n\geq1. \tag*{(22)} g n ( t , 1 , … , n ) = P : ( 1 , … , n ) = ⋃ j X j ∑ A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ }) X i ⊂ P ⨂ g ∣ X i ∣ 0 ( X i ) , n ≥ 1. (22)
where generating operators are cumulants of the corresponding order of operator groups (9) for ∣ P ∣ |P| ∣ P ∣ clusters of n n n quantum particles is defined by the expansion (13) . The simplest examples of correlation operators (22) are represented by such expansions:
g 1 ( t , 1 ) = A 1 ( t , 1 ) g 1 0 ( 1 ) , g_1(t,1)=\mathcal{A}_1(t,1)g_1^0(1), g 1 ( t , 1 ) = A 1 ( t , 1 ) g 1 0 ( 1 ) ,
g 2 ( t , 1 , 2 ) = A 1 ( t , { 1 , 2 } ) g 2 0 ( 1 , 2 ) + A 2 ( t , 1 , 2 ) g 1 0 ( 1 ) ⊗ g 1 0 ( 2 ) , g_2(t,1,2)=\mathcal{A}_1(t,\{1,2\})g_2^0(1,2)+\mathcal{A}_2(t,1,2)g_1^0(1)\otimes g_1^0(2), g 2 ( t , 1 , 2 ) = A 1 ( t , { 1 , 2 }) g 2 0 ( 1 , 2 ) + A 2 ( t , 1 , 2 ) g 1 0 ( 1 ) ⊗ g 1 0 ( 2 ) ,
In particular, for separable initial states, the sequence of initial correlation operators takes the form g ( 0 ) = ( 0 , g 1 0 , 0 , … , 0 , … ) g(0)=(0,g_1^0,0,\ldots,0,\ldots) g ( 0 ) = ( 0 , g 1 0 , 0 , … , 0 , … ) . Then the evolution of correlation operators is represented as follows:
g n ( t , 1 , … , n ) = A n ( t , 1 , … , n ) ⨂ j = 1 n g 1 0 ( j ) , n ≥ 1 , g_n(t,1,\ldots,n)=\mathcal{A}_n(t,1,\ldots,n)\bigotimes_{j=1}^{n}g_1^0(j),\quad n\geq1, g n ( t , 1 , … , n ) = A n ( t , 1 , … , n ) j = 1 ⨂ n g 1 0 ( j ) , n ≥ 1 ,
where the generating operator A n ( t ) \mathcal{A}_n(t) A n ( t ) is the n n n th-order cumulant of operator groups (9) determined by expansion (11) .
Taking into consideration the explicit form of a one-particle correlation operator (22)
g 1 ( t , j ) = A 1 ( t , j ) g 1 0 ( j ) = g_1(t,j)=\mathcal{A}_1(t,j)g_1^0(j)= g 1 ( t , j ) = A 1 ( t , j ) g 1 0 ( j ) =
G 1 ∗ ( t , j ) g 1 0 ( j ) = e − i ℏ t K ( j ) g 1 0 ( j ) e i ℏ t K ( j ) , \mathcal{G}_1^*(t,j)g_1^0(j)=e^{-\frac{i}{\hbar}tK(j)}g_1^0(j)e^{\frac{i}{\hbar}tK(j)}, G 1 ∗ ( t , j ) g 1 0 ( j ) = e − ℏ i t K ( j ) g 1 0 ( j ) e ℏ i t K ( j ) ,
we derive the law of the creation of correlations (the process of creating correlations from chaos)
g n ( t , 1 , … , n ∣ g 1 ( t ) ) ≔ A n ( t , 1 , … , n ) ⨂ j = 1 n ( A 1 ( t , j ) ) − 1 g 1 ( t , j ) = A ^ n ( t , 1 , … , n ) ⨂ j = 1 n g 1 ( t , j ) , n ≥ 2 , (23) g_n(t,1,\ldots,n\mid g_1(t))\coloneqq\mathcal{A}_n(t,1,\ldots,n)\bigotimes_{j=1}^{n}\left(\mathcal{A}_1(t,j)\right)^{-1}g_1(t,j)=
\widehat{\mathcal{A}}_n(t,1,\ldots,n)\bigotimes_{j=1}^{n}g_1(t,j),\quad n\geq2, \tag*{(23)} g n ( t , 1 , … , n ∣ g 1 ( t )) : = A n ( t , 1 , … , n ) j = 1 ⨂ n ( A 1 ( t , j ) ) − 1 g 1 ( t , j ) = A n ( t , 1 , … , n ) j = 1 ⨂ n g 1 ( t , j ) , n ≥ 2 , (23)
where the cumulant A ^ n ( t ) \widehat{\mathcal{A}}_n(t) A n ( t ) of the groups of scattering operators was introduced
G ^ n ( t ) ≔ G n ∗ ( t , 1 , … , n ) ∏ i = 1 n ( G 1 ∗ ( t , i ) ) − 1 , n ≥ 1. (24) \widehat{\mathcal{G}}_n(t)\coloneqq\mathcal{G}_n^*(t,1,\ldots,n)\prod_{i=1}^{n}\left(\mathcal{G}_1^*(t,i)\right)^{-1},\quad n\geq1. \tag*{(24)} G n ( t ) : = G n ∗ ( t , 1 , … , n ) i = 1 ∏ n ( G 1 ∗ ( t , i ) ) − 1 , n ≥ 1. (24)
Note that the non-perturbative solution (22) of the Cauchy problem of the von Neumann hierarchy (27) ,(28) is reduced to an iterative (perturbative) series as a result of applying analogues of the Duhamel equations for cumulants (11) of operator groups [21 ] , [32 ] .
For f ∈ L α 1 ( F H ) f\in\mathcal{L}_{\alpha}^{1}(\mathcal{F}_{\mathcal{H}}) f ∈ L α 1 ( F H ) , the one-parameter family of nonlinear mappings (22)
G ( t ; 1 , … , n ∣ f ) ≔ ∑ P : ( 1 , … , n ) = ⋃ i X i G ∣ P ∣ ∗ ( t , { X 1 } , … , { X ∣ P ∣ } ) ⨂ X i ⊂ P f ∣ X i ∣ ( X i ) , (25) \mathcal{G}(t;1,\ldots,n\mid f)\coloneqq\sum_{\mathsf{P}:(1,\ldots,n)=\bigcup_iX_i}\mathcal{G}_{|P|}^*(t,\{X_1\},\ldots,\{X_{|P|}\})\bigotimes_{X_i\subset P}f_{|X_i|}(X_i), \tag*{(25)} G ( t ; 1 , … , n ∣ f ) : = P : ( 1 , … , n ) = ⋃ i X i ∑ G ∣ P ∣ ∗ ( t , { X 1 } , … , { X ∣ P ∣ }) X i ⊂ P ⨂ f ∣ X i ∣ ( X i ) , (25)
is defined, and the following estimate holds for the correlation operators g ( t , 1 , … , n ) = G ( t ; 1 , … , n ∣ g ( 0 ) ) g(t,1,\ldots,n)=\mathcal{G}(t;1,\ldots,n\mid g(0)) g ( t , 1 , … , n ) = G ( t ; 1 , … , n ∣ g ( 0 )) , n ≥ 1 n\geq1 n ≥ 1 :
∥ G ( t ; 1 , … , n ∣ g ( 0 ) ) ∥ L 1 ( H ⊗ n ) ≤ n ! e 2 n c n , \left\|\mathcal{G}(t;1,\ldots,n\mid g(0))\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes n})}\leq n!e^{2n}c^n, ∥ G ( t ; 1 , … , n ∣ g ( 0 )) ∥ L 1 ( H ⊗ n ) ≤ n ! e 2 n c n ,
where c ≡ e 3 max ( 1 , max P : ( 1 , … , n ) = ⋃ i X i ∥ g ∣ X i ∣ 0 ∥ L 1 ( H ⊗ ∣ X i ∣ ) ) c\equiv e^3\max\left(1,\max_{\mathsf{P}:(1,\ldots,n)=\bigcup_iX_i}\left\|g_{|X_i|}^0\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes|X_i|})}\right) c ≡ e 3 max ( 1 , max P : ( 1 , … , n ) = ⋃ i X i g ∣ X i ∣ 0 L 1 ( H ⊗ ∣ X i ∣ ) ) , and we denoted g ( 0 ) = ( g 0 , g 1 0 , … , g n 0 , … ) g(0)=(g_0,g_1^0,\ldots,g_n^0,\ldots) g ( 0 ) = ( g 0 , g 1 0 , … , g n 0 , … ) . In the space f ∈ L α 1 ( F H ) f\in\mathcal{L}_{\alpha}^{1}(\mathcal{F}_{\mathcal{H}}) f ∈ L α 1 ( F H ) , the one-parameter mapping (25) forms a strongly continuous group of nonlinear operators. On the subspace L 0 1 ( F H ) \mathcal{L}^{1}_{0}(\mathcal{F}_{\mathcal{H}}) L 0 1 ( F H ) the infinitesimal generator of the group of nonlinear operators (25) coincides with the following operator:
N ( 1 , … , n ∣ f ) = N n ∗ f n ( 1 , … , n ) + ∑ P : ( 1 , … , n ) = X 1 ∪ X 2 ∑ i 1 ∈ X 1 ∑ i 2 ∈ X 2 N i n t ∗ ( i 1 , i 2 ) f ∣ X 1 ∣ ( t , X 1 ) ⊗ f ∣ X 2 ∣ ( t , X 2 ) , (26) \mathcal{N}(1,\ldots,n\mid f)=\mathcal{N}^{*}_{n}f_{n}(1,\ldots,n)+\sum_{\mathrm{P}:(1,\ldots,n)=X_{1}\cup X_{2}}\sum_{i_{1}\in X_{1}}\sum_{i_{2}\in X_{2}}\mathcal{N}^{*}_{\mathrm{int}}(i_{1},i_{2})f_{\lvert X_{1}\rvert}(t,X_{1})\otimes f_{\lvert X_{2}\rvert}(t,X_{2}), \tag*{(26)} N ( 1 , … , n ∣ f ) = N n ∗ f n ( 1 , … , n ) + P : ( 1 , … , n ) = X 1 ∪ X 2 ∑ i 1 ∈ X 1 ∑ i 2 ∈ X 2 ∑ N int ∗ ( i 1 , i 2 ) f ∣ X 1 ∣ ( t , X 1 ) ⊗ f ∣ X 2 ∣ ( t , X 2 ) , (26)
where the symbol ∑ P : ( 1 , … , n ) = X 1 ∪ X 2 \sum_{\mathrm{P}:(1,\ldots,n)=X_{1}\cup X_{2}} ∑ P : ( 1 , … , n ) = X 1 ∪ X 2 denotes the sum over all possible partitions P of the index set ( 1 , … , n ) (1,\ldots,n) ( 1 , … , n ) into two non-empty disjoint subsets X 1 X_{1} X 1 and X 2 X_{2} X 2 , and the operators N n ∗ \mathcal{N}^{*}_{n} N n ∗ and N i n t ∗ \mathcal{N}^{*}_{\mathrm{int}} N int ∗ are defined by the formula (10) . The evolution of a state of non-fixed, but finite average number of quantum particles, i.e., the sequence of correlation operators (22) , is determined by the Cauchy problem for the sequence of the von Neumann hierarchy [33 ] :
∂ ∂ t g n ( t , 1 , … , n ) = N n ∗ ( 1 , … , n ) g n ( t , 1 , … , n ) + ∑ P : ( 1 , … , n ) = X 1 ∪ X 2 ∑ i 1 ∈ X 1 ∑ i 2 ∈ X 2 N i n t ∗ ( i 1 , i 2 ) g ∣ X 1 ∣ ( t , X 1 ) ⊗ g ∣ X 2 ∣ ( t , X 2 ) , (27) \frac{\partial}{\partial t}g_{n}(t,1,\ldots,n)=\mathcal{N}^{*}_{n}(1,\ldots,n)g_{n}(t,1,\ldots,n)+\sum_{\mathrm{P}:(1,\ldots,n)=X_{1}\cup X_{2}}\sum_{i_{1}\in X_{1}}\sum_{i_{2}\in X_{2}}\mathcal{N}^{*}_{\mathrm{int}}(i_{1},i_{2})g_{\lvert X_{1}\rvert}(t,X_{1})\otimes g_{\lvert X_{2}\rvert}(t,X_{2}), \tag*{(27)} ∂ t ∂ g n ( t , 1 , … , n ) = N n ∗ ( 1 , … , n ) g n ( t , 1 , … , n ) + P : ( 1 , … , n ) = X 1 ∪ X 2 ∑ i 1 ∈ X 1 ∑ i 2 ∈ X 2 ∑ N int ∗ ( i 1 , i 2 ) g ∣ X 1 ∣ ( t , X 1 ) ⊗ g ∣ X 2 ∣ ( t , X 2 ) , (27)
g n ( t ) ∣ t = 0 = g n 0 , n ≥ 1 , (28) g_{n}(t)\big\vert_{t=0}=g^{0}_{n},\qquad n\geq1, \tag*{(28)} g n ( t ) t = 0 = g n 0 , n ≥ 1 , (28)
where the symbol ∑ P : ( 1 , … , n ) = X 1 ∪ X 2 \sum_{\mathrm{P}:(1,\ldots,n)=X_{1}\cup X_{2}} ∑ P : ( 1 , … , n ) = X 1 ∪ X 2 means the sum over all possible partitions P of the set ( 1 , … , n ) (1,\ldots,n) ( 1 , … , n ) into two nonempty mutually disjoint subsets X 1 X_{1} X 1 and X 2 X_{2} X 2 , as was accepted above. It should be noted that the von Neumann hierarchy (27) for correlation operators constitutes a set of nonlinear recurrence evolution equations. We present examples of typical equations of the hierarchy (27) :
∂ ∂ t g 1 ( t , 1 ) = N 1 ∗ ( 1 ) g 1 ( t , 1 ) , ∂ ∂ t g 2 ( t , 1 , 2 ) = N 2 ∗ ( 1 , 2 ) g 2 ( t , 1 , 2 ) + N i n t ∗ ( 1 , 2 ) g 1 ( t , 1 ) ⊗ g 1 ( t , 2 ) , ⋮ \begin{aligned}
\frac{\partial}{\partial t}g_{1}(t,1)&=\mathcal{N}^{*}_{1}(1)g_{1}(t,1),\\
\frac{\partial}{\partial t}g_{2}(t,1,2)&=\mathcal{N}^{*}_{2}(1,2)g_{2}(t,1,2)+\mathcal{N}^{*}_{\mathrm{int}}(1,2)g_{1}(t,1)\otimes g_{1}(t,2),\\
&\vdots
\end{aligned} ∂ t ∂ g 1 ( t , 1 ) ∂ t ∂ g 2 ( t , 1 , 2 ) = N 1 ∗ ( 1 ) g 1 ( t , 1 ) , = N 2 ∗ ( 1 , 2 ) g 2 ( t , 1 , 2 ) + N int ∗ ( 1 , 2 ) g 1 ( t , 1 ) ⊗ g 1 ( t , 2 ) , ⋮
As a consequence of the properties of operator groups (25) , an existence theorem holds in the space of sequences of trace-class operators [21 ] .
Theorem 3. For bounded interaction potentials, a non-perturbative solution to the Cauchy problem for the von Neumann hierarchy (27) ,(28) is given by the sequence of correlation operators represented by the expansions (22) . For initial data D ( 0 ) ∈ L 0 1 ( F H ) D(0)\in\mathcal{L}^{1}_{0}(\mathcal{F}_{\mathcal{H}}) D ( 0 ) ∈ L 0 1 ( F H ) , the sequence of correlation operators (8) is the globalglobal-in-time classical solution to the Cauchy problem for the von Neumann hierarchy, whereas for arbitrary initial data from the space L α 1 ( F H ) \mathcal{L}^{1}_{\alpha}(\mathcal{F}_{\mathcal{H}}) L α 1 ( F H ) , it is a global-in-time generalized solution.
Thus, the cumulant genesis of the operators for state correlations determines a cumulant structure of the generating operators (13) for expansions (22) that describe the propagation of correlations in systems of many quantum particles.
Note that the dynamics of correlations underlie the description of the state evolution, both for finitely and infinitely many quantum particles.
On the evolution of observables and states of an infinite number of particles
This subsection provides the classification of hierarchies of evolution equations for quantum many-particle systems. There are two primary approaches to describing the evolution of many-particle quantum systems. The first approach involves observables whose evolution is governed by the hierarchy of evolution equations for reduced observables. The second approach is based on a notion of the state whose evolution is determined by the BBGKY hierarchy. For systems with a finite number of particles, these hierarchies of evolution equations are equivalent to the Heisenberg equations (4) for observables and the von Neumann equations (16) for density operators, respectively. The advantage of these hierarchies from a mathematical viewpoint lies in their capacity to rigorously describe the evolution of infinitely many particles whose collective behavior is characterized by statistical features, namely, the existence of an equilibrium state, as well as the kinetic or hydrodynamic behavior in appropriate scaling approximations.
Representations of the mean-value functional of observables (6) underlie the classification of possible equivalent methods for describing observables and the state of quantum many-particle systems. To construct new representations of the mean-value functional (6) , it is convenient to introduce the following operators acting on sequences of operators. In the space L γ ( F H ) \mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}}) L γ ( F H ) , we define an operator analogous to the creation operator
( a + b ) n ( 1 , … , n ) ≔ ∑ j = 1 n b n − 1 ( 1 , … , n ∖ j ) ⊗ 1 ( j ) , n ≥ 1 , (29) (a^{+}b)_n(1,\ldots,n) \coloneqq\sum_{j=1}^{n} b_{n-1}(1,\ldots,n\setminus j)\otimes\mathbb{1}_{(j)}, \qquad n\geq1, \tag*{(29)} ( a + b ) n ( 1 , … , n ) : = j = 1 ∑ n b n − 1 ( 1 , … , n ∖ j ) ⊗ 1 ( j ) , n ≥ 1 , (29)
where the symbol 1 ( j ) \mathbb{1}_{(j)} 1 ( j ) denotes the identity operator in the j j j Hilbert subspace of the space H ⊗ n \mathcal{H}^{\otimes n} H ⊗ n . On sequences of trace-class operators from the space L α 1 ( F H ) \mathcal{L}^{1}_{\alpha}(\mathcal{F}_{\mathcal{H}}) L α 1 ( F H ) we define an adjoint operator to the operator (29) in the sense of the functional (6) , which is an analogue to the annihilation operator,
( a f ) n ( 1 , … , n ) ≔ Tr n + 1 f n + 1 ( 1 , … , n , n + 1 ) , n ≥ 1. (30) (af)_n(1,\ldots,n) \coloneqq\operatorname{Tr}_{n+1} f_{n+1}(1,\ldots,n,n+1), \qquad n\geq1. \tag*{(30)} ( a f ) n ( 1 , … , n ) : = Tr n + 1 f n + 1 ( 1 , … , n , n + 1 ) , n ≥ 1. (30)
Then, for b ∈ L γ ( F H ) b\in\mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}}) b ∈ L γ ( F H ) and f ∈ L α 1 ( F H ) f\in\mathcal{L}^{1}_{\alpha}(\mathcal{F}_{\mathcal{H}}) f ∈ L α 1 ( F H ) by virtue of the equalities: ( b , f ) = ( e a + e − a + b , f ) = ( e − a + b , e a f ) (b,f)=(e^{a^{+}}e^{-a^{+}}b,f)=(e^{-a^{+}}b,e^{a}f) ( b , f ) = ( e a + e − a + b , f ) = ( e − a + b , e a f ) , for mean-value functional (6) , we derive the following representation:
⟨ O ⟩ ( t ) = ( I , D ( 0 ) ) − 1 ( O ( t ) , D ( 0 ) ) = ( B ( t ) , F ( 0 ) ) , (31) \langle O\rangle(t)=(I,D(0))^{-1}(O(t),D(0))=(B(t),F(0)), \tag*{(31)} ⟨ O ⟩ ( t ) = ( I , D ( 0 ) ) − 1 ( O ( t ) , D ( 0 )) = ( B ( t ) , F ( 0 )) , (31)
where the coefficient ( I , D ( 0 ) ) − 1 (I,D(0))^{-1} ( I , D ( 0 ) ) − 1 is the normalizing factor and the sequence of reduced observables is given by the formula
B ( t ) = e − a + G ( t ) O ( 0 ) , (32) B(t)=e^{-a^{+}}\mathcal{G}(t)O(0), \tag*{(32)} B ( t ) = e − a + G ( t ) O ( 0 ) , (32)
and the sequence of reduced density operators is given by the following expression
F ( 0 ) = ( I , D ( 0 ) ) − 1 e a D ( 0 ) . (33) F(0)=(I,D(0))^{-1}e^{a}D(0). \tag*{(33)} F ( 0 ) = ( I , D ( 0 ) ) − 1 e a D ( 0 ) . (33)
Note that the description of observables and the system state in terms of the corresponding reduced operators is a result of the series resummation in expression (6) , taking into account the normalization factor ( I , D ( 0 ) ) − 1 (I,D(0))^{-1} ( I , D ( 0 ) ) − 1 .
