Automating the formulation and resolution of convex variational problems with the fenics_optim package

Jeremy Bleyer (Laboratoire Navier, Ecole des Ponts ParisTech, Universit√© Gustave Eiffel, 🇫🇷)
Friday session 3 (Zoom) (17:00–18:30 GMT)
View slides (pdf) (available under a CC BY 4.0 license)

In this work, we present a Python package called fenics_optim which enables to easily formulate convex variational problems as conic optimization problems within FEniCS. It relies on the conic optimization solver Mosek which uses state-of-the-art interior-point methods for solving large-scale linear, second-order cone and semi-definite programming problems. These are particularly suited for solving non-smooth optimization problems which arise in contact or elasto-plasticity problems for instance but also in the image processing community. In particular, we will present an application to solving limit analysis problems, ie computing directly the limit load of a perfectly plastic structure as a convex optimization problem and without relying on an incremental elasto-plastic procedure until final collapse. I will finish by a recent extension towards limit-analysis based topology optimization.