PACE: Pareto–AICc Consensus Estimator for Automated λ Selection in SINDy

This repository contains MATLAB code for the PACE λ-selection framework used in Sparse Identification of Nonlinear Dynamics (SINDy).

Workflow

1. Generate datasets (run once)

Datagenerator

1. Run experiments

runBatch

This performs 200 identification trials and stores results in a .mat file.

1. Optional baseline comparison

generate_baselines

This step reproduces the baseline comparisons used in the paper.

Files

buildLibrary.m – library construction
computeDerivative.m – derivative estimation
stlsq.m – sparse regression
tuneLambda.m – PACE λ-selection
runSINDyID.m – SINDy identification
runBatch.m – batch experiments
Datagenerator.m – dataset generation
generate_baselines.m – baseline comparison