Thus, according to the definition of operator (29) , the sequence of reduced observables (32) in component-wise form is given by the following expansions:
B s ( t , 1 , … , s ) = ∑ n = 0 s ( − 1 ) n n ! ∑ j 1 ≠ ⋯ ≠ j n = 1 s ( G ( t ) O ( 0 ) ) s − n ( ( 1 , … , s ) ∖ ( j 1 , … , j n ) ) ⊗ I ( j 1 ) ⊗ ⋯ ⊗ I ( j n ) , s ≥ 1. (34) B_s(t,1,\ldots,s)=\sum_{n=0}^{s}\frac{(-1)^n}{n!}\sum_{j_1\ne\cdots\ne j_n=1}^{s}\left(\mathcal{G}(t)\mathcal{O}(0)\right)_{s-n}\left((1,\ldots,s)\setminus(j_1,\ldots,j_n)\right)\otimes\mathbb{I}_{(j_1)}\otimes\cdots\otimes\mathbb{I}_{(j_n)},\qquad s\ge1. \tag*{(34)} B s ( t , 1 , … , s ) = n = 0 ∑ s n ! ( − 1 ) n j 1 = ⋯ = j n = 1 ∑ s ( G ( t ) O ( 0 ) ) s − n ( ( 1 , … , s ) ∖ ( j 1 , … , j n ) ) ⊗ I ( j 1 ) ⊗ ⋯ ⊗ I ( j n ) , s ≥ 1. (34)
It should be emphasized that for the mean-value functional (31) for the reduced observables, a fundamental relation holds with its equivalent representation
( B ( t ) , F ( 0 ) ) = ( B ( 0 ) , F ( t ) ) , (35) (B(t),F(0))=(B(0),F(t)), \tag*{(35)} ( B ( t ) , F ( 0 )) = ( B ( 0 ) , F ( t )) , (35)
where according to the definition of operators (30) and (33) , the sequence of reduced density operators at instant t ∈ R t\in\mathbb{R} t ∈ R in component-wise form is represented by the following expansions into series:
F s ( t , 1 , … , s ) = ( I , D ( 0 ) ) − 1 ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n ( G ∗ ( t ) D ( 0 ) ) s + n ( 1 , … , s + n ) , s ≥ 1. (36) F_s(t,1,\ldots,s)=(I,D(0))^{-1}\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\left(\mathcal{G}^{*}(t)D(0)\right)_{s+n}(1,\ldots,s+n),\qquad s\ge1. \tag*{(36)} F s ( t , 1 , … , s ) = ( I , D ( 0 ) ) − 1 n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n ( G ∗ ( t ) D ( 0 ) ) s + n ( 1 , … , s + n ) , s ≥ 1. (36)
As was noted above, observables and a state of a system with a finite number of particles are described by one-component sequences. Consequently, the reduced observables and reduced density operators are represented by finite sequences of operators.
The mean-value functionals (35) define a duality between reduced observables and the reduced density operators. For F n 0 ∈ L 1 ( H ⊗ n ) F_n^0\in\mathcal{L}^1(\mathcal{H}^{\otimes n}) F n 0 ∈ L 1 ( H ⊗ n ) , and B n 0 ∈ L ( H ⊗ n ) B_n^0\in\mathcal{L}(\mathcal{H}^{\otimes n}) B n 0 ∈ L ( H ⊗ n ) , n ≥ 1 n\ge1 n ≥ 1 , they exist.
Hierarchy of evolution equations for reduced observables
As noted above, one of the approaches to describing the evolution of quantum many-particle systems is based on sequences B ( t ) = ( B 0 , B 1 ( t ) , … , B s ( t ) , … ) ∈ L γ ( F H ) B(t)=(B_0,B_1(t),\ldots,B_s(t),\ldots)\in\mathcal{L}_\gamma(\mathcal{F}_{\mathcal{H}}) B ( t ) = ( B 0 , B 1 ( t ) , … , B s ( t ) , … ) ∈ L γ ( F H ) of reduced observables (32) . By applying the cluster expansions of operator groups (2) :
G s ( t , ( 1 , … , s ) ∖ ( j 1 , … , j n ) , j 1 , … , j n ) = ∑ P : { ( 1 , … , s ) ∖ ( j 1 , … , j n ) , j 1 , … , j n } = ⋃ i X i X i ⊂ P ∏ X i ∈ P A ∣ X i ∣ ( t , X i ) , n ≥ 0. (37) \mathcal{G}_s\left(t,(1,\ldots,s)\setminus(j_1,\ldots,j_n),j_1,\ldots,j_n\right)=\sum_{\substack{\mathrm{P}:\{(1,\ldots,s)\setminus(j_1,\ldots,j_n),j_1,\ldots,j_n\}=\bigcup_i X_i\\X_i\subset\mathrm{P}}}\prod_{X_i\in\mathrm{P}}\mathfrak{A}_{|X_i|}(t,X_i),\qquad n\ge0. \tag*{(37)} G s ( t , ( 1 , … , s ) ∖ ( j 1 , … , j n ) , j 1 , … , j n ) = P : {( 1 , … , s ) ∖ ( j 1 , … , j n ) , j 1 , … , j n } = ⋃ i X i X i ⊂ P ∑ X i ∈ P ∏ A ∣ X i ∣ ( t , X i ) , n ≥ 0. (37)
to the operator groups in the definition of reduced observables (34) , and by considering the definition of reduced observables at the initial instant, we derive that the evolution of reduced observables for a quantum system with a non-fixed number of particles is described by the following sequence of operators:
B s ( t , 1 , … , s ) = ∑ n = 0 s 1 n ! ∑ j 1 ≠ ⋯ ≠ j n = 1 s A 1 + n ( t , { ( 1 , … , s ) ∖ ( j 1 , … , j n ) } , j 1 , … , j n ) B_s(t,1,\ldots,s)=\sum_{n=0}^{s}\frac{1}{n!}\sum_{j_1\ne\cdots\ne j_n=1}^{s}\mathfrak{A}_{1+n}\left(t,\{(1,\ldots,s)\setminus(j_1,\ldots,j_n)\},j_1,\ldots,j_n\right) B s ( t , 1 , … , s ) = n = 0 ∑ s n ! 1 j 1 = ⋯ = j n = 1 ∑ s A 1 + n ( t , {( 1 , … , s ) ∖ ( j 1 , … , j n )} , j 1 , … , j n )
× B s − n 0 ( ( 1 , … , s ) ∖ ( j 1 , … , j n ) ) ⊗ I ( j 1 ) ⋯ ⊗ I ( j n ) , s ≥ 1 , (38) \qquad{}\times B_{s-n}^{0}\left((1,\ldots,s)\setminus(j_1,\ldots,j_n)\right)\otimes\mathbb{I}_{(j_1)}\cdots\otimes\mathbb{I}_{(j_n)},\qquad s\ge1, \tag*{(38)} × B s − n 0 ( ( 1 , … , s ) ∖ ( j 1 , … , j n ) ) ⊗ I ( j 1 ) ⋯ ⊗ I ( j n ) , s ≥ 1 , (38)
where the generating operators are the cumulants of the corresponding order for the operator groups (2) associated with a cluster of s − n s-n s − n quantum particles and n n n individual particles:
A 1 + n ( t , { ( 1 , … , s ) ∖ ( j 1 , … , j n ) } , j 1 , … , j n ) ≔ ∑ P : { ( 1 , … , s ) ∖ ( j 1 , … , j n ) , j 1 , … , j n } = ⋃ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! ∏ X i ∈ P G ∣ θ ( X i ) ∣ ( t , θ ( X i ) ) , n ≥ 0 , (39) \mathfrak{A}_{1+n}\left(t,\{(1,\ldots,s)\setminus(j_1,\ldots,j_n)\},j_1,\ldots,j_n\right)\coloneqq\sum_{\substack{\mathrm{P}:\{(1,\ldots,s)\setminus(j_1,\ldots,j_n),j_1,\ldots,j_n\}=\bigcup_i X_i}}\left(-1\right)^{|\mathrm{P}|-1}(|\mathrm{P}|-1)!\prod_{X_i\in\mathrm{P}}\mathcal{G}_{|\theta(X_i)|}\left(t,\theta(X_i)\right),\qquad n\ge0, \tag*{(39)} A 1 + n ( t , {( 1 , … , s ) ∖ ( j 1 , … , j n )} , j 1 , … , j n ) : = P : {( 1 , … , s ) ∖ ( j 1 , … , j n ) , j 1 , … , j n } = ⋃ i X i ∑ ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 )! X i ∈ P ∏ G ∣ θ ( X i ) ∣ ( t , θ ( X i ) ) , n ≥ 0 , (39)
and the notation used here is similar to that in formula (13) . We give the simplest examples of expansions (38) for reduced observables:
B 1 ( t , 1 ) = A 1 ( t , 1 ) B 1 0 ( 1 ) , B_1(t,1)=\mathcal{A}_1(t,1)B_1^0(1), B 1 ( t , 1 ) = A 1 ( t , 1 ) B 1 0 ( 1 ) ,
B 2 ( t , 1 , 2 ) = A 1 ( t , { 1 , 2 } ) B 2 0 ( 1 , 2 ) + A 2 ( t , 1 , 2 ) ( B 1 0 ( 1 ) ⊗ 1 ( 2 ) + 1 ( 1 ) ⊗ B 1 0 ( 2 ) ) , B_2(t,1,2)=\mathcal{A}_1(t,\{1,2\})B_2^0(1,2)+\mathcal{A}_2(t,1,2)\left(B_1^0(1)\otimes\mathbb{1}_{(2)}+\mathbb{1}_{(1)}\otimes B_1^0(2)\right), B 2 ( t , 1 , 2 ) = A 1 ( t , { 1 , 2 }) B 2 0 ( 1 , 2 ) + A 2 ( t , 1 , 2 ) ( B 1 0 ( 1 ) ⊗ 1 ( 2 ) + 1 ( 1 ) ⊗ B 1 0 ( 2 ) ) ,
It should also be noted that observables of many-particle quantum systems have a certain structure. For example, the k k k -ary type observable has the structure O n ( k ) ( 1 , … , n ) = ∑ i 1 < … < i k = 1 n o k ( i 1 , … , i k ) O_n^{(k)}(1,\ldots,n)=\sum_{i_1<\ldots<i_k=1}^{n}o_k(i_1,\ldots,i_k) O n ( k ) ( 1 , … , n ) = ∑ i 1 < … < i k = 1 n o k ( i 1 , … , i k ) , n ≥ 1 n\geq1 n ≥ 1 . Therefore, the one-component sequences of reduced observables B ( 1 ) ( 0 ) = ( 0 , o 1 ( 1 ) , 0 , … ) B^{(1)}(0)=(0,o_1(1),0,\ldots) B ( 1 ) ( 0 ) = ( 0 , o 1 ( 1 ) , 0 , … ) correspond to additive-type observables [37 ] . If the initial reduced observable is specified by a sequence of additive-type operators B ( 1 ) ( 0 ) = ( 0 , b 1 , 0 , … ) B^{(1)}(0)=(0,b_1,0,\ldots) B ( 1 ) ( 0 ) = ( 0 , b 1 , 0 , … ) , then the structure of expansions (38) takes the form:
B s ( 1 ) ( t , 1 , … , s ) = A s ( t , 1 , … , s ) ∑ j = 1 s 1 ( 1 ) ⊗ … ⊗ 1 ( j − 1 ) ⊗ b 1 ( j ) ⊗ 1 ( j + 1 ) ⊗ … ⊗ 1 ( s ) , s ≥ 1 , B_s^{(1)}(t,1,\ldots,s)=\mathcal{A}_s(t,1,\ldots,s)\sum_{j=1}^{s}\mathbb{1}_{(1)}\otimes\ldots\otimes\mathbb{1}_{(j-1)}\otimes b_1(j)\otimes\mathbb{1}_{(j+1)}\otimes\ldots\otimes\mathbb{1}_{(s)},\qquad s\geq1, B s ( 1 ) ( t , 1 , … , s ) = A s ( t , 1 , … , s ) j = 1 ∑ s 1 ( 1 ) ⊗ … ⊗ 1 ( j − 1 ) ⊗ b 1 ( j ) ⊗ 1 ( j + 1 ) ⊗ … ⊗ 1 ( s ) , s ≥ 1 ,
whose generating operator A s ( t ) \mathcal{A}_s(t) A s ( t ) is the s s s th-order cumulant of operator groups (2.2) associated with s s s individual particles. In the case of a sequence of initial reduced observables of k k k -ary type, k ≥ 2 k\geq2 k ≥ 2 , operators B ( k ) ( 0 ) = ( 0 , … , 0 , b k , 0 , … ) B^{(k)}(0)=(0,\ldots,0,b_k,0,\ldots) B ( k ) ( 0 ) = ( 0 , … , 0 , b k , 0 , … ) , the structure of expansions (38) takes the form:
B s ( k ) ( t ) = 0 , 1 ≤ s < k , B_s^{(k)}(t)=0,\qquad1\leq s<k, B s ( k ) ( t ) = 0 , 1 ≤ s < k ,
B s ( k ) ( t , 1 , … , s ) = 1 ( s − k ) ! ∑ j 1 ≠ … ≠ j s − k = 1 s A 1 + s − k ( t , { ( 1 , … , s ) ∖ ( j 1 , … , j s − k ) } ) , B_s^{(k)}(t,1,\ldots,s)=\frac{1}{(s-k)!}\sum_{j_1\ne\ldots\ne j_{s-k}=1}^{s}\mathcal{A}_{1+s-k}\left(t,\{(1,\ldots,s)\setminus(j_1,\ldots,j_{s-k})\}\right), B s ( k ) ( t , 1 , … , s ) = ( s − k )! 1 j 1 = … = j s − k = 1 ∑ s A 1 + s − k ( t , {( 1 , … , s ) ∖ ( j 1 , … , j s − k )} ) ,
j 1 , … , j s − k b k ( ( 1 , … , s ) ∖ ( j 1 , … , j s − k ) ) ⊗ 1 ( j 1 ) ⊗ … ⊗ 1 ( j s − k ) , s ≥ k , j_1,\ldots,j_{s-k}\,b_k\left((1,\ldots,s)\setminus(j_1,\ldots,j_{s-k})\right)\otimes\mathbb{1}_{(j_1)}\otimes\ldots\otimes\mathbb{1}_{(j_{s-k})},\qquad s\geq k, j 1 , … , j s − k b k ( ( 1 , … , s ) ∖ ( j 1 , … , j s − k ) ) ⊗ 1 ( j 1 ) ⊗ … ⊗ 1 ( j s − k ) , s ≥ k ,
where the generating operator A 1 + s − k ( t ) \mathcal{A}_{1+s-k}(t) A 1 + s − k ( t ) is the ( 1 + s − k ) (1+s-k) ( 1 + s − k ) th-order cumulant of operator groups (2.2). Provided that γ < e − 1 \gamma<e^{-1} γ < e − 1 , the following inequality holds for the sequence of reduced observables (38) in the space L γ ( F H ) L_\gamma(\mathcal{F}_{\mathcal{H}}) L γ ( F H ) :
∥ B ( t ) ∥ L γ ( F H ) ≤ e 2 ( 1 − γ e ) − 1 ∥ B ( 0 ) ∥ L γ ( F H ) . \lVert B(t)\rVert_{L_\gamma(\mathcal{F}_{\mathcal{H}})}\leq e^2(1-\gamma e)^{-1}\lVert B(0)\rVert_{L_\gamma(\mathcal{F}_{\mathcal{H}})}. ∥ B ( t ) ∥ L γ ( F H ) ≤ e 2 ( 1 − γ e ) − 1 ∥ B ( 0 ) ∥ L γ ( F H ) .
The evolution of observables of non-fixed, but finite average number of quantum particles, i.e., the sequence of reduced observables (38) , is governed by the Cauchy problem to the following hierarchy of evolution equations [37 ] :
d d t B ( t ) = N B ( t ) + [ N , a + ] B ( t ) , (40) \frac{d}{dt}B(t)=\mathcal{N}B(t)+[\mathcal{N},a^+]B(t), \tag*{(40)} d t d B ( t ) = N B ( t ) + [ N , a + ] B ( t ) , (40)
B ( t ) ∣ t = 0 = B ( 0 ) , (41) B(t)\big|_{t=0}=B(0), \tag*{(41)} B ( t ) t = 0 = B ( 0 ) , (41)
where the generator of this hierarchy of evolution equations is the Heisenberg generator (2.3) together with the commutator of this generator and the operator (2.29), which component-wise has the form
( [ N , a + ] B ( t ) ) s ( 1 , … , s ) = ∑ j 1 ≠ j 2 = 1 s N i n t ( j 1 , j 2 ) B s − 1 ( t , ( 1 , … , s ) ∖ j 1 ) ⊗ 1 ( j 1 ) , s ≥ 2. \left([\mathcal{N},a^+]B(t)\right)_s(1,\ldots,s)=\sum_{j_1\ne j_2=1}^{s}\mathcal{N}_{\mathrm{int}}(j_1,j_2)B_{s-1}\left(t,(1,\ldots,s)\setminus j_1\right)\otimes\mathbb{1}_{(j_1)},\qquad s\geq2. ( [ N , a + ] B ( t ) ) s ( 1 , … , s ) = j 1 = j 2 = 1 ∑ s N int ( j 1 , j 2 ) B s − 1 ( t , ( 1 , … , s ) ∖ j 1 ) ⊗ 1 ( j 1 ) , s ≥ 2.
