Parametrized systems of generalized polynomial inequalitites via linear algebra and convex geometry

Abstract

We provide fundamental results on positive solutions to parametrized systems of generalized polynomial $\textit{inequalities}$ (with real exponents and positive parameters), including generalized polynomial $\textit{equations}$. In doing so, we also offer a new perspective on fewnomials and (generalized) mass-action systems. We find that geometric objects, rather than matrices, determine generalized polynomial systems: a bounded set/"polytope" $P$ (arising from the coefficient matrix) and two subspaces representing monomial differences and dependencies (arising from the exponent matrix). The dimension of the latter subspace, the monomial dependency $d$, is crucial. As our main result, we rewrite $\textit{polynomial inequalities}$ in terms of $d$ $\textit{binomial equations}$ on $P$, involving $d$ monomials in the parameters. In particular, we establish an explicit bijection between the original solution set and the solution set on $P$ via exponentiation. (i) Our results apply to any generalized polynomial system. (ii) The dependency $d$ and the dimension of $P$ indicate the complexity of a system. (iii) Our results are based on methods from linear algebra and convex/polyhedral geometry, and the solution set on $P$ can be further studied using methods from analysis such as sign-characteristic functions (introduced in this work). We illustrate our results (in particular, the relevant geometric objects) through three examples from real fewnomial and reaction network theory. For two mass-action systems, we parametrize the set of equilibria and the region for multistationarity, respectively, and even for univariate trinomials, we offer new insights: We provide a "solution formula" involving discriminants and "roots".