Dataset for "Parameter-Aware Ensemble SINDy with Dimensional Filtering: Application to Subgrid-Scale Closure"

Dataset for “Parameter-Aware Ensemble SINDy with Dimensional Filtering: Application to Subgrid-Scale Closure”


This repository provides the postprocessed datasets used to generate the figures and quantitative results in the manuscript Parameter-Aware Ensemble SINDy with Dimensional Filtering.


Contents


Sheet 1 - Sample Case (Heat Diffusion).csv

A sample reference solution for a heat diffusion problem.


Sheet 2 - Sample Case (Burgers).csv

A sample reference solution for a Burgers equation problem.


Sheet 3 - Sample Case (KdV–Burgers).csv

A sample reference solution for a KdV–Burgers system.


Note: All sample cases correspond to the same case index and therefore share consistent random seed and initial condition structure.


Sheet 4 - Two Stages KdV.csv

Two-stage SINDy regression results for the KdV–Burgers system. Includes both first and second stage coefficients for reproducing the bar plots in the manuscript.


Sheet 5 - SGS Heatmap (Sine).csv

Reference Burgers solution (sine initial condition) with SGS model outputs including SINDy SGS predictions.


Sheet 6 - SGS Heatmap (Cubic).csv

Reference Burgers solution (cubic sine initial condition) with SGS model outputs including SINDy SGS predictions.


Sheet 7 - SGS Line (Sine).csv

Linewise SGS comparison at t = 0.5 for the sine case.


Sheet 8 - SGS Line (Cubic).csv

Linewise SGS comparison at t = 0.5 for the cubic sine case.


Sheet 9 - SGS Model Comparison.csv

Metric comparison (MAE, MSE, R2) of SGS models across all Perlin noise-based test cases.


Plotting


All datasets are provided in CSV format and can be plotted using any standard environment including Python or MATLAB. The published figures were generated using DataGraph on macOS.


Notes on Code Availability


The complete data generation and modelling pipeline forms part of an active research line extending this work to higher-dimensional and reacting-flow systems. To protect this ongoing work, the full codebase is not released at this stage. The datasets provided here are sufficient for full reproduction of the results reported in the manuscript.


Contact


For questions regarding dataset structure or reproducibility, please contact

Hanseul Kang

Aalto University

Email: [email protected]