In component-wise form, the hierarchy of evolution equations (40) for reduced observables constitutes, in fact, a system of recurrence evolution equations. Let us present examples of typical equations from this hierarchy:
∂ ∂ t B 1 ( t , 1 ) = N 1 ( 1 ) B 1 ( t , 1 ) , \frac{\partial}{\partial t}B_1(t,1)=\mathcal{N}_1(1)B_1(t,1), ∂ t ∂ B 1 ( t , 1 ) = N 1 ( 1 ) B 1 ( t , 1 ) ,
∂ ∂ t B 2 ( t , 1 , 2 ) = ( ∑ j = 1 2 N ( j ) + N i n t ( 1 , 2 ) ) B 2 ( t , 1 , 2 ) + N i n t ( 1 , 2 ) ( B 1 ( t , 1 ) ⊗ 1 ( 2 ) + 1 ( 1 ) ⊗ B 1 ( t , 2 ) ) , \frac{\partial}{\partial t}B_2(t,1,2)=\left(\sum_{j=1}^{2}\mathcal{N}(j)+\mathcal{N}_{\mathrm{int}}(1,2)\right)B_2(t,1,2)+\mathcal{N}_{\mathrm{int}}(1,2)\left(B_1(t,1)\otimes1_{(2)}+1_{(1)}\otimes B_1(t,2)\right), ∂ t ∂ B 2 ( t , 1 , 2 ) = ( j = 1 ∑ 2 N ( j ) + N int ( 1 , 2 ) ) B 2 ( t , 1 , 2 ) + N int ( 1 , 2 ) ( B 1 ( t , 1 ) ⊗ 1 ( 2 ) + 1 ( 1 ) ⊗ B 1 ( t , 2 ) ) ,
Note that the structure of the generator of the hierarchy of evolution equations (40) for quantum particle systems with an arbitrary n n n -body interaction potential is given by: e − a + N e a + e^{-\mathfrak{a}^{+}}\mathcal{N}e^{\mathfrak{a}^{+}} e − a + N e a + . For a two-body interaction potential, the following equality holds [21 ] : e − a + N e a + = N + + [ N , a + ] e^{-\mathfrak{a}^{+}}\mathcal{N}e^{\mathfrak{a}^{+}}=\mathcal{N}_{+}+\left[\mathcal{N},\mathfrak{a}^{+}\right] e − a + N e a + = N + + [ N , a + ] .
In the space L γ ( F H ) \mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}}) L γ ( F H ) the following existence theorem holds [32 ] .
Theorem 4. For bounded interaction potentials, a non-perturbative solution of the Cauchy problem (40) ,(41) is represented by expansions (38) , in which the generating operators are the cumulants of the operator groups (2) of the corresponding order. Provided that γ < e − 1 \gamma< e^{-1} γ < e − 1 , for initial data B ( 0 ) ∈ L 0 ( F H ) B(0) \in\mathcal{L}_0(\mathcal{F}_{\mathcal{H}}) B ( 0 ) ∈ L 0 ( F H ) , the sequence of reduced observables (38) is the unique globalglobal-in-time classical solution, whereas for arbitrary initial data B ( 0 ) ∈ L γ ( F H ) B(0) \in\mathcal{L}_{\gamma}(\mathcal{F}_{\mathcal{H}}) B ( 0 ) ∈ L γ ( F H ) , it is the unique global-in-time generalized solution.
The proof of this statement is analogous to that of the existence theorem for the Heisenberg equations (4) in the space of sequences of bounded operators. A similar result also holds for sequences of unbounded operators [35 ] .
Thus, the cumulant background of the cluster expansions (37) for operator groups of a non-perturbative solution to the Heisenberg equations (4) determines a cumulant structure of the generating operators of expansions (38) for the non-perturbative solution of the hierarchy of evolution equations (40) for reduced observables. It should be emphasized that the uniqueness of the cluster expansion structure is determined by the symmetry of the infinitesimal generator (3) of the group of operators (2) .
Evolution of states of an infinite number of quantum particles
As noted above, one more approach to describing the evolution of quantum many-particle systems is based on sequences of reduced density operators (33) . By applying the cluster expansions (12) of operator groups (9) :
G s + n ∗ ( t , 1 , … , s , s + 1 , … , s + n ) = ∑ P : { 1 , … , s } , s + 1 , … , s + n = ⋃ i X i X i ⊂ P ∏ X i A ∣ X i ∣ ( t , X i ) , n ≥ 0. (42) \mathcal{G}^{*}_{s+n}(t,1,\ldots,s,s+1,\ldots,s+n)=\sum_{\substack{\mathrm{P}: \{1,\ldots,s\},s+1,\ldots,s+n=\bigcup_i X_i\\X_i\subset\mathrm{P}}}\prod_{X_i}\mathfrak{A}_{|X_i|}(t,X_i),\qquad n\geq0. \tag*{(42)} G s + n ∗ ( t , 1 , … , s , s + 1 , … , s + n ) = P : { 1 , … , s } , s + 1 , … , s + n = ⋃ i X i X i ⊂ P ∑ X i ∏ A ∣ X i ∣ ( t , X i ) , n ≥ 0. (42)
to the operator groups in the definition of reduced density operators (36) , and by considering the definition of reduced density operators at the initial instant, we derive that the evolution of the reduced density operators for a quantum system with a non-fixed number of particles is described by the following sequence of operators:
F s ( t , 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) F s + n 0 ( 1 , … , s + n ) , s ≥ 1 , (43) F_s(t,1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\mathcal{A}_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)F^0_{s+n}(1,\ldots,s+n),\quad s\geq1, \tag*{(43)} F s ( t , 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) F s + n 0 ( 1 , … , s + n ) , s ≥ 1 , (43)
where the generating operators are the cumulants of the corresponding order for the operator groups (2.9) associated with a cluster of s s s particles and n n n individual particles
A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) ≔ ∑ P : ( { 1 , … , s } , s + 1 , … , s + n ) = ⋃ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! ∏ X i ⊂ P g ∣ θ ( X i ) ∣ ∗ ( t , θ ( X i ) ) , (44) \mathcal{A}_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)\coloneqq
\sum_{\substack{\mathrm{P}:\,(\{1,\ldots,s\},s+1,\ldots,s+n)=\bigcup_i X_i}}
(-1)^{|\mathrm{P}|-1}(|\mathrm{P}|-1)!\prod_{X_i\subset\mathrm{P}}g^*_{|\theta(X_i)|}(t,\theta(X_i)), \tag*{(44)} A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) : = P : ({ 1 , … , s } , s + 1 , … , s + n ) = ⋃ i X i ∑ ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 )! X i ⊂ P ∏ g ∣ θ ( X i ) ∣ ∗ ( t , θ ( X i )) , (44)
and the notation used here is similar to that in formula (2.11).
Provided that α > e \alpha>e α > e , by estimate (2.15), the series (2.43) converges in the norm of the space L α 1 ( F H ) L^1_\alpha(\mathcal{F}_{\mathcal{H}}) L α 1 ( F H ) and the following inequality holds:
∥ F ( t ) ∥ L α 1 ( F H ) ≤ e 2 ( 1 − e α ) − 1 ∥ F ( 0 ) ∥ L α 1 ( F H ) . \left\|F(t)\right\|_{L^1_\alpha(\mathcal{F}_{\mathcal{H}})}
\leq e^2\left(1-\frac{e}{\alpha}\right)^{-1}\left\|F(0)\right\|_{L^1_\alpha(\mathcal{F}_{\mathcal{H}})}. ∥ F ( t ) ∥ L α 1 ( F H ) ≤ e 2 ( 1 − α e ) − 1 ∥ F ( 0 ) ∥ L α 1 ( F H ) .
The parameter α \alpha α can be interpreted as the inverse of the average number of particles.
Based on an alternative approach to describing state evolution by the von Neumann hierarchy (2.27), the reduced density operators can also be defined through of correlation operators (2.22). The definition of the reduced density operators (2.36) represented in terms of the correlation operators (2.22) by the following series expansion [3 ] :
F s ( t , 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n g 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) , s ≥ 1 , (45) F_s(t,1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}g_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n),\quad s\geq1, \tag*{(45)} F s ( t , 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n g 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) , s ≥ 1 , (45)
where the correlation operators g 1 + n ( t ) g_{1+n}(t) g 1 + n ( t ) , n ≥ 0 n\geq0 n ≥ 0 , of a cluster of s s s particles and n n n individual particles are determined similar to expansions (2.22), namely
g 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) = ∑ P : ( { 1 , … , s } , s + 1 , … , s + n ) = ⋃ j X j A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ } ) ⨂ X j ⊂ P g ∣ X j ∣ 0 ( X j ) , n ≥ 0 , (46) g_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)=
\sum_{\substack{\mathrm{P}:\,(\{1,\ldots,s\},s+1,\ldots,s+n)=\bigcup_j X_j}}
\mathcal{A}_{|\mathrm{P}|}(t,\{X_1\},\ldots,\{X_{|\mathrm{P}|}\})
\bigotimes_{X_j\subset\mathrm{P}}g^0_{|X_j|}(X_j),\quad n\geq0, \tag*{(46)} g 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) = P : ({ 1 , … , s } , s + 1 , … , s + n ) = ⋃ j X j ∑ A ∣ P ∣ ( t , { X 1 } , … , { X ∣ P ∣ }) X j ⊂ P ⨂ g ∣ X j ∣ 0 ( X j ) , n ≥ 0 , (46)
and the notation adopted in formula (2.11) is used. The redefinition of reduced density operators (2.45) naturally arises from their definition within the grand canonical ensemble (2.36) when explicitly taking into account the normalization factor [3 ] .
Traditionally, the evolution of many-particle systems is described within the framework of the evolution of the state governed by the BBGKY hierarchy for reduced density operators [6 ] , [21 ] , [23 ] .
The cluster expansions (2.12) of operator groups (2.9) underlie the classification of possible solution representations [32 ] of the Cauchy problem of the BBGKY hierarchy for quantum many-particle system. In particular, the non-perturbative solution (2.43) of the BBGKY hierarchy for many quantum particles can be represented in the form of the perturbation (iteration) series as a result of applying analogs of the Duhamel equation to cumulants (2.11) of operator groups (2.9)
F ( t ) = ∑ n = 0 ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n G ∗ ( t − t 1 ) [ a , N ∗ ] G ∗ ( t 1 − t 2 ) … G ∗ ( t n − 1 − t n ) [ a , N ∗ ] G ∗ ( t n ) F ( 0 ) . F(t)=\sum_{n=0}^{\infty}\int_{0}^{t}dt_{1}\ldots\int_{0}^{t_{n-1}}dt_{n}\mathcal{G}^{*}(t-t_{1})[a,\mathcal{N}^{*}]\mathcal{G}^{*}(t_{1}-t_{2})\ldots\mathcal{G}^{*}(t_{n-1}-t_{n})[a,\mathcal{N}^{*}]\mathcal{G}^{*}(t_{n})F(0). F ( t ) = n = 0 ∑ ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n G ∗ ( t − t 1 ) [ a , N ∗ ] G ∗ ( t 1 − t 2 ) … G ∗ ( t n − 1 − t n ) [ a , N ∗ ] G ∗ ( t n ) F ( 0 ) .
Conventionally, perturbation theory methods are used to study the problem of the rigorous derivation of quantum kinetic equations in the scaling limits of the solution of the BBGKY hierarchy [21 ] , [23 ] .
The BBGKY hierarchy for reduced density operators
The evolution of all possible states of non-fixed but finite average number of quantum particles, i.e., the sequence of reduced density operators (2.43) is governed by the Cauchy problem of the BBGKY hierarchy [21 ] :
d d t F ( t ) = N ∗ F ( t ) + [ a , N ∗ ] F ( t ) , (47) \frac{d}{dt}F(t)=\mathcal{N}^{*}F(t)+[a,\mathcal{N}^{*}]F(t), \tag*{(47)} d t d F ( t ) = N ∗ F ( t ) + [ a , N ∗ ] F ( t ) , (47)
F ( t ) ∣ t = 0 = F ( 0 ) , (48) F(t)|_{t=0}=F(0), \tag*{(48)} F ( t ) ∣ t = 0 = F ( 0 ) , (48)
where the generator of the BBGKY hierarchy is the von Neumann generator (2.10) together with the commutator of this generator and the operator (2.30), which component-wise has the form
( [ a , N ∗ ] F ( t ) ) s ( 1 , … , s ) = ∑ j = 1 s Tr s + 1 N i n t ∗ ( j , s + 1 ) F s + 1 ( t , 1 , … , s + 1 ) , s ≥ 1. \left([a,\mathcal{N}^{*}]F(t)\right)_{s}(1,\ldots,s)=\sum_{j=1}^{s}\operatorname{Tr}_{s+1}\mathcal{N}_{\mathrm{int}}^{*}(j,s+1)F_{s+1}(t,1,\ldots,s+1),\qquad s\geq1. ( [ a , N ∗ ] F ( t ) ) s ( 1 , … , s ) = j = 1 ∑ s Tr s + 1 N int ∗ ( j , s + 1 ) F s + 1 ( t , 1 , … , s + 1 ) , s ≥ 1.
Note that the structure of the generator of the BBGKY hierarchy for quantum particle systems with an arbitrary n n n -body interaction potential is given by: e a N ∗ e − a e^{a}\mathcal{N}^{*}e^{-a} e a N ∗ e − a [21 ] . For the two-body interaction potential, the following equality holds: e a N ∗ e − a = N ∗ + [ a , N ∗ ] e^{a}\mathcal{N}^{*}e^{-a}=\mathcal{N}^{*}+[a,\mathcal{N}^{*}] e a N ∗ e − a = N ∗ + [ a , N ∗ ] .
In the space L α 1 ( F H ) L_{\alpha}^{1}(\mathcal{F}_{\mathcal{H}}) L α 1 ( F H ) the following statement holds [32 ] .
Theorem 5. For bounded interaction potentials, a non-perturbative solution of the Cauchy problem to the Cauchy problem of the BBGKY hierarchy (2.47),(2.48) is determined by the sequence of reduced density operators represented by series expansions (2.43). Provided that α > e \alpha>e α > e , for initial data F ( 0 ) ∈ L α 1 ( F H ) F(0)\in\mathcal{L}_{\alpha}^{1}(\mathcal{F}_{\mathcal{H}}) F ( 0 ) ∈ L α 1 ( F H ) , the sequence of reduced density operators (2.43) is the unique strong solution, and for arbitrary initial data F ( 0 ) ∈ L α 1 ( F H ) F(0)\in L_{\alpha}^{1}(\mathcal{F}_{\mathcal{H}}) F ( 0 ) ∈ L α 1 ( F H ) , it is a weak global-in-time solution.
It should be emphasized that the cumulant background of the cluster expansions (2.42) for operator groups of a non-perturbative solution to the von Neumann equations (2.16) determines a cumulant structure for the generating operators of the series expansions (2.43) for the non-perturbative solution of the BBGKY hierarchy (2.47). Based on an alternative approach to describing state evolution by the von Neumann hierarchy (2.27), the reduced density operators are determined by the correlation operators (2.45). Then the cumulant structure of the non-perturbative solution to the von Neumann hierarchy (2.22) induces the cumulant structure of the series expansion of the solution (2.43) to the Cauchy problem for the BBGKY hierarchy (2.47), which is adopted to describe systems with an infinite number of particles; in other words, the dynamics of quantum systems with an infinite number of particles is generated by the dynamics of correlations.
The hierarchy of nonlinear evolution equations for reduced correlation operators
It is well known that, on a microscopic scale, the macroscopic characteristics of fluctuations of observables are directly determined by the reduced correlation operators [21 ] .
Traditionally reduced correlation operators are introduced by the cluster expansions of the reduced density operators:
F s ( t , 1 , … , s ) = ∑ P : ( 1 , … , s ) = ⋃ i X i ⨂ X i ⊂ P G ∣ X i ∣ ( t , X i ) , s ≥ 1 , (49) F_s(t,1,\ldots,s)=\sum_{P:(1,\ldots,s)=\bigcup_i X_i}\bigotimes_{X_i\subset P}G_{\lvert X_i\rvert}(t,X_i),\qquad s\geq1, \tag*{(49)} F s ( t , 1 , … , s ) = P : ( 1 , … , s ) = ⋃ i X i ∑ X i ⊂ P ⨂ G ∣ X i ∣ ( t , X i ) , s ≥ 1 , (49)
where, as above, the symbol ∑ P : ( 1 , … , s ) = ⋃ i X i \sum_{P:(1,\ldots,s)=\bigcup_i X_i} ∑ P : ( 1 , … , s ) = ⋃ i X i is the sum over all possible partitions P P P of the set ( 1 , … , s ) (1,\ldots,s) ( 1 , … , s ) into ∣ P ∣ \lvert P\rvert ∣ P ∣ nonempty mutually disjoint subsets X i ⊂ ( 1 , … , s ) X_i\subset(1,\ldots,s) X i ⊂ ( 1 , … , s ) . As a consequence of this, the solutions of recurrence relations (49) are represented through reduced density operators as follows:
G s ( t , 1 , … , s ) = ∑ P : ( 1 , … , s ) = ⋃ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! ⨂ X i ⊂ P F ∣ X i ∣ ( t , X i ) , s ≥ 1. (50) G_s(t,1,\ldots,s)=\sum_{P:(1,\ldots,s)=\bigcup_i X_i}(-1)^{\lvert P\rvert-1}(\lvert P\rvert-1)!\bigotimes_{X_i\subset P}F_{\lvert X_i\rvert}(t,X_i),\qquad s\geq1. \tag*{(50)} G s ( t , 1 , … , s ) = P : ( 1 , … , s ) = ⋃ i X i ∑ ( − 1 ) ∣ P ∣ − 1 (∣ P ∣ − 1 )! X i ⊂ P ⨂ F ∣ X i ∣ ( t , X i ) , s ≥ 1. (50)
The structure of this expansions (50) is such that the reduced correlation operators are cumulants (semi-invariants) of the reduced density operators (43) . Consequently, these operators (50) are interpreted as correlations between the states of individual particles.
Assuming as a basis an alternative approach to the description of the evolution of the state within the framework of correlation operators (27) , then the reduced correlation operators (50) are defined through of correlation operators [21 ] :
G s ( t , 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n g s + n ( t , 1 , … , s + n ) , s ≥ 1 , (51) G_s(t,1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}g_{s+n}(t,1,\ldots,s+n),\qquad s\geq1, \tag*{(51)} G s ( t , 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n g s + n ( t , 1 , … , s + n ) , s ≥ 1 , (51)
where the generating operators g s + n ( t , 1 , … , s + n ) g_{s+n}(t,1,\ldots,s+n) g s + n ( t , 1 , … , s + n ) , n ≥ 0 n\geq0 n ≥ 0 , are determined by expansions (22) for correlation operators of s + n s+n s + n individual particles. In particular, for the reduced correlation operators (51) with a separable initial state, the following equality holds:
G s 0 ( 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n g 1 0 δ s + n , 1 = g 1 0 ( 1 ) δ s , 1 , s ≥ 1 , G_s^0(1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}g_1^0\delta_{s+n,1}=g_1^0(1)\delta_{s,1},\qquad s\geq1, G s 0 ( 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n g 1 0 δ s + n , 1 = g 1 0 ( 1 ) δ s , 1 , s ≥ 1 ,
where δ s , 1 \delta_{s,1} δ s , 1 is a Kronecker symbol.
