Extending Chubanov algorithm to generic linear program: strong polynomial time in not degenerated cases ?

Adrien Chan-Hon-Tong (1)

1 ONERA, Université Paris Saclay (COmUE) [Palaiseau]

This short paper presents an algorithm which aims to solve generic linear program using simple projections and call to a sub solver dedicated to linear feasibility, recently, proven to be a strongly polynomial problem thank to Chubanov algorithm. Then, a complexity analysis states that this algorithm is strongly polynomial in not degenerated cases. Despite surely containing faults, this complexity analysis and this linear feasibility based algorithm call for a discussion.

Mots clés: Linear programming; Strongly polynomial

Domaines: Recherche opérationnelle [math.OC]

Adrien Chan-Hon-Tong. Extending Chubanov algorithm to generic linear program: strong polynomial time in not degenerated cases ?. 2019. ⟨hal-00722920v12⟩