A passivity-based stability criterion for a class of biochemical reaction networks (Q932455)

A network of interconnected systems is studied, where each single system is governed by an ordinary differential equation of the form \[ \dot{x} = -f(x) + g(x) h(u), \] where \(h(u)\) is the sum of (nonlinear) influences of other single systems. Based on the network structure and assumptions on the functions \(f,g,h\) a dissipativity matrix \(E\in\mathbb{R}^{l\times l}\) is definied, where \(l\in\mathbb{N}\) is the number of overall connections. This matrix contains outside the diagonal only the values \(-1,0,1\), which represant the (directed) connections, and the diagonal contains some upper bound of the ``influence'' of the other systems on the correspoding single system. Sufficient conditions (for example a condition which ensures positivity of the variables) are given for which the equillibrium of the interconnected system is asymptotically stable. Further conditions are given for which the equillibrium is globally asymptotically stable. The results are illustrated with extensive examples from biological cell processes. Finally, the results are applied to study the robustness of stability for compartmental models.