Physics-Preserving Compression of High-Dimensional Plasma Turbulence Simulations

ICLR 2026 Conference Submission18255 Authors

Keywords: physics-inspired, turbulence, plasma, neural compression, autoencoders, neural fields

TL;DR: Neural compression methods enable extreme compression of plasma turbulence simulation data while maintaining low reconstruction error and preserving key physical characteristics.

Abstract:

High-fidelity scientific simulations are now producing unprecedented amounts of data, creating a storage and analysis bottleneck. A single simulation can generate tremendous data volumes, often forcing researchers to discard valuable information.
A prime example of this is plasma turbulence described by the Gyrokinetic equations: nonlinear, multiscale, and 5D in phase space.
They represent one of the most computationally demanding frontiers of modern science, with runs taking weeks and resulting in tens of terabytes of data dumps.
The increasing storage demands underscore the importance of compression, however, compressed snapshots might not preserve essential physical characteristics after reconstruction.
To assess whether such characteristics are captured, we propose a unified evaluation pipeline which accounts for spatial phenomena and multi-scale transient fluctuations.
Indeed, we find that various compression techniques lack preservation of temporal turbulence characteristics.
Therefore, we explore Physics-Inspired Neural Compression (PINC), which incorporates physics-informed losses tailored to gyrokinetics and enables extreme compressions of up to 70,000x.
This direction provides a viable and scalable solution to the prohibitive storage demands of gyrokinetics, enabling post-hoc analyses that were previously infeasible.

Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)

Submission Number: 18255