Path-space variational inference for non-equilibrium coarse-grained systems

详细信息    查看全文

• 作者：Vagelis Harmandaris^c ; ^d ; ^{harman@uoc.gr" class="auth_mail" title="E-mail the corresponding author} ; Evangelia Kalligiannaki^c ; ^{ekalligian@tem.uoc.gr" class="auth_mail" title="E-mail the corresponding author} ; Markos Katsoulakis^b ; ^{markos@math.umass.edu" class="auth_mail" title="E-mail the corresponding author} ; Petr Plechá ; č^a ; ^{plechac@math.udel.edu" class="auth_mail" title="E-mail the corresponding author}
• 关键词：Coarse graining ; Non-equilibrium ; Information metrics ; Machine learning ; Variational inference ; Stochastic optimization ; Time series ; Langevin dynamics
• 刊名：Journal of Computational Physics
• 出版年：2016
• 出版时间：1 June 2016
• 年：2016
• 卷：314
• 期：Complete
• 页码：355-383
• 全文大小：722 K

文摘

In this paper we discuss information-theoretic tools for obtaining optimized coarse-grained molecular models for both equilibrium and non-equilibrium molecular simulations. The latter are ubiquitous in physicochemical and biological applications, where they are typically associated with coupling mechanisms, multi-physics and/or boundary conditions. In general the non-equilibrium steady states are not known explicitly as they do not necessarily have a Gibbs structure.

The presented approach can compare microscopic behavior of molecular systems to parametric and non-parametric coarse-grained models using the relative entropy between distributions on the path space and setting up a corresponding path-space variational inference problem. The methods can become entirely data-driven when the microscopic dynamics are replaced with corresponding correlated data in the form of time series. Furthermore, we present connections and generalizations of force matching methods in coarse-graining with path-space information methods. We demonstrate the enhanced transferability of information-based parameterizations to different observables, at a specific thermodynamic point, due to information inequalities.

We discuss methodological connections between information-based coarse-graining of molecular systems and variational inference methods primarily developed in the machine learning community. However, we note that the work presented here addresses variational inference for correlated time series due to the focus on dynamics. The applicability of the proposed methods is demonstrated on high-dimensional stochastic processes given by overdamped and driven Langevin dynamics of interacting particles.