CNOPsolution Tester: a web application for result validation in constrained optimization problems

Abstract

In numerical optimization, solutions to optimization problems are developed using methods or techniques that generate approximate results for complex problems, such as metaheuristics. When these techniques generate results, it is essential to verify whether these values meet the optimization problem requirements, for example, values of the variables within the minimum and maximum range and the satisfaction of the problem constraints.
Doing it by hand can be very difficult, especially considering optimization problems have different characteristics. This paper presents the development of a web application called CNOPsolution Tester, which allows researchers and end-users to validate the values of decision variables based on their ranges and constraints. The application was designed based on diagrams created with the Unified Modeling Language and developed in JavaScript. We tested the web application with multiple users and 24 benchmark optimization problems to evaluate its usability, measuring the number of clicks needed to solve each problem. We also validated the functionality using 20% of the problems. The results demonstrate that the CNOPsolution Tester is an efficient tool and easy to use by any end-user, as shown by the response time and the number of clicks used on the application. We applied the Pearson correlation to identify the relationship between the number of clicks, problem variables, and constraints.