Welcome to visit Communications in Theoretical Physics,
Quantum Physics and Quantum Information

Wick’s theorem for superoperators linear in bosonic operators and its applications

  • Yang Yang 1, 2 ,
  • Rui Zhang 1, 2 ,
  • Jian Ma 3 ,
  • Zhaoyu Fei 2 ,
  • Xiaoguang Wang , 2, *
Expand
  • 1School of Physics, Zhejiang University, Hangzhou 310027, China
  • 2Zhejiang Key Laboratory of Quantum State Control and Optical Field Manipulation, Department of Physics, Zhejiang Sci-Tech University, 310018 Hangzhou, China
  • 3Shenzhen Jingtai Technology Co., Ltd. (XtalPi), Fubao Community, Shenzhen 518045, China

*Author to whom any correspondence should be addressed.

Received date: 2026-03-26

  Revised date: 2026-05-12

  Accepted date: 2026-05-13

  Online published: 2026-06-30

Copyright

© 2026 Institute of Theoretical Physics CAS, Chinese Physical Society and IOP Publishing. All rights, including for text and data mining, AI training, and similar technologies, are reserved.
This article is available under the terms of the IOP-Standard License.

Abstract

For the superoperators that are linear in bosonic operators, we prove Wick’s theorem rigorously by applying stable partition theory. Higher-order time-ordered superoperator terms can be decomposed into sums of products of second-order time-ordered terms. By applying the theorem, explicit expressions for all even-order terms of the reduced density matrix are derived, with vanishing odd-order terms. We obtain the set of equations for the reduced dynamics of a system interacting with a bosonic bath initially in thermal equilibrium.

Cite this article

Yang Yang , Rui Zhang , Jian Ma , Zhaoyu Fei , Xiaoguang Wang . Wick’s theorem for superoperators linear in bosonic operators and its applications[J]. Communications in Theoretical Physics, 2026 , 78(9) : 095101 . DOI: 10.1088/1572-9494/ae6ccb

1. Introduction

Wick’s theorem is a powerful mathematical tool in quantum field theory, providing a systematic method for evaluating time-ordered products of field operators [1, 2]. By expressing such products in terms of normal-ordered components and contractions, the theorem enables the derivation of Feynman rules and facilitates the computation of scattering amplitudes [37]. In finite-temperature field theory, Wick’s theorem plays a central role in computing thermal Green’s functions, which are essential for analyzing equilibrium and nonequilibrium thermodynamic properties [811]. Furthermore, in statistical mechanics and condensed matter systems, it is routinely employed to decompose multi-point correlation functions into sums of products of two-point correlation functions, thereby enabling analytical progress in the study of interacting many-body systems [1217]. Further applications include the evaluation of quantum coherence [18], the characterization of order parameters [19], and the analysis of spin correlation functions [20].
Beyond the standard formulation, Wick’s theorem for superoperators has been proposed in Liouville space, avoiding the need for backward propagations and analytic continuations using artificial times required in Hilbert space [21]. Wick’s theorem for superoperators allows the factorization of higher-order response functions into products of two fundamental Green’s functions in Liouville space [21]. This method motivates us to propose a derivation of Wick’s theorem for superoperators in the study of reduced dynamics which involves higher-order superoperators based on operator algebra.
In this work, we prove Wick’s theorem for superoperators that are linear in bosonic operators rigorously. This theorem enables the exact solution of the reduced dynamics for systems interacting with a bosonic bath. Our approach employs superoperators labeled by upper indices to explicitly record whether system and bath operators multiply the density matrix from the left or the right, thus making their relative ordering with respect to the density matrix manifest. We apply the stable partition theory, which rearranges a sequence while preserving the relative ordering of elements with identical labels [22]. Using this framework, we show that any higher-order time-ordered superoperator term can be decomposed into sums of products of second-order terms. We derive explicit expressions for all even-order terms of the reduced density matrix, while all odd-order terms vanish. Higher-order terms of reduced density matrix can be expressed in terms of second-order terms. Consequently, the reduced dynamics can be constructed by summing these even-order contributions, yielding a formally exact expression.
In section 2, we study the reduced dynamics of the system coupled to a bosonic bath and highlight the challenge of evaluating time-ordered higher-order superoperator terms. In section 3, we prove Wick’s theorem for superoperators rigorously by applying a stable partition theory and demonstrate how higher-order time-ordered superoperators can be systematically decomposed into second-order terms. In section 4, we apply Wick’s theorem for superoperators, leading to an exact expression for the reduced density matrix. In section 5, we make our conclusions.

