v4.0.0 release

1. Improve and rename APIs for SLA, MLA, TLA, SA and their crowd-tune interfaces.
2. Update example driver scripts for SuperLU_DIST[_RCI], NIMROD, IMPACT-Z, ButterflyPACK, heffte, Hypre, etc.
3. Improve the setting of random seeds.
4. Improve/add TLA algorithms.
5. Add support for recording and loading surrogate model info for Model_GPy_LCM.
6. Add/update build/run scripts for Perlmutter, Crusher, and Summit.
7. Update FAQ.
8. Add the user-guide for v4.0.0.
9. Fix bugs in gptuneband, batched function evaluation, duplicated samples and MLA.
10. Add an example to demonstrate the use of suprocess for launching MPI applications in the lite mode.
11. Add an example to support GPTune + Downhill-SIMPLEX (Nelder–Mead) for improved tuning quality.
12. Integrate gptunehybrid (hybridMinimization of Hengrui Luo), and add support for handling black-box input constraints.
13. Improve the sample/model/search algorithm options when the lite mode is used.
14. Add support for NIX installers.

v3.0.0 release

Improved cmake files for spack build
Improved TLA algorithms for tuning across machines and/or node counts
Updated FAQ
Improved cGP interfaces
Improved historyDB functions for shared-memory parallel sampling and RCI
Added programmable APIs and examples, improved sensitivity analysis
Added support for non-equal samples sizes when Model_GPy_LCM is used -> improved TLA and GPTuneBand algorithms
Tested RCI for Perlmutter, added a few more GPU tuning examples
Added an env GPTUNE_LITE_MODE for installing and running GPTune. The lite mode removes dependencies on openturns, pygmo and openmpi>4.0.
Added SearchSciPy and SearchPyMoo for the searcher classes
Added kernel types when Model_GPy_LCM is used

v2.1.0 release

1. Added more tuning examples
2. Improved multi-objective tuning
3. Improved transfer learning
4. Improved database
5. Added the cGP interface
6. Added sensitivity analysis

release 2.0

2.0.0

Smoothing Splines Tuning Example for Household PowerConsumption Data