What question did this study set out to answer?

This work aims to approximate an unknown probability distribution using a sequential algorithm that iterates density estimates.

September 20, 2024Open Access

Stochastic mirror descent for nonparametric adaptive importance sampling

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved density approximation in multi-modal distributions, suggesting enhanced sampling efficacy.

This work aims to approximate an unknown probability distribution using a sequential algorithm that iterates density estimates.
Utilizes a mirror descent (MD) algorithm optimizing the Kullback-Leibler divergence on density functions.
Implements a stochastic approximation technique to balance algorithm complexity and variability.
Proposes a particle subsampling method for enhanced computational efficiency.
The sequence of estimated densities converges almost surely to the target density uniformly on compact sets.
Fixing the learning rate η significantly boosts algorithm performance in multi-modal distributions.
Numerical tests show improved performance compared to other approaches.

Bianchi et al. (2024) studied this question.

View Full Paper

Ask AI

Bookmark