2. Reduced dynamics

We consider the system interacting with a bosonic bath, described by the total Hamiltonian
$\begin{align} H = H_{S}+H_{B}+H_{SB},\end{align}$
where $H_{S}$ and $H_{B}$ are the free Hamiltonians for the system and bath, respectively. The system-bath interaction Hamiltonian $H_{SB}$ in the interaction picture becomes
$\begin{align} H_{SB}\left(t\right) = \textrm{e}^{\frac{\textrm{i}}{\hbar}\left(H_{S}+H_{B}\right)t}H_{SB}\textrm{e}^{- \frac{\textrm{i}}{\hbar}\left(H_{S}+H_{B}\right)t},\end{align}$
and the total density matrix in the interaction picture is
$\begin{align} \tilde{\rho}_{\text{tot}}\left(t\right) = &\textrm{e}^{\frac{\textrm{i}}{\hbar} \left(H_{S}+H_{B}\right)t}\rho_{\text{tot}}\left(t\right)\textrm{e}^{-\frac{\textrm{i}}{\hbar} \left(H_{S}+H_{B}\right)t} \nonumber\\ = &\textrm{e}^{\frac{\textrm{i}}{\hbar}\left(H_{S}+H_{B}\right)t^{ \times}}\rho_{\text{tot}}\left(t\right),\end{align}$
where the superoperator is defined by $A^{\times}B\equiv\left[A,\, B\right]$ and ${\rho}_{\text{tot}}$ is the total density matrix in the Schrödinger picture. The time evolution of $\tilde{\rho}_{\text{tot}}\left(t\right)$ follows the equation
$\begin{align} \frac{\partial}{\partial t}\tilde{\rho}_{\text{tot}}\left(t\right) = -\frac{\textrm{i}}{ \hbar}\left[H_{SB}\left(t\right),\tilde{\rho}_{\text{tot}}\left(t\right) \right].\end{align}$
Therefore, $\tilde{\rho}_{\text{tot}}\left(t\right)$ can be written as
$\begin{align} \tilde{\rho}_{\text{tot}}\left(t\right) = & \tilde{\rho}_{\text{tot} }\left(0\right)-\frac{\textrm{i}}{\hbar}\int_{0}^{t}\textrm{d}\tau_{1}\, H_{SB}^{\times}\left(\tau_{1}\right)\tilde{\rho}_{\text{tot}}\left(0\right) \nonumber\\ & +\dots+\left(-\frac{\textrm{i}}{\hbar}\right)^{n}\int_{0}^{t}\textrm{d}\tau_{n}\dots \int_{0}^{\tau_{2}}\textrm{d}\tau_{1}\, \nonumber\\ &\times H_{SB}^{\times}\left(\tau_{n}\right)\dots H_{SB}^{\times}\left(\tau_{1}\right)\tilde{\rho}_{\text{tot}}\left(0\right) \nonumber\\ = & \sum_{n = 0}^{\infty}\frac{1}{n!}\mathcal{T}\left[-\frac{\textrm{i}}{\hbar} \int_{0}^{t}\textrm{d}\tau\, H_{SB}^{\times}\left(\tau\right)\right]^{n}\tilde{\rho} _{\text{tot}}\left(0\right),\end{align}$
where $\mathcal{T}$ is the time-ordering operator.
In general, we assume a factorized initial state of the system and bath, $\rho_{ \text{tot}}\left(0\right) = \rho_{S}\left(0\right)\rho_{B}\left(0\right)$. Substituting this into equation (5) and tracing out the bath degrees of freedom, $ \tilde{\rho}_{S}\left(t\right) = \text{Tr}_{B}\left[\tilde{\rho}_{\text{tot}}\left(t\right)\right] $, we obtain the reduced density matrix as follows:
$\begin{align} \tilde{\rho}_{S}\left(t\right) = & \sum_{n = 0}^{\infty}\varrho_{n}\left(t\right),\end{align}$
where the $n$th order contribution $\varrho_{n}$ is given by
$\begin{align} \varrho _{n}\left( t\right) = &\frac{\left(-\textrm{i}\right)^n}{n!\hbar ^{n}}\int_{0}^{t}\textrm{d}\tau _{n}\dots\int_{0}^{t}\textrm{d}\tau _{1} \nonumber\\ &\times\text{Tr}_{B}\left[ \mathcal{T} H_{SB}^{\times}\left(\tau_n\right)\dots H_{SB}^{\times}\left(\tau_1\right) \rho_{B}\left( 0\right) \rho _{S}\left( 0\right) \right].\end{align}$
The system-bath interaction is bilinear,
$\begin{align} H_{SB}\left(t\right) = V\left(t\right)B\left(t\right),\end{align}$
where $V\left(t\right)$ and $B\left(t\right)$ are the system and bath operators, respectively. The system operator $V\left(t\right)$ could be generalized coordinates or spin operators. Below, we take the bath operator $B\left(t\right)$ to have the form
$\begin{align} B\left(t\right) = \sum_{k}\left[\xi_{k}\left(t\right)a_{k}^{\dagger}+ \xi_{k}^{*}\left(t\right)a_{k}\right],\end{align}$
