2025 Volume 38 Issue 6
Article Contents

Jia-Lin Pan, Zi-Fan Zhu, Shixuan Chen, Yu Su, Yao Wang. Fermionic Influence Superoperator for Transport through Majorana Zero Modes†[J]. Chinese Journal of Chemical Physics, 2025, 38(6): 819-832. doi: 10.1063/1674-0068/cjcp2509147
Citation: Jia-Lin Pan, Zi-Fan Zhu, Shixuan Chen, Yu Su, Yao Wang. Fermionic Influence Superoperator for Transport through Majorana Zero Modes†[J]. Chinese Journal of Chemical Physics, 2025, 38(6): 819-832. doi: 10.1063/1674-0068/cjcp2509147

Fermionic Influence Superoperator for Transport through Majorana Zero Modes†

通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(2)

Article Metrics

Article views(78) PDF downloads(0) Cited by(0)

Access History

Fermionic Influence Superoperator for Transport through Majorana Zero Modes†

Abstract: In recent years, the study of Majorana signatures in quantum transport has become a central focus in condensed matter physics. Here, we present a rigorous and systematic derivation of the fermionic superoperator describing the open quantum dynamics of electron transport through Majorana zero modes, building on the techniques introduced in Phys. Rev. B 105, 035121 (2022). The numerical implementation of this superoperator is to construct its differential equivalence, the hierarchical equations of motion (HEOM). The HEOM approach describes the system-bath correlated dynamics. Furthermore, we also develop a functional derivative scheme that provides exact expressions for the transport observables in terms of the auxiliary density operators introduced in the HEOM formulation. The superoperator formalism establishes a solid theoretical foundation for analyzing key transport signatures that may uncover the unique characteristics of Majorana physics in mesoscopic systems.

    • The search for Majorana zero modes (MZMs), quasiparticles with non-Abelian statistics, has become a major frontier in condensed matter physics, driven by their potential applications in topological quantum computation and fault-tolerant quantum information processing [1−4]. These exotic states, localized at the ends of one-dimensional topological superconductors, exhibit robustness against local perturbations and are expected to manifest unique signatures in quantum transport experiments, such as zero-bias conductance peaks and fractional Josephson effects [5, 6].

      Recent advances in hybrid nanostructures, including semiconductor nanowires with strong spin-orbit coupling and proximity-induced superconductivity, have enabled experimental observations consistent with MZMs [1, 2]. However, distinguishing genuine Majorana signatures from trivial Andreev bound states remains a critical challenge [7−12]. Theoretical frameworks that accurately describe open quantum dynamics in these systems are thus essential for interpreting transport data and guiding future experiments [13].

      In this work, we develop a canonical fermionic superoperator formalism to model the nonequilibrium quantum transport through MZMs, building on the operator-space techniques established in Ref.[14]. We express the environmental influence on the system in terms of a fermionic influence superoperator, where the bath properties are fully characterized by their two-time correlation functions. For numerical solving the influence superoperator, we construct its differential equivalence, the hierarchical equations of motion (HEOM) formalism. The HEOM approach describes the dynamics of the reduced density operator and a set of auxiliary density operators (ADOs), which encode the system-bath correlations [15−21].

      Although the derivation basically follows the framework established in Ref.[14], different from Ref.[14], we propose (i) a simplified proof of Wick’s theorem (cf. Appendices C and D), exploiting the techniques developed in Ref.[22]; (ii) the functional derivative scheme on the influence superoperator, which allows us to derive exact expressions for transport observables in terms of the ADOs (Eqs. (82)–(89)). As a result, our approach provides a rigorous foundation for analyzing characteristics in Majorana-based devices, offering insights beyond Markovian treatments.

      Further generalizations of our scheme lead to the inner relations among the ADOs. This lays the foundation for the concept of the statistical quasi-particle, dissipaton, proposed in Ref.[19]. Although originating from the intrinsic algebraic structure of HEOM, the dissipaton formalism offers a powerful route to extend open quantum system theory to nonlinear system-bath couplings [23−26].

