Polyhedral Lyapunov functions structurally ensure global asymptotic stability of dynamical networks iff the Jacobian is non-singular

Franco Blanchini, Giulia Giordano

Research output: Contribution to journalArticleScientificpeer-review

27 Citations (Scopus)
69 Downloads (Pure)

Abstract

For a vast class of dynamical networks, including chemical reaction networks (CRNs) with monotonic reaction rates, the existence of a polyhedral Lyapunov function (PLF) implies structural (i.e., parameter-free) local stability. Global structural stability is ensured under the additional assumption that each of the variables (chemical species concentrations in CRNs) is subject to a spontaneous infinitesimal dissipation. This paper solves the open problem of global structural stability in the absence of the infinitesimal dissipation, showing that the existence of a PLF structurally ensures global convergence if and only if the system Jacobian passes a structural non-singularity test. It is also shown that, if the Jacobian is structurally non-singular, under positivity assumptions for the system partial derivatives, the existence of an equilibrium is guaranteed. For systems subject to positivity constraints, it is shown that, if the system admits a PLF, under structural non-singularity assumptions, global convergence within the positive orthant is structurally ensured, while the existence of an equilibrium can be proven by means of a linear programming test and the computation of a piecewise-linear-in-rate Lyapunov function.

Original languageEnglish
Pages (from-to)183-191
JournalAutomatica
Volume86
DOIs
Publication statusPublished - 2017

Bibliographical note

Accepted Author Manuscript

Keywords

  • Chemical reaction networks
  • Dynamical networks
  • Global stability
  • Piecewise-linear Lyapunov functions
  • Structural stability

Fingerprint

Dive into the research topics of 'Polyhedral Lyapunov functions structurally ensure global asymptotic stability of dynamical networks iff the Jacobian is non-singular'. Together they form a unique fingerprint.

Cite this