where $a_{k}$ and $a_{k}^{\dagger}$ are the annihilation and creation bosonic operators in mode $k$. $\xi_{k}(t)$ is the complex coefficient that determines the contribution of each mode $k$ to $B(t)$. To perform the partial trace of the bath, we assume the bath is in thermal equilibrium. The density matrix of the bath is $\rho_{B} = \frac{\exp\left(-\beta H_{B}\right)}{\text{Tr} \exp\left(-\beta H_{B}\right)},$ where $ H_B = \sum_{k}\omega _{k}a_{k}^{\dagger }a_{k}$. Since $\text{Tr}\left[a\rho_{B}\right] = \text{Tr}\left[a^\dagger\rho_{B}\right] = 0$ and more generally $\text{Tr}\left[(a+a^\dagger)^{2n+1}\rho_{B}\right] = 0$, we conclude that all odd-order terms $\varrho_{2n+1}(t) = 0$. Thus, we are only required to compute the even-order terms $\varrho_{2n}(t)$.
To express the commutation structure resulting from the form of the interaction Hamiltonian, we introduce the following superoperators:
$\begin{align} L_{S}^{\left( 1\right) }\rho _{S} = &V\rho _{S},\;L_{S}^{\left( 2\right) }\rho _{S} = -\rho _{S}V, \nonumber\\ L_{B}^{\left( 1\right) }\rho _{B} = &B\rho _{B},\;L_{B}^{\left( 2\right) }\rho _{B} = \rho _{B}B.\end{align}$
Then $H_{SB}^{\times}$ acting on the density matrix can be written as
$\begin{align}H_{SB}^{\times }\rho = \sum_{i = 1}^{2}\left( L_{S}^{\left( i\right) }\rho _{S}\right) \otimes \left( L_{B}^{\left( i\right) }\rho _{B}\right).\end{align}$
Substituting these definitions into equation (7), the second-order contribution $\varrho_2(t)$ becomes
$\begin{align} \varrho _{2}\left( t\right) = &-\frac{1}{2\hbar ^{2}}\int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{t}\textrm{d}\tau _{1} \sum_{i,j = 1}^{2}\text{Tr}_{B}\left[ \mathcal{T}_{B} \left(L_{B_{2}}^{\left( i\right)}L_{B_{1}}^{\left( j\right)}\right)\rho_{B}\right]\nonumber\\ &\times\left[\mathcal{T}_{S}\left( L_{S_{2}}^{\left( i\right)}L_{S_{1}}^{\left( j\right)}\right)\rho _{S}\left( 0\right) \right],\end{align}$
where $\mathcal{T}_{S}$ and $\mathcal{T}_{B}$ are the time-ordering operators for the system and bath, respectively. Instead of integrating over the full domain $0 \unicode{x2A7D} \tau_1, \tau_2 \unicode{x2A7D} t$ with the time-ordering operator, which requires the use of Heaviside step functions to enforce proper ordering, we simplify the calculation by restricting the integration to the ordered domain $0 \unicode{x2A7D} \tau_1 \unicode{x2A7D} \tau_2 \unicode{x2A7D} t$ after removing the time-ordering operator. Thus, we have
$\begin{align} \varrho _{2}\left( t\right) = &-\frac{1}{\hbar ^{2}}\text{Tr}_{B}\left\{ \int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{\tau _{2}}\textrm{d}\tau _{1} \right. \nonumber\\ &\times \left. \sum_{i,j = 1}^{2}\left[ L_{B_{2}}^{\left( i\right) }L_{B_{1}}^{\left( j\right) }\rho _{B} \right] \left[ L_{S_{2}}^{\left( i\right) }L_{S_{1}}^{\left( j\right) }\rho _{S}\left( 0\right) \right] \right\} .\end{align}$
For convenience, we introduce the symmetric correlation function $G\left( \tau \right) $, and the response function $ \chi \left( \tau \right) $, given by
$\begin{align}G\left( \tau _{2}-\tau _{1}\right) \equiv &\frac{1}{2}\left\langle \left\{ B_{2},\,B_{1}\right\} \right\rangle _{B} = \text{Re}\left\langle B_{2}B_{1}\right\rangle _{B}, \nonumber\\ \chi \left( \tau _{2}-\tau _{1}\right) \equiv &\frac{\textrm{i}}{\hbar }\left\langle \left[ B_{2},\,B_{1}\right] \right\rangle _{B} = -\frac{2}{\hbar }\text{Im} \left\langle B_{2}B_{1}\right\rangle _{B}.\end{align}$
Indeed, $G$ and $\chi $ are related to the real and imaginary parts of the bath time correlation function $C\left( \tau _{2}-\tau _{1}\right) = \left\langle B_{2}B_{1}\right\rangle _{B}. $ Then
$\begin{align} \varrho _{2}\left( t\right) = &-\frac{1}{\hbar ^{2}}\int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{\tau _{2}}\textrm{d}\tau _{1}\,\left[ G\left( \tau _{2}-\tau _{1}\right) V_{2}^{\times }V_{1}^{\times }\right. \nonumber\\ & \left. -\textrm{i}\frac{\hbar }{2}\chi \left( \tau _{2}-\tau _{1}\right) V_{2}^{\times }V_{1}^{\circ }\right] \rho _{S}\left( 0\right) ,\end{align}$
where $V^{\circ }\rho = \left\{V,\,\rho \right\} $. Compared to second-order term $\varrho _{2}$, higher-order terms are extremely difficult to express using symmetric correlation functions and response functions directly. In the next section, we propose a method to simplify higher-order terms.

