Geometric Priors for Scientific Generative Models in Inertial Confinement Fusion

TL;DR

This paper introduces a Wasserstein autoencoder with a hyperspherical prior tailored for multimodal data in inertial confinement fusion, improving sampling efficiency and leveraging scientific constraints for validation.

Contribution

It proposes a novel hyperspherical prior sampling method and integrates scientific constraints to enhance generative modeling in fusion research.

Findings

Efficient sampling from a normal distribution with projection improves model performance.

Incorporating scientific constraints ensures the validity of generated samples.

The model effectively captures multimodal data in inertial confinement fusion.

Abstract

In this paper, we develop a Wasserstein autoencoder (WAE) with a hyperspherical prior for multimodal data in the application of inertial confinement fusion. Unlike a typical hyperspherical generative model that requires computationally inefficient sampling from distributions like the von Mis Fisher, we sample from a normal distribution followed by a projection layer before the generator. Finally, to determine the validity of the generated samples, we exploit a known relationship between the modalities in the dataset as a scientific constraint, and study different properties of the proposed model.