Repository files navigation

This contains code used to generate figures in the paper "Deep Learning Moment Closure Approximations using Dynamic Boltzmann Distributions":

arXiv 1905.12122

The dependencies are automatically managed using theCPM.cmake package manager.There is no need to manually install dependencies; just proceed toContents below.

For completeness, the dependencies downloaded automatically are:

• DynamicBoltzmann library v4.5here.
• Q3 C1 Finite Elements library v3.0here.
• LatticeGillespie C++ library v2.0here.
• Armadillo library v9.300.2here.

• stoch_sims contains code to generate the stochastic simulations.
• learn contains code to train the dynamic Boltzmann distribution.

Other (not used in paper) contents:

• ode is a Mathematica notebook to solve the ODE system for a well-mixed Lotka-Volterra system.
• ssa is a Mathematica notebook to generate stochastic simulations using the Gillespie algorithm for the well-mixed Lotka-Volterra system. This requires the Gillesipe module for Mathematica, availablehere.