220. Global Optimization Algorithm through High-Resolution Sampling
Invited abstract in session TB-4: Stochastic and Deterministic Global Optimization, stream Global optimization.
Tuesday, 10:30-12:30Room: B100/5013
Authors (first author is the speaker)
| 1. | Daniel Cortild
|
| Faculty of Science and Engineering, University of Groningen | |
| 2. | Claire Delplancke
|
| EDF Lab Paris-Saclay | |
| 3. | Nadia Oudjane
|
| EDF R&D Palaiseau, France | |
| 4. | Juan Peypouquet
|
| Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Groningen |
Abstract
We present an optimization algorithm that can identify a global minimum of a potentially nonconvex smooth function with high probability, assuming the Gibbs measure of the potential satisfies a logarithmic Sobolev inequality. Our contribution is twofold: on the one hand we propose said global optimization method, which is built on an oracle sampling algorithm producing arbitrarily accurate samples from a given Gibbs measure. On the other hand, we propose a new sampling algorithm, drawing inspiration from both overdamped and underdamped Langevin dynamics, as well as from the high-resolution differential equation known for its acceleration in deterministic settings. While the focus of the work is primarily theoretical, we demonstrate the effectiveness of our algorithms on the Rastrigin function, where it outperforms recent approaches.
Keywords
- Global optimization
Status: accepted
Back to the list of papers