3. Wick’s theorem for superoperators

Since all odd-order contributions vanish, we focus exclusively on even-order terms. This naturally suggests applying Wick’s theorem, which is commonly employed to evaluate expectation values of time-ordered even numbers of operators [1]. However, the conventional form of Wick’s theorem, which applies to operator products, is not applicable to the superoperators involved in our formulation.
To overcome this limitation, we propose Wick’s theorem for superoperators that simplifies the analysis of higher-order contributions. This theorem states that the bath trace of any time-ordered product of $2n$ superoperators can be expressed as a sum of products over all pairings of the $2n$ superoperators into $n$ pairs, with each pair taken in internal time order,
$\begin{align} &\mathrm{Tr}_B\left[ \mathcal{T}_BL_{B_{2n}}^{\left( i_{2n}\right)}L_{B_{2n-1}}^{\left( i_{2n-1}\right) }\dots L_{B_{2}}^{\left( i_{2}\right)}L_{B_{1}}^{\left( i_{1}\right) }\rho _{B}\right] \nonumber\\ & \quad = \sum_{j = 1}^m \prod_{\left(u,v\right)\in Q_{1,j}} \mathrm{Tr}_B\left[\mathcal{T}_B\,L_{B_{u}}^{\left(i_{u}\right)}L_{B_{v}}^{\left(i_{v}\right)}\rho_B\right],\end{align}$
where $m = \frac{(2n)!}{2^nn!}$ is the total number of such pairings. Following the left-hand side of equation (16), we have the base index sequence
$\begin{align} &\mathrm{SEQ}_1 = \left[ 2n,2n-1,\dots,2,1\right].\end{align}$
Here $Q_{l,j}$ (for $l = 1,2,3,4$ and $j = 1,\dots,m$) denotes the $j$th way of pairing up all $2n$ elements of the $l$th sequence $\mathrm{SEQ}_l$ into $n$ pairs $(u_l, v_l)$. In the theorem we only consider the case $l = 1$. The indices $l = 2,3,4$ are used only to describe the sequences $\mathrm{SEQ}_l$ that appear in the proof. For simplicity, we write $u,v$ in place of $u_1, v_1$. The case $2n = 4$ is provided as an example,
$\begin{align} &\mathrm{SEQ}_1 = \left[4,3,2,1\right].\end{align}$
There are $m = 3$ ways of picking pairs,
$\begin{align} Q_{1,1}& = \begin{Bmatrix}\left(4,3\right),\left(2,1\right)\end{Bmatrix}, \nonumber\\ Q_{1,2}& = \begin{Bmatrix}\left(4,2\right),\left(3,1\right)\end{Bmatrix}, \nonumber\\ Q_{1,3}& = \begin{Bmatrix}\left(4,1\right),\left(3,2\right)\end{Bmatrix}.\end{align}$
We now consider the case of $2n$ elements. Without loss of generality, one way to pair up all elements in this sequence is
$\begin{align} Q_{1,1} = \begin{Bmatrix}\left(2n,2n-1\right),\dots,\left(4,3\right),\left(2,1\right)\end{Bmatrix}.\end{align}$
After performing the time ordering, the left side of equation (16) can be expressed as
$\begin{align} \mathrm{Tr}_B\left[L_{B_{\sigma\left(2n\right)}}^{\left(i_{\sigma\left(2n\right)}\right)}L_{B_{\sigma\left(2n-1\right)}}^{\left(i_{\sigma\left(2n-1\right)}\right)}\dots L_{B_{\sigma\left(2\right)}}^{\left(i_{\sigma\left(2\right)}\right)}L_{B_{\sigma\left(1\right)}}^{\left(i_{\sigma\left(1\right)}\right)}\rho _{B}\right],\end{align}$
where $\sigma$ is a permutation that sorts the time labels in nonincreasing order:
$\begin{align} \tau_{\sigma\left(2n\right)} \unicode{x2A7E} \tau_{\sigma\left(2n-1\right)} \unicode{x2A7E} \cdots \unicode{x2A7E} \tau_{\sigma\left(2\right)}\unicode{x2A7E} \tau_{\sigma\left(1\right)}.\end{align}$
Equation (21) involves $2n$ superoperators, among which $p$ carries upper index 2 and the remaining $d = 2n-p$ carry upper index 1. The time-ordering operator arranges these superoperators chronologically, resulting in a sequence where $L^{(1)}$ and $L^{(2)}$ terms are mixed. So from equation (21), after the action of the superoperators on the density matrix, whether each $B_\sigma$ is on the right side of $\rho$ or on the left is obscured.
To address this, we gather superoperators with the same upper indices together while keeping their relative order unchanged. This rearrangement is justified by the superoperator’s property: superoperators with different upper indices commute, whereas those with the same upper indices do not.
