A Massively Parallel Exact Solution Algorithm for the Balanced Minimum Evolution Problem

Catanzaro, Daniele [UCL] Frohn, Martin Gascuel, Olivier Pesenti, Raffaele

We build upon recent theoretical advances on the Balanced Minimum Evolution Problem (BMEP) to design a new massively parallel exact solution algorithm that proves to be up to one order of magnitude faster than the current state-of-the-art under the same computing settings and environment. We also investigate, for the first time, the theoretical connections between numerical stability and statistical consistency of the BMEP and we show that some rescaling techniques introduced to numerically stabilize the problem may affect negatively the statistical consistency of the optimal solution to the problem.