Classical quasi-steady state reduction -- A mathematical characterization

TL;DR

This paper provides a rigorous mathematical framework for classical quasi-steady state (QSS) reduction in chemical reaction networks, characterizing when it is valid and how it relates to singular perturbation methods.

Contribution

It characterizes QSS parameter values using polynomial equations, compares QSS and singular perturbation reductions, and offers an algorithmic approach to determine valid reduction ranges.

Findings

QSS reduction is accurate at QSS parameter values.

QSS parameter ranges can be computed algorithmically.

QSS and singular perturbation reductions agree at lowest order for many cases.

Abstract

We discuss parameter dependent polynomial ordinary differential equations that model chemical reaction networks. By classical quasi-steady state (QSS) reduction we understand the following familiar heuristic: Set the rate of change for certain (a priori chosen) variables equal to zero and use the resulting algebraic equations to obtain a system of smaller dimension for the remaining variables. This procedure will generally be valid only for certain parameter ranges. We start by showing that the reduction is accurate if and only if the corresponding parameter is what we call a QSS parameter value, and that the reduction is approximately accurate if and only if the corresponding parameter is close to a QSS parameter value. These QSS parameter values can be characterized by polynomial equations and inequations, hence parameter ranges for which QSS reduction is valid are accessible in an algorithmic manner. A closer investigation of QSS parameter values and the associated systems shows the existence of certain invariant sets; here singular perturbations enter the picture in a natural manner. We compare QSS reduction and singular perturbation reduction, and show that, while they do not agree in general, they do, up to lowest order in a small parameter, for a quite large and relevant class of examples. This observation, in turn, allows the computation of QSS reductions even in cases where an explicit resolution of the polynomial equations is not possible.