Model Reduction of Multivariate Geometric Brownian Motions and Localization in a Two-State Quantum System

TL;DR

This paper introduces a systematic method for reducing multivariate geometric Brownian motions using invariant manifolds and adiabatic elimination, with an application to quantum systems demonstrating effective localization modeling.

Contribution

It presents a novel framework combining invariant manifolds and adiabatic elimination for model reduction of multivariate GBMs, including an extended fluctuation-dissipation theorem formulation.

Findings

Reduced equations accurately capture localization in quantum systems

Significantly simplifies analysis of complex stochastic models

Framework applicable to various fields like finance and physics

Abstract

We develop a systematic framework for the model reduction of multivariate geometric Brownian motions (GBMs), a fundamental class of stochastic processes with broad applications in mathematical finance, population biology, and statistical physics. Our approach leverages the interplay between the method of invariant manifolds and adiabatic elimination to derive closed-form reduced equations for the deterministic drift. An extended formulation of the fluctuation-dissipation theorem is subsequently employed to characterize the stochastic component of the reduced description. As a concrete application, we apply our reduction scheme to a GBM arising from a two-state quantum system, showing that the reduced dynamics accurately capture the localization properties of the original model while significantly simplifying the analysis.