English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Modern computational science and engineering applications are being improvedby the advances in scientific machine learning. Data-driven methods such asDynamic Mode Decomposition (DMD) can extract coherent structures fromspatio-temporal data generated from dynamical systems and infer differentscenarios for said systems. The spatio-temporal data comes as snapshotscontaining spatial information for each time instant. In modern engineeringapplications, the generation of high-dimensional snapshots can be time and/orresource-demanding. In the present study, we consider two strategies forenhancing DMD workflow in large numerical simulations: (i) snapshotscompression to relieve disk pressure; (ii) the use of in situ visualizationimages to reconstruct the dynamics (or part of) in runtime. We evaluate ourapproaches with two 3D fluid dynamics simulations and consider DMD toreconstruct the solutions. Results reveal that snapshot compressionconsiderably reduces the required disk space. We have observed that lossycompression reduces storage by almost $50\%$ with low relative errors in thesignal reconstructions and other quantities of interest. We also extend ouranalysis to data generated on-the-fly, using in-situ visualization tools togenerate image files of our state vectors during runtime. On large simulations,the generation of snapshots may be slow enough to use batch algorithms forinference. Streaming DMD takes advantage of the incremental SVD algorithm andupdates the modes with the arrival of each new snapshot. We use streaming DMDto reconstruct the dynamics from in-situ generated images. We show that thisprocess is efficient, and the reconstructed dynamics are accurate.

Enhancing Dynamic Mode Decomposition Workflow with In-Situ Visualization and Data Compression