Formally, this process corresponds to a stable partition of the superoperator sequence based on the upper indices. A stable partition algorithm is presented in appendix A. Without a stable partition, the proof of Wick’s theorem for superoperators can hardly be presented clearly. Stable partition groups terms with the same index into contiguous blocks while preserving the original relative order within each group. In this way, we introduce a canonical reordering of a general sequence under a partial commutativity constraint: commuting factors can be permuted to form contiguous index blocks, while the relative order of non-commuting factors is preserved.
With these explicit upper indices, we can then determine the ordering of the operators resulting from the action of the superoperators in the subsequent calculation. Specifically, the sequence $\mathrm{SEQ}_1$ can be partitioned into two subsequences, $\mathrm{SUBSEQ}_1$ and $\mathrm{SUBSEQ}_2$,
$\begin{align} &\mathrm{SUBSEQ}_1 = \left[k_d,\dots,k_2,k_1\right], k_d \gt \dots \gt k_2 \gt k_1, \nonumber\\ &\mathrm{SUBSEQ}_2 = \left[k_p^{^{\prime}},\dots,k_2^{^{\prime}},k_1^{^{\prime}}\right], k_p^{^{\prime}} \gt \dots \gt k_2^{^{\prime}} \gt k_1^{^{\prime}},\end{align}$
where
$\begin{align} i_{\sigma\left(k_j\right)} = 1\;\left(1\unicode{x2A7D} j\unicode{x2A7D} d\right),\ i_{\sigma\left(k^{^{\prime}}_j\right)} = 2\;\left(1\unicode{x2A7D} j\unicode{x2A7D} p\right).\end{align}$
Here $k_{j}$ (respectively $k_{j}^{^{\prime}}$) denotes the index in the permutation $\sigma$ such that the superoperator $L_{B_{\sigma(k_{j})}}^{i_{\sigma(k_{j})}}$ $\left({\text{respectively}}L_{B_{\sigma(k^{^{\prime}}_{j})}}^{i_{\sigma(k^{^{\prime}}_{j})}}\right)$ carries upper index 1 (respectively 2). Under this reordering, equation (21) can be expressed as
$\begin{align} \mathrm{Tr}_B\left[ L_{B_{\sigma\left(k_d\right)}}^{\left(1\right)} \cdots L_{B_{\sigma\left(k_1\right)}}^{\left(1\right)} L_{B_{\sigma\left(k^{^{\prime}}_p\right)}}^{\left(2\right)} \cdots L_{B_{\sigma\left(k^{^{\prime}}_1\right)}}^{\left(2\right)} \rho_B \right].\end{align}$
This allows us to express the corresponding expression in terms of operators rather than superoperators. When these superoperators act on the density matrix, we obtain
$\begin{align} &\mathrm{Tr}_B\left[B_{\sigma\left(k_{d}\right)} \dots B_{\sigma\left(k_{1}\right)}\rho _{B} B_{\sigma\left(k_{1}^{^{\prime}}\right)}\dots B_{\sigma\left(k_{p}^{^{\prime}}\right)}\right] \nonumber\\ & \quad = \mathrm{Tr}_B\left[B_{\sigma\left(k_{1}^{^{\prime}}\right)}\dots B_{\sigma\left(k_{p}^{^{\prime}}\right)} B_{\sigma\left(k_{d}\right)} \dots B_{\sigma\left(k_{1}\right)}\rho _{B} \right].\end{align}$
Applying Wick’s theorem, we have
$\begin{align} \sum_{j = 1}^m \prod_{\left(u_{2},v_{2}\right)\in Q_{2,j}} \mathrm{Tr}_B\left[ B_{u_{2}} B_{v_{2}} \rho_B \right],\end{align}$
where
$\begin{align} \mathrm{SEQ}_2 &= \left[ \sigma\left(k_{1}^{^{\prime}}\right),\sigma\left(k_{2}^{^{\prime}}\right),\dots,\sigma\left(k_{p}^{^{\prime}}\right), \right. \nonumber\\ &\quad \left. \sigma\left(k_{d}\right),\sigma\left(k_{d-1}\right),\dots,\sigma\left(k_{1}\right) \right] .\end{align}$
However, our target is to pair superoperators. By applying the cyclic property of trace again, equation (27) can be rewritten into the superoperator representation,
$\begin{align} \sum_{j = 1}^m \prod_{\left(u_{3},v_{3}\right)\in Q_{3,j}} \mathrm{Tr}_B\!\left[ L_{B_{u_{3}}}^{\left(i_{u_{3}}\right)} L_{B_{v_{3}}}^{\left(i_{v_{3}}\right)} \rho_B \right],\end{align}$
where
$\begin{align} \mathrm{SEQ}_3 &= \left[ \sigma\left(k_{p}^{^{\prime}}\right),\dots,\sigma\left(k_{2}^{^{\prime}}\right),\sigma\left(k_{1}^{^{\prime}}\right), \right. \nonumber\\ &\quad \left. \sigma\left(k_{d}\right),\sigma\left(k_{d-1}\right),\dots,\sigma\left(k_{1}\right) \right].\end{align}$
Next, we consider the right-hand side of equation (16). By applying the time-ordering operator, we obtain
$\begin{align} \sum_{j = 1}^m \prod_{\left(u_{4},v_{4}\right)\in Q_{4,j}} \mathrm{Tr}_B\!\left[ L_{B_{u_{4}}}^{\left(i_{u_{4}}\right)} L_{B_{v_{4}}}^{\left(i_{v_{4}}\right)} \rho_B \right],\end{align}$
where
$\begin{align} &\mathrm{SEQ}_4 = \left[\sigma\left(2n\right),\dots,\sigma\left(1\right)\right].\end{align}$
After applying the stable partition operation, equation (29) is consistent with equation (31). This completes the proof of equation (16).