Considering that the evolution of correlations for a separable initial state is governed by formula (23) , namely
G s ( t , 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n A ^ s + n ( t , 1 , … , s + n ) ⨂ j = 1 s + n G 1 0 ( j ) , G_s(t,1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\widehat{\mathcal{A}}_{s+n}(t,1,\ldots,s+n)\bigotimes_{j=1}^{s+n}G_1^0(j), G s ( t , 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n A s + n ( t , 1 , … , s + n ) j = 1 ⨂ s + n G 1 0 ( j ) ,
where A ^ n ( t ) \widehat{\mathcal{A}}_n(t) A n ( t ) is the cumulant of the groups of scattering operators (24) , we derive the law of creation correlations from chaos in a system with either infinitely or finitely many quantum particles
G s ( t , 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n A s + n ( t , 1 , … , s + n ) ⨂ j = 1 s + n G 1 0 ( j ) . (52) G_s(t,1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\mathcal{A}_{s+n}(t,1,\ldots,s+n)\bigotimes_{j=1}^{s+n}G_1^0(j). \tag*{(52)} G s ( t , 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n A s + n ( t , 1 , … , s + n ) j = 1 ⨂ s + n G 1 0 ( j ) . (52)
where the generating operator A s + n ( t ) \mathcal{A}_{s+n}(t) A s + n ( t ) of series (2.52) is the ( s + n ) (s+n) ( s + n ) th-order cumulant (2.11) of operator groups (2.9).
The evolution of all possible states of a quantum system, as an equivalent method compared with the reduced density operators, is described by means of the sequence G ( t ) = ( I , G 1 ( t ) , … , G s ( t ) , … ) ∈ L 1 ( F H ) G(t)=(I,G_1(t),\ldots,G_s(t),\ldots)\in\mathcal{L}^{1}(\mathcal{F}_{\mathcal{H}}) G ( t ) = ( I , G 1 ( t ) , … , G s ( t ) , … ) ∈ L 1 ( F H ) of the reduced correlation operators G s ( t ) = G s ( t , 1 , … , s ) G_s(t)=G_s(t,1,\ldots,s) G s ( t ) = G s ( t , 1 , … , s ) , s ≥ 1 s\ge1 s ≥ 1 , governed by the Cauchy problem of the following hierarchy of nonlinear evolution equations [21 ] :
d d t G ( t ) = N ( ⋅ ∣ G ( t ) ) + α N ( ⋅ ∣ G ( t ) ) − N ( ⋅ ∣ α G ( t ) ) , (53) \frac{d}{dt}G(t)=\mathcal{N}(\cdot\mid G(t))+\alpha\mathcal{N}(\cdot\mid G(t))-\mathcal{N}(\cdot\mid\alpha G(t)), \tag*{(53)} d t d G ( t ) = N ( ⋅ ∣ G ( t )) + α N ( ⋅ ∣ G ( t )) − N ( ⋅ ∣ α G ( t )) , (53)
G ( t ) ∣ t = 0 = G ( 0 ) , (54) G(t)\big|_{t=0}=G(0), \tag*{(54)} G ( t ) t = 0 = G ( 0 ) , (54)
where the generator of this hierarchy of nonlinear evolution equations is the generator of the von Neuman hierarchy (26) together with the commutator of this generator and the operator (30) , which component-wise has the form
( α N ( ⋅ ∣ G ( t ) ) ) s ( 1 , … , s ) − N ( 1 , … , s ∣ α G ( t ) ) = Tr s + 1 ∑ j = 1 s N i n t ∗ ( j , s + 1 ) ( G s + 1 ( t , 1 , … , s + 1 ) + ∑ P : ( 1 , … , s + 1 ) = X 1 ∪ X 2 , j ∈ X 1 ; s + 1 ∈ X 2 G ∣ X 1 ∣ ( t , X 1 ) ⊗ G ∣ X 2 ∣ ( t , X 2 ) ) . \begin{aligned}
(\alpha\mathcal{N}(\cdot\mid G(t)))_s(1,\ldots,s)-\mathcal{N}(1,\ldots,s\mid\alpha G(t))
={}&\operatorname{Tr}_{s+1}\sum_{j=1}^{s}\mathcal{N}_{\mathrm{int}}^{*}(j,s+1)(G_{s+1}(t,1,\ldots,s+1)\\
&+\sum_{\substack{\mathrm{P}:(1,\ldots,s+1)=X_1\cup X_2,\\j\in X_1;\,s+1\in X_2}}
G_{\lvert X_1\rvert}(t,X_1)\otimes G_{\lvert X_2\rvert}(t,X_2)).
\end{aligned} ( α N ( ⋅ ∣ G ( t )) ) s ( 1 , … , s ) − N ( 1 , … , s ∣ α G ( t )) = Tr s + 1 j = 1 ∑ s N int ∗ ( j , s + 1 ) ( G s + 1 ( t , 1 , … , s + 1 ) + P : ( 1 , … , s + 1 ) = X 1 ∪ X 2 , j ∈ X 1 ; s + 1 ∈ X 2 ∑ G ∣ X 1 ∣ ( t , X 1 ) ⊗ G ∣ X 2 ∣ ( t , X 2 )) .
where the notations is similar to the formula (26) , and the symbol ∑ P : ( 1 , … , s + 1 ) = X 1 ∪ X 2 , j ∈ X 1 ; s + 1 ∈ X 2 \sum_{\substack{\mathrm{P}:(1,\ldots,s+1)=X_1\cup X_2,\\j\in X_1;\,s+1\in X_2}} ∑ P : ( 1 , … , s + 1 ) = X 1 ∪ X 2 , j ∈ X 1 ; s + 1 ∈ X 2 means the sum over all possible partitions of the set ( 1 , … , s + 1 ) (1,\ldots,s+1) ( 1 , … , s + 1 ) into two mutually disjoint subsets X 1 X_1 X 1 and X 2 X_2 X 2 such that j j j particle belongs to the subset X 1 X_1 X 1 and s + 1 s+1 s + 1 particle belongs to X 2 X_2 X 2 .
The statement below is true [38 ] .
Theorem 6. For quantum many-particle systems with bounded interaction potentials and separable initial data G ( 0 ) = ( I , G 1 0 , 0 , … , 0 , … ) G(0)=(I,G_1^0,0,\ldots,0,\ldots) G ( 0 ) = ( I , G 1 0 , 0 , … , 0 , … ) , the non-perturbative solution to the Cauchy problem for the hierarchy of nonlinear evolution equations (53) , (54) is determined by a sequence of reduced correlation operators represented by the expansions (52) . Provided that α > e \alpha>e α > e , for initial data G ( 0 ) ∈ L α 1 ( F H ) G(0)\in\mathcal{L}^{1}_{\alpha}(\mathcal{F}_{\mathcal{H}}) G ( 0 ) ∈ L α 1 ( F H ) , the sequence of reduced correlation operators (52) is the unique global-in-time strong solution, and for arbitrary initial data G ( 0 ) ∈ L α 1 ( F H ) G(0)\in\mathcal{L}^{1}_{\alpha}(\mathcal{F}_{\mathcal{H}}) G ( 0 ) ∈ L α 1 ( F H ) , it is a weak global-in-time solution.
The outline of the proof for this assertion is analogous to the proof of the existence theorem for the von Neumann hierarchy (27) in the space of sequences of trace-class operators.
Mathematical origins of quantum kinetic equations
One of the challenges of kinetic theory is to describe the evolution of a system compose of many particles within the framework of a typical particle’s state, which is governed by a kinetic equation.
In circumstances when the initial state is specified by a one-particle density operator, the mean-value functional of observables at any given moment can be represented by a one-particle density operator governed by a non-Markovian nonlinear evolution equation. In other words, for such initial data, the Cauchy problem for the BBGKY hierarchy is equivalent to the Cauchy problem for the nonlinear kinetic equation, and a sequence of reduced correlation functionals of a state is determined explicitly by the solution of this kinetic equation. The generating operators for expansions of the reduced correlation functionals of a state are determined by a solution of the kinetic cluster expansions (analogs of equilibrium virial expansions) of cumulants of operator groups of the von Neumann equations.
Approaches to deriving quantum kinetic equations
Typically, when describing the evolution of many-particle quantum systems by kinetic equations, it is assumed that the initial state is separable
F ( c ) ( 0 ) ≡ ( I , F 1 0 ( 1 ) , … , ⨂ i = 1 n F 1 0 ( i ) , … ) , (55) F^{(c)}(0) \equiv\left(I,F_1^0(1),\ldots,\bigotimes_{i=1}^{n}F_1^0(i),\ldots\right), \tag*{(55)} F ( c ) ( 0 ) ≡ ( I , F 1 0 ( 1 ) , … , i = 1 ⨂ n F 1 0 ( i ) , … ) , (55)
which means the absence of correlations between particles, or, in other words, a state satisfying the chaos condition [7 , 21 ] . We highlight that the assumption about the initial state is a crucial hypothesis in current rigorous kinetic theory.
It should be observed that, for an initial state specified by the one-particle density operator (55) , the Cauchy problem for the BBGKY hierarchy is not well-posed. Consequently, this Cauchy problem can be reformulated as a new Cauchy problem for the one-particle density operator and a sequence of explicitly defined functionals of the solution to the nonlinear evolution equation for the one-particle density operator, known as the kinetic equation. The method for constructing such state functionals relies on the application of kinetic cluster expansions [39 ] .
The approach to describing the evolution of the state for separable initial states of quantum many-body systems is based on sequences of reduced state functionals.
Indeed, we apply to the generating operators of the non-perturbative solution F ( t ) ≡ ( I , F 1 ( t ) , F 2 ( t ) , … , F s ( t ) , … ) F(t) \equiv(I,F_1(t),F_2(t),\ldots,F_s(t),\ldots) F ( t ) ≡ ( I , F 1 ( t ) , F 2 ( t ) , … , F s ( t ) , … ) of the BBGKY hierarchy (47)
F s ( t , 1 , … , s ) = ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) ⨂ i = 1 s + n F 1 0 ( i ) , (56) F_s(t,1,\ldots,s)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\mathcal{A}_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)\bigotimes_{i=1}^{s+n}F_1^0(i), \tag*{(56)} F s ( t , 1 , … , s ) = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) i = 1 ⨂ s + n F 1 0 ( i ) , (56)
s ≥ 1 s \ge1 s ≥ 1 ,
the following cluster expansions, referred to as kinetic cluster expansions [39 ] :
A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) = ∑ n 1 = 0 n n ! ( n − n 1 ) ! V 1 + n − n 1 ( t , { 1 , … , s } , s + 1 , … , s + n − n 1 ) ∑ D : ( s + n − n 1 + 1 , … , s + n ) = ⋃ k X k , ∣ D ∣ ≤ s + n − n 1 1 ∣ D ∣ ! ∑ i 1 ≠ ⋯ ≠ i ∣ D ∣ = 1 s + n − n 1 ∏ X k ⊂ D 1 ∣ X k ∣ ! A 1 + ∣ X k ∣ ( t , i k , X k ) ∏ m = 1 m ≠ i 1 , … , i ∣ D ∣ s + n − n 1 A 1 ( t , m ) , n ≥ 0 , (57) \mathcal{A}_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)=\sum_{n_1=0}^{n}\frac{n!}{(n-n_1)!}\mathcal{V}_{1+n-n_1}(t,\{1,\ldots,s\},s+1,\ldots,s+n-n_1)
\sum_{\substack{D:(s+n-n_1+1,\ldots,s+n)=\bigcup_{k}X_k,\\ |D|\le s+n-n_1}}\frac{1}{|D|!}
\sum_{\substack{i_1\ne\cdots\ne i_{|D|}=1}}^{s+n-n_1}
\prod_{X_k\subset D}\frac{1}{|X_k|!}\mathcal{A}_{1+|X_k|}(t,i_k,X_k)
\prod_{\substack{m=1\\m\ne i_1,\ldots,i_{|D|}}}^{s+n-n_1}\mathcal{A}_1(t,m),\qquad n\ge0, \tag*{(57)} A 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) = n 1 = 0 ∑ n ( n − n 1 )! n ! V 1 + n − n 1 ( t , { 1 , … , s } , s + 1 , … , s + n − n 1 ) D : ( s + n − n 1 + 1 , … , s + n ) = ⋃ k X k , ∣ D ∣ ≤ s + n − n 1 ∑ ∣ D ∣ ! 1 i 1 = ⋯ = i ∣ D ∣ = 1 ∑ s + n − n 1 X k ⊂ D ∏ ∣ X k ∣ ! 1 A 1 + ∣ X k ∣ ( t , i k , X k ) m = 1 m = i 1 , … , i ∣ D ∣ ∏ s + n − n 1 A 1 ( t , m ) , n ≥ 0 , (57)
where the symbol ∑ D : ( s + n − n 1 + 1 , … , s + n ) = ⋃ k X k , ∣ D ∣ ≤ s + n − n 1 \sum_{\substack{D:(s+n-n_1+1,\ldots,s+n)=\bigcup_k X_k,\\ |D|\le s+n-n_1}} ∑ D : ( s + n − n 1 + 1 , … , s + n ) = ⋃ k X k , ∣ D ∣ ≤ s + n − n 1 means the sum over all possible dissections D D D of the linearly ordered set ( s + n − n 1 + 1 , … , s + n ) (s+n-n_1+1,\ldots,s+n) ( s + n − n 1 + 1 , … , s + n ) on no more than s + n − n 1 s+n-n_1 s + n − n 1 linearly ordered subsets. The first few of the reformulated cluster expansions (3.3) take the form
A ^ 1 ( t , { 1 , … , s } ) = V 1 ( t , { 1 , … , s } ) , \widehat{\mathfrak A}_1(t,\{1,\ldots,s\})=\mathfrak V_1(t,\{1,\ldots,s\}), A 1 ( t , { 1 , … , s }) = V 1 ( t , { 1 , … , s }) ,
A ^ 2 ( t , { 1 , … , s } , s + 1 ) = V 2 ( t , { 1 , … , s } , s + 1 ) + V 1 ( t , { 1 , … , s } ) ∑ i 1 = 1 s A ^ 2 ( t , i 1 , s + 1 ) , \widehat{\mathfrak A}_2(t,\{1,\ldots,s\},s+1)=\mathfrak V_2(t,\{1,\ldots,s\},s+1)+\mathfrak V_1(t,\{1,\ldots,s\})\sum_{i_1=1}^{s}\widehat{\mathfrak A}_2(t,i_1,s+1), A 2 ( t , { 1 , … , s } , s + 1 ) = V 2 ( t , { 1 , … , s } , s + 1 ) + V 1 ( t , { 1 , … , s }) i 1 = 1 ∑ s A 2 ( t , i 1 , s + 1 ) ,
As a result, we establish that a sequence of reduced state functionals F ( t ∣ F 1 ( t ) ) ≡ ( I , F 1 ( t ) , F 2 ( t ∣ F 1 ( t ) ) , … , F s ( t ∣ F 1 ( t ) ) , … ) F(t\mid F_1(t))\equiv(I,F_1(t),F_2(t\mid F_1(t)),\ldots,F_s(t\mid F_1(t)),\ldots) F ( t ∣ F 1 ( t )) ≡ ( I , F 1 ( t ) , F 2 ( t ∣ F 1 ( t )) , … , F s ( t ∣ F 1 ( t )) , … ) is represented by following expansions into series:
F 1 ( t , 1 ) = ∑ n = 0 ∞ 1 n ! Tr 2 , … , 1 + n A ^ 1 + n ( t , 1 , … , 1 + n ) ⨂ i = 1 1 + n F 1 0 ( i ) , (58) F_1(t,1)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{2,\ldots,1+n}\widehat{\mathfrak A}_{1+n}(t,1,\ldots,1+n)\bigotimes_{i=1}^{1+n}F_1^0(i), \tag*{(58)} F 1 ( t , 1 ) = n = 0 ∑ ∞ n ! 1 Tr 2 , … , 1 + n A 1 + n ( t , 1 , … , 1 + n ) i = 1 ⨂ 1 + n F 1 0 ( i ) , (58)
F s ( t , 1 , … , s ∣ F 1 ( t ) ) ≐ ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n V 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) ⨂ i = 1 s + n F 1 ( t , i ) , s ≥ 2. (59) F_s(t,1,\ldots,s\mid F_1(t))\doteq
\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\mathfrak V_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)\bigotimes_{i=1}^{s+n}F_1(t,i),\qquad s\geq2. \tag*{(59)} F s ( t , 1 , … , s ∣ F 1 ( t )) ≐ n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n V 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) i = 1 ⨂ s + n F 1 ( t , i ) , s ≥ 2. (59)
The generating operators of reduced state functionals (3.5) are solutions of kinetic cluster expansions:
V 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) ≐ ∑ k = 0 n ( − 1 ) k ∑ n 1 = 1 n ⋯ ∑ n k = 1 n − n 1 − ⋯ − n k − 1 n ! ( n − n 1 − ⋯ − n k ) ! × A ^ 1 + n − n 1 − ⋯ − n k ( t , { 1 , … , s } , s + 1 , … , s + n − n 1 − ⋯ − n k ) × ∏ j = 1 k ∑ D j : X j = ⋃ l j Y l j ∣ D j ∣ ≤ s + n − n 1 − ⋯ − n j 1 ∣ D j ∣ ! ∑ i 1 ≠ ⋯ ≠ i ∣ D j ∣ i 1 , … , i ∣ D j ∣ = 1 s + n − n 1 − ⋯ − n j ∏ Y l j ⊂ D j 1 ∣ Y l j ∣ ! A ^ 1 + ∣ Y l j ∣ ( t , i l j , Y l j ) , n ≥ 0 , (60) \begin{aligned}
\mathfrak V_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)&\doteq\sum_{k=0}^{n}(-1)^k\sum_{n_1=1}^{n}\cdots\sum_{n_k=1}^{n-n_1-\cdots-n_{k-1}}\frac{n!}{(n-n_1-\cdots-n_k)!}\\
&\quad\times\widehat{\mathfrak A}_{1+n-n_1-\cdots-n_k}(t,\{1,\ldots,s\},s+1,\ldots,s+n-n_1-\cdots-n_k)\\
&\quad\times\prod_{j=1}^{k}\sum_{\substack{\mathrm D_j:X_j=\bigcup_{l_j}Y_{l_j}\\ |D_j|\leq s+n-n_1-\cdots-n_j}}\frac{1}{|D_j|!}
\sum_{\substack{i_1\ne\cdots\ne i_{|D_j|}\\ i_1,\ldots,i_{|D_j|}=1}}^{s+n-n_1-\cdots-n_j}
\prod_{Y_{l_j}\subset D_j}\frac{1}{|Y_{l_j}|!}\widehat{\mathfrak A}_{1+|Y_{l_j}|}(t,i_{l_j},Y_{l_j}),\qquad n\geq0, \tag*{(60)}
\end{aligned} V 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) ≐ k = 0 ∑ n ( − 1 ) k n 1 = 1 ∑ n ⋯ n k = 1 ∑ n − n 1 − ⋯ − n k − 1 ( n − n 1 − ⋯ − n k )! n ! × A 1 + n − n 1 − ⋯ − n k ( t , { 1 , … , s } , s + 1 , … , s + n − n 1 − ⋯ − n k ) × j = 1 ∏ k D j : X j = ⋃ l j Y l j ∣ D j ∣ ≤ s + n − n 1 − ⋯ − n j ∑ ∣ D j ∣ ! 1 i 1 = ⋯ = i ∣ D j ∣ i 1 , … , i ∣ D j ∣ = 1 ∑ s + n − n 1 − ⋯ − n j Y l j ⊂ D j ∏ ∣ Y l j ∣ ! 1 A 1 + ∣ Y l j ∣ ( t , i l j , Y l j ) , n ≥ 0 , (60)
where, as above, we denoted X j ≡ ( s + n − n 1 − ⋯ − n j + 1 , … , s + n − n 1 − ⋯ − n j − 1 ) X_j\equiv(s+n-n_1-\cdots-n_j+1,\ldots,s+n-n_1-\cdots-n_{j-1}) X j ≡ ( s + n − n 1 − ⋯ − n j + 1 , … , s + n − n 1 − ⋯ − n j − 1 ) , the symbol ∑ D : Y = ⋃ l X l , ∣ D ∣ ≤ s + n − n 1 \sum_{\mathrm D:Y=\bigcup_l X_l,\ |D|\leq s+n-n_1} ∑ D : Y = ⋃ l X l , ∣ D ∣ ≤ s + n − n 1 means the sum over all possible dissections D of the linearly ordered set Y ≡ ( s + n − n 1 + 1 , … , s + n ) Y\equiv(s+n-n_1+1,\ldots,s+n) Y ≡ ( s + n − n 1 + 1 , … , s + n ) on no more than s + n − n 1 s+n-n_1 s + n − n 1 linearly ordered subsets and the generating operator A ^ 1 + n ( t ) \widehat{\mathfrak A}_{1+n}(t) A 1 + n ( t ) of these expansions is the ( 1 + n ) (1+n) ( 1 + n ) th-order cumulant of the groups of scattering operators (2.24). Here are a few simple examples of the generating operators used in these expansions:
V 1 ( t , { 1 , … , s } ) = A ^ 1 ( t , { 1 , … , s } ) , \mathfrak V_1(t,\{1,\ldots,s\})=\widehat{\mathfrak A}_1(t,\{1,\ldots,s\}), V 1 ( t , { 1 , … , s }) = A 1 ( t , { 1 , … , s }) ,
V 1 + 1 ( t , { 1 , … , s } , s + 1 ) = A ^ 2 ( t , { 1 , … , s } , s + 1 ) − A ^ 1 ( t , { 1 , … , s } ) ∑ i = 1 s A ^ 2 ( t , i , s + 1 ) , \mathfrak V_{1+1}(t,\{1,\ldots,s\},s+1)=\widehat{\mathfrak A}_2(t,\{1,\ldots,s\},s+1)-\widehat{\mathfrak A}_1(t,\{1,\ldots,s\})\sum_{i=1}^{s}\widehat{\mathfrak A}_2(t,i,s+1), V 1 + 1 ( t , { 1 , … , s } , s + 1 ) = A 2 ( t , { 1 , … , s } , s + 1 ) − A 1 ( t , { 1 , … , s }) i = 1 ∑ s A 2 ( t , i , s + 1 ) ,
Thus, the equality holds F ( t ) = F ( t ∣ F 1 ( t ) ) F(t)=F(t\mid F_1(t)) F ( t ) = F ( t ∣ F 1 ( t )) , where the sequence F ( t ) F(t) F ( t ) is a non-perturbative solution (3.2) of the BBGKY hierarchy (2.47) for the initial data (3.1), and the sequence F ( t ∣ F 1 ( t ) ) F(t\mid F_1(t)) F ( t ∣ F 1 ( t )) is a sequence of reduced state functionals (3.5). This relationship essentially elucidates the links among the equations that govern the state evolution of many-particle quantum systems.
