Structural obstruction to the simplicity of the eigenvalue zero in chemical reaction networks

TL;DR

This paper demonstrates that certain chemical reaction networks inherently prevent the eigenvalue zero of their Jacobian from being simple, posing a fundamental obstacle to typical saddle-node bifurcations regardless of kinetic choices.

Contribution

It introduces a structural obstruction in chemical reaction networks that hinders the eigenvalue zero from being simple, independent of concentrations and kinetics.

Findings

A specific network example with non-simple eigenvalue zero

Structural properties cause eigenvalue zero multiplicity

Obstruction exists regardless of kinetic parameters

Abstract

Multistationarity is the property of a system to exhibit two distinct equilibria (steady-states) under otherwise identical conditions, and it is a phenomenon of recognized importance for biochemical systems. Multistationarity may appear in the parameter space as a consequence of saddle-node bifurcations, which necessarily require a simple eigenvalue zero of the Jacobian, at the bifurcating equilibrium. Matrices with a simple eigenvalue zero are generic in the set of singular matrices: any system whose Jacobian has an algebraically multiple eigenvalue zero can be perturbed to a system whose Jacobian has a simple eigenvalue zero. Thus, one would expect that in applications singular Jacobians are always with a simple eigenvalue zero. However, chemical reaction networks typically consider a fixed network structure, while the freedom rests with the various and different choices of kinetics. Here we present an example of a chemical reaction network, whose Jacobian is either nonsingular or has an algebraically multiple eigenvalue zero. The structural obstruction to the simplicity of the eigenvalue zero is based on the network alone, and it is independent of the value of concentrations and the choice of kinetics. This in particular constitutes an obstruction to standard saddle-node bifurcations.