A Projection Argument for Differential Inclusions, with Applications to Persistence of Mass-Action Kinetics

Motivated by questions in mass-action kinetics, we introduce the notion of vertexical family of differential inclusions. Defined on open hypercubes, these families are characterized by particular good behavior under projection maps. The motivating examples are certain families of reaction networks – including reversible, weakly reversible, endotactic, and strongly endotactic reaction networks – that give rise to vertexical families of mass-action differential inclusions. We prove that vertexical families are amenable to structural induction. Consequently, a trajectory of a vertexical family approaches the boundary if and only if either the trajectory approaches a vertex of the hypercube, or a trajectory in a lower-dimensional member of the family approaches the boundary. With this technology, we make progress on the global attractor conjecture, a central open problem concerning mass-action kinetics systems. Additionally, we phrase mass-action kinetics as a functor on reaction networks with variable rates.

💡 Research Summary

The paper introduces a novel structural concept for families of differential inclusions called “vertexical families.” These are collections of differential inclusions defined on open hypercubes (0,1)^n that behave well under coordinate‑projection maps. The defining property is that whenever a trajectory of a member of the family approaches a boundary face of the hypercube, its projection onto the coordinates that remain interior yields a trajectory of a lower‑dimensional member of the same family. This projection compatibility makes the families amenable to a form of structural induction: one can reduce questions about the full‑dimensional system to analogous questions about systems of smaller dimension.

The authors then show that several important classes of reaction networks in mass‑action kinetics give rise to vertexical families of mass‑action differential inclusions. The classes include reversible networks, weakly reversible networks, endotactic networks, and strongly endotactic networks. By allowing the kinetic rate constants to be arbitrary positive functions of time (variable rates), the usual mass‑action ODEs are recast as differential inclusions whose right‑hand side at each point consists of all possible velocity vectors compatible with the admissible rates. The geometric conditions defining endotacticity (all reaction vectors lie in a common half‑space determined by each supporting hyperplane) are preserved under coordinate projection, which is the key to establishing the vertexical property for these networks.

The central technical result, the “Boundary‑Approach Theorem,” states that for any vertexical family, a trajectory can approach the boundary of the hypercube only in two ways: (1) it converges to a vertex of the hypercube (i.e., all coordinates tend to 0 or 1), or (2) its projection onto a proper subset of coordinates yields a trajectory of a lower‑dimensional member that itself approaches the boundary. Consequently, to prove persistence (the property that no species concentration tends to zero) for a high‑dimensional system it suffices to prove persistence for all its lower‑dimensional projections. This eliminates the need for intricate “boundary‑flow” analyses that have traditionally been required in the study of chemical reaction network dynamics.

Using this machinery, the authors make progress on the Global Attractor Conjecture (GAC), a long‑standing open problem in chemical reaction network theory. The GAC asserts that for any complex‑balanced mass‑action system, every positive trajectory converges to the unique complex‑balanced equilibrium within its stoichiometric compatibility class. By applying the vertexical framework to weakly reversible and strongly endotactic networks, the paper establishes both persistence and boundedness for these classes. Persistence prevents trajectories from approaching the coordinate hyperplanes, while boundedness prevents escape to infinity. Together with standard Lyapunov‑function arguments (e.g., the Horn–Jackson function) and LaSalle’s invariance principle, these properties imply convergence to the complex‑balanced equilibrium, thereby confirming the GAC for the considered subclasses.

Finally, the authors formalize the relationship between reaction networks and their associated differential inclusions as a functor between appropriate categories. Objects are reaction networks equipped with variable rate functions; morphisms are network transformations (such as adding or deleting reactions or re‑parameterizing rates) that induce natural maps between the corresponding differential inclusions. This categorical viewpoint clarifies how structural modifications of a network affect the dynamical system and opens the door to systematic network design and control using algebraic‑topological tools.

In summary, the paper provides a powerful projection‑based inductive framework for analyzing differential inclusions arising from mass‑action kinetics. By defining vertexical families and proving their key boundary‑approach property, it reduces high‑dimensional persistence problems to lower‑dimensional ones, thereby advancing the understanding of the Global Attractor Conjecture and offering a functorial perspective on reaction‑network dynamics. This approach is likely to influence future work on stability, robustness, and control of complex biochemical and ecological networks.