In the paper [20 ] , Bogolyubov proposed a different approach to deriving Markovian kinetic equations. This method relies on the construction of reduced state functionals F s ( t , 1 , … , s ∣ F 1 ( t ) ) F_s(t,1,\ldots,s\mid F_1(t)) F s ( t , 1 , … , s ∣ F 1 ( t )) using perturbation theory methods. To compare, we briefly outline this approach.
First, consider the first terms of the series expansion (59) . For a bounded interaction potential, an analogue of the Duhamel equation holds for the second-order cumulant of the scattering operators (24)
A ^ 2 ( t , { 1 , … , s } , s + 1 ) = ∫ 0 t d τ G s ∗ ( τ , 1 , … , s ) G 1 ∗ ( τ , s + 1 ) ∑ i 1 = 1 s N i n t ∗ ( i 1 , s + 1 ) × G s + 1 ∗ ( t − τ , 1 , … , s + 1 ) ∏ i 2 = 1 s + 1 G 1 ∗ ( − τ , i 2 ) . \widehat{\mathfrak{A}}_2(t,\{1,\ldots,s\},s+1)=\int_0^t d\tau\,\mathcal{G}_s^*(\tau,1,\ldots,s)\mathcal{G}_1^*(\tau,s+1)\sum_{i_1=1}^{s}\mathcal{N}_{\mathrm{int}}^*(i_1,s+1)\times
\mathcal{G}_{s+1}^*(t-\tau,1,\ldots,s+1)\prod_{i_2=1}^{s+1}\mathcal{G}_1^*(-\tau,i_2). A 2 ( t , { 1 , … , s } , s + 1 ) = ∫ 0 t d τ G s ∗ ( τ , 1 , … , s ) G 1 ∗ ( τ , s + 1 ) i 1 = 1 ∑ s N int ∗ ( i 1 , s + 1 ) × G s + 1 ∗ ( t − τ , 1 , … , s + 1 ) i 2 = 1 ∏ s + 1 G 1 ∗ ( − τ , i 2 ) .
and, consequently, the following representation holds for the second-order generating operator of series (59) :
V 2 ( t , { 1 , … , s } , s + 1 ) = ( A ^ 2 ( t , { 1 , … , s } , s + 1 ) − A ^ 1 ( t , { 1 , … , s } ) ) ∑ i 1 = 1 s A ^ 2 ( t , i 1 , s + 1 ) = \mathfrak{V}_2(t,\{1,\ldots,s\},s+1)=\left(\widehat{\mathfrak{A}}_2(t,\{1,\ldots,s\},s+1)-\widehat{\mathfrak{A}}_1(t,\{1,\ldots,s\})\right)\sum_{i_1=1}^{s}\widehat{\mathfrak{A}}_2(t,i_1,s+1)= V 2 ( t , { 1 , … , s } , s + 1 ) = ( A 2 ( t , { 1 , … , s } , s + 1 ) − A 1 ( t , { 1 , … , s }) ) i 1 = 1 ∑ s A 2 ( t , i 1 , s + 1 ) =
∫ 0 t d τ G s ∗ ( τ , 1 , … , s ) G 1 ∗ ( τ , s + 1 ) ( ∑ i 1 = 1 s N i n t ∗ ( i 1 , s + 1 ) ) G ^ s + 1 ∗ ( τ − t , 1 , … , s + 1 ) − \int_0^t d\tau\,\mathcal{G}_s^*(\tau,1,\ldots,s)\mathcal{G}_1^*(\tau,s+1)\left(\sum_{i_1=1}^{s}\mathcal{N}_{\mathrm{int}}^*(i_1,s+1)\right)\widehat{\mathcal{G}}_{s+1}^*(\tau-t,1,\ldots,s+1)- ∫ 0 t d τ G s ∗ ( τ , 1 , … , s ) G 1 ∗ ( τ , s + 1 ) ( i 1 = 1 ∑ s N int ∗ ( i 1 , s + 1 ) ) G s + 1 ∗ ( τ − t , 1 , … , s + 1 ) −
G ^ s ∗ ( τ − t , 1 , … , s ) ∑ i 1 = 1 s N i n t ∗ ( i 1 , s + 1 ) G ^ 2 ∗ ( τ − t , i 1 , s + 1 ) ∏ i 2 = 1 s + 1 G 1 ∗ ( − τ , i 2 ) . \widehat{\mathcal{G}}_s^*(\tau-t,1,\ldots,s)\sum_{i_1=1}^{s}\mathcal{N}_{\mathrm{int}}^*(i_1,s+1)\widehat{\mathcal{G}}_2^*(\tau-t,i_1,s+1)\prod_{i_2=1}^{s+1}\mathcal{G}_1^*(-\tau,i_2). G s ∗ ( τ − t , 1 , … , s ) i 1 = 1 ∑ s N int ∗ ( i 1 , s + 1 ) G 2 ∗ ( τ − t , i 1 , s + 1 ) i 2 = 1 ∏ s + 1 G 1 ∗ ( − τ , i 2 ) .
The characteristic scale for kinetic phenomena described, for example, by the quantum Boltzmann equation is determined by the mean free path. On the kinetic scale, the variation of the operator group variables (9) for a finite number of particles depends on the microscopic time variable ε − 1 t \varepsilon^{-1}t ε − 1 t , where ε ≥ 0 \varepsilon\geq0 ε ≥ 0 is a scaling parameter, and the dimensionless reduced state functionals are represented in the form: F s ( ε − 1 t , 1 , … , s ∣ F 1 ( t ) ) F_s(\varepsilon^{-1}t,1,\ldots,s\mid F_1(t)) F s ( ε − 1 t , 1 , … , s ∣ F 1 ( t )) . Then, according to the Duhamel equation for the second-order cumulant of scattering operators, in the formal Markovian limit ε → 0 \varepsilon\to0 ε → 0 , the first two terms of the expansions of the dimensionless reduced state functionals coincide with the corresponding terms obtained via perturbation theory using the correlation weakening condition in [20 ]
lim ε → 0 F s ( ε − 1 t , 1 , … , s ∣ F 1 ( t ) ) = G ^ s ( ∞ , 1 , … , s ) ∏ i = 1 s F 1 ( t , i ) + \lim_{\varepsilon\to0}F_s(\varepsilon^{-1}t,1,\ldots,s\mid F_1(t))=\widehat{\mathcal{G}}_s(\infty,1,\ldots,s)\prod_{i=1}^{s}F_1(t,i)+ ε → 0 lim F s ( ε − 1 t , 1 , … , s ∣ F 1 ( t )) = G s ( ∞ , 1 , … , s ) i = 1 ∏ s F 1 ( t , i ) +
∫ 0 ∞ d τ G s ∗ ( τ , 1 , … , s ) Tr s + 1 ( ∑ i 1 = 1 s N i n t ∗ ( i 1 , s + 1 ) G ^ s + 1 ( ∞ , 1 , … , s , s + 1 ) − \int_0^\infty d\tau\,\mathcal{G}_s^*(\tau,1,\ldots,s)\operatorname{Tr}_{s+1}(\sum_{i_1=1}^{s}\mathcal{N}_{\mathrm{int}}^*(i_1,s+1)\widehat{\mathcal{G}}_{s+1}(\infty,1,\ldots,s,s+1)- ∫ 0 ∞ d τ G s ∗ ( τ , 1 , … , s ) Tr s + 1 ( i 1 = 1 ∑ s N int ∗ ( i 1 , s + 1 ) G s + 1 ( ∞ , 1 , … , s , s + 1 ) −
G ^ s ( ∞ , 1 , … , s ) ∑ i 1 = 1 s N i n t ∗ ( i 1 , s + 1 ) G ^ 2 ( ∞ , i 1 , s + 1 ) ) ∏ i 2 = 1 s + 1 G 1 ∗ ( − τ , i 2 ) F 1 ( t , i 2 ) + e t c . \widehat{\mathcal{G}}_s(\infty,1,\ldots,s)\sum_{i_1=1}^{s}\mathcal{N}_{\mathrm{int}}^*(i_1,s+1)\widehat{\mathcal{G}}_2(\infty,i_1,s+1))\prod_{i_2=1}^{s+1}\mathcal{G}_1^*(-\tau,i_2)F_1(t,i_2)+\mathrm{etc.} G s ( ∞ , 1 , … , s ) i 1 = 1 ∑ s N int ∗ ( i 1 , s + 1 ) G 2 ( ∞ , i 1 , s + 1 )) i 2 = 1 ∏ s + 1 G 1 ∗ ( − τ , i 2 ) F 1 ( t , i 2 ) + etc.
Thus, at the specified kinetic scale, the collision integral of the generalized kinetic equation (61) takes the form of the Bogolyubov collision integral [20 ] . It should be noted that, in the homogeneous case, the collision integral in the first approximation of the resulting expression has a more general form than the collision integral of the quantum Boltzmann equation.
Evolution of correlations governed by the kinetic equation
Let us formulate the evolution equation for the one-particle reduced density operator representing by series (3.4). By differentiating the series expansion (3.4) with respect to the time variable in the sense of pointwise convergence in the space L 1 ( H ) \mathcal{L}^{1}(\mathcal{H}) L 1 ( H ) and applying the kinetic cluster expansions (3.3) for the cumulants of operator groups (2.44), we derive an identity for the one-particle density operator. The resulting identity for the one-particle density operator is then treated as an evolution equation governing the one-particle state of quantum systems.
If F 1 0 ∈ L 1 ( H ) F_1^0 \in\mathcal{L}^{1}(\mathcal{H}) F 1 0 ∈ L 1 ( H ) , then for an arbitrary t ∈ R t \in\mathbb{R} t ∈ R , the one-particle reduced density operator (3.4) is governed by the Cauchy problem for the following generalized quantum kinetic equation [39 ] :
d d t F 1 ( t , 1 ) = N 1 ∗ ( 1 ) F 1 ( t , 1 ) + Tr 2 N i n t ∗ ( 1 , 2 ) F 2 ( t , 1 , 2 ∣ F 1 ( t ) ) , (61) \frac{d}{dt}F_1(t,1)=\mathcal{N}_1^*(1)F_1(t,1)+\operatorname{Tr}_2\mathcal{N}_{\mathrm{int}}^*(1,2)F_2(t,1,2\mid F_1(t)), \tag*{(61)} d t d F 1 ( t , 1 ) = N 1 ∗ ( 1 ) F 1 ( t , 1 ) + Tr 2 N int ∗ ( 1 , 2 ) F 2 ( t , 1 , 2 ∣ F 1 ( t )) , (61)
F 1 ( t , 1 ) ∣ t = 0 = F 1 0 ( 1 ) , (62) F_1(t,1)\big|_{t=0}=F_1^0(1), \tag*{(62)} F 1 ( t , 1 ) t = 0 = F 1 0 ( 1 ) , (62)
where the collision integral in this kinetic equation is determined by the two-particle reduced functional of the state (3.5), and the operators N 1 ∗ \mathcal{N}_1^* N 1 ∗ and N i n t ∗ \mathcal{N}_{\mathrm{int}}^* N int ∗ are defined by the formula (2.10).
In the space of trace class operators L 1 ( H ) \mathcal{L}^{1}(\mathcal{H}) L 1 ( H ) the following statement holds [39 ] .
Theorem 7. For bounded interaction potentials, the global-in-time non-perturbative solution to the Cauchy problem (3.7),(3.8) is determined by a one-particle reduced density operator represented by the series expansion (3.4). If ∥ F 1 0 ∥ L 1 ( H ) < e − 2 \lVert F_1^0\rVert_{\mathcal{L}^{1}(\mathcal{H})}<e^{-2} ∥ F 1 0 ∥ L 1 ( H ) < e − 2 , then, provided F 1 0 ∈ L 0 1 ( H ) F_1^0\in\mathcal{L}_0^1(\mathcal{H}) F 1 0 ∈ L 0 1 ( H ) , it is a strong (classical) solution, whereas for arbitrary initial data F 1 0 ∈ L 1 ( H ) F_1^0\in\mathcal{L}^1(\mathcal{H}) F 1 0 ∈ L 1 ( H ) , it is a weak (generalized) solution.
The outline of the proof for this assertion is analogous to the proof of the existence theorem for the BBGKY hierarchy (2.47) in the space of sequences of trace-class operators.
Thus, for a separable initial state (3.1), the evolution of the state of a system of many quantum particles is described in terms of the state of a typical particle governed by a non-Markovian nonlinear kinetic equation (3.7), and a sequence of functionals (3.5) of this state (3.4) without any scaling approximations. The state functionals (3.5) describe the process of creating correlations in a quantum system of particles by employing the kinetic equation (3.7).
The process of creating correlations from chaos
To describe the process of correlation propagation for the separable initial state (3.1), we use the reduced correlation operators (2.51), defined in terms of the correlation operators (2.23). As a result, the reduced correlation functionals are represented by the following series expansions:
G s ( t , 1 , … , s ∣ G 1 ( t ) ) ≔ ∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n g s + n ( t , 1 , … , s + n ∣ g 1 ( t ) ) = G_s(t,1,\ldots,s\mid G_1(t))\coloneqq\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}g_{s+n}(t,1,\ldots,s+n\mid g_1(t))= G s ( t , 1 , … , s ∣ G 1 ( t )) : = n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n g s + n ( t , 1 , … , s + n ∣ g 1 ( t )) =
∑ n = 0 ∞ 1 n ! Tr s + 1 , … , s + n A s + n ( t , 1 , … , s + n ) ⨂ j = 1 s + n G 1 ( t , j ) , s ≥ 2 , (63) \sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{s+1,\ldots,s+n}\mathfrak{A}_{s+n}(t,1,\ldots,s+n)\bigotimes_{j=1}^{s+n}G_1(t,j),\qquad s\geq2, \tag*{(63)} n = 0 ∑ ∞ n ! 1 Tr s + 1 , … , s + n A s + n ( t , 1 , … , s + n ) j = 1 ⨂ s + n G 1 ( t , j ) , s ≥ 2 , (63)
where the generating operators V s + n ( t ) \mathfrak{V}_{s+n}(t) V s + n ( t ) , n ≥ 0 n \ge0 n ≥ 0 , are ( s + n ) (s+n) ( s + n ) th-order operators represented by the expansions (3.6) for s + n s+n s + n individual particles and
G 1 ( t , 1 ) = ∑ n = 0 ∞ 1 n ! Tr 2 , … , 1 + n A 1 + n ( t , 1 , … , 1 + n ) ⨂ j = 1 1 + n G 1 0 ( j ) . G_1(t,1)=\sum_{n=0}^{\infty}\frac{1}{n!}\operatorname{Tr}_{2,\ldots,1+n}\mathfrak{A}_{1+n}(t,1,\ldots,1+n)\bigotimes_{j=1}^{1+n}G_1^0(j). G 1 ( t , 1 ) = n = 0 ∑ ∞ n ! 1 Tr 2 , … , 1 + n A 1 + n ( t , 1 , … , 1 + n ) j = 1 ⨂ 1 + n G 1 0 ( j ) .
The sequence of reduced correlation functionals (3.9) describes the process of creating correlations from chaos in a quantum system with infinitely many particles.
It should be emphasized that the reduced correlation functionals (3.9) are cumulants of the reduced state functionals (3.5). The generating operators for the series expansions (3.9) and (3.5) differ in their order. The relationship between the reduced correlation functionals (3.9) and reduced state functionals (3.5) elucidates the links among the equations that govern the state evolution of many-particle quantum systems.
The reduced correlation functionals (3.9) on the microscopic scale directly determine the macroscopic characteristics of fluctuations of observables. For example, the functional representing the dispersion of an additive-type observable, i.e., the sequence O ( 1 ) = ( 0 , o 1 ( 1 ) , … , ∑ i 1 = 1 n o 1 ( i 1 ) , … ) O^{(1)}=(0,o_1(1),\ldots,\sum_{i_1=1}^{n}o_1(i_1),\ldots) O ( 1 ) = ( 0 , o 1 ( 1 ) , … , ∑ i 1 = 1 n o 1 ( i 1 ) , … ) , is given by the formula
⟨ ( O ( 1 ) − ⟨ O ( 1 ) ⟩ ) 2 ⟩ ( t ) = Tr 1 ( o 1 2 ( 1 ) − ⟨ O ( 1 ) ⟩ 2 ( t ) ) G 1 ( t , 1 ) + Tr 1 , 2 o 1 ( 1 ) o 1 ( 2 ) G 2 ( t , 1 , 2 ∣ G 1 ( t ) ) . \left\langle\left(O^{(1)}-\left\langle O^{(1)}\right\rangle\right)^2\right\rangle(t)=\operatorname{Tr}_1\left(o_1^2(1)-\left\langle O^{(1)}\right\rangle^2(t)\right)G_1(t,1)+\operatorname{Tr}_{1,2}o_1(1)o_1(2)G_2(t,1,2\mid G_1(t)). ⟨ ( O ( 1 ) − ⟨ O ( 1 ) ⟩ ) 2 ⟩ ( t ) = Tr 1 ( o 1 2 ( 1 ) − ⟨ O ( 1 ) ⟩ 2 ( t ) ) G 1 ( t , 1 ) + Tr 1 , 2 o 1 ( 1 ) o 1 ( 2 ) G 2 ( t , 1 , 2 ∣ G 1 ( t )) .
where ⟨ O ( 1 ) ⟩ ( t ) = Tr 1 o 1 ( 1 ) G 1 ( t , 1 ) \left\langle O^{(1)}\right\rangle(t)=\operatorname{Tr}_1 o_1(1)G_1(t,1) ⟨ O ( 1 ) ⟩ ( t ) = Tr 1 o 1 ( 1 ) G 1 ( t , 1 ) is the mean-value functional of an additive-type observable.