      The paper is organized as follows. In Sec. II, we introduce the theoretical model. In Sec. III, we present a canonical derivation of the fermionic influence superoperator governing the transport through MZMs. Then in Sec. IV, we construct the differentiate equivalence of the influence functional, the HEOM. Furthermore, we derive the formula for evaluating the transport quantities via the HEOM. Numerical demonstrations are carried out in Sec. V. Finally, in Sec. VI, we summarize our results and outline potential future work. Throughout this work, we set $ \hbar=1 $ and $\beta_{\alpha}=1/(k_{\rm B}T_{\alpha})$, with $ k\mathrm{_B} $ being the Boltzmann constant and $ T_{\alpha} $ being the temperature of $ \alpha $-lead (with $ \alpha = {\rm{L}} $ representing the left one and $ \alpha = {\rm{R}} $ for the right one).

    II.   THEORETICAL MODELS
    • The transport setup is the same as that in Ref.[9]. The total system-bath Hamiltonian reads

      Here, the central system consists of a pair of MZMs,

      where $ \hat \gamma_{{\rm{L}}}/\hat \gamma_{{\rm{R}}} $ are two MZMs, satisying $ \hat \gamma^{ \dagger}_{{\rm{L}}/{\rm{R}}}=\hat \gamma_{{\rm{L}}/{\rm{R}}} $, coupled to each other with coupling constant $ \varepsilon_{\rm{M}} $. The bath Hamiltonian reads

      with $ \hat{c}_{\alpha k}^{ \dagger} $ and $ \hat{c}_{\alpha k} $ being the creation and annihilation operators of the $ k $-state electron in the $ \alpha $-lead. For imposing the current transport, we apply an electric potential $ \varphi_{\alpha} $ on each lead, leading to

      with $\hat{N}_{\alpha} \equiv\displaystyle \sum_{k}\hat{c}_{\alpha k}^\dagger\hat{c}_{\alpha k}$ being the bath particle number operator. The system-bath interaction reads

    • The Majorana fermions can be expressed as

      Here, $ \hat{f} $ and $ \hat{f}^{ \dagger} $ are regular fermions, satisfying $ \{\hat{f}, \hat{f}^{ \dagger}\}=1 $ and $ \hat{f}^2=(\hat{f}^{ \dagger})^2=0 $. Eq.(6) reproduces the properties of Majoranan fermions,

      and

      The system Hamiltonian in Eq.(2) can then be recast as

      And the system-bath interaction Hamiltonian in Eq.(5) is reformulated as

      with $\hat{F}_{\alpha}\equiv\displaystyle \sum_k t_{\alpha k}^{\ast}\hat{c}_{\alpha k}$, and

      In the next section, we will give the canonical derivation of the influence superoperator based on Eqs. (9), (10), and (3), based on the techniques established in Ref.[14].

    III.   CANONICAL DERIVATION OF THE INFLUENCE SUPEROPERATOR
    • The quantum system-bath hybridization dynamics is described by quantum states living in the composite Hilbert space $ {\cal{H}}_{\rm{T}} = {\cal{H}}_{\rm{B}} \times {\cal{H}}_{\rm{S}} $. The total space operator generated by the system subspace operator $ \hat{A}_{\rm{S}} $, and bath one $ \hat{F}_{\rm{B}} $ is defined via the tensor product $ \hat{F}_{\rm{B}}\otimes\hat{A}_{\rm{S}} $, satisfying the multiplication rule

      However, for the fermionic operators, such a definition cannot lead to the anti-commutation relation between the system and bath operators, i.e., $ \hat{f}\hat{c}_{\alpha k} = -\hat{c}_{\alpha k}\hat{f} $. Consequently, we have to define the graded tensor product $ \otimes_g $, with [27, 28]

      Here, the degree of a system operator is defined as

      with $ \hat{P}_{\rm{S}} \equiv \exp(i\pi\hat{f}^\dagger\hat{f}) $ being the system parity operator. The definition for the bath operator is similar, also with the bath parity being

      Note that in defining the graded tensor product, we only consider those operators with definite parity. For an arbitrary operator, we can first decomposite it into the even (+1 parity) and odd ($ -1 $ parity) parts, and the proceed with the graded tensor product. See the detailed discussions on the parity in Appendix A.

