We develop a data-driven method to learn chemical reaction networks from trajectory data. Modeling the reaction system as a continuous-time Markov chain and assuming the system is fully observed, our method learns the propensity functions of the system with predetermined basis functions by maximizing the likelihood function of the trajectory data under l1 sparse regularization. We demonstrate our method with numerical examples using synthetic data and carry out an asymptotic analysis of the proposed learning procedure in the infinite-data limit.
@article{arxiv.1902.04920,
title = {Learning chemical reaction networks from trajectory data},
author = {Wei Zhang and Stefan Klus and Tim Conrad and Christof Schütte},
journal= {arXiv preprint arXiv:1902.04920},
year = {2019}
}