The mean-field scaling approximation for a separable initial state
The widely accepted approach to describing evolution through kinetic equations is as follows [7 , 11 ] . If the initial state of a system composed of many particles is specified by the state of a typical particle of the system, meaning there are no correlations between the particles at the initial instant, then under certain scaling approximations, the evolution of the system can be effectively described as the evolution of a typical particle state. This approach is formulated in terms of a one-particle reduced density operator, which is governed by a Markovian nonlinear kinetic equation.
The specific quantum kinetic equations, such as the quantum Boltzmann equation and others, can be derived from the generalized quantum kinetic equation (3.7) under appropriate scaling approximations. Below, we will argue the asymptotic behavior in the mean-field scaling limit of the solution (3.2) of the generalized quantum kinetic equation (3.7) and the sequence of reduced state functionals (3.5).
To determine the scaling parameter, we rewrite the kinetic equation (3.7) in dimensionless form. Then the collision integral of this kinetic equation takes the form: ε Tr 2 N i n t ∗ ( 1 , 2 ) F 2 ( t , 1 , 2 ∣ F 1 ( t ) ) \varepsilon\operatorname{Tr}_2\mathcal{N}_{\mathrm{int}}^{*}(1,2)F_2(t,1,2\mid F_1(t)) ε Tr 2 N int ∗ ( 1 , 2 ) F 2 ( t , 1 , 2 ∣ F 1 ( t )) , where ε > 0 \varepsilon>0 ε > 0 , is a scaling parameter, which is a consequence of the dimensionless form of the Hamiltonian H n = ∑ j = 1 n K ( j ) + ε ∑ j 1 < j 2 = 1 n Φ ( j 1 , j 2 ) H_n=\sum_{j=1}^{n}K(j)+\varepsilon\sum_{j_1<j_2=1}^{n}\Phi(j_1,j_2) H n = ∑ j = 1 n K ( j ) + ε ∑ j 1 < j 2 = 1 n Φ ( j 1 , j 2 ) .
Since for an arbitrary finite time interval for an asymptotically perturbed first-order cumulant of strongly continuous operator groups (2.9) such equality holds
lim ε → 0 ∥ G s ∗ ( t , 1 , … , s ) f s − ∏ j = 1 s G 1 ∗ ( t , j ) f s ∥ L 1 ( H ⊗ s ) = 0. (64) \lim_{\varepsilon\to0}\left\|\mathcal{G}_s^{*}(t,1,\ldots,s)f_s-\prod_{j=1}^{s}\mathcal{G}_1^{*}(t,j)f_s\right\|_{\mathcal{L}^{1}(\mathcal{H}^{\otimes s})}=0. \tag*{(64)} ε → 0 lim G s ∗ ( t , 1 , … , s ) f s − j = 1 ∏ s G 1 ∗ ( t , j ) f s L 1 ( H ⊗ s ) = 0. (64)
then for the ( 1 + n ) (1+n) ( 1 + n ) th-order cumulant of asymptotically perturbed operator groups (2.9) the equality holds
lim ϵ → 0 ∥ 1 ϵ n 1 n ! A 1 + n ∗ ( t , { 1 , … , s } , s + 1 , … , s + n ) f s + n − ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n ∏ j = 1 s G 1 ∗ ( t − t 1 , j ) ∑ i 1 = 1 s N i n t ∗ ( i 1 , s + 1 ) ∏ j 1 = 1 s + 1 G 1 ∗ ( t 1 − t 2 , j 1 ) … ∏ j n − 1 = 1 s + n − 1 G 1 ∗ ( t n − 1 − t n , j n − 1 ) ∑ i n = 1 s + n N i n t ∗ ( i n , s + n ) ∏ j n = 1 s + n G 1 ∗ ( t n , j n ) f s + n ∥ L 1 ( H ⊗ ( s + n ) ) = 0 , n ≥ 1. (65) \lim_{\epsilon\to0}\left\|\frac{1}{\epsilon^n}\frac{1}{n!}\mathfrak{A}_{1+n}^*\left(t,\{1,\ldots,s\},s+1,\ldots,s+n\right)f_{s+n}-
\int_0^t dt_1\ldots\int_0^{t_{n-1}}dt_n\prod_{j=1}^{s}G_1^*(t-t_1,j)\sum_{i_1=1}^{s}\mathcal{N}_{\mathrm{int}}^*(i_1,s+1)\prod_{j_1=1}^{s+1}G_1^*(t_1-t_2,j_1)\ldots
\prod_{j_{n-1}=1}^{s+n-1}G_1^*(t_{n-1}-t_n,j_{n-1})\sum_{i_n=1}^{s+n}\mathcal{N}_{\mathrm{int}}^*(i_n,s+n)\prod_{j_n=1}^{s+n}G_1^*(t_n,j_n)f_{s+n}\right\|_{\mathfrak{L}_1(\mathcal{H}^{\otimes(s+n)})}=0,
\quad n\ge1. \tag*{(65)} ϵ → 0 lim ϵ n 1 n ! 1 A 1 + n ∗ ( t , { 1 , … , s } , s + 1 , … , s + n ) f s + n − ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n j = 1 ∏ s G 1 ∗ ( t − t 1 , j ) i 1 = 1 ∑ s N int ∗ ( i 1 , s + 1 ) j 1 = 1 ∏ s + 1 G 1 ∗ ( t 1 − t 2 , j 1 ) … j n − 1 = 1 ∏ s + n − 1 G 1 ∗ ( t n − 1 − t n , j n − 1 ) i n = 1 ∑ s + n N int ∗ ( i n , s + n ) j n = 1 ∏ s + n G 1 ∗ ( t n , j n ) f s + n L 1 ( H ⊗ ( s + n ) ) = 0 , n ≥ 1. (65)
The last equalities are a consequence of the validity for the bounded interaction potential of analogues of the Duhamel equation for cumulants (2.11) of asymptotically perturbed operator groups (2.9).
According to equalities (3.10) and (3.11) for a one-particle reduced density operator (3.4) the following limit theorem holds [39 ] .
Theorem 8. Let the mean-field limit exist for the initial one-particle reduced density operator in the following sense:
lim ϵ → 0 ∥ ϵ F 1 0 − f 1 0 ∥ L 1 ( H ) = 0 , \lim_{\epsilon\to0}\left\|\epsilon F_1^0-f_1^0\right\|_{\mathfrak{L}_1(\mathcal{H})}=0, ϵ → 0 lim ϵ F 1 0 − f 1 0 L 1 ( H ) = 0 , then for an arbitrary finite time interval for series expansion (3.4) the equality holds
lim ϵ → 0 ∥ ϵ F 1 ( t ) − f 1 ( t ) ∥ L 1 ( H ) = 0 , \lim_{\epsilon\to0}\left\|\epsilon F_1(t)-f_1(t)\right\|_{\mathfrak{L}_1(\mathcal{H})}=0, ϵ → 0 lim ∥ ϵ F 1 ( t ) − f 1 ( t ) ∥ L 1 ( H ) = 0 , where the limiting one-particle density operator f 1 ( t ) f_1(t) f 1 ( t ) is represented by a series converging in the norm of the space L 1 ( H ) \mathfrak{L}^1(\mathcal{H}) L 1 ( H )
f 1 ( t , 1 ) = ∑ n = 0 ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n Tr 2 , … , n + 1 G 1 ∗ ( t − t 1 , 1 ) N i n t ∗ ( 1 , 2 ) ∏ j 1 = 1 2 G 1 ∗ ( t 1 − t 2 , j 1 ) … ∏ i n = 1 n G 1 ∗ ( t n − 1 − t n , i n ) ∑ k n = 1 n N i n t ∗ ( k n , n + 1 ) ∏ j n = 1 n + 1 G 1 ∗ ( t n , j n ) ⨂ i = 1 n + 1 f 1 0 ( i ) . (66) f_1(t,1)=\sum_{n=0}^{\infty}\int_0^t dt_1\ldots\int_0^{t_{n-1}}dt_n\operatorname{Tr}_{2,\ldots,n+1}G_1^*(t-t_1,1)\mathcal{N}_{\mathrm{int}}^*(1,2)\prod_{j_1=1}^{2}G_1^*(t_1-t_2,j_1)\ldots
\prod_{i_n=1}^{n}G_1^*(t_{n-1}-t_n,i_n)\sum_{k_n=1}^{n}\mathcal{N}_{\mathrm{int}}^*(k_n,n+1)\prod_{j_n=1}^{n+1}G_1^*(t_n,j_n)\bigotimes_{i=1}^{n+1}f_1^0(i). \tag*{(66)} f 1 ( t , 1 ) = n = 0 ∑ ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n Tr 2 , … , n + 1 G 1 ∗ ( t − t 1 , 1 ) N int ∗ ( 1 , 2 ) j 1 = 1 ∏ 2 G 1 ∗ ( t 1 − t 2 , j 1 ) … i n = 1 ∏ n G 1 ∗ ( t n − 1 − t n , i n ) k n = 1 ∑ n N int ∗ ( k n , n + 1 ) j n = 1 ∏ n + 1 G 1 ∗ ( t n , j n ) i = 1 ⨂ n + 1 f 1 0 ( i ) . (66) For bounded interaction potentials, the series (3.12) converges in the norm of the space L 1 ( H ) \mathfrak{L}^1(\mathcal{H}) L 1 ( H ) provided that t < t 0 ≡ ( 2 ∥ Φ ∥ L ( H ⊗ 2 ) ∥ f 1 0 ∥ L 1 ( H ) ) − 1 t<t_0\equiv\left(2\left\|\Phi\right\|_{\mathfrak{L}(\mathcal{H}^{\otimes2})}\left\|f_1^0\right\|_{\mathfrak{L}^1(\mathcal{H})}\right)^{-1} t < t 0 ≡ ( 2 ∥ Φ ∥ L ( H ⊗ 2 ) f 1 0 L 1 ( H ) ) − 1 .
For the initial state f 1 0 ∈ L 1 ( H ) f_1^0\in\mathfrak{L}^1(\mathcal{H}) f 1 0 ∈ L 1 ( H ) the limiting operator (3.12) satisfies the Cauchy problem for the quantum Vlasov kinetic equation [21 ]
∂ ∂ t f 1 ( t , 1 ) = N 1 ∗ ( 1 ) f 1 ( t , 1 ) + Tr 2 N i n t ∗ ( 1 , 2 ) f 1 ( t , 1 ) ⊗ f 1 ( t , 2 ) , (67) \frac{\partial}{\partial t}f_1(t,1)=\mathcal{N}_1^*(1)f_1(t,1)+\operatorname{Tr}_2\mathcal{N}_{\mathrm{int}}^*(1,2)f_1(t,1)\otimes f_1(t,2), \tag*{(67)} ∂ t ∂ f 1 ( t , 1 ) = N 1 ∗ ( 1 ) f 1 ( t , 1 ) + Tr 2 N int ∗ ( 1 , 2 ) f 1 ( t , 1 ) ⊗ f 1 ( t , 2 ) , (67) f 1 ( t , 1 ) ∣ t = 0 = f 1 0 ( 1 ) . (68) f_1(t,1)\big|_{t=0}=f_1^0(1). \tag*{(68)} f 1 ( t , 1 ) t = 0 = f 1 0 ( 1 ) . (68)
Note that for pure states—that is, for f 1 ( t ) = ∣ ψ ( t ) ⟩ ⟨ ψ ( t ) ∣ f_1(t)=|\psi(t)\rangle\langle\psi(t)| f 1 ( t ) = ∣ ψ ( t )⟩ ⟨ ψ ( t ) ∣ or, in terms of the kernel of such an operator, f 1 ( t , q , q ′ ) = ψ ( t , q ) ψ ∗ ( t , q ′ ) f_1(t,q,q')=\psi(t,q)\psi^*(t,q') f 1 ( t , q , q ′ ) = ψ ( t , q ) ψ ∗ ( t , q ′ ) —the quantum Vlasov kinetic equation (67) reduces to the Hartree equation
i ∂ ∂ t ψ ( t , q ) = − 1 2 Δ q ψ ( t , q ) + ∫ d q ′ Φ ( q − q ′ ) ∣ ψ ( q ′ ) ∣ 2 ψ ( t , q ) . i\frac{\partial}{\partial t}\psi(t,q)=-\frac{1}{2}\Delta_q\psi(t,q)+\int dq'\Phi(q-q')|\psi(q')|^2\psi(t,q). i ∂ t ∂ ψ ( t , q ) = − 2 1 Δ q ψ ( t , q ) + ∫ d q ′ Φ ( q − q ′ ) ∣ ψ ( q ′ ) ∣ 2 ψ ( t , q ) .
In particular, if the interaction potential kernel Φ ( q ) = δ ( q ) \Phi(q)=\delta(q) Φ ( q ) = δ ( q ) is a Dirac measure, then the Hartree kinetic equation reduces to the nonlinear Schrödinger equation with cubic nonlinearity.
∂ ∂ t ψ ( t , q ) = − 1 2 Δ q ψ ( t , q ) + ∣ ψ ( t , q ) ∣ 2 ψ ( t , q ) . \frac{\partial}{\partial t}\psi(t,q)=-\frac{1}{2}\Delta_q\psi(t,q)+|\psi(t,q)|^2\psi(t,q). ∂ t ∂ ψ ( t , q ) = − 2 1 Δ q ψ ( t , q ) + ∣ ψ ( t , q ) ∣ 2 ψ ( t , q ) .
By virtue of property (65) of the cumulants of asymptotically perturbed operator groups, the following equalities hold for the generating operators (60) of reduced state functionals (59) :
lim ε → 0 ∥ ( V 1 ( t , { 1 , … , s } ) − I ) f s ∥ L 1 ( H ⊗ s ) = 0 , \lim_{\varepsilon\to0}\left\|\left(\mathfrak{V}_1(t,\{1,\ldots,s\})-I\right)f_s\right\|_{\mathfrak{L}^1(\mathcal{H}^{\otimes s})}=0, ε → 0 lim ∥ ( V 1 ( t , { 1 , … , s }) − I ) f s ∥ L 1 ( H ⊗ s ) = 0 ,
and
lim ε → 0 ∥ 1 ε n V 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) f s + n ∥ L 1 ( H ⊗ ( s + n ) ) = 0 , n ≥ 1 , \lim_{\varepsilon\to0}\left\|\frac{1}{\varepsilon^n}\mathfrak{V}_{1+n}(t,\{1,\ldots,s\},s+1,\ldots,s+n)f_{s+n}\right\|_{\mathfrak{L}^1(\mathcal{H}^{\otimes(s+n)})}=0,\qquad n\geq1, ε → 0 lim ε n 1 V 1 + n ( t , { 1 , … , s } , s + 1 , … , s + n ) f s + n L 1 ( H ⊗ ( s + n ) ) = 0 , n ≥ 1 ,
respectively. Due to such equalities and convergence in the mean-field limit of the solution of the Cauchy problem of the generalized quantum kinetic equation (61) ,(62) to the solution of the Cauchy problem of the Vlasov quantum kinetic equation (67) ,(68) for reduced state functionals (59) the following limit theorem holds.
Theorem 9. For the reduced state functionals (59) , the following equalities hold:
lim ε → 0 ∥ ε s F s ( t , 1 , … , s ) − ⨂ j = 1 s f 1 ( t , j ) ∥ L 1 ( H ⊗ s ) = 0 , s ≥ 2 , \lim_{\varepsilon\to0}\left\|\varepsilon^s F_s(t,1,\ldots,s)-\bigotimes_{j=1}^{s}f_1(t,j)\right\|_{\mathfrak{L}^1(\mathcal{H}^{\otimes s})}=0,\qquad s\geq2, ε → 0 lim ε s F s ( t , 1 , … , s ) − j = 1 ⨂ s f 1 ( t , j ) L 1 ( H ⊗ s ) = 0 , s ≥ 2 , where the limit operator f 1 ( t ) f_1(t) f 1 ( t ) is determined by series (66) , which represents the perturbative solution of the quantum Vlasov kinetic equation (67) .
The last assertion describes how initial chaos propagates in the mean-field limit. If the system is initially uncorrelated, then no correlations arise during its evolution within this approximation.
It should be noted that the traditional approach to the problem of the propagation of initial chaos is based on constructing the asymptotic behavior of the solution to the quantum BBGKY hierarchy (47) for reduced density operators within the framework of perturbation theory [21 ] , [23 ] .
Kinetic evolution founded on observables
This section will clarify the meaning of kinetic evolution in terms of observables for many quantum particles. A rigorous description of the kinetic evolution in terms of observables will be achieved by examining the mean-field asymptotics of the non-perturbative solution (38) to the Cauchy problem for the hierarchy of evolution equations for observables (40) . We will also establish how the evolution described by observables for a system of many quantum particles in the scaling approximation is related to the kinetic evolution of the state of a typical particle in that system.
Note that, for the separable initial state (55) , the kinetic representation of the mean-value functional for observables (31) holds
( B ( t ) , F ( c ) ) = ( B ( 0 ) , F ( t ∣ F 1 ( t ) ) ) , (69) \left(B(t),F^{(c)}\right)=\left(B(0),F\left(t\mid F_1(t)\right)\right), \tag*{(69)} ( B ( t ) , F ( c ) ) = ( B ( 0 ) , F ( t ∣ F 1 ( t ) ) ) , (69)
where the sequence F ( t ∣ F 1 ( t ) ) F\left(t\mid F_1(t)\right) F ( t ∣ F 1 ( t ) ) of reduced state functionals defined by formulas (58) and (59) . For additive-type reduced observables B ( 1 ) ( 0 ) = ( 0 , b 1 ( 1 ) , 0 , … ) B^{(1)}(0)=(0,b_1(1),0,\ldots) B ( 1 ) ( 0 ) = ( 0 , b 1 ( 1 ) , 0 , … ) the mean-value functional for observables (69) takes the form
( B ( 1 ) ( t ) , F ( c ) ) = Tr 1 b 1 ( 1 ) F 1 ( t ) , \left(B^{(1)}(t),F^{(c)}\right)=\operatorname{Tr}_1 b_1(1)F_1(t), ( B ( 1 ) ( t ) , F ( c ) ) = Tr 1 b 1 ( 1 ) F 1 ( t ) ,
where the one-particle reduced density operator F 1 ( t ) F_1(t) F 1 ( t ) is given by the series (58) . For k k k -ary reduced observables B ( k ) ( 0 ) = ( 0 , … , 0 , b k ( 1 , … , k ) , 0 , … ) B^{(k)}(0)=(0,\ldots,0,b_k(1,\ldots,k),0,\ldots) B ( k ) ( 0 ) = ( 0 , … , 0 , b k ( 1 , … , k ) , 0 , … ) , k ≥ 2 k\ge2 k ≥ 2 , the kinetic representation (69) takes the form
( B ( k ) ( t ) , F ( c ) ) = 1 k ! Tr 1 , … , k b k ( 1 , … , k ) F k ( t , 1 , … , k ∣ F 1 ( t ) ) , \left(B^{(k)}(t),F^{(c)}\right)=\frac{1}{k!}\operatorname{Tr}_{1,\ldots,k}b_k(1,\ldots,k)F_k\left(t,1,\ldots,k\mid F_1(t)\right), ( B ( k ) ( t ) , F ( c ) ) = k ! 1 Tr 1 , … , k b k ( 1 , … , k ) F k ( t , 1 , … , k ∣ F 1 ( t ) ) ,
where the reduced state functional F k ( t , 1 , … , k ∣ F 1 ( t ) ) F_k\left(t,1,\ldots,k\mid F_1(t)\right) F k ( t , 1 , … , k ∣ F 1 ( t ) ) is given by the series (59) .