      However, for constructing the canonical representation of the influence functional, the graded tensor product presents additional algebraic complexities. Thus, we have to map the graded algebra into the normal one ($ \otimes $). This is achieved by the extension

      The extension mapping is isomorphic to the original one, that is

      The proof is presented in Appendix B. As a special case of Eq.(17), we see that for an arbitrary bath operator $ \hat{F}_{\rm{B}} $,

      which is valid regardless of the parity of $ \hat{F}_{\rm{B}} $. For an arbitrary system operator,

      with $ \hat{A}_{\rm{S}}^{(\pm)} $ being the even and odd parts of $ \hat{A}_{\rm{S}} $ defined by Eq.(A6). As a result, we denote $ \hat{F}_{\rm{B}} $ representing a bath operator in both the pure-bath space $ {\cal{H}}_{\rm{B}} $ and composite space $ {\cal{H}}_{\rm{T}} $. And we denote

      to represent the system operator $ \hat{A}_{\rm{S}} $ in the total Hilbert space. Then we have

      by noting that $ \hat{s}_{\alpha} $ is of odd parity.

    • The open quantum system formalism focuses on the reduced system dynamics, which involves the tracing out of the bath degrees of freedom. We define the bath partial trace for a total space operator $ \hat{O}_{\rm{T}} $ as

      Here, we use the bath occupation number basis,

      Within this definition, we know that for any system and bath operators $ \hat{A}_{\rm{S}} $ and $ \hat{F}_{\rm{B}} $,

      More complicatedly,

      This can be proven by noting

      In the last identity, we have used the fact that

      with $ \hat{P}_{\rm{S}}^{\pm} \equiv \dfrac{1}{2}(1 \pm \hat{P}_{\rm{S}}) $; see details in Appendix A. Using Eq.(26), we obtain

    • The system-plus-bath composite forms a closed quantum system, with the total density operator $ \rho_{\rm{T}}(t) $ satisfying the Liouville-von Neumann equation

      The reduced system density operator is defined via

      for any system operator $ \hat{A}_{\rm{S}} $. Quantum dissipation process always assumes the initial state as separate state,

      with

      and

      being the grand canonical ensemble of the bath. Here, $\beta_{\alpha} = 1/(k_{\rm B}T_{\alpha})$ and $ \mu_{\alpha} $ ($\alpha = {\rm{L}}, {\rm{R}}$) represent the inverse temperature and the chemical potential, respectively. From Eq.(29), we can verify that the definition in Eq.(31) holds for the separate state. However, it is worth noting that directly calculating the partial trace over the separate state gives

      This indicates that $ {\rm{tr}}_{\rm{B}}\rho_{\rm{T}}(t) $ is incompatible to the definition of reduced system density operator (Eq.(31)), which is different from the bosonic scenario [14].

    • To proceed, we turn to the interaction picture evolution which is defined via

      leading to

      with

      and

      Here, $\hat{F}_{\alpha}(t) = {\rm e}^{i\tilde h_{\rm{B}}t}\hat{F}_{\alpha} {\rm e}^{-i\tilde h_{\rm B}t}$ and $\hat{s}_{\alpha}(t) = {\rm e}^{iH_{\rm{S}}t}\hat{s}_{\alpha} {\rm e}^{-iH_{\rm{S}}t}$. For convenience of notation, we denote the following superoperators for an operator $ \hat{O} $,

      As an example, we can rewrite Eq.(38) as

      Its formal solution reads

      where we impose the time ordering for superoperators,

      For deriving the reduced system density dynamics, our strategy goes by analyzing the structure of Eq.(42) and expressing it in terms of sum of direct product of bath and system parts,

      From Eq.(31), we have

      where the last equality should be seen as the definition of $ \rho_{\rm{S}}^I(t) $. For simplifying the notation, we recast Eq.(39) as

      Here, we introduce the notation $ \sigma = \pm $, with $ \hat{F}^{+}_{\alpha} = \hat{F}^\dagger_{\alpha} $ and $ \hat{F}^-_{\alpha} = \hat{F}_{\alpha} $, and the system part being the same. We also denote $ \bar{\sigma}\equiv -\sigma $.

