Learning Dynamic Boltzmann Distributions as Reduced Models of Spatial Chemical Kinetics

Abstract

Finding reduced models of spatially-distributed chemical reaction networks requires an estimation of which effective dynamics are relevant. We propose a machine learning approach to this coarse graining problem, where a maximum entropy approximation is constructed that evolves slowly in time. The dynamical model governing the approximation is expressed as a functional, allowing a general treatment of spatial interactions. In contrast to typical machine learning approaches which estimate the interaction parameters of a graphical model, we derive Boltzmann-machine like learning algorithms to estimate directly the functionals dictating the time evolution of these parameters. By incorporating analytic solutions from simple reaction motifs, an efficient simulation method is demonstrated for systems ranging from toy problems to basic biologically relevant networks. The broadly applicable nature of our approach to learning spatial dynamics suggests promising applications to multiscale methods for spatial networks, as well as to further problems in machine learning.

References in corpus (8)

• Spatial Networks
• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• Solving the 3D Ising Model with the Conformal Bootstrap
• Theory of Branching and Annihilating Random Walks
• Conformal Field Theories in Fractional Dimensions
• A Mean-Field Approach to Evolving Spatial Networks, with an Application to Osteocyte Network Formation
• Time-Ordered Product Expansions for Computational Stochastic Systems Biology
• Brownian dynamics simulations of an idealized chemical reaction network under spatial confinement and crowding conditions

Cited by in corpus (4)

• TMI: Thermodynamic inference of data manifolds
• Prospects for Declarative Mathematical Modeling of Complex Biological Systems
• Learning Moment Closure in Reaction-Diffusion Systems with Spatial Dynamic Boltzmann Distributions
• Deep Learning Moment Closure Approximations using Dynamic Boltzmann Distributions