The hierarchy of quantum kinetic equations for observables
To determine the scaling parameter, we rewrite the hierarchy of evolution equations (40) in dimensionless form, as before. Then the generator of this hierarchy takes the form:
( N s b ) s ( 1 , … , s ) = ∑ j = 1 s N ( j ) + ϵ ∑ j 1 < j 2 = 1 s N i n t ( j 1 , j 2 ) , \left(\mathcal{N}_s b\right)_s(1,\ldots,s)=\sum_{j=1}^{s}\mathcal{N}(j)+\epsilon\sum_{j_1<j_2=1}^{s}\mathcal{N}_{\mathrm{int}}(j_1,j_2), ( N s b ) s ( 1 , … , s ) = j = 1 ∑ s N ( j ) + ϵ j 1 < j 2 = 1 ∑ s N int ( j 1 , j 2 ) ,
( [ N , a + ] b ) s ( 1 , … , s ) = ϵ ∑ j 1 ≠ j 2 = 1 s N i n t ( j 1 , j 2 ) b s − 1 ( ( 1 , … , s ) ∖ j 1 ) ⊗ 1 ( j 1 ) , s ≥ 2. \left([\mathcal{N},a^+]b\right)_s(1,\ldots,s)=\epsilon\sum_{j_1\ne j_2=1}^{s}\mathcal{N}_{\mathrm{int}}(j_1,j_2)b_{s-1}\left((1,\ldots,s)\setminus j_1\right)\otimes1_{(j_1)},\qquad s\ge2. ( [ N , a + ] b ) s ( 1 , … , s ) = ϵ j 1 = j 2 = 1 ∑ s N int ( j 1 , j 2 ) b s − 1 ( ( 1 , … , s ) ∖ j 1 ) ⊗ 1 ( j 1 ) , s ≥ 2.
where ϵ > 0 \epsilon>0 ϵ > 0 is a scaling parameter, which is a consequence of the dimensionless form of the Hamiltonian.
Suppose that at the initial moment there exists the mean-field limit of reduced observables (41) in the sense of ∗ \ast ∗ -weak convergence in the space of bounded operators L ( H ⊗ s ) \mathcal{L}(\mathcal{H}^{\otimes s}) L ( H ⊗ s )
w ∗ − lim ϵ → 0 ( ϵ − s B s ϵ , 0 − b s 0 ) = 0 , (70) \mathop{\mathrm{w}^{*}\!-\!\lim}_{\epsilon\to0}\left(\epsilon^{-s}B_s^{\epsilon,0}-b_s^0\right)=0, \tag*{(70)} w ∗ − lim ϵ → 0 ( ϵ − s B s ϵ , 0 − b s 0 ) = 0 , (70)
where ϵ > 0 \epsilon>0 ϵ > 0 is a scaling parameter. Then the following limit theorem holds for the non-perturbative solution (38) of the hierarchy of evolution equations (40) .
Theorem 10. If condition (70) is satisfied, then for a finite time interval, the mean-field limit exists in the same sense for the sequence of reduced observables (38) as well [40 , 41 ]
w ∗ − lim ϵ → 0 ( ϵ − s B s ( t ) − b s ( t ) ) = 0 , s ≥ 1 , \mathop{\mathrm{w}^{*}\!-\!\lim}_{\epsilon\to0}\left(\epsilon^{-s}B_s(t)-b_s(t)\right)=0,\qquad s\ge1, w ∗ − lim ϵ → 0 ( ϵ − s B s ( t ) − b s ( t ) ) = 0 , s ≥ 1 ,
where the limiting reduced observables b s ( t ) b_s(t) b s ( t ) , s ≥ 1 s \ge1 s ≥ 1 , are defined by the following expansions:
b s ( t , 1 , … , s ) = ∑ n = 0 s − 1 ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n G s 0 ( t − t 1 ) ∑ i 1 ≠ j 1 = 1 s N i n t ( i 1 , j 1 ) G s − 1 0 ( t 1 − t 2 ) … G s − n + 1 0 ( t n − 1 − t n ) ∑ i n ≠ j n = 1 , i n , j n ∉ ( j 1 , … , j n − 1 ) s N i n t ( i n , j n ) G s − n 0 ( t n ) × b s − n 0 ( ( 1 , … , s ) ∖ ( j 1 , … , j n ) ) ⊗ 1 ( j 1 ) … ⊗ 1 ( j n ) , s ≥ 1 , (71) \begin{aligned}
b_s(t,1,\ldots,s)={}&\sum_{n=0}^{s-1}\int_0^t dt_1\ldots\int_0^{t_{n-1}}dt_n\,\mathcal{G}_s^0(t-t_1)\sum_{i_1\ne j_1=1}^{s}\mathcal{N}_{\mathrm{int}}(i_1,j_1)\mathcal{G}_{s-1}^0(t_1-t_2)\ldots\\
&\mathcal{G}_{s-n+1}^0(t_{n-1}-t_n)\sum_{\substack{i_n\ne j_n=1,\\ i_n,j_n\notin(j_1,\ldots,j_{n-1})}}^{s}\mathcal{N}_{\mathrm{int}}(i_n,j_n)\mathcal{G}_{s-n}^0(t_n)\times\\
&b_{s-n}^0\left((1,\ldots,s)\setminus(j_1,\ldots,j_n)\right)\otimes\mathbb{1}_{(j_1)}\ldots\otimes\mathbb{1}_{(j_n)},\qquad s\ge1, \tag*{(71)}
\end{aligned} b s ( t , 1 , … , s ) = n = 0 ∑ s − 1 ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n G s 0 ( t − t 1 ) i 1 = j 1 = 1 ∑ s N int ( i 1 , j 1 ) G s − 1 0 ( t 1 − t 2 ) … G s − n + 1 0 ( t n − 1 − t n ) i n = j n = 1 , i n , j n ∈ / ( j 1 , … , j n − 1 ) ∑ s N int ( i n , j n ) G s − n 0 ( t n ) × b s − n 0 ( ( 1 , … , s ) ∖ ( j 1 , … , j n ) ) ⊗ 1 ( j 1 ) … ⊗ 1 ( j n ) , s ≥ 1 , (71)
and for the operator groups of non-interacting particles, the notation was used
G s − n + 1 0 ( t n − 1 − t n ) ≡ ∏ j ∈ ( 1 , … , s ) ∖ ( j 1 , … , j n − 1 ) G 1 ( t n − 1 − t n , j ) . \mathcal{G}_{s-n+1}^0(t_{n-1}-t_n)\equiv\prod_{j\in(1,\ldots,s)\setminus(j_1,\ldots,j_{n-1})}\mathcal{G}_1(t_{n-1}-t_n,j). G s − n + 1 0 ( t n − 1 − t n ) ≡ j ∈ ( 1 , … , s ) ∖ ( j 1 , … , j n − 1 ) ∏ G 1 ( t n − 1 − t n , j ) .
For a certain type of reduced observable, the structure of expansions (4.3) takes a specific form. For example, in the case of reduced observables of k k k -ary type, we have:
b s ( k ) ( t , 1 , … , s ) = ∫ 0 t d t 1 … ∫ 0 t s − k − 1 d t s − k G s 0 ( t − t 1 ) ∑ i 1 ≠ j 1 = 1 s N i n t ( i 1 , j 1 ) G s − 1 0 ( t 1 − t 2 ) … G s − n + 1 0 ( t s − k − 1 − t s − k ) ∑ i s − k ≠ j s − k = 1 , i s − k , j s − k ∉ ( j 1 , … , j s − k − 1 ) s N i n t ( i s − k , j s − k ) G s − n 0 ( t s − k ) × b k 0 ( ( 1 , … , s ) ∖ ( j 1 , … , j s − k ) ) ⊗ 1 ( j 1 ) … ⊗ 1 ( j s − k ) , 1 ≤ s ≤ k . (72) \begin{aligned}
b_s^{(k)}(t,1,\ldots,s)={}&\int_0^t dt_1\ldots\int_0^{t_{s-k-1}}dt_{s-k}\,\mathcal{G}_s^0(t-t_1)\sum_{i_1\ne j_1=1}^{s}\mathcal{N}_{\mathrm{int}}(i_1,j_1)\mathcal{G}_{s-1}^0(t_1-t_2)\ldots\\
&\mathcal{G}_{s-n+1}^0(t_{s-k-1}-t_{s-k})\sum_{\substack{i_{s-k}\ne j_{s-k}=1,\\ i_{s-k},j_{s-k}\notin(j_1,\ldots,j_{s-k-1})}}^{s}\mathcal{N}_{\mathrm{int}}(i_{s-k},j_{s-k})\mathcal{G}_{s-n}^0(t_{s-k})\times\\
&b_k^0\left((1,\ldots,s)\setminus(j_1,\ldots,j_{s-k})\right)\otimes\mathbb{1}_{(j_1)}\ldots\otimes\mathbb{1}_{(j_{s-k})},\qquad1\le s\le k. \tag*{(72)}
\end{aligned} b s ( k ) ( t , 1 , … , s ) = ∫ 0 t d t 1 … ∫ 0 t s − k − 1 d t s − k G s 0 ( t − t 1 ) i 1 = j 1 = 1 ∑ s N int ( i 1 , j 1 ) G s − 1 0 ( t 1 − t 2 ) … G s − n + 1 0 ( t s − k − 1 − t s − k ) i s − k = j s − k = 1 , i s − k , j s − k ∈ / ( j 1 , … , j s − k − 1 ) ∑ s N int ( i s − k , j s − k ) G s − n 0 ( t s − k ) × b k 0 ( ( 1 , … , s ) ∖ ( j 1 , … , j s − k ) ) ⊗ 1 ( j 1 ) … ⊗ 1 ( j s − k ) , 1 ≤ s ≤ k . (72)
In the space b 0 ∈ L γ ( F H ) b^0\in\mathcal{L}_\gamma(\mathcal{F}_H) b 0 ∈ L γ ( F H ) the sequence b ( t ) = ( b 0 , b 1 ( t ) , … , b s ( t ) , … ) b(t)=(b_0,b_1(t),\ldots,b_s(t),\ldots) b ( t ) = ( b 0 , b 1 ( t ) , … , b s ( t ) , … ) of limiting reduced observables (4.3) satisfies the Cauchy problem for the so-called dual Vlasov hierarchy:
∂ ∂ t b s ( t , 1 , … , s ) = ∑ j = 1 s N ( j ) b s ( t , 1 , … , s ) + ∑ j 1 ≠ j 2 = 1 s N i n t ( j 1 , j 2 ) b s − 1 ( t , ( 1 , … , s ) ∖ ( j 1 ) ) ⊗ 1 ( j 1 ) , (73) \frac{\partial}{\partial t}b_s(t,1,\ldots,s)=\sum_{j=1}^{s}\mathcal{N}(j)b_s(t,1,\ldots,s)+\sum_{j_1\ne j_2=1}^{s}\mathcal{N}_{\mathrm{int}}(j_1,j_2)b_{s-1}\left(t,(1,\ldots,s)\setminus(j_1)\right)\otimes\mathbb{1}_{(j_1)}, \tag*{(73)} ∂ t ∂ b s ( t , 1 , … , s ) = j = 1 ∑ s N ( j ) b s ( t , 1 , … , s ) + j 1 = j 2 = 1 ∑ s N int ( j 1 , j 2 ) b s − 1 ( t , ( 1 , … , s ) ∖ ( j 1 ) ) ⊗ 1 ( j 1 ) , (73)
b s ( t , 1 , … , s ) ∣ t = 0 = b s 0 ( 1 , … , s ) , s ≥ 1. (74) b_s(t,1,\ldots,s)\big|_{t=0}=b_s^0(1,\ldots,s),\qquad s\ge1. \tag*{(74)} b s ( t , 1 , … , s ) t = 0 = b s 0 ( 1 , … , s ) , s ≥ 1. (74)
It should be noted that the sequence of evolution equations (4.5) has the structure of recurrence evolution equations.
Let us give a few examples of equations of the dual Vlasov hierarchy (4.5) in terms of the kernels of operators to the limiting reduced observables:
i ∂ ∂ t b 1 ( t , q 1 ; q 1 ′ ) = − 1 2 ( − Δ q 1 + Δ q 1 ′ ) b 1 ( t , q 1 ; q 1 ′ ) , i\frac{\partial}{\partial t}b_1(t,q_1;q'_1)=-\frac{1}{2}(-\Delta_{q_1}+\Delta_{q'_1})b_1(t,q_1;q'_1), i ∂ t ∂ b 1 ( t , q 1 ; q 1 ′ ) = − 2 1 ( − Δ q 1 + Δ q 1 ′ ) b 1 ( t , q 1 ; q 1 ′ ) ,
i ∂ ∂ t b 2 ( t , q 1 , q 2 ; q 1 ′ , q 2 ′ ) = ( − 1 2 ∑ i = 1 2 ( − Δ q i + Δ q i ′ ) + ( Φ ( q 1 ′ − q 2 ′ ) − Φ ( q 1 − q 2 ) ) ) b 2 ( t , q 1 , q 2 ; q 1 ′ , q 2 ′ ) + i\frac{\partial}{\partial t}b_2(t,q_1,q_2;q'_1,q'_2)=
\left(-\frac{1}{2}\sum_{i=1}^{2}(-\Delta_{q_i}+\Delta_{q'_i})+\left(\Phi(q'_1-q'_2)-\Phi(q_1-q_2)\right)\right)b_2(t,q_1,q_2;q'_1,q'_2)+ i ∂ t ∂ b 2 ( t , q 1 , q 2 ; q 1 ′ , q 2 ′ ) = ( − 2 1 i = 1 ∑ 2 ( − Δ q i + Δ q i ′ ) + ( Φ ( q 1 ′ − q 2 ′ ) − Φ ( q 1 − q 2 ) ) ) b 2 ( t , q 1 , q 2 ; q 1 ′ , q 2 ′ ) +
( Φ ( q 1 ′ − q 2 ′ ) − Φ ( q 1 − q 2 ) ) ( b 1 ( t , q 1 ; q 1 ′ ) + b 1 ( t , q 2 ; q 2 ′ ) ) , \left(\Phi(q'_1-q'_2)-\Phi(q_1-q_2)\right)\left(b_1(t,q_1;q'_1)+b_1(t,q_2;q'_2)\right), ( Φ ( q 1 ′ − q 2 ′ ) − Φ ( q 1 − q 2 ) ) ( b 1 ( t , q 1 ; q 1 ′ ) + b 1 ( t , q 2 ; q 2 ′ ) ) ,
Thus, in the mean-field limit, the kinetic behavior of quantum systems of many particles is described in terms of the sequence of limiting reduced observables (71) , in particular, by the sequence (72) , whose evolution is governed by the Cauchy problem of the dual Vlasov hierarchy (4.5),(4.6).
The propagation of the initial chaos
Let us consider the relationship between the kinetic behavior within the mean-field approximation for a many-particle quantum system described by the dual Vlasov hierarchy (73) for limiting reduced observables, and the Vlasov kinetic equation (67) for the state of a typical particle of the system.
Let the initial state be separable, i.e., it is specified by such a sequence of limiting reduced density operators
f ( c ) ≡ ( I , f 1 0 ( 1 ) , … , ⨂ i = 1 ∞ f 1 0 ( i ) , … ) . (75) f^{(c)}\equiv\left(I,f_1^0(1),\ldots,\bigotimes_{i=1}^{\infty}f_1^0(i),\ldots\right). \tag*{(75)} f ( c ) ≡ ( I , f 1 0 ( 1 ) , … , i = 1 ⨂ ∞ f 1 0 ( i ) , … ) . (75)
As previously noted, this assumption regarding the initial state is characteristic of the kinetic description of a gas, since in this case the state is completely determined by the one-particle density operator.
For the mean-value functional for limiting reduced additive-type observables and a separable initial state, the following equality holds on a finite time interval:
( b ( 1 ) ( t ) , f ( c ) ) = ∑ s = 0 ∞ 1 s ! Tr 1 , … , s b s ( 1 ) ( t , 1 , … , s ) ⨂ i = 1 s f 1 0 ( i ) = Tr 1 b 1 0 ( 1 ) f 1 ( t , 1 ) , \left(b^{(1)}(t),f^{(c)}\right)=\sum_{s=0}^{\infty}\frac{1}{s!}\operatorname{Tr}_{1,\ldots,s}b_s^{(1)}(t,1,\ldots,s)\bigotimes_{i=1}^{s}f_1^0(i)=\operatorname{Tr}_1 b_1^0(1)f_1(t,1), ( b ( 1 ) ( t ) , f ( c ) ) = s = 0 ∑ ∞ s ! 1 Tr 1 , … , s b s ( 1 ) ( t , 1 , … , s ) i = 1 ⨂ s f 1 0 ( i ) = Tr 1 b 1 0 ( 1 ) f 1 ( t , 1 ) ,
where the limiting one-particle density operator is represented as a perturbation theory series (66)
f 1 ( t , 1 ) = ∑ n = 0 ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n Tr 2 , … , n + 1 G 1 ∗ ( t − t 1 , 1 ) N i n t ∗ ( 1 , 2 ) ∏ j 1 = 1 2 G 1 ∗ ( t 1 − t 2 , j 1 ) … f_1(t,1)=\sum_{n=0}^{\infty}\int_{0}^{t}dt_1\ldots\int_{0}^{t_{n-1}}dt_n\operatorname{Tr}_{2,\ldots,n+1}\mathcal{G}_1^*(t-t_1,1)\mathcal{N}_{\mathrm{int}}^*(1,2)\prod_{j_1=1}^{2}\mathcal{G}_1^*(t_1-t_2,j_1)\ldots f 1 ( t , 1 ) = n = 0 ∑ ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n Tr 2 , … , n + 1 G 1 ∗ ( t − t 1 , 1 ) N int ∗ ( 1 , 2 ) j 1 = 1 ∏ 2 G 1 ∗ ( t 1 − t 2 , j 1 ) …
∏ i n = 1 n G 1 ∗ ( t n − 1 − t n , i n ) ∑ k n = 1 n N i n t ∗ ( k n , n + 1 ) ∏ j n = 1 n + 1 G 1 ∗ ( t n , j n ) ⨂ i = 1 n + 1 f 1 0 ( i ) . (76) \prod_{i_n=1}^{n}\mathcal{G}_1^*(t_{n-1}-t_n,i_n)\sum_{k_n=1}^{n}\mathcal{N}_{\mathrm{int}}^*(k_n,n+1)\prod_{j_n=1}^{n+1}\mathcal{G}_1^*(t_n,j_n)\bigotimes_{i=1}^{n+1}f_1^0(i). \tag*{(76)} i n = 1 ∏ n G 1 ∗ ( t n − 1 − t n , i n ) k n = 1 ∑ n N int ∗ ( k n , n + 1 ) j n = 1 ∏ n + 1 G 1 ∗ ( t n , j n ) i = 1 ⨂ n + 1 f 1 0 ( i ) . (76)
The one-particle density operator represented by the series (76) satisfies the quantum Vlasov kinetic equation (67) .
