EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2294. VIVA: A web-based tool for analyzing and visualizing optimization solver benchmarking results

Invited abstract in session WA-30: Optimization Tools, stream Software for Optimization.

Wednesday, 8:30-10:00
Room: 064 (building: 208)

Authors (first author is the speaker)

1. Andreas Lundell
Department of Information Technologies, Åbo Akademi University
2. Oscar Härtull
Department of Information Technologies, Åbo Akademi University
3. Alireza Olama
Department of Information Technology, Åbo Akademi University

Abstract

When developing software for solving various classes of optimization problems, e.g., MILP, and MINLP, it is crucial to have tools for comparing the impact of new functionality. The performance can be measured using many different metrics including solution time and quality of the solution, so it is not always trivial to assess the added value. It is also important for end users to be able to compare the performance of available optimization solvers when selecting which one to use. So far, available tools for performing such analysis have been mainly static, i.e., there have been limited options for the users to interact, filter, and sort the results.

In this presentation, we introduce a tool called VIVA (Visualize, Interact, Verify, and Analyze Benchmarking Data for Optimization Solvers) for comparing the performance of optimization solvers in an efficient and user-friendly way. VIVA is an open-source project implemented in TypeScript and generally based on standardized web technologies. This means that VIVA can be used directly in a web browser without installing and configuring additional dependencies. Currently, VIVA implements most of the functionality of the Python-based PAVER tool, including its visualization capabilities, but the open framework of VIVA means it can be extended further in the future.

To illustrate the capabilities of VIVA, the tool is also demonstrated on recent benchmark results from convex and nonconvex MINLP solvers during the presentation.

Keywords

Status: accepted

Back to the list of papers