4. Applying Wick’s theorem for superoperators

Having established the theorem, we now apply Wick’s theorem for superoperators to evaluate the even-order contributions $\varrho_{2n}(t)$. First, we consider the fourth-order term $\varrho_{4}(t)$. From equation (7) we have
$\begin{align} \varrho _{4}\left( t\right) = &\frac{1}{4!\hbar ^{4}}\int_{0}^{t}\textrm{d}\tau _{4}\int_{0}^{t}\textrm{d}\tau _{3}\int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{t}\textrm{d}\tau_{1} \nonumber\\ &\quad\times\text{Tr}_{B} \left[ \mathcal{T} H_{SB}^{\times}\left(\tau_4\right)H_{SB}^{\times}\left(\tau_3\right)H_{SB}^{\times}\left(\tau_2\right) \right. \nonumber\\ &\quad \left. H_{SB}^{\times}\left(\tau_1\right) \rho_{B}\rho _{S}\left( 0\right) \right] .\end{align}$
The integrand can be written as
$\begin{align} &\sum_{i,j,k,l = 1}^2\mathrm{Tr}_B\left[\mathcal{T}_BL_{B_4}^{\left(i\right)}L_{B_3}^{\left(j\right)}L_{B_2}^{\left(k\right)}L_{B_1}^{\left(l\right)}\rho_B\right] \nonumber\\ &\quad \times \left[\mathcal{T}_SL_{S_4}^{\left(i\right)}L_{S_3}^{\left(j\right)}L_{S_2}^{\left(k\right)}L_{S_1}^{\left(l\right)}\right]\rho_S\left(0\right) = A_1+A_2+A_3,\end{align}$
where we have identified three terms $A_1, A_2, A_3$ corresponding to the different ways of pairing the bath superoperators. These terms are given by
$\begin{align} A_{1} &= \sum_{i,j,k,l = 1}^{2}\text{Tr}_{B}\left[\mathcal{T}_BL_{B_{4}}^{\left(i \right)}L_{B_{3}}^{\left(j\right)}\rho_{B}\right]\text{Tr}_{B} \left[\mathcal{T}_BL_{B_{2}}^{\left(k\right)}L_{B_{1}}^{\left(l\right)}\rho_{B}\right] \nonumber\\ &\quad\times\left[\mathcal{T}_SL_{S_{4}}^{\left(i\right)}L_{S_{3}}^{\left(j\right)}L_{S_{2}}^{ \left(k\right)}L_{S_{1}}^{\left(l\right)}\right]\rho_{S}\left(0\right),\end{align}$
$\begin{align} A_{2} & = \sum_{i,j,k,l = 1}^{2}\text{Tr}_{B}\left[\mathcal{T}_BL_{B_{4}}^{\left(i \right)}L_{B_{2}}^{\left(k\right)}\rho_{B}\right]\text{Tr}_{B}\left[\mathcal{ T}_BL_{B_{3}}^{\left(j\right)}L_{B_{1}}^{\left(l\right)}\rho_{B}\right] \nonumber\\ &\quad\times\left[\mathcal{T}_SL_{S_{4}}^{\left(i\right)}L_{S_{3}}^{\left(j\right)}L_{S_{2}}^{ \left(k\right)}L_{S_{1}}^{\left(l\right)}\right]\rho_{S}\left(0\right),\end{align}$
$\begin{align}A_{3} &= \sum_{i,j,k,l = 1}^{2}\text{Tr}_{B}\left[\mathcal{T}_BL_{B_{4}}^{\left(i \right)}L_{B_{1}}^{\left(l\right)}\rho_{B}\right]\text{Tr}_{B}\left[\mathcal{ T}_BL_{B_{3}}^{\left(j\right)}L_{B_{2}}^{\left(k\right)}\rho_{B}\right] \nonumber\\ &\quad\times\left[\mathcal{T}_SL_{S_{4}}^{\left(i\right)}L_{S_{3}}^{\left(j\right)}L_{S_{2}}^{ \left(k\right)}L_{S_{1}}^{\left(l\right)}\right]\rho_{S}\left(0\right).\end{align}$
Up to this point, we have factorized the bath superoperator contributions using Wick’s theorem for superoperators. We now take into account the system superoperators. Importantly, system superoperators can be reordered freely without altering the result because under the time-ordering operator, system superoperators commute. Therefore, the system parts of $A_1$, $A_2$, and $A_3$ can be rearranged to align with the corresponding bath superoperator orderings. Implementing this reordering, $A_1$ can be written as
$\begin{align} A_1& = \mathcal{T}_{S}\left\{\sum_{i,j = 1}^{2}\text{Tr}_{B}\left[\mathcal{T}_B L_{B_{4}}^{\left(i\right)}L_{B_{3}}^{\left(j\right)}\rho_{B}\right] L_{S_{4}}^{\left(i\right)}L_{S_{3}}^{\left(j\right)}\right\} \nonumber\\ &\quad\times\left\{ \sum_{k,l = 1}^{2}\text{Tr}_{B}\left[\mathcal{T}_BL_{B_{2}}^{\left(k \right)}L_{B_{1}}^{\left(l\right)}\rho_{B}\right]L_{S_{2}}^{\left(k \right)}L_{S_{1}}^{\left(l\right)}\right\} \rho_{S}\left(0\right),\end{align}$
and $A_2,A_3$ can be written analogously. The order of integration can be exchanged to align with the superoperator order without altering the result. Therefore, the integrals for $A_1$, $A_2$, and $A_3$ are identical, since they differ only in dummy indices,
$\begin{align} \varrho _{4}\left( t\right) = &\frac{1}{4!\hbar ^{4}}\int_{0}^{t}\textrm{d}\tau _{4}\int_{0}^{t}\textrm{d}\tau _{3}\int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{t}\textrm{d}\tau _{1} \left(A_1+A_2+A_3\right) \nonumber\\ = &\frac{3}{4!\hbar ^{4}}\int_{0}^{t}\textrm{d}\tau _{4}\int_{0}^{t}\textrm{d}\tau _{3}\int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{t}\textrm{d}\tau _{1} A_1.\end{align}$
By removing the time-ordering operator $\mathcal{T}_{B}$ acting on bath superoperators in $A_1$, we obtain
$\begin{align} \varrho_{4}\left(t\right) & = \frac{1}{2\hbar^{4}}\int_{0}^{t}\textrm{d}\tau_{4}\int_{0}^{\tau_{4}}\textrm{d}\tau_{3} \int_{0}^{t}\textrm{d}\tau_{2}\int_{0}^{\tau_{2}}\textrm{d}\tau_{1} \nonumber\\ &\quad\times \mathcal{T}_{S}\left\{\sum_{i,j = 1}^{2}\text{Tr}_{B}\left[ L_{B_{4}}^{\left(i\right)}L_{B_{3}}^{\left(j\right)}\rho_{B}\right] L_{S_{4}}^{\left(i\right)}L_{S_{3}}^{\left(j\right)}\right\} \nonumber\\ &\quad\times\left\{ \sum_{k,l = 1}^{2}\text{Tr}_{B}\left[L_{B_{2}}^{\left(k \right)}L_{B_{1}}^{\left(l\right)}\rho_{B}\right]L_{S_{2}}^{\left(k \right)}L_{S_{1}}^{\left(l\right)}\right\} \rho_{S}\left(0\right).\end{align}$
An interesting feature of equation (40) is that the double integrals over $(\tau_{2}, \tau_{1})$ and $(\tau_{4}, \tau_{3})$ play identical roles. Equation (40) can be factorized as the square of a single double-integral expression. In fact, by using the second-order result equation (15), equation (40) can be rewritten in a compact form,
$\begin{align} \varrho_{4}\left(t\right) = & \frac{1}{2\hbar^{4}}\mathcal{T}_{S}\left[\int_{0}^{t}\textrm{d}\tau_{2} \int_{0}^{\tau_{2}}\textrm{d}\tau_{1}\right. \nonumber\\ & \quad\times\left. \left(G_{21}V_{2}^{\times}V_{1}^{\times}-\textrm{i}\frac{ \hbar}{2}\chi_{21}V_{2}^{\times}V_{1}^{\circ}\right)\right] ^{2}\rho_{S}\left(0\right),\end{align}$
where $G_{kl} = \text{Re}\left\langle B_{k}B_{l}\right\rangle ,\;\chi_{kl} = -\frac{2}{ \hbar}\text{Im}\left\langle B_{k}B_{l}\right\rangle $.
Using a similar procedure, we obtain the $2n$-th term in appendix B.For the $2n$-th term, we can express it with a power of $n$,
$\begin{align} \varrho_{2n}\left( t\right) = &\frac{1}{n!\hbar ^{2n}}\mathcal{T}_S\left[ \int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{\tau _{2}}\textrm{d}\tau _{1} \right. \nonumber\\ &\quad\times \left. \left(G_{21}V_{2}^{\times }V_{1}^{\times }-\textrm{i}\frac{\hbar }{2}\chi _{21}V_{2}^{\times }V_{1}^{\circ }\right) \right] ^{n}\rho _{S}\left( 0\right),\end{align}$
establishing a correspondence between the term order $2n$ and the power exponent $n$. Finally, we derive the exact expression for the reduced dynamics,
$\begin{align} \tilde{\rho}_{S}\left( t\right) & = \sum_{n = 0}^{\infty }\varrho _{2n}\left( t\right) \nonumber\\ & = \mathcal{T}_S\exp \left\{-\frac{1}{\hbar ^{2}}\int_{0}^{t}\textrm{d}\tau _{2}V_{2}^{\times }\left[ \int_{0}^{\tau _{2}}\textrm{d}\tau _{1}\right. \right. \nonumber\\ &\quad \times \left. \left. G\left( \tau_{2}-\tau _{1}\right) V_{1}^{\times }-\textrm{i}\frac{\hbar }{2}\chi \left( \tau _{2}-\tau _{1}\right) V_{1}^{\circ }\right] \right\} \rho _{S}\left( 0\right).\end{align}$
Our result, derived using Wick’s theorem for superoperators, is consistent with that obtained from the Feynman and Vernon influence functional method, which yields a continued fraction expression for the reduced density matrix [2325].