      Now we consider how the superoperator $ H_{\rm{SB}}^\times(t) $ acts. Calculate

      For $ \hat{O}_{\rm{B}}\hat{O}_{\rm{S}} $ being even, we have

      Denote the following superoperators

      and

      with $ \mathscr{P}_{\rm{S}}(\cdot) \equiv \hat{P}_{\rm{S}}(\cdot)\hat{P}_{\rm{S}} $ and $ \mathscr{P}_{\rm{B}}(\cdot) \equiv \hat{P}_{\rm{B}}(\cdot)\hat{P}_{\rm{B}} $. For later use, we use the index $ \lambda $ to label the left and right action, that is $ \lambda = +1 $ for $ \gt $ and $ \lambda = -1 $ for $ \lt $, and $ \bar{\lambda} = -\lambda $. Then we have,

      for $ \hat{O}_{\rm{B}}\hat{O}_{\rm{S}} $ even, and

      for $ \hat{O}_{\rm{B}}\hat{O}_{\rm{S}} $ odd. As a result, the first order contribution reads

      Since each action of $ H^\times_{\rm{SB}}(t) $ remains the parity of operators, we can rewrite Eq.(42) as

      Here, we separate the total time ordering to $ {\cal{T}} = {\cal{T}}_{\rm{B}}{\cal{T}}_{\rm{S}} $, with the system and bath time ordering being of fermionic type, i.e.,

      and the system one is similar. By comparing Eq.(54) with Eq.(44) and using Eq.(45), we readily have

      For further proceeding, we have to consider the multi-point correlation functions $ {\rm{tr}}_{\rm{B}}\big[ {\cal{T}}_{\rm{B}}\mathscr{B}_{\alpha_n}^{\sigma_n\lambda_n}(t_n) \cdots \mathscr{B}_{\alpha_1}^{\sigma_1\lambda_1}(t_1)\rho_{\rm{B}}^{\rm{eq}} \big] $ and $ {\rm{tr}}_{\rm{B}}\big[{\cal{T}}_{\rm{B}} \hat{P}_{\rm{B}}\mathscr{B}_{\alpha_n}^{\sigma_n\lambda_n}(t_n) \cdots \mathscr{B}_{\alpha_1}^{\sigma_1\lambda_1}(t_1)(\rho_{\rm{B}}^{\rm{eq}}\hat{P}_{\rm{B}}) \big] $. Note that for $ n $ being odd, the correlation functions vanish. Thus, we only focus on the even $ n $ case. Based on $ \{\hat{P}_{\rm{B}},\hat{F}_{\alpha}^{\sigma}(t)\} = 0 $ and $ \hat{P}_{\rm{B}}^2 = 1 $, those even $ n $-point correlations reduce to $ (-i)^n{\rm{tr}}_{\rm{B}}\big[{\cal{T}}_{\rm{B}} \mathscr{F}_{\alpha_n}^{\sigma_n\lambda_n}(t_n) \cdots \mathscr{F}_{\alpha_1}^{\sigma_1\lambda_1}(t_1)\rho_{\rm{B}}^{\rm{eq}} \big] $ and $ (-i)^n {\rm{tr}}_{\rm{B}}\big[ {\cal{T}}_{\rm{B}}\hat{P}_{\rm{B}}\mathscr{F}_{\alpha_n}^{\sigma_n\lambda_n}(t_n) \cdots \mathscr{F}_{\alpha_1}^{\sigma_1\lambda_1}(t_1) (\rho_{\rm{B}}^{\rm{eq}}\hat{P}_{\rm{B}}) \big] $, with

      In order to combine the two large terms in the summation of Eq.(56), we observe that the system part of the second term differs from the first one only by minus signs of the number of the right actions $ \bar{\mathscr{s}}_{\alpha}^{\sigma <}(t) $. On the other hand, in order to take the $ \hat{P}_{\rm{B}} $ within $ {\rm{tr}}_{\rm{B}}\big[{\cal{T}}_{\rm{B}} \hat{P}_{\rm{B}}\mathscr{F}_{\alpha_n}^{\sigma_n\lambda_n}(t_n) \cdots \mathscr{F}_{\alpha_1}^{\sigma_1\lambda_1}(t_1)(\rho_{\rm{B}}^{\rm{eq}}\hat{P}_{\rm{B}}) \big] $, we have to compensate the minus signs of the number of the left actions $ \mathscr{F}^{\sigma>}_{\alpha}(t) $. Since the total contribution is equal to $ (-1)^n $ with $ n $ being even (only even $ n $ is considered), then no additional sign is introduced. As a result, we have

      with noting $ \rho_{\rm{S}}(0) = \rho_{\rm{S}}^{(+)}(0) + \rho_{\rm{S}}^{(-)}(0) $ (cf. Eq.(A6)). So far, we derive the reduced system density operator in terms of the Dyson series. Nextly, we apply the Wick's theorem for fermionic superoperators to resolve the multi-point correlation functions.

