Data-driven construction of chemical reaction network graph using constrained LASSO

Abstract

This paper proposes a data-driven method for the automatic construction of chemical reaction network graph. For a given set of chemical species measurements and their (approximate) time derivatives, the proposed method first estimates the nonlinear dynamic model of the reaction using an L1 penalty type sparse identification approach called ’constrained least absolute shrinkage and selection operator’ (Constrained LASSO). Using the estimated model, a network graph construction approach that takes into account the dynamical constraints (e.g. non-negativity and law of mass action kinetics) of chemical reaction networks is then presented. Simulation result is also presented to illustrate the proposed method.