Source data and computational artifacts for long-time KdV soliton propagation using co-moving conservation-regularized PINNs

Read full text
Authors: Ahmed Fathi
Cited by: 2

Abstract

This record contains the source data and computational artifacts supporting the manuscript “Long-Time KdV Soliton Propagation Using Co-Moving Conservation-Regularized Physics-Informed Neural Networks.” It includes the source data underlying Figures 1–6 and Supplementary Figures S1–S3, laboratory- and co-moving-frame numerical archives, ETDRK4 numerical-reference arrays, field-error and crest diagnostics, invariant-drift diagnostics, local residual diagnostics, frozen configuration metadata, trained-model construction bundles, and final PyTorch state dictionaries. The deposited archives support reproduction of the reported figures and numerical values without retraining. The controlled frame-by-stage comparison uses one fixed seed and does not make a statistical multi-seed robustness claim.

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 July 2026
Document type
Dataset