Enabling wave-based inversion on GPUs with randomized trace estimation

TL;DR

This paper introduces a GPU-based wave inversion method using randomized trace estimation, enabling high-fidelity 3D imaging with reduced memory requirements suitable for GPU clusters.

Contribution

It presents a novel GPU-exclusive wave-based inversion approach leveraging randomized trace estimation to significantly lower memory usage.

Findings

High-fidelity images on synthetic datasets including salt and anisotropy.

Memory footprint suitable for GPU clusters, enabling fully GPU-based processing.

Potential for scalable 3D wave inversion technology.

Abstract

By building on recent advances in the use of randomized trace estimation to drastically reduce the memory footprint of adjoint-state methods, we present and validate an imaging approach that can be executed exclusively on accelerators. Results obtained on field-realistic synthetic datasets, which include salt and anisotropy, show that our method produces high-fidelity images. These findings open the enticing perspective of 3D wave-based inversion technology with a memory footprint that matches the hardware and that runs exclusively on clusters of GPUs without the undesirable need to offload certain tasks to CPUs.