Coordinate-independent model reductions of chemical reaction networks based on geometric singular perturbation theory (2508.03304v1)

Abstract: The quasi-steady-state approximation (QSSA) is a standard technique for reducing the complexity of chemical reaction networks (CRNs). The validity of any QSSA-based model is restricted to specific parameter regimes, which often overlap, meaning multiple different reduced models can be simultaneously valid. This ambiguity complicates unnecessarily the analysis, as selecting the appropriate reduction is not always straightforward. Here, we employ a more powerful alternative: coordinate-independent geometric singular perturbation theory (ci-GSPT) accompanied by the parametrization method. A key advantage of this framework is its ability to derive reduced models independent of a clear timescale separation in the variables which, in turn, means less-restricted parameter regimes. We demonstrate our approach on two benchmark systems. For the Michaelis-Menten (MM) reaction, we derive a unique model reduction for normally hyperbolic and attracting critical manifolds of a specific parameter configuration. We systematically explore parameter configurations across three orders of magnitude: asymptotically large, small, and `order one'. Hence, we provide distinct model reductions for: (i) 14 relevant parameter configurations of the irreversible reaction scheme and (ii) 25 for the reversible reaction scheme (other parameter configurations where there are trivial reductions or degeneracies in critical manifolds, are discussed briefly). For the more complex Kim-Forger model, we provide a new reduction without the need of a coordinate transformation showcasing the method's applicability to larger systems.