    • The Wick’s theorem for fermionic superoperators is proved in Appendix D. It states that the multi-point correlation function can be expressed as the sum of all possible products of two-point ones. As a result, for even $ n $, we have

      with

      Here, $ C_n $ is the set of all possible time-ordered pairings and $ \#_\mathscr{c} $ counts the crossing number of the pairing configuration $ \mathscr{c} $. Thus, the reduced system density operator is recast as

      where

      with

      and the bare-bath correlation function being

      Note that the minus signs from the pairing configuration $ \mathscr{c} $ are exactly compensated by the fermionic time ordering of system superoperators. Since all terms in the product of Eq.(61) are identical and the total number of those terms is $ (2n-1)!! $, we have

      with the influence superoperator being

      In the last identity of Eq.(66), we have used the fact that $ (2n)!/(2^n n!) = (2n-1)!! $.

    IV.   HIERARCHICAL EQUATIONS OF MOTION
    • Quantum dissipation dynamics aims at constructing the dynamics of reduced system density operator. Based on the influence functional formalism, one intuitive way is to differentiate Eq.(66). However, such operations generally lead to more auxiliary quantities which are not closed. The key to resolve this issue is to expand the bath correlation functions in exponential series for $ t>0 $,

      with the prefactor $ \eta_{\alpha\kappa}^{\sigma} $ and exponential $ \gamma_{\alpha\kappa}^{\sigma} $ being complex. The exponential decomposition can be achieved via various schemes, such as the Matsubara decomposition [16, 18], the Padé spectrum decomposition [29, 30], time-domain Prony scheme [31], the numerically analytical continuation [32], Fano spectrum decomposition scheme [33], and so on. All the decomposition schemes require [20, 34]

      This leads to

      In practice, we determine the bath correlation functions via the fluctuation–dissipation theorem,

      where the spectral density function is defined via

      As a result, we recast Eq.(62) as

      with

    • We are now in the position to construct the hierarchical equations of motion (HEOM). We introduce the index abbreviation

      Define the auxiliary density operators (ADOs) as

      where

      and $ {\bf{j}} \equiv j_1j_2\cdots j_n $. Using

      we have

      with $ {\bf{j}}_r^- $ being the index string obtained by removing $ j_r $ from j. The reduced system density operator is just the zeroth-tier ADO, i.e., $ \rho_{\rm{S}}(t) = \rho^{(0)}(t) $. The initial conditions are $ \rho^{(0)}(0) = \rho_{\rm{S}}(0) $ and $ \rho_{}^{(n>0)}(0) = 0 $. Here, the superoperators are defined as

      Generally, since the initial reduced system density operator is physical, it must be an even one. In such a case, the superoperators are given by

    • For discussing the transport property of the impurity system, we have to consider the dynamics of electronic current, defined as [9, 20, 35]

      with $ \hat{N}_{\alpha} = \sum\nolimits_k\hat{c}_{\alpha k}^\dagger\hat{c}_{\alpha k} $ being the particle number of the $ \alpha $-th bath. We are interested in the transient mean value of the current,

      By defining the interaction picture Eq.(36), we have

      where $ \varrho_{\rm{S}}^{\sigma\alpha}(t) $ is obtained via tracing out the bath degrees of freedom of

      Using Eq.(53), we have

      where $ {\delta}/{\delta\mathscr{s}^{\bar{\sigma}>}_{\alpha}(t)} $ is the functional derivative over the superoperator $ \mathscr{s}^{\bar{\sigma}>}_{\alpha}(t) $. Consequently, we have

      Substituting Eq.(66) into Eq.(87), we readily obtain

      Then, the current is evaluated by using the first tier of the ADOs, namely,

      Thus, we have finished the HEOM formalism for evaluating the transport current from the canonical algebra.

