EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1192. CasADi as the "bread and butter" in optimization for problem-solving engineers

Invited abstract in session WC-30: Optimization Frameworks, stream Software for Optimization.

Wednesday, 12:30-14:00
Room: 064 (building: 208)

Authors (first author is the speaker)

1. Alejandro Astudillo
Department of Mechanical Engineering, KU Leuven
2. Joris Gillis
Department of mechanical engineering, KU Leuven

Abstract

As an open-source software framework available in C++, Python and MATLAB since 2010, CasADi has been empowering academic and industrial researchers to rapidly prototype and efficiently solve challenging (dynamic) optimization problems.

Demonstrated through 3700+ citing papers, such optimization problems serve as an answer to many design, planning, estimation and control needs arising in engineering fields such as heat & energy, automotive, chemistry & pharmacy, or robotics.

The focus of this talk is on how CasADi facilitates vertical integration. The key facilitator is CasADi's versatile expression graph, capable of efficiently representing engineering specifications, dynamic models and transcription methods. A plugin system maps these expression graphs to the needs of a range of open-source and commercial optimization solvers, often including sparsity detection, (repeated) algorithmic differentiation, and linearity detection. By manipulating, communicating and code-generating the expression graphs, one can introduce solid software abstractions that pair readability and maintainability with excellent runtime performance. In particular, the talk presents the case of robot motion skill tool and follows the journey of the involved expression graphs all the way down to the CasADi solver plugins.

Keywords

Status: accepted

Back to the list of papers