Optimization as a fundamental technology is a universal concept that finds application in almost every discipline, including business, engineering, economics. It is defined as a process of modifying a system to make certain features more efficient.
Traditionally, users write a low-level code that directly generates the input data structures for the optimization problem. The already existing optimization libraries of Python were user-friendly but introduced significant performance bottlenecks. Hence, researchers felt the need for a software package to solve constrained optimization problems like linear programming and integer programming problems.
To improve the quality of optimization, modeling languages that are easy to use and fast and powerful came up. The researchers at MIT’s Operation Research Center began experimenting with a new programming language-Julia. The early experiments proved that Julia could bring out the best of both worlds: speed and flexibility. While similar libraries of Python were slower, Google’s prototype JuMP was competitive with the state-of-art libraries.
JuMP is an open-source modeling language that allows users to express various optimization problems in high-level algebraic syntax. It takes the help of the advanced features of Julia to offer exceptional functionality and achieve an at-par performance with the commercial modeling tools available.
One of the first benchmarks achieved by JuMP is its ability to produce quadratic and conic-quadratic optimization models in a suitable format for the users as fast as the state-of-the-art commercial models. Since then, the team has continued to raise its standards. In order to find a home for its long-term sustainability, the researchers took the support of NumFOCUS, a non-profit organization assisting open-source scientific software.
Amongst many applications of JuMP, one of the studies, Sepulveda et al. on cost-effective ways to decarbonize the grid, is a significant breakthrough. The goal is to operate data centers and campuses entirely on carbon-free energy by the year 2030.
JuMP 1.0 is set to release soon, and the researchers rely on JuMP’s strong culture of consensus-driven development to take on the subsequent challenges.