    V.   NUMERICAL DEMONSTRATION
    • In this section, we apply the HEOM method to demonstrate the transport phenomena induced by a Majorana impurity. For both two baths, we adopt the spectral density function to be the Lorentz type, namely

      We set the chemical potential as the zero energy point, $ \mu_{\rm{L}} = \mu_{\rm{R}} = 0 $. The band width is set to be $ W = 10\Delta $. For illustrating the unique Majorana transport property, we will compare the results with those of a usual fermionic Brownian oscillator (FBO) model. The FBO model is a spinless, non-interacting variant of the single-impurity Anderson model, describing a fermionic particle (e.g., an electron) coupled to a thermal fermionic bath, where its dissipative motion emerges from hybridization with the bath degrees of freedom. Its Hamiltonian reads

      with $ \hat{a} $ ($ \hat{a}^\dagger $) being the annihilation (creation) operator of the impurity fermion and $ \varepsilon_{\rm{F}} $ being its energy level.

      In FIG. 1, we present the transient current dynamics. The parameters are set to $ \varepsilon_{\rm{M}} = \varepsilon_{\rm{F}} = 10\Delta $ and $ \beta_{\rm{L}} = \beta_{\rm{R}} = 1000\Delta^{-1} $. The external bias voltages are chosen as $ \varphi_{\rm{L}} = 0, 5\Delta, 10\Delta $ while $ \varphi_{\rm{R}} = 0 $. In both models, the transient currents exhibit oscillations at a frequency around the system eigenenergy, and the magnitude of the steady-state current $ |I_{\rm{L}}^{\rm{st}}| $ increases with $ \varphi_{\rm{L}} $. In contrast, the Majorana impurity displays behavior distinct from the FBO model: (i) In the absence of an applied bias, the transient current vanishes for the Majorana impurity, whereas it remains finite in the FBO model but eventually decays to zero at long times. This indicates that the Majorana impurity is effectively decoupled from the lead when no voltage is applied, a behavior in MZM transport configurations typically depending on the coupling geometry. (ii) The steady-state total current, $ I_{\rm{L}}^{\rm{st}}+I_{\rm{R}}^{\rm{st}} $, is nonzero for the Majorana case but vanishes for the FBO at long times. This originates from the fact that, for the Majorana impurity, the total current is not a conserved quantity of the total Hamiltonian.

      FIG. 2 presents the steady-state differential conductance, $ G \equiv -{ {\rm{d}} I_{\rm{L}}^{\rm{st}}}/{ {\rm{d}} \varphi_{\rm{L}}} $, at various values of system energy level. The FBO model exhibits a single conductance peak located at the system energy level, $ \varepsilon_{\rm{F}} $. By contrast, the Majorana model shows more intricate behavior: as $ \varepsilon_{\rm{M}} $ increases, the conductance peak splits into two, with positions approximately at $ \pm\varepsilon_{\rm{M}} $. Notably, for the non-interacting zero mode ($ \varepsilon_{\rm{M}}=0 $), the peak height agrees with the Landauer–Büttiker result, $ G(\varphi_{\rm{L}})= 1/\pi $, a hallmark of the Majorana zero mode [7].

    VI.   CONCLUSION
    • In summary, we have established the HEOM formalism for quantum transport through a Majorana impurity system. The present theory is constructed based on the canonical algebra and the fermionic superoperators. The key step is to introduce the Wick’s theorem for fermionic superoperators. Based on this, we are allowed to sum over the contributions of all the bath degrees of freedom and obtain the influence functional. The HEOM formalism is then constructed via the exponential decomposition of bath correlation functions. Besides, we use the functional derivative technique to construct the relation between the transport current and the first-tier ADOs. Generalizing the technique would lead to the influence functional representation of the generalized Wick’s theorem for fermionic dissipatons [19, 20, 25, 36].

      For numerical demonstration, we have investigated the transient transport dynamics and steady-state differential conductance of a Majorana impurity system. The results are compared with those of a regular fermionic impurity model. The unique transport properties induced by the Majorana mode are clearly exhibited. The present HEOM formalism is numerically exact and applicable to arbitrary system–bath coupling strength, external bias voltage, and temperature. It provides a powerful tool to explore the exotic transport phenomena in Majorana systems.

