2025-11-21T15:43:16.363497

Weak-Memory Dynamics in Discrete Time

Meyer, Brandner
Discrete dynamics arise naturally in systems with broken temporal translation symmetry and are typically described by first-order recurrence relations representing classical or quantum Markov chains. When memory effects induced by hidden degrees of freedom are relevant, however, higher-order discrete evolution equations are generally required. Focusing on linear dynamics, we identify a well-delineated weak-memory regime where such equations can, on an intermediate time scale, be systematically reduced to a unique first-order counterpart acting on the same state space. We formulate our results as a mathematical theorem and work out two examples showing how they can be applied to stochastic Floquet dynamics under coarse-grained and quantum collisional models.
academic

Weak-Memory Dynamics in Discrete Time

Basic Information

  • Paper ID: 2510.26325
  • Title: Weak-Memory Dynamics in Discrete Time
  • Authors: Hugues Meyer and Kay Brandner (University of Nottingham)
  • Classification: cond-mat.stat-mech, cond-mat.mes-hall, quant-ph
  • Publication Date: October 31, 2025
  • Paper Link: https://arxiv.org/abs/2510.26325

Abstract

Discrete dynamics naturally emerge in systems with broken time-translation symmetry, typically described by first-order recurrence relations representing classical or quantum Markov chains. However, when memory effects arising from hidden degrees of freedom become significant, higher-order discrete evolution equations are generally required. This paper focuses on linear dynamics and identifies a well-defined weak-memory regime in which these equations can be systematically reduced to a unique first-order counterpart acting on the same state space at intermediate timescales. The authors formulate results as mathematical theorems and provide two examples demonstrating applications to coarse-grained stochastic Floquet dynamics and quantum collision models.

Research Background and Motivation

Problem Definition

  1. Core Problem: Physical systems naturally produce discrete dynamics when time-translation symmetry is broken (e.g., periodically driven systems). When observers can only access coarse-grained variables without observing all microscopic details, hidden degrees of freedom introduce memory effects, requiring the system evolution to be described by higher-order recurrence relations: Xn+1=VXn+∑m=1nKmXn−mX_{n+1} = VX_n + \sum_{m=1}^{n} K_m X_{n-m} where KmK_m is the memory kernel.
  2. Problem Significance:
    • Higher-order recurrence relations, though more accurate, lack physical transparency and are difficult to solve
    • Memory effects are ubiquitous in mesoscopic systems (e.g., quantum dots, molecular dynamics)
    • Need to balance accuracy with tractability
  3. Limitations of Existing Approaches:
    • First-order Markovian approximation ignores memory effects, insufficient accuracy
    • Complete higher-order equations are computationally expensive and lack physical intuition
    • Lack of systematic theory to determine when simplification is possible
  4. Research Motivation:
    • Generalize recently developed continuous-time weak-memory theory to discrete time
    • Establish rigorous mathematical framework determining boundaries of weak-memory regime
    • Provide systematic method to construct effective first-order approximations

Core Contributions

  1. Mathematical Theorem: Establishes main theorem for discrete-time weak-memory dynamics, providing necessary and sufficient conditions for reducing higher-order recurrence relations to first-order form
  2. Effective Generator Theory: Proposes concepts of effective generator GG and shift matrix DD, enabling long-time approximation Yn=GnDX0Y_n = G^n D X_0 to capture asymptotic behavior of original system
  3. Error Bounds: Establishes rigorous exponential decay error bounds: ∣Xn−Yn∣≤ζ−kη−ζ∣X0∣ζn|X_n - Y_n| \leq \frac{\zeta - k}{\eta - \zeta} |X_0| \zeta^n where ζ<1\zeta < 1 ensures error decays exponentially with time
  4. Constructive Algorithm: Provides explicit scheme for solving effective generator through fixed-point equation iteration
  5. Physical Applications: Validates theory in two important physical systems:
    • Coarse-grained dynamics of mesoscopic charge pump
    • Quantum collision model with memory

Methodology Details

Task Definition

Input: Higher-order recurrence relation Xn+1=VXn+∑m=1nKmXn−mX_{n+1} = VX_n + \sum_{m=1}^{n} K_m X_{n-m} where Xn∈CNX_n \in \mathbb{C}^N is state vector, V∈CN×NV \in \mathbb{C}^{N \times N} is free generator, KmK_m is memory kernel.

Output: Effective first-order recurrence relation Yn+1=GYn,Y0=DX0Y_{n+1} = GY_n, \quad Y_0 = DX_0 such that YnY_n approximates XnX_n in long-time limit.