5. Conclusion

We have rigorously proved Wick’s theorem for superoperators that are linear in bosonic operators that decompose higher-order time-ordered superoperators into a sum of products of second-order terms. This derivation employs stable partition theory to reorder superoperators according to their upper indices, thereby enabling a systematic factorization of higher-order contributions.
Applying Wick’s theorem for superoperators, explicit expressions for all even-order terms of the reduced density matrix are derived. We have derived a formally exact expression for the reduced density matrix of a system coupled to a bosonic bath initially in thermal equilibrium. The Wick’s theorem for superoperators offers new insights into the reduced dynamics of open quantum systems. We believe there will be other applications for the Wick’s theorem for superoperators.

Appendix A. Stable partition

A stable partition algorithm rearranges the elements of a sequence such that elements satisfying a given condition precede those that do not, while preserving the original relative order of the elements within each subsequence.For example, consider the sequence:
$\begin{align}S = \left[2n,2n-1,\dots,2,1\right].\end{align}$
Applying the condition that an element is odd, we identify two subsequences. The subsequence of the elements satisfying the condition is:
$\begin{align} S_{\text{odd}} = \left[2n-1,2n-3,\dots,1\right],\end{align}$
and the subsequence of elements not satisfying the condition is:
$\begin{align} S_{\text{even}} = \left[2n,2n-2,\dots,2\right].\end{align}$
A stable partition of the original sequence $S$ with respect to this condition yields the sequence:
$\begin{align} S_{\text{partitioned}} = \left[2n-1,\dots,1,2n,2n-2,\dots,2\right].\end{align}$

Appendix B. $2n$-th Order integral

The expression for the $2n$-th order terms is
$\begin{align} \varrho_{2n}\left(t\right) = & \frac{1}{\left(2n\right)!\hbar^{2n}} \int_{0}^{t}\textrm{d}\tau_{2n}\dots\int_{0}^{t}\textrm{d}\tau_{1} \nonumber\\ &\times \text{Tr}_{B} \left[\mathcal{T}H_{SB}^{\times}\left(\tau_{2n}\right)\dots H_{SB}^{\times}\left(\tau_{1}\right)\rho_{B}\rho_{S}\left(0\right)\right].\end{align}$
Employing Wick’s theorem for superoperators and equation (11), the integrand can be written as
$\begin{align} &\sum_{j = 1}^m \left\{\prod_{\left(u,v\right)\in Q_{1,j}}\sum_{i_{v},i_{u} = 1}^2\mathrm{Tr}_B\left[\mathcal{T}_B\,L_{B_u}^{\left(i_{u}\right)}L_{B_v}^{\left(i_{v}\right)}\rho_B\right] \right\} \nonumber\\ &\quad \times\left[\mathcal{ T}_SL_{S_{2n}}^{\left(i_{2n}\right)}\dots L_{S_{2}}^{\left(i_{2}\right)}L_{S_{1}}^{\left(i_{1}\right)}\right] \rho_{S}\left(0\right) = \sum_{j = 1}^{m}A_{j},\end{align}$
where
$\begin{align} A_{j} = & \left\{\prod_{\left(u,v\right)\in Q_{1,j}}\sum_{i_{u},i_{v} = 1}^2\mathrm{Tr}_B\left[\mathcal{T}_B\,L_{B_{u}}^{\left(i_{u}\right)}L_{B_{v}}^{\left(i_{v}\right)}\rho_B\right] \right\} \nonumber\\ &\times \left[ \mathcal{T}_SL_{S_{2n}}^{\left( i_{2n}\right) }\dots L_{S_{2}}^{\left( i_{2}\right) }L_{S_{1}}^{\left( i_{1}\right) }\right] \rho _{S}\left( 0\right).\end{align}$
Due to the presence of the time-ordering operator, we can always rearrange the order of the superoperators $L_{S}^{\left( i\right) }$ corresponding to the pairing of superoperators $ L_{B}^{\left( i\right)} $, we have
$\begin{align} A_j = &\mathcal{T}_S\left\{\prod_{\left(u,v\right)\in Q_{1,j}}\sum_{i_u,i_v = 1}^2\mathrm{Tr}_B \left[\mathcal{T}_B\,L_{B_u}^{\left(i_u\right)}L_{B_v}^{\left(i_v\right)}\rho_B\right] \right. \nonumber\\ &\times\left.\left[L_{S_u}^{\left(i_u\right)}L_{S_v}^{\left(i_v\right)}\right] \vphantom{\prod_{\left(u,v\right)\in Q_{1,j}}}\right\} \rho_{S}\left( 0\right).\end{align}$
Since the $A_{j} ( 1\unicode{x2A7D} j \unicode{x2A7D} m)$ differ only by dummy indices, they become identical after integration,
$\begin{align} \int_{0}^{t}\textrm{d}\tau _{2n}\dots \int_{0}^{t}\textrm{d}\tau_{1} \sum_{j = 1}^{m}A_j = m \int_{0}^{t}\textrm{d}\tau _{2n}\dots \int_{0}^{t}\textrm{d}\tau_{1} A_j.\end{align}$
Therefore, $\varrho _{2n}$ can be derived as
$\begin{align} \varrho _{2n}\left( t\right) \! &= \!\frac{m}{\hbar ^{2n}\left(2n\right)!} \int_{0}^{t}\textrm{d}\tau _{2n}\dots \int_{0}^{t}\textrm{d}\tau_{1} \mathcal{T}_S \nonumber\\ &\quad\times \left\{\prod_{\left(u,v\right)\in Q_{1,j}} \sum_{i_u,i_v = 1}^2 \mathrm{Tr}_B\left[\mathcal{T}_B\,L_{B_u}^{\left(i_u\right)}L_{B_v}^{\left(i_v\right)}\rho_B\right] \left[L_{S_u}^{\left(i_u\right)}L_{S_v}^{\left(i_v\right)}\right]\vphantom{\prod_{\left(u,v\right)\in Q_{1,j}}} \right\} \rho_{S}\left( 0\right).\end{align}$
Removing the time-ordering operator $\mathcal{T}_B$ acting on the bath superoperator, we have
$\begin{align} \varrho _{2n}\left( t\right) = &\frac{1}{\hbar ^{2n}n!}\int_{0}^{t}\textrm{d}\tau _{2n}\int_{0}^{\tau _{2n}}\textrm{d}\tau _{2n-1}\dots \nonumber\\ & \int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{\tau _{2}}\textrm{d}\tau _{1} \mathcal{T}_S \left\{\prod_{\left(u,v\right)\in Q_{1,1}}\mathrm{Tr}_B \left[\mathcal{T}_B\,L_{B_u}^{\left(i_u\right)}L_{B_v}^{\left(i_v\right)}\rho_B\right] \right. \nonumber\\ &\times\left. \left[L_{S_u}^{\left(i_u\right)}L_{S_v}^{\left(i_v\right)}\right] \vphantom{\prod_{\left(u,v\right)\in Q_{1,j}}}\right\} \rho_{S}\left( 0\right) \nonumber\\ = &\frac{1}{\hbar ^{2n}n!}\mathcal{T}_S\left[ \int_{0}^{t}\textrm{d}\tau _{2}\int_{0}^{\tau _{2}}\textrm{d}\tau _{1} \right. \nonumber\\ &\times\left. \left( G_{21}V_{2}^{\times }V_{1}^{\times } -i\frac{\hbar }{2}\chi _{21}V_{2}^{\times }V_{1}^{\circ }\right) \right] ^{n}\rho _{S}\left( 0\right) .\end{align}$

