Authors :
Joseph Gogodze.
Volume/Issue :
Volume 3 - 2018, Issue 12 - December
Google Scholar :
https://goo.gl/DF9R4u
Scribd :
https://goo.gl/YuWknv
Thomson Reuters ResearcherID :
https://goo.gl/KTXLC3
Abstract :
In this note, we propose a new approach for benchmarking computational problems and their solvers. The proposed methods are based on special construction of the paired comparison matrices of the problems and solvers; in other words, the scorematrices of the problems and the solvers. Having these matrices at our disposal, different rating/ranking methods can be used. We illustrate our approach using rating/ranking methods originated for ranking sports teams. The proposed approach is illustrated using an example to demonstrate its viability and suitability for applications.
Keywords :
Benchmarking; Software; Solvers; Problems; Testing; Ranking Method.
In this note, we propose a new approach for benchmarking computational problems and their solvers. The proposed methods are based on special construction of the paired comparison matrices of the problems and solvers; in other words, the scorematrices of the problems and the solvers. Having these matrices at our disposal, different rating/ranking methods can be used. We illustrate our approach using rating/ranking methods originated for ranking sports teams. The proposed approach is illustrated using an example to demonstrate its viability and suitability for applications.
Keywords :
Benchmarking; Software; Solvers; Problems; Testing; Ranking Method.
