Perprof-py: A Python Package for Performance Profile of Mathematical Optimization Software

A very important area of research in the field of Mathematical Optimization is the benchmarking of optimization packages to compare solvers. During benchmarking, one usually collects a large amount of information like CPU time, number of functions evaluations, number of iterations, and much more. Th...

Full description

Saved in:
Bibliographic Details
Published inJournal of open research software Vol. 4; no. 1; pp. 12 - e12
Main Authors Soares Siqueira, Abel, Costa da Silva, Raniere Gaia, Santos, Luiz-Rafael
Format Journal Article
LanguageEnglish
Published Ubiquity Press 22.04.2016
Subjects
mathematical optimization
performance profile
Python 3
software benchmarking
Online AccessGet full text

Cover

More Information
Summary:A very important area of research in the field of Mathematical Optimization is the benchmarking of optimization packages to compare solvers. During benchmarking, one usually collects a large amount of information like CPU time, number of functions evaluations, number of iterations, and much more. This information, if presented as tables, can be difficult to analyze and compare due to large amount of data. Therefore tools to better process and understand optimization benchmark data have been developed. One of the most widespread tools is the Performance Profile graphics proposed by Dolan and Moré [2]. In this context, this paper describes perprof-py, a free/open source software that creates 'Performance Profile' graphics. This software produces graphics in PDF using LaTeX with PGF/TikZ [22] and PGFPLOTS [4] packages, in PNG using matplotlib [9], and in HTML using Bokeh [1]. Perprof-py can also be easily extended to be used with other plot libraries. It is implemented in Python 3 with support for internationalization, and is under the General Public License Version 3 (GPLv3).
ISSN:2049-9647
2049-9647
DOI:10.5334/jors.81