This work is supported by the Quantum Science and Technology-National Science and Technology Major Project (Grant No. 2024ZD0301000), Science Challenge Project (Grant No. TZ2025017), and the Science Foundation of Zhejiang Sci-Tech University (Grant No. 23062088-Y).

1
Wick G C 1950 The evaluation of the collision matrix Phys. Rev. 80 268

DOI

2
Bruus H, Flensberg K 2004 Many-Body Quantum Theory in Condensed Matter Physics: an Introduction Oxford University Press

3
Peskin M E 2018 An Introduction to Quantum Field Theory CRC Press

4
van Leeuwen R, Stefanucci G 2012 Wick theorem for general initial states Phys. Rev. B 85 115119

DOI

5
Mulkerin B C, Liu X-J, Hu H 2016 Beyond Gaussian pair fluctuation theory for strongly interacting Fermi gases Phys. Rev. A 94 013610

DOI

6
Corson J P, Peatross J 2011 Quantum-electrodynamic treatment of photoemission by a single-electron wave packet Phys. Rev. A 84 053832

DOI

7
Bialynicki-Birula I, Sowiński T 2007 Quantum electrodynamics of qubits Phys. Rev. A 76 062106

DOI

8
Fetter A L, Walecka J D 1971 Quantum Theory of Many-Particle Systems McGraw-Hill

9
Chakraborty A, Gorantla P, Sensarma R 2019 Nonequilibrium field theory for dynamics starting from arbitrary athermal initial conditions Phys. Rev. B 99 054306

DOI

10
Nietner C, Pelster A 2012 Ginzburg-Landau theory for the Jaynes-Cummings-Hubbard model Phys. Rev. A 85 043831

DOI

11
Fujimoto J, Matsuo M 2019 Alternating current-induced interfacial spin-transfer torque Phys. Rev. B 100 220402

DOI

12
Piroli L, Calabrese P 2017 Exact dynamics following an interaction quench in a one-dimensional anyonic gas Phys. Rev. A 96 023611

DOI

13
Kanungo S K, Lu Y, Dunning F B, Yoshida S, Burgdörfer J, Killian T C 2023 Measuring nonlocal three-body spatial correlations with Rydberg trimers in ultracold quantum gases Phys. Rev. A 107 033322

DOI

14
Pérez F T B, Matera J M 2024 Quantum covariance scalar products and efficient estimation of maximum-entropy projections Phys. Rev. A 109 022401

DOI

15
Fogedby H C 2022 Field-theoretical approach to open quantum systems and the Lindblad equation Phys. Rev. A 106 022205

DOI

16
Müller S, Novaes M 2018 Full perturbative calculation of spectral correlation functions for chaotic systems in the unitary symmetry class Phys. Rev. E 98 052208

DOI

17
Wang Y, Miao J-J, Jin H-K, Chen S 2017 Characterization of topological phases of dimerized Kitaev chain via edge correlation functions Phys. Rev. B 96 205428

DOI

18
Na D-Y, Zhu J, Chew W C, Teixeira F L 2020 Quantum information preserving computational electromagnetics Phys. Rev. A 102 013711

DOI

19
Guo Z-X, Yu X-J, Hu X-D, Li Z 2022 Emergent phase transitions in a cluster Ising model with dissipation Phys. Rev. A 105 053311

DOI

20
Jafari R, Akbari A 2020 Dynamics of quantum coherence and quantum Fisher information after a sudden quench Phys. Rev. A 101 062105

DOI

21
Mukamel S 2003 Superoperator representation of nonlinear response: Unifying quantum field and mode coupling theories Phys. Rev. E 68 021111

DOI

22
Katajainen J, Pasanen T 1992 Stable minimum space partitioning in linear time BIT Numer. Math. 32 580

DOI

23
Feynman R, Vernon F 1963 The theory of a general quantum system interacting with a linear dissipative system Ann. Phys. 24 118

DOI

24
Tanimura Y 2006 Stochastic Liouville, Langevin, Fokker–Planck and Master equation approaches to quantum dissipative systems J. Phys. Soc. Japan 75 082001

DOI

25
Breuer H-P, Petruccione F 2007 The Theory of Open Quantum Systems Oxford University Press

Outlines

/