      As a perspective, in realistic nanowire systems, MZMs are usually composite quasi-particles with internal degrees of freedom. While the model used in this work treats them as idealized fermionic modes, such internal structure could lead to quantitative corrections in transport behavior. Exploring how these internal degrees of freedom influence experimentally observable signatures represents an important direction for future work.

    Appendix A.   STRUCTURE OF FERMIONIC HILBERT SPACE
    • This appendix discusses the algebraic structure of a fermionic Hilbert space. The $ N $-fermion Hilbert space is spanned by the occupation number basis,

      Here, we introduce the no-particle vacuum state $ |{\bf{0}} \rangle $ and the creation and annihilation operators with anti-commutation relations,

      The fermionic statistics requires that the occupation number for each state can only be 0 or 1. We can separate the total Hilbert space into two parts via determining whether the total particle number $n \equiv\displaystyle \sum_k n_k$ is odd or even. To achieve this, we define the parity operator

      leading to

      We can always project a quantum state into the odd and even part by introducing the projection operators

      with noting $ \hat{P}_{\pm}^2 = \hat{P}_{\pm} $, $ \hat{P}_+\hat{P}_- = 0 $, and $ \hat{P}_+ + \hat{P}_- = \hat{I} $.

      Consequently, an operator $ \hat{O} $ is also decomposed into the even and odd parts,

      with

      Evidently, $ {\rm{Tr}}\,\hat{O}^{(+)} = {\rm{Tr}}\,\hat{O} $ and $ {\rm{Tr}}\,\hat{O}^{(-)} = 0 $. It is easy to see from Eq.(A7) that an operator $ \hat{O} $ is an even one if and only if $ [\hat{O},\hat{P}] = 0 $; conversely, $ \hat{O} $ is odd if and only if $ \{\hat{O},\hat{P}\} = 0 $. As a corollary, the product of two even or two odd operators is an even one; the product of an even and an odd operators is odd. As an example, $ \hat{c}_k $ is odd and $ \hat{c}_k^\dagger\hat{c}_k $ is even.

    Appendix B.   PROOF OF Eq.(18)
    • We prove it directly. The left-hand side of Eq.(18) is

      The right-hand side is

      Using the identity

      and discussing case by case according to the parity of $ \hat{A}_{\rm{S}} $ and $ \hat{G}_{\rm{B}} $, one is easy to show the equality.

    Appendix C.   GENERAL WICK’S THEOREM FOR FERMIONIC OPERATORS
    • This section presents the general Wick's theorem for fermionic operators, which will be readily used in the Appendix D. The Wick’s theorem generally discusses the relation between two orderings of a set of operators [22]. We define the ordering as

      Here, $ \#_{\mathscr{p}_n} $ is the number of permutations to arrange the sequence $ \{1,2,\cdots,n\} $ into the sequence $ \{\mathscr{p}_1,\mathscr{p}_2,\cdots,\mathscr{p}_n\} $. We denote the ordered label as $ \mathscr{p}_1 \succ \mathscr{p}_2 \succ\cdots\succ \mathscr{p}_n $. Denote the linear transformation

      And the other ordering $ {\cal{O}}' $ is defined on $ \{\hat\varphi_k\} $. The core quantity is the contraction,

      where we introduce the ordering step function, $ \theta_{\alpha\succ\alpha'} $ with $ \theta_{\alpha\succ\alpha'} = 1 $ if $ \alpha\succ\alpha' $ and $ \theta_{\alpha\succ\alpha'} = 0 $ otherwise. And $ \theta_{k\succ k'} $ is similar. Note that the second term should be explained as

      The general Wick’s theorem states that if all $ \{\hat\phi_{\alpha},\hat\phi_{\alpha'}\} $ are c-numbers, changing the ordering from $ {\cal{O}} $ to $ {\cal{O}}' $ gives [22]

      with

      Here, $ \partial_{\alpha'} $ is the Grassmann derivative with respect to $ \hat\phi_{\alpha'} $, satisfying

      We prove the theorem by induction. Firstly, the cases that $ n=0 $ and $ n=1 $ are trivial. Assume the theorem holds for $ n $. We will show that

      Assume that $ \alpha_n\succ\cdots\succ\alpha_1 $, and we have

      Then,

      To apply the case $ n $, we move $ \hat\phi_{\alpha} $ to the leftmost side. The first move gives