Thus, the hierarchy of evolution equations (73) for limiting reduced observables of the additive-type and for a separable initial state (75) describes the evolution of quantum many-particle systems in a way equivalent to the Vlasov quantum kinetic equation (67) .
Accordingly, for the mean-value functional of the limiting reduced observables of nonadditive-type (72) and a separable initial state (75) , the following equality holds on a finite time interval:
( b ( k ) ( t ) , f ( c ) ) = ∑ s = 0 ∞ 1 s ! Tr 1 , … , s b s ( k ) ( t , 1 , … , s ) ∏ i = 1 s f 1 0 ( i ) = 1 k ! Tr 1 , … , k b k 0 ( 1 , … , k ) ⨂ i = 1 k f 1 ( t , i ) , k ≥ 2 , (77) \left(b^{(k)}(t),f^{(c)}\right)=\sum_{s=0}^{\infty}\frac{1}{s!}\operatorname{Tr}_{1,\ldots,s}b_s^{(k)}(t,1,\ldots,s)\prod_{i=1}^{s}f_1^0(i)=
\frac{1}{k!}\operatorname{Tr}_{1,\ldots,k}b_k^0(1,\ldots,k)\bigotimes_{i=1}^{k}f_1(t,i),\quad k\geq2, \tag*{(77)} ( b ( k ) ( t ) , f ( c ) ) = s = 0 ∑ ∞ s ! 1 Tr 1 , … , s b s ( k ) ( t , 1 , … , s ) i = 1 ∏ s f 1 0 ( i ) = k ! 1 Tr 1 , … , k b k 0 ( 1 , … , k ) i = 1 ⨂ k f 1 ( t , i ) , k ≥ 2 , (77)
where the one-particle reduced density operator is represented as a perturbation theory series (76) , i.e., satisfies the Cauchy problem for the Vlasov quantum kinetic equation (67) ,(68) . Equality (77) describes the propagation of initial chaos (75) in the mean-field limit via the solution (76) to the quantum Vlasov kinetic equation (67) . Specifically, for any finite time interval, the state is represented by a sequence of limiting reduced operators:
f k ( t , 1 , … , k ) = ⨂ i = 1 k f 1 ( t , i ) , k ≥ 2. (78) f_k(t,1,\ldots,k)=\bigotimes_{i=1}^{k}f_1(t,i),\quad k\geq2. \tag*{(78)} f k ( t , 1 , … , k ) = i = 1 ⨂ k f 1 ( t , i ) , k ≥ 2. (78)
That is, if there are no correlations in the particle system at the initial time, then within this approximation, correlations between particle states do not arise during the evolution process, and individual particle states evolve according to the kinetic equation (67) .
On the propagation of initial correlations
It should be noted that the approach to deriving kinetic equations discussed above allows for their formulation for more general initial states, describing not only gases of quantum particles but also many-particle systems in condensed states.
Let the initial state of a quantum system of many particles be specified by a one-particle density operator and correlation operators [41 ]
f ( c c ) ≡ ( I , f 1 0 ( 1 ) , ⨂ i = 1 2 f 1 0 ( i ) ⊗ g 2 0 ( 1 , 2 ) , … , ⨂ i = 1 n f 1 0 ( i ) ⊗ g n 0 ( 1 , … , n ) , … ) . (79) f^{(\mathrm{cc})}\equiv\left(I,f_1^0(1),\bigotimes_{i=1}^{2}f_1^0(i)\otimes g_2^0(1,2),\ldots,\bigotimes_{i=1}^{n}f_1^0(i)\otimes g_n^0(1,\ldots,n),\ldots\right). \tag*{(79)} f ( cc ) ≡ ( I , f 1 0 ( 1 ) , i = 1 ⨂ 2 f 1 0 ( i ) ⊗ g 2 0 ( 1 , 2 ) , … , i = 1 ⨂ n f 1 0 ( i ) ⊗ g n 0 ( 1 , … , n ) , … ) . (79)
where the correlations of the initial states of the particles are specified by the operators g n 0 ∈ L 1 ( H ⊗ n ) g_n^0\in\mathcal{L}^{1}(\mathcal{H}^{\otimes n}) g n 0 ∈ L 1 ( H ⊗ n ) , n ≥ 2 n\geq2 n ≥ 2 . It should be emphasized that this assumption (79) regarding the initial state is characteristic of the kinetic description of many-particle systems in condensed states, which are inherently marked by the presence of correlations.
Then, using a method for deriving kinetic equations based on the hierarchy of kinetic equations for reduced observables, we establish that, for an arbitrary finite time interval, the state is described by a sequence f ( t ) = ( I , f 1 ( t ) , … , f n ( t , 1 , … , n ) , … ) f(t)=\left(I,f_1(t),\ldots,f_n(t,1,\ldots,n),\ldots\right) f ( t ) = ( I , f 1 ( t ) , … , f n ( t , 1 , … , n ) , … ) of limiting reduced density operators, with the one-particle density operator being represented by a corresponding perturbation theory series expansion
f 1 ( t , 1 ) = ∑ n = 0 ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n Tr 2 , … , n + 1 ∏ i 1 = 1 1 G 1 ∗ ( t − t 1 , i 1 ) N i n t ∗ ( 1 , 2 ) ∏ j 1 = 1 2 G 1 ∗ ( t 1 − t 2 , j 1 ) … ∏ i n = 1 n G 1 ∗ ( t n − t n , i n ) ∑ k n = 1 n N i n t ∗ ( k n , n + 1 ) ∏ j n = 1 n G 1 ∗ ( t n , j n ) ⨂ i = 1 n + 1 f 1 0 ( i ) ⊗ g n + 1 0 ( 1 , … , n + 1 ) , (80) \begin{aligned}
f_1(t,1)={}&\sum_{n=0}^{\infty}\int_0^t dt_1\ldots\int_0^{t_{n-1}}dt_n\operatorname{Tr}_{2,\ldots,n+1}\prod_{i_1=1}^{1}G_1^*(t-t_1,i_1)\mathcal{N}_{\mathrm{int}}^*(1,2)\prod_{j_1=1}^{2}G_1^*(t_1-t_2,j_1)\ldots\\
&\prod_{i_n=1}^{n}G_1^*(t_n-t_n,i_n)\sum_{k_n=1}^{n}\mathcal{N}_{\mathrm{int}}^*(k_n,n+1)\prod_{j_n=1}^{n}G_1^*(t_n,j_n)\bigotimes_{i=1}^{n+1}f_1^0(i)\otimes g_{n+1}^0(1,\ldots,n+1), \tag*{(80)}
\end{aligned} f 1 ( t , 1 ) = n = 0 ∑ ∞ ∫ 0 t d t 1 … ∫ 0 t n − 1 d t n Tr 2 , … , n + 1 i 1 = 1 ∏ 1 G 1 ∗ ( t − t 1 , i 1 ) N int ∗ ( 1 , 2 ) j 1 = 1 ∏ 2 G 1 ∗ ( t 1 − t 2 , j 1 ) … i n = 1 ∏ n G 1 ∗ ( t n − t n , i n ) k n = 1 ∑ n N int ∗ ( k n , n + 1 ) j n = 1 ∏ n G 1 ∗ ( t n , j n ) i = 1 ⨂ n + 1 f 1 0 ( i ) ⊗ g n + 1 0 ( 1 , … , n + 1 ) , (80)
and the limiting reduced density operators f n ( t , 1 , … , n ) f_n(t,1,\ldots,n) f n ( t , 1 , … , n ) , n ≥ 2 n \ge2 n ≥ 2 , are determined by the following expressions:
f n ( t , 1 , … , n ) = ∏ i 1 = 1 n G 1 ∗ ( t , i 1 ) g n 0 ( 1 , … , n ) ⊗ ∏ i 2 = 1 n ( G 1 ∗ ) − 1 ( t , i 2 ) ⊗ ⨂ j = 1 n f 1 ( t , j ) , n ≥ 2 , (81) f_n(t,1,\ldots,n)=\prod_{i_1=1}^{n}\mathcal{G}_1^*(t,i_1)g_n^0(1,\ldots,n)\otimes\prod_{i_2=1}^{n}(\mathcal{G}_1^*)^{-1}(t,i_2)\otimes\bigotimes_{j=1}^{n}f_1(t,j),\qquad n\ge2, \tag*{(81)} f n ( t , 1 , … , n ) = i 1 = 1 ∏ n G 1 ∗ ( t , i 1 ) g n 0 ( 1 , … , n ) ⊗ i 2 = 1 ∏ n ( G 1 ∗ ) − 1 ( t , i 2 ) ⊗ j = 1 ⨂ n f 1 ( t , j ) , n ≥ 2 , (81)
where the inverse operator group to group (9) is denoted by ( G 1 ∗ ) − 1 ( t ) (\mathcal{G}_1^*)^{-1}(t) ( G 1 ∗ ) − 1 ( t ) .
Indeed, for the mean-value functionals of the limiting k k k -ary reduced observables (72) and the initial state specified by the one-particle density operator and the correlation operators (79) , the following representation holds on a finite time interval:
( b ( k ) ( t ) , f ( c c ) ) = 1 k ! Tr 1 , … , k b k 0 ( 1 , … , k ) f k ( t , 1 , … , k ) , k ≥ 2. \left(b^{(k)}(t),f^{(cc)}\right)=\frac{1}{k!}\operatorname{Tr}_{1,\ldots,k}b_k^0(1,\ldots,k)f_k(t,1,\ldots,k),\qquad k\ge2. ( b ( k ) ( t ) , f ( cc ) ) = k ! 1 Tr 1 , … , k b k 0 ( 1 , … , k ) f k ( t , 1 , … , k ) , k ≥ 2.
For initial states specified by the sequence of limiting correlation operators (79) , the process of initial correlation propagation is described by the following sequence of correlation operators:
g n ( t , 1 , … , n ) = ∏ i 1 = 1 n G 1 ∗ ( t , i 1 ) ∑ P : ( 1 , … , n ) = ∪ i X i ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 ) ! ⨂ X i ⊂ P g ∣ X i ∣ 0 ( X i ) ⊗ g_n(t,1,\ldots,n)=\prod_{i_1=1}^{n}\mathcal{G}_1^*(t,i_1)\sum_{\mathrm{P}:(1,\ldots,n)=\cup_i X_i}(-1)^{|\mathrm{P}|-1}(|\mathrm{P}|-1)!\bigotimes_{X_i\subset\mathrm{P}}g_{|X_i|}^0(X_i)\otimes g n ( t , 1 , … , n ) = i 1 = 1 ∏ n G 1 ∗ ( t , i 1 ) P : ( 1 , … , n ) = ∪ i X i ∑ ( − 1 ) ∣ P ∣ − 1 ( ∣ P ∣ − 1 )! X i ⊂ P ⨂ g ∣ X i ∣ 0 ( X i ) ⊗
∏ i 2 = 1 n ( G 1 ∗ ) − 1 ( t , i 2 ) ⨂ j = 1 n f 1 ( t , j ) , n ≥ 2. \prod_{i_2=1}^{n}(\mathcal{G}_1^*)^{-1}(t,i_2)\bigotimes_{j=1}^{n}f_1(t,j),\qquad n\ge2. i 2 = 1 ∏ n ( G 1 ∗ ) − 1 ( t , i 2 ) j = 1 ⨂ n f 1 ( t , j ) , n ≥ 2.
The limiting one-particle density operator (80) is governed by the quantum Vlasov kinetic equation with initial correlations:
∂ ∂ t f 1 ( t , 1 ) = N 1 ∗ ( 1 ) f 1 ( t , 1 ) + (82) \frac{\partial}{\partial t}f_1(t,1)=\mathcal{N}_1^*(1)f_1(t,1)+ \tag*{(82)} ∂ t ∂ f 1 ( t , 1 ) = N 1 ∗ ( 1 ) f 1 ( t , 1 ) + (82)
Tr 2 N i n t ∗ ( 1 , 2 ) ∏ i 1 = 1 2 G 1 ∗ ( t , i 1 ) g 2 0 ( 1 , 2 ) ⊗ ∏ i 2 = 1 2 ( G 1 ∗ ) − 1 ( t , i 2 ) f 1 ( t , 1 ) ⊗ f 1 ( t , 2 ) , \operatorname{Tr}_2\mathcal{N}_{\mathrm{int}}^*(1,2)\prod_{i_1=1}^{2}\mathcal{G}_1^*(t,i_1)g_2^0(1,2)\otimes\prod_{i_2=1}^{2}(\mathcal{G}_1^*)^{-1}(t,i_2)f_1(t,1)\otimes f_1(t,2), Tr 2 N int ∗ ( 1 , 2 ) i 1 = 1 ∏ 2 G 1 ∗ ( t , i 1 ) g 2 0 ( 1 , 2 ) ⊗ i 2 = 1 ∏ 2 ( G 1 ∗ ) − 1 ( t , i 2 ) f 1 ( t , 1 ) ⊗ f 1 ( t , 2 ) ,
f 1 ( t , 1 ) ∣ t = 0 = f 1 0 ( 1 ) , (83) f_1(t,1)\big|_{t=0}=f_1^0(1), \tag*{(83)} f 1 ( t , 1 ) t = 0 = f 1 0 ( 1 ) , (83)
where the notation adopted in equation (67) is used.
It should be noted that the resulting kinetic equation (82) is non-Markovian. For pure states, kinetic equation (82) reduces to the Hartree equation with initial correlations. In the case of a separable initial state (75) , kinetic equation (82) coincides with the quantum Vlasov kinetic equation (67) , and the reduced density operators (81) describe the propagation of initial chaos (78) .
We also note that, for arbitrary initial states in the mean-field limit, the sequence of limiting reduced density operators, as well as the sequence (80) ,(81) , is a solution to the Cauchy problem for the Vlasov hierarchy:
∂ ∂ t f s ( t , 1 , … , s ) = ∑ i = 1 s N 1 ∗ ( i ) f s ( t , 1 , … , s ) + \frac{\partial}{\partial t}f_s(t,1,\ldots,s)=\sum_{i=1}^{s}\mathcal{N}_1^*(i)f_s(t,1,\ldots,s)+ ∂ t ∂ f s ( t , 1 , … , s ) = i = 1 ∑ s N 1 ∗ ( i ) f s ( t , 1 , … , s ) +
∑ i = 1 s Tr s + 1 N i n t ∗ ( i , s + 1 ) f s + 1 ( t , 1 , … , s + 1 ) , \sum_{i=1}^{s}\operatorname{Tr}_{s+1}\mathcal{N}_{\mathrm{int}}^*(i,s+1)f_{s+1}(t,1,\ldots,s+1), i = 1 ∑ s Tr s + 1 N int ∗ ( i , s + 1 ) f s + 1 ( t , 1 , … , s + 1 ) ,
f s ( t , 1 , … , s ) ∣ t = 0 = f s 0 ( 1 , … , s ) , s ≥ 1 , f_s(t,1,\ldots,s)\big|_{t=0}=f_s^0(1,\ldots,s),\qquad s\ge1, f s ( t , 1 , … , s ) t = 0 = f s 0 ( 1 , … , s ) , s ≥ 1 ,
where the operators N 1 ∗ \mathcal{N}^{*}_{1} N 1 ∗ and N i n t ∗ \mathcal{N}^{*}_{\mathrm{int}} N int ∗ are defined according to formulas (10) . This hierarchy of evolution equations is dual to the hierarchy (73) in the sense of the mean-value functional for the limiting reduced observables (71) .
Thus, for initial states specified by a one-particle density operator and correlation operators (79) , the dual Vlasov hierarchy (73) for the limiting reduced additive-type observables describes the evolution of quantum many-particle systems, just as the non-Markovian Vlasov quantum kinetic equation with the initial correlations (82) does. In the case of limiting reduced nonadditive-type observables, the dual Vlasov hierarchy (73) describes equivalently, in the sense of equality (81) , the propagation of initial correlations in terms of reduced density operators (81) . In other words, the alternative method of describing the evolution of the state of quantum systems in the mean-field approximation is based on the non-Markovian Vlasov kinetic equation with initial correlations (82) . The results can be extended to many-body bosonic or fermionic systems, as in [34 ] , to study the dynamics of correlations.
Conclusion and outlook
This article examines the evolution equations of quantum many-particle systems and, in essence, classifies them. Hierarchies of such evolution equations describe the evolution of observables and the state of systems with both finite and infinite numbers of particles, which underlie the description of kinetic evolution.
The article introduces the concepts of cluster (37) ,(42) and cumulant (39) ,(44) expansions of operator groups, which underlie the description of the evolution of observables (38) and the state (43) , respectively, in quantum many-particle systems. The generating operators for the non-perturbative representation of solutions to the hierarchies of evolution equations for the state (44) and for the hierarchy of evolution equations for observables (38) are determined, respectively, by the cumulants (semi-invariants) of dual operator groups (9) and the cumulants of operator groups (2) for quantum systems with a finite number of particles.
Non-perturbative solutions to the Cauchy problem for hierarchies of evolution equations for the state have clarified their relationship (45) and (51) to the dynamics of correlations (22) in quantum particle systems. In passing, laws (63) and (23) governing processes of correlation creation from chaos were established.
The notion of cumulants of operator groups also underlies the kinetic description of the collective behavior of quantum systems with an infinite number of particles. The methods for deriving kinetic equations using non-perturbative solutions of hierarchies of evolution equations of quantum many-particle systems are based on the notion of kinetic cluster expansions (57) of cumulants of operator groups.
For separable initial states (55) , we established the equivalence of the Cauchy problems for the hierarchies of evolution equations for reduced density operators and reduced observables to the Cauchy problem for the generalized quantum kinetic equation (61) and a sequence of correlation functionals (59) describing the propagation of correlations in quantum many-particle systems. Proving this result involves applying kinetic cluster expansions (57) to the generating operators of a non-perturbative representation of solutions of these hierarchies of evolution equations.
This article also addresses the challenge of describing kinetic evolution in terms of observables (73) , in contrast to the conventional approach based on describing the effective evolution of the state of many quantum particles. One of the advantages of the developed approach is the opportunity to construct kinetic equations (82) that account for particle correlations in the initial state, as well as to describe the propagation of initial correlations (81) within scaling approximations. An alternative perspective on the nature of kinetic equations based on describing the evolution of correlations in systems of many quantum particles is also presented. The approach describes the evolution of a state through the von Neumann hierarchy of evolution equations (27) for a sequence of cumulants of the density operators (19) .
Acknowledgments. The key findings of this article were discussed during the Year of Quantum Science. The author expresses heartfelt gratitude to the organizers of StatPhys29 for the inspiring conference atmosphere and financial support during my visit to Florence. The author also thanks the Isaac Newton Institute (INI) and the London Mathematical Society (LMS) for their financial support, which made essential workshops and meetings possible, as well as the School of Mathematics and Physics at the University of Surrey for their warm hospitality.