Constraints (weak-memory conditions): ∥V−1∥≤1/v,∥Kn∥≤Mkn−1\|V^{-1}\| \leq 1/v, \quad \|K_n\| \leq Mk^{n-1}

(v-k)^2/4, & v+k < 2\\ (1-k)(v-1), & v+k \geq 2 \end{cases}$$ ### Theoretical Framework #### 1. Fixed-Point Equation for Effective Generator The effective generator $G$ satisfies fixed-point equation: $$G = V + \sum_{n=1}^{\infty} K_n G^{-n} = \mathcal{T}(G)$$ The physical intuition is that in the long-time limit, hidden degrees of freedom reach quasi-stationarity, and their feedback on observable variables gradually vanishes. #### 2. Adjoint Generator Define adjoint effective generator $H$ satisfying: $$H = V + \sum_{n=1}^{\infty} H^{-n} K_n = \mathcal{U}(H)$$ $G$ and $H$ are related through shift matrix $D$ via intertwining relation: $DG = HD$ #### 3. Shift Matrix The shift matrix describes effective renormalization of initial state: $$D^{-1} = \mathbb{1} + \sum_{m=1}^{\infty} \sum_{n=1}^{\infty} H^{-m} K_{m+n-1} G^{-n}$$ Physical meaning: Short-time transient processes are effectively absorbed into correction of initial conditions. ### Mathematical Proof Strategy #### Existence and Uniqueness 1. **Banach Fixed-Point Theorem**: Define set $\mathcal{B} = \{X \in \mathbb{C}^{N \times N}: \|X^{-1}\| \leq 1/\eta\}$, where $$\eta = \frac{v+k+\sqrt{(v-k)^2-4M}}{2}$$ 2. **Contraction Mapping**: Prove $\mathcal{T}$ is contraction mapping on $\mathcal{B}$ under metric $d(X,Y) = \|X-Y\|$, with contraction coefficient $$q = \frac{M}{(\eta-k)^2} < 1$$ 3. **Iterative Solution**: Starting from initial guess $G_0$ (e.g., $G_0 = V$), iteration $G_{\ell+1} = \mathcal{T}(G_\ell)$ converges to unique solution. #### Error Analysis 1. **Propagator Decomposition**: Define $Z_n = X_n/X_0$ (propagator), decompose as $Z_n = G^n D + R_n$ 2. **Remainder Estimate**: Prove by recursion $\|R_n\| \leq \|D\|q p^n$, where $$p = \frac{k(\eta-k)+M}{\eta(\eta-k)} < 1$$ 3. **Memory Function Bound**: Define memory function $E_n = G^{n+1}(A_{n+1}-A_n)$, prove $\|E_n\| \leq (\zeta-k)\zeta^n$, where $$\zeta = \frac{v+k-\sqrt{(v-k)^2-4M}}{2} < 1$$ ### Technical Innovations 1. **Discrete-Time Generalization**: First to generalize weak-memory theory from continuous-time systems to discrete time, overcoming difficulties from lack of differential structure in discrete systems 2. **Optimality Proof**: Prove uniqueness of effective generator—any other generator satisfying asymptotic conditions leads to slower convergence rate 3. **Constructive Methods**: Provide two approaches to construct effective generator: - Direct iteration of fixed-point equation - Perturbation expansion via memory function 4. **Physical Parameter Identification**: Clarify physical meaning of three timescales: - $v$: slowest mode of adiabatic dynamics - $k$: relaxation rate of hidden degrees of freedom - $M$: coupling strength between observable and unobservable parts ## Experimental Setup ### Application 1: Mesoscopic Charge Pump #### System Description - **Physical System**: Two quantum dots coupled to thermalized reservoirs, three microscopic states (left dot occupied, right dot occupied, neither occupied) - **Operating Cycle**: Three-step process - (a) Left dot absorbs particle (probability $L_+$) - (b) Tunneling between dots (probability $L_0$) - (c) Right dot releases particle (probability $L_-$) #### Coarse-Graining - **Microscopic States**: $P_n = [P_n^{◦◦}, P_n^{•◦}, P_n^{◦•}]^T$ - **Mesoscopic States**: $X_n = [X_n^◦, X_n^•]^T$, where $X_n^◦ = P_n^{◦◦}$, $X_n^• = P_n^{•◦} + P_n^{◦•}$ - **Evolution Equation**: Derived via projection operator method #### Parameter Settings - Free generator: $V = \mathbb{1} + L_+(1-L_0L_-)J_1 + \frac{L_-}{2}J_2$ - Memory kernel: $K_n = \frac{L_+L_-(1-2L_0+2L_-)k^n}{2-L_-}J_1 - \frac{L_-^2k^n}{2(2-L_-)}J_2$ - Decay rate: $k = \frac{(1-2L_0)(2-L_-)}{2}$ where $J_1 = \begin{bmatrix}-1 & 0\\ 1 & 0\end{bmatrix}$, $J_2 = \begin{bmatrix}0 & 1\\ 0 & -1\end{bmatrix}$ ### Application 2: Quantum Collision Model #### System Description - **Composition**: System qubit $S$ sequentially collides with series of auxiliary qubits $A_i$ - **Memory Mechanism**: Interactions exist between auxiliary qubits ($A_1$ collides with $S$ then with $A_2$) - **Interaction Form**: - System-auxiliary: unitary partial swap $\mathcal{U}_i ◦ = U_i ◦ U_i^\dagger$, $U_i = \sqrt{1-u} - i\sqrt{u}S_{Si}$ - Auxiliary-auxiliary: incoherent partial swap $\mathcal{Q}_{ij}◦ = (1-k)◦ + kS_{ij}◦S_{ij}$ #### Evolution Equation Density matrix $\rho_n$ satisfies: $$\rho_{n+1} = K_0\rho_n + \sum_{m=1}^{n} K_m\rho_{n-m} + F_{n+1}\rho_0$$ where: - $K_n = (1/k-1)F_{n+1}$ - $F_n ◦ = k^n \text{Tr}_1[U_1^n(◦ \otimes \xi_1)]$ #### Long-Time Approximation Includes correction for inhomogeneous term: $$\chi_n = G^n D\rho_0 + \sum_{m=1}^{n} G^{n-m} D F_m \rho_0$$ ## Experimental Results ### Application 1: Charge Pump Results #### Weak-Memory Regime - **Parameter Space**: On $(L_0, L_+)$ plane, weak-memory conditions define clear region - **Boundary Characteristics**: $L_+$ and $L_-$ must be sufficiently small (weak coupling), $L_0$ far from $1/2$ (non-equilibrium) - **Physical Interpretation**: When $L_0 = 1/2$, charge distribution fully equilibrates within one period, memory kernel vanishes #### Error Analysis - **Exponential Decay**: Error $|X_n - Y_n|$ decays exponentially with $n$ - **Tightness of Theoretical Bounds**: Numerical results closely follow theoretical error bounds (dashed lines) - **Typical Parameters**: With $L_+ = L_- = 0.3$, $L_0 = 0.3$, error drops to $10^{-3}$ order after $n \sim 5$ #### Effective Generator Solution - **Analytical Solution**: Due to simple structure of memory kernel $K_n = k^{n-1}K_1$, fixed-point equation reduces to quadratic matrix equation - **Parametrization**: Using Ansatz $G = \mathbb{1} + g_1J_1 + g_2J_2$, solve for $(g_1, g_2)$ - **Shift Matrix**: Computed explicitly via formula (10) ### Application 2: Collision Model Results #### Weak-Memory Regime - **Parameter Space**: $(u, k)$ plane, where $u = \sin^2\theta$ is system-auxiliary swap probability, $k$ is auxiliary-auxiliary swap probability - **Boundary Shape**: Approximately rectangular region, requiring both $u$ and $k$ small - **Physical Conditions**: $v = 1/\|K_0^{-1}\|$, $M = k - k^2$ #### Iterative Convergence - **Iteration Scheme**: $G_{\ell+1} = \mathcal{T}(G_\ell)$, initial guess $G_0 = K_0$ - **Convergence Behavior**: - Early iterations: Error decays rapidly exponentially, following theoretical bounds - Plateau reached: Error stops at $\sim 10^{-\ell}$ order, where $\ell$ is iteration count - Plateau height: Decreases exponentially with iteration count #### Numerical Efficiency - **Typical Parameters**: $u = 0.15$, $k = 0.2$, $\alpha = 0.3$ (auxiliary initial state parameter) - **Accuracy Achieved**: 3-4 iterations suffice for $10^{-3}$ precision - **Computational Cost**: Each iteration requires summing infinite series, but series converges rapidly (exponential decay) ### Key Findings 1. **Tightness of Theoretical Bounds**: In both applications, numerical errors closely follow theoretical upper bounds, indicating error estimates are optimal 2. **Universality of Weak-Memory Regime**: Even when $M$, $k$, $v$ are of same order, weak-memory conditions can be satisfied—strong timescale separation not required 3. **Effectiveness of Iterative Method**: Direct iteration of fixed-point equation provides efficient numerical solution scheme 4. **Physical Transparency**: Effective generator $G$ preserves physical structure of original system (e.g., stochasticity, positive-definiteness) ## Related Work ### Continuous-Time Memory Dynamics - **Mori-Zwanzig Formalism**: Classical projection operator method yielding generalized Langevin equations - **Recent Progress**: Continuous-time weak-memory theory developed by Brandner et al. (2025) [59,60] - **Relation to This Work**: Generalizes continuous-time framework to discrete time, overcoming difficulties from lack of differential structure ### Coarse-Graining Methods - **Classical Systems**: - Pigolotti & Vulpiani (2008): Master equation coarse-graining with fast-slow state separation [32] - Esposito (2012): Stochastic thermodynamics under coarse-graining [33] - Hummer & Szabo (2015): Optimal dimensionality reduction for Markovian state models [35] - **Quantum Systems**: - Lapolla & Godec (2019, 2021): Manifestation of projection-induced memory [37,45] - Hartich & Godec (2021, 2023): Emergent memory in strongly driven networks [44,46] - **Advantages of This Work**: Provides rigorous mathematical conditions and constructive algorithms rather than heuristic approaches ### Floquet Systems - **Periodic Driving**: Bukov et al. (2015), Eckardt (2017), Oka & Kitamura (2019) [1-3] - **Open Systems**: Menczel & Brandner (2019), Mori (2023) [6,7] - **Contribution of This Work**: First systematic treatment of memory effects in coarse-grained Floquet dynamics ### Collision Models - **Foundational Theory**: Karevski & Platini (2009), Ciccarello et al. (2022) [27,68] - **Strong Coupling**: Strasberg (2019), Strasberg et al. (2017) [28,29] - **Extension in This Work**: Addresses memory effects introduced by interactions between auxiliary qubits ### Discrete-Time Quantum Circuits - **Entanglement Propagation**: Nahum et al. (2017, 2018), Khemani et al. (2018) [13-16] - **Solvable Models**: Rule 54 quantum cellular automaton [21-24] - **Potential Applications**: Theory may apply to coarse-graining of many-body circuit models ## Conclusions and Discussion ### Main Conclusions 1. **Main Theorem**: Establishes necessary and sufficient conditions for reducing discrete-time higher-order recurrence relations to first-order form, including: - Weak-memory conditions: $k < v$, $M < M^*(k,v)$ - Unique effective generator: satisfying fixed-point equation $G = \mathcal{T}(G)$ - Exponential error bounds: $|X_n - Y_n| \leq C|X_0|\zeta^n$, $\zeta < 1$ 2. **Constructive Methods**: Provide three solution approaches: - Fixed-point equation iteration - Analytical solution based on Ansatz - Perturbation expansion via memory function 3. **Physical Verification**: Validates theory in two important physical systems ### Limitations 1. **Linearity Restriction**: Current theory applies only to linear dynamics ($X_n \in \mathbb{C}^N$ linear evolution) - Nonlinear systems require more complex mathematical tools - Many physical systems are inherently nonlinear 2. **Weak-Memory Assumption**: Requires $M$ sufficiently small relative to $v-k$ - Strong-memory systems (e.g., glassy states, long-range correlations) not applicable - May violate conditions near critical points 3. **Finite Dimensionality**: State space $\mathbb{C}^N$ is finite-dimensional - Infinite-dimensional systems (e.g., field theory) require functional analysis tools - Behavior in thermodynamic limit not addressed 4. **Initial-State Dependence**: Physical meaning of shift matrix $D$ depends on initial state choice - Different initial states may require different effective descriptions - Treatment of non-equilibrium steady states requires further research ### Future Directions 1. **Nonlinear Generalization**: - Develop nonlinear weak-memory theory - Explore possibilities of Volterra series expansion 2. **Many-Body Systems**: - Apply to coarse-graining of quantum circuits - Study memory effects in entanglement propagation 3. **Thermodynamic Applications**: - Entropy production under coarse-graining - Thermodynamic cost of hidden pumps 4. **Metastable States**: - Metastable states induced by hidden degrees of freedom - Memory effects in dynamical phase transitions 5. **Experimental Verification**: - Floquet engineering in cold atom systems - Collision model implementation on quantum computing platforms ## In-Depth Evaluation ### Strengths #### 1. Mathematical Rigor - **Complete Proofs**: Main theorem proof uses Banach fixed-point theorem with rigorous logic - **Optimal Error Bounds**: Not only proves convergence but provides optimal exponential decay rate - **Uniqueness**: Proves uniqueness of effective generator, eliminating arbitrariness #### 2. Physical Insight - **Timescale Separation**: Clearly identifies meaning of three physical timescales $(v, k, M)$ - **Memory Mechanism**: Reveals physical origin of memory effects—slow relaxation of hidden degrees of freedom - **Universality**: Theory applies to both classical and quantum systems #### 3. Practical Value - **Constructive Algorithm**: Provides directly implementable numerical method - **Parameter Identification**: Gives explicit criteria for determining weak-memory regime - **Model Validation**: Two detailed examples demonstrate practical application workflow #### 4. Theoretical Innovation - **Discrete-Time Generalization**: First systematic treatment of discrete-time memory dynamics - **Optimal Approximation**: Proves effective generator corresponds to fastest-decaying memory function - **Intertwining Relation**: Discovers dual structure