[Paper Review] Numerical analysis for inchworm Monte Carlo method: Sign problem and error growth

Zhenning Cai, Jianfeng Lu+1 more|arXiv (Cornell University)|Jun 13, 2020

Quantum and electron transport phenomena32 references4 citations

TL;DR

This paper provides a rigorous numerical analysis of the inchworm Monte Carlo method for open quantum systems, identifying and distinguishing two sources of error growth: the numerical sign problem (variance explosion in stochastic estimation) and error amplification (from time integration). It demonstrates that partial resummation in the inchworm method effectively suppresses the sign problem, leading to polynomially bounded error growth instead of exponential divergence, thereby enabling stable long-time simulations.

ABSTRACT

We consider the numerical analysis of the inchworm Monte Carlo method, which is proposed recently to tackle the numerical sign problem for open quantum systems. We focus on the growth of the numerical error with respect to the simulation time, for which the inchworm Monte Carlo method shows a flatter curve than the direct application of Monte Carlo method to the classical Dyson series. To better understand the underlying mechanism of the inchworm Monte Carlo method, we distinguish two types of exponential error growth, which are known as the numerical sign problem and the error amplification. The former is due to the fast growth of variance in the stochastic method, which can be observed from the Dyson series, and the latter comes from the evolution of the numerical solution. Our analysis demonstrates that the technique of partial resummation can be considered as a tool to balance these two types of error, and the inchwormMonte Carlo method is a successful case where the numerical sign problem is effectively suppressed by such means. We first demonstrate our idea in the context of ordinary differential equations, and then provide complete analysis for the inchworm Monte Carlo method. Several numerical experiments are carried out to verify our theoretical results.

Motivation & Objective

• To understand the mechanism by which the inchworm Monte Carlo method mitigates the numerical sign problem in real-time simulations of open quantum systems.
• To distinguish and analyze two distinct sources of error growth: the numerical sign problem (variance in Monte Carlo estimation) and error amplification (from time integration).
• To demonstrate that partial resummation in the inchworm method balances these two error types, leading to controlled, polynomially growing error instead of exponential divergence.
• To provide a theoretical foundation for the observed stability of the inchworm method in long-time simulations, validated by numerical experiments.

Proposed method

• The authors analyze the inchworm Monte Carlo method through a decomposition of error into two components: the numerical sign problem arising from high-dimensional, oscillatory integrals in the Dyson series, and error amplification from the time integration scheme.
• They apply a formal asymptotic analysis to the second-order derivatives of the propagator with respect to stochastic variables, identifying the leading-order contributions based on time interval overlaps and grid alignment.
• The method involves deriving scaling laws for the derivatives of the propagator functional with respect to Green's functions, using piecewise-constant approximations of the time-ordered exponential.
• The analysis distinguishes four cases based on the location of time nodes relative to the integration domain, deriving order-of-magnitude estimates (O(h^k)) for each derivative under different configurations.
• The authors use a Runge-Kutta time integrator combined with Monte Carlo estimation of the right-hand side, modeling the scheme as a stochastic differential equation with expectation terms.
• They validate the theoretical error scaling through numerical experiments on model systems, confirming the predicted polynomial growth of error with time.

Experimental results

Research questions

• RQ1 How does the inchworm Monte Carlo method suppress the numerical sign problem compared to direct Monte Carlo application to the Dyson series?
• RQ2 What are the distinct contributions of the numerical sign problem and error amplification to overall error growth in the inchworm method?
• RQ3 Why does the inchworm method exhibit flatter error growth curves than standard Monte Carlo methods in long-time simulations?
• RQ4 To what extent can partial resummation balance the two error sources, and how does this affect the long-time stability of the simulation?
• RQ5 What is the precise scaling of the error growth in terms of time and time step size in the inchworm Monte Carlo framework?

Key findings

• The numerical sign problem causes error to grow as exp(O(t²)), which is the dominant source of instability in standard Monte Carlo methods for open quantum systems.
• Error amplification from time integration contributes a slower, polynomially growing error component, which is less severe than the sign problem in isolation.
• The inchworm method reduces the effective exponent of the sign problem by partial resummation, resulting in overall error growth bounded by exp(O(t^p)) for p < 2.
• Theoretical analysis shows that the leading-order derivatives of the propagator scale as O(h^2) to O(h^4), depending on the relative positioning of time nodes and integration intervals.
• Numerical experiments confirm that the error in the inchworm method grows as a polynomial in time, not exponentially, validating the theoretical prediction of suppressed sign problem.
• The method achieves stable long-time simulations by balancing the two error sources through strategic resummation, making it effective for non-Markovian open quantum systems.