These are scripts and data we used to compare GTApprox against several popular approximation libraries, as described in the paper GTApprox: surrogate modeling for industrial design (journal publication, arXiv preprint).
runAllTests.py to save results for all individual tests. (Each test is run in a separate process by first restricting available virtual memory (with
ulimit, Linux-only) and then calling the script
runSingleTest.py.) After completion, run
plotResults.py to collect results and plot performance profiles.