Computing time-dependent reduced models for classical and quantum dynamics
Tommaso Grigoletto · 2025
Read Computing time-dependent reduced models for classical and quantum dynamics by Tommaso Grigoletto (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-20675.txt.
Introduction
Tommaso Grigoletto, working alone, offers a new way to build small stand in models for large systems whose full motion is too costly to track. A fixed smaller space holds the stand in, but its driving rule must change over time to stay faithful to the full system.
His recursive method builds that time changing rule as a polynomial in time, drawn from the growth of the exponential map and a new result on ordered exponentials of such rules. The approach avoids assuming weak links between parts, a common crutch in quantum master equations. On test models from spin systems it stays accurate at short times, beats plain cut off methods at low orders, and keeps key physical guarantees.
Catalog record
| CREATOR | Tommaso Grigoletto |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 68 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-20675
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Computing time-dependent reduced models for classical and quantum dynamics Tommaso Grigoletto Abstract This paper introduces a novel method for approximating the dynamics of a large autonomous system projected onto a fixed subspace. The core contribution is a novel recursive algorithm to construct an effective time-dependent generator that is polynomial in the time variable, ensuring accuracy for short time scales. The derivation is based on the Taylor expansion of the exponential map and a new result for computing the time-ordered exponential of polynomial generators. This work is motivated by the challenge of deriving time-convolutionless master equations in quantum physics and the proposed method offers an alternative to typical derivations based on expansions in the coupling strength. The resulting approximation is accurate for small times, does not require a weak-coupling assumption, performs better than a truncation of the exponential map at low orders, and crucially, guarantees a completely positive and trace-preserving map at the lowest orders. The proposed method is validated against several prototypical models: a dephasing spin-boson model, a central spin model, and an Ising spin chain. I Introduction Whenever a dynamical model is too large to be simulated efficiently, one should resort to model reduction techniques to find a smaller model that is simulatable and that retains fundamental characteristics of the original system. A dominant approach in control systems theory – including, for example, minimal realization techniques rosenbrock1970state , balanced truncation antoulas ; gugercin2004survey , and moment matching astolfi_moment_matching ; padoanmomentmatching – consists of finding a low-dimensional subspace where to project the dynamics, obtaining a linear time-invariant (LTI) model. This work considers a different approach: the space onto which the dynamics is projected is assumed to be given a priori and fixed , and, instead, the dynamics is approximated by seeking a linear time-varying (LTV) model. This approach is inspired by quantum physics. To obtain realistic models of open quantum systems, one shall consider the effects of the interaction between the system and its environment breuer2002theory . Unfortunately, simulating the system-environment couple is impractical (if not impossible), as the scaling of the Hilbert space makes such a simulation prohibitive even for small environments. This implies that, in order to study the effect of the coupling with the environment, one needs to find a reduced model that (hopefully) captures the interesting features of the interaction. A common method to reduce the joint system-environment model is known in the physics literature with the name of time-convolutionless master equations (TCL ME) shibata_generalized_1977 ; chaturvedi_time-convolutionless_1979 . This method aims at obtaining an LTV model of the same size of the system that replicates the effects of the interaction with the environment. Although a formal exact solution to this problem exists, computing this exact LTV generator is generally considered as difficult as solving the original model. The main approach found in the physics literature approximates this formal solution by leveraging an expansion in the strength of the coupling between the system and the environment, which is typically assumed to be small. For more details, see breuer2002theory and references therein. In this work an alternative approach is proposed: instead of an expansion in the coupling strength, an expansion in time is considered. In particular, since the formal solution provides an analytic LTV generator in the time variable t , the exact incomputable generator is approximated with its truncation to a finite order N . Under these assumptions a recursive formula for the coefficients of the polynomial generator is found, and the approximation error of the resulting reduced model scales as O(t^{N+1}) . Note that while time-dependent dynamical systems are typically more difficult to study than time-independent ones, this approach finds an important application in the simulation of very large systems, as the Runge-Kutta methods allow one to simulate time-dependent models ascher1997implicit . In this regard, under the assumption of a time-dependent generator that is polynomial in t , this paper proposes a novel recursive formula to compute the propagator of this LTV model. Another benefit of the proposed reduction method consists in the fact that, when computing the reduced LTV generator for unitarily evolving quantum systems, the reduced polynomial generator at first and second order is provably in Lindblad form for all times t thus ensuring that the reduced model at the lowest orders generates a completely positive (CP) and trace preserving (TP) semigroup. It is important to point out that the method proposed in this work is not intended to substitute the perturbative approach for approximating TCL MEs shibata_generalized_1977 ; chaturvedi_time-convolutionless_1979 but rather to provide an alternative that can be of use whenever the main assumptions of the standard method do not hold, e.g. in the strong-coupling regime or whenever the time-ordered integrals are difficult to assess. Structure and paper contributions: For the reader’s convenience, the structure of this work and its main contributions are summarized next: • In Sec. II the problem treated in this work is defined rigorously. The known formal solution of the problem is derived in detail using the Nakajima-Swanzig equations nakajima ; zwanzig , and the key difficulties associated with computing this LTV model are explained. • In Sec. III the main contribution of the paper is presented: Theorem 2 derives the polynomial approximation of the formal LTV model and the recursive equation that determines its coefficients. This result is based on Theorem 1 where a novel method to compute the time-ordered exponential is derived. The reduction procedure is also validated on a linear model and compared against the truncation of the exponential series. • Sec. IV focuses on the application of the proposed reduction procedure to quantum dynamical models. The proposed method is connected and compared with the typical physics approach for the approximation of TCL ME which is based on the expansion of the coupling strength. The connections between the recursive formula obtained here and other similar recursive equations known in the physics literature nestmann2019time ; cerrillo2014non ; cygorek2025timenonlocalversustimelocallongtime ; gasbarri2018recursive are discussed, highlighting how the expansion in t greatly simplifies the associated calculations. Most importantly, leveraging a few recent results grigoletto2024exact ; grigoletto2025quantummodelreductioncontinuoustime it is proven that when the original generatos is purely Hamiltonian, the proposed approximation at first and second order provides a CPTP two-parameter semigroup. • In Sec. V the proposed procedure is validated against several prototypical quantum examples: a dephasing spin boson model, a central spin model, and and Ising spin chain. Note that the majority of the applications presented in this work are in the context of quantum physics since, up to this point, this is where these LTV-based model reduction techniques have found most applications. Throughout the paper, pedagogical examples are included to intuitively explain the most involved proofs. Furthermore, both the strengths and weaknesses of the proposed approximation procedure are highlighted in detail, namely: the recursive formula used to compute the terms of the generators at different orders makes the procedure numerically easier to access at higher orders than the alternatives; allows for an initial model that is in Lindblad form (not only purely Hamiltonian); does not require the model to be put in rotating frame; and does not rely on a weak coupling assumption, thus allowing to study systems in strong coupling regimes. On the other hand, the proposed reduction