In this work we adapt recent model reduction approaches to predict the solutions of time-dependentparametrized problems describing crowd motion in the presence of obstacles. A prototypical exampleof problem of interest is a discrete contact model. In this problem, in a discrete timesetting, the set of velocities of each agent in the crowd is the solution at each time step to a constrainedleast-squares optimization statement. The parametric variations in the problem (associated with thegeometric configuration of the system) have a very strong impact on the variability of the solution,both in terms of positions of the agents and contact forces between them, the latter being given bythe Lagrange multipliers of the underlying saddle-point problem. More precisely, the Kolmogorov n-width of the solution set is very slowly decaying. Motivated by this observation and inspired by recent works, we investigate new developments and combinations of the reduced-basis methodand supervised machine-learning techniques to effectively estimate primal and dual solutions. Theproposed nonlinear compressive strategy is numerically validated by comparisons with more standardlinear and nonlinear approximations.
A nonlinear reduced basis approximation of discrete contact problems in crowd motion
Research Group:
Speaker:
Giulia Sambataro
Institution:
Ecole des Ponts ParisTech
Schedule:
Friday, June 14, 2024 - 14:00
Abstract:
