Benchmark datasets for the PCE-GPR Toolbox

This archive contains benchmark datasets used to validate and compare the performance of the PCE-GPR Toolbox. Each benchmark test case includes input/output data for surrogate model training and evaluation. The test cases comprise analytically defined or numerically evaluated functions implemented in MATLAB, as well as SPICE-based simulations of electronic circuits. For each test case, the MATLAB functions or SPICE netlists used to generate the data are provided. 📁 Structure Each test case is organized in the following format: ``` test_case_name/ ├── data/ │ ├── data_train.mat # Training data for the surrogate model │ └── data_mc.mat # Reference (Monte Carlo) test data for validation ├── function/ │ └── function_name.m # MATLAB function(s) to generate the data (when applicable) └── netlists/ ├── netlist_name_train.sp # SPICE netlist to generate training data (when applicable) ├── netlist_name_mc.sp # SPICE netlist to generate reference (Monte Carlo) data └── *.lib # Other SPICE files (library models, sweep definitions) ``` 🧪 Test Cases `01D_Forrester_function`: Forrester function (1D analytical example) `03D_Ishigami_function`: Ishigami function (3D analytical example) `05D_Duffing_van_der_Pol_oscillator`: Duffing-van der Pol oscillator (5D numerical example in MATLAB) `05D_Friedman_function`: Friedman function (5D analytical example) `06D_Undamped_oscillator`: Undamped oscillator (6D analytical example) `08D_Borehole function`: Borehole function (8D analytical example) `10D_Wingweight_function`: Wingweight function (10D analytical example) `20D_Morris_function`: Morris function (20D analytical example) `25D_Amplifier_gain`: Low-noise amplifier gain (25D numerical example in SPICE) `26D_Maximum_crosstalk`: Transient maximum crosstalk in coupled transmission lines (26D numerical example in SPICE) `54D_Interconnect_delay`: Delay in a network of transmission lines (54D numerical example in SPICE) Each `.mat` file contains MATLAB variables ready to load into your analysis workflow. The MATLAB functions or SPICE netlists can be used to reproduce the training and test data, if needed. 📝 Notes Data files are saved in MATLAB format (`.mat`) and are compatible with R2023b and newer. All test cases are intended for reproducibility and benchmarking. Refer to the PCE-GPR Toolbox documentation for usage examples. 🔗 Related This dataset is associated with the [PCE-GPR Toolbox](https://doi.org/10.5281/zenodo.15348860).