Compression of Tokamak Simulation Data

Journal of Computational Physics

Tokamak simulation visualization

Simulating the plasma inside a tokamak fusion reactor produces enormous volumes of high-dimensional data — enough that storing and moving the output becomes a bottleneck for the science. Because the underlying physics has strong low-rank structure, that data is highly compressible without sacrificing the quantities physicists care about.

This work, published in the Journal of Computational Physics, applies low-rank matrix and tensor compression to tokamak plasma simulation data, dramatically reducing its storage footprint while preserving physical fidelity in the reconstructed fields.

Read the paper in the Journal of Computational Physics.