between $G$ and $H$ ### Weaknesses #### 1. Limited Application Scope - **Linearity Assumption**: Many physical systems (nonlinear oscillators, turbulence) do not satisfy this - **Finite Dimensionality**: Continuous spectrum systems (quantum field theory) cannot directly apply - **Weak-Memory Limitation**: Strongly correlated systems may fall outside theory's applicability #### 2. Numerical Implementation Details - **Infinite Series Truncation**: Practical computation requires truncation, but truncation error analysis insufficient - **Matrix Inversion**: For large-scale systems, matrix inversion may be numerically unstable - **Convergence Speed**: Iteration convergence speed depends on compression coefficient $q$, may be slow near boundary #### 3. Physical Interpretation - **Shift Matrix**: Physical meaning of $D$, particularly off-diagonal elements, not sufficiently intuitive - **Non-Equilibrium Steady States**: Theory mainly addresses transient dynamics, steady-state properties discussed less - **Thermodynamic Consistency**: Entropy production under coarse-graining not discussed #### 4. Experimental Verifiability - **Parameter Measurement**: How to precisely measure $v, k, M$ experimentally not specified - **Error Verification**: Experimental errors may mask theoretically predicted exponential decay - **System Implementation**: Particularly collision model experimental implementation is challenging ### Impact #### Contribution to Field 1. **Theoretical Framework**: Provides systematic theoretical foundation for discrete-time non-Markovian dynamics 2. **Bridge Role**: Connects microscopic complete description to mesoscopic effective theory 3. **Methodology**: Demonstrates how to generalize continuous-time techniques to discrete time #### Practical Value 1. **Numerical Simulation**: Provides dimensionality reduction method for molecular dynamics, quantum circuit simulation 2. **Model Simplification**: Helps identify when Markovian approximation can be safely used 3. **Parameter Estimation**: Enables inference of microscopic parameters from coarse-grained data #### Reproducibility - **Mathematical Formulas**: All formulas explicit, easy to implement - **Example Code**: While code not provided, algorithm steps are clear - **Supplementary Material**: Detailed supplementary material contains complete proofs ### Applicable Scenarios #### Ideal Application Scenarios 1. **Periodically Driven Systems**: - Coarse-grained description in Floquet engineering - Mesoscopic models of quantum pumps 2. **Sequential Interactions**: - Quantum collision models - Cascaded quantum systems 3. **Fast-Slow Separation**: - Coarse time-stepping in molecular dynamics - Multiscale simulation #### Scenarios Requiring Caution 1. **Strongly Correlated Systems**: Near critical points, quantum phase transitions 2. **Long-Range Memory**: Power-law decaying memory kernels 3. **Nonlinear Systems**: Chaotic systems, soliton equations 4. **Open Boundaries**: Non-periodic driving, non-stationary processes ## References ### Key Citations 1. **Brandner (2025)**: Phys. Rev. Lett. 134, 037101 - Continuous-time weak-memory theory 2. **Ciccarello et al. (2022)**: Phys. Rep. 954, 1 - Quantum collision models review 3. **Esposito & Parrondo (2015)**: Phys. Rev. E 91, 052114 - Stochastic thermodynamics of hidden pumps 4. **Schilling (2022)**: Phys. Rep. 972, 1 - Non-equilibrium coarse-graining modeling ### Mathematical Tools 5. **Burton (2005)**: Volterra Integral and Differential Equations - Volterra equation theory 6. **Agarwal et al. (2018)**: Fixed Point Theory in Metric Spaces - Fixed-point theory 7. **Horn & Johnson (2013)**: Matrix Analysis - Matrix analysis --- **Overall Assessment**: This is an excellent paper with rigorous mathematics and clear physical intuition. The main theorem's proof is complete using Banach fixed-point theorem, and physical applications are convincing. While limited by linearity and weak-memory assumptions, within applicable scope it provides powerful tools. For researchers studying mesoscopic systems, periodically driven systems, and quantum information, this represents important theoretical progress. The constructive methods make theory operationally practical, with expected applications in numerical simulation and experimental design.