      Iteratively,

      Using Eq.(C6) and the induction assumption, we obtain

      Define

      with

      Here, the derivative is $ \tilde\partial_{k} \equiv {\partial}/{\partial\hat\varphi_k} $ and the contraction is defined as ${\cal{C}}_{kk'}\equiv\displaystyle \sum_{\alpha\alpha'}g_{\alpha k}g_{\alpha k'}{\cal{C}}_{\alpha\alpha'}$. It is easy to verify

      Then we have

      Consequently,

      Substituting Eq.(C18) into Eq.(C13) and using Eq.(C3), we arrive at the desired result.

    Appendix D.   WICK’S THEOREM FOR FERMIONIC SUPEROPERATORS
    • In this Appendix, we simply review Wick’s theorem for fermionic superoperators. The Wick’s theorem is applied to reduce the multi-point average over the thermal state of fermionic operators to a two-point one. Here, we discuss the general case with arbitrary ordering of the involved operators.

      Generally, we are interested in evaluating

      Here, the average is over the equilibrium state $ \rho_{\rm{eq}} = {\rm{e}}^{-\beta H} / Z $ with $ Z = {\rm{Tr}}\big( {\rm{e}}^{-\beta H} \big) $ and

      And the involved superoperators are defined as

      with $ \sigma = \pm $ representing the creation/annihilation operators. In Eq.(D1), we use the index $ \lambda $ to label the left and right action. Using the fermionic commutators, we have ($ \bar{\sigma} \equiv -\sigma $)

      and

      To proceed, we denote [28]

      with $ \mathscr{P}(\cdot) \equiv \hat{P}(\cdot)\hat{P} $ and $\hat{P} \equiv \displaystyle\prod_k\exp(i\pi\hat{c}_k^\dagger\hat{c}_k)$. The corresponding anti-commutation relations read

      and

      which resemble the conventional fermionic algebra. For simplicity, we denote $ \lambda=1 $ for $ \gt $ and $ \lambda=-1 $ for <, and $ \bar{\lambda} \equiv -\lambda $. Then we have

      To present the Wick’s theorem, we define a certain ordering of a set of $ \{\mathscr{j}_{k}^{\sigma p}\} $,

      Here, $ \#_{\mathscr{p}_n} $ is the number of permutations to arrange the sequence $ \{1,2,\cdots,n\} $ into the ordered one $ \{\mathscr{p}_1,\mathscr{p}_2,\cdots,\mathscr{p}_n\} $ with $ \mathscr{p}_1\succ \mathscr{p}_2 \succ \cdots \succ \mathscr{p}_n $. Thus, the Wick's theorem reads (for $ n $ even) [28]

      Here, $ {\cal{P}}_n $ is the set of all possible ordered pairs $ (i,j) $ with $ i\succ j $ and $ (-1)^{\#_\mathscr{p}} $ is the sign for permuting the original order into the order $ \mathscr{p} $. For $ n $ odd, the average is zero.

    • We focus on the quantity

      Note that for an operator $ \hat{O} $,

      Thus we denote the normal ordering $ {\cal{N}} $ for superoperators by putting all $ \mathscr{j}_k^{\sigma >} $ to the right side of $ \mathscr{j}_k^{\sigma <} $. And for $ \sigma $ and $ k $, we order them by firstly arranging $ \sigma = + $ to the left side of $ \sigma = - $, and then arranging $ k $ in an ascending order. For example, we have

      where the minus sign is from odd number of permutations. From Eq.(D13), we know that

      Since the superoperators $ \{\mathscr{j}_k^{\sigma\lambda}\} $ satisfy the fermionic anti-commutation relation (Eq.(D9)), we can apply the general Wick’s theorem directly. Using Eqs. (C5) and (D15),

      Eq.(D11) is readily proved. Because the bath superoperators involved in Eq.(58) are linear combination of $ \{\mathscr{j}_k^{\sigma\lambda}\} $, Eq.(D11) also holds for them. Let $ {\cal{O}} $ be the time-ordering operator $ {\cal{T}}_{\rm{B}} $, then we arrive at Eq.(59).

    Figure (2)  Reference (36)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return