Piecewise linearization of bivariate nonlinear functions: minimizing the number of pieces under a bounded approximation error - LAAS - Laboratoire d'Analyse et d'Architecture des Systèmes Access content directly
Proceedings Year : 2022

Piecewise linearization of bivariate nonlinear functions: minimizing the number of pieces under a bounded approximation error

Abstract

This work focuses on the approximation of bivariate functions into piecewise linear ones with a minimal number of pieces and under a bounded approximation error. Applications include the approximation of mixed integer nonlinear optimization problems into mixed integer linear ones that are in general easier to solve. A framework to build dedicated linearization algorithms is introduced, and a comparison to the state of the art heuristics shows their efficiency.
Fichier principal
Vignette du fichier
tech_report_corridor_fitting_problem_heuristics.pdf (469.77 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03629850 , version 1 (04-04-2022)
hal-03629850 , version 2 (16-06-2022)

Identifiers

Cite

Aloïs Duguet, Sandra Ulrich Ngueveu. Piecewise linearization of bivariate nonlinear functions: minimizing the number of pieces under a bounded approximation error. 13526, Springer International Publishing, pp.117-129, 2022, Lecture Notes in Computer Science, ⟨10.1007/978-3-031-18530-4_9⟩. ⟨hal-03629850v2⟩
111 View
176 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More