ATHENA
ATHENA copied to clipboard
mathLab
ATHENA: Advanced Techniques for High dimensional parameter spaces to Enhance Numerical Analysis
**Describe the bug** The GPy dependency does not work anymore for python>=3.8 **To Reproduce** Run LocalAS **Output** `AttributeError: module 'numpy' has no attribute 'bool'` **Proposed solution** Use scikit implementation instead:...
There are broken latex formulas in the notebook.
Remove the custom scaler and use scikit-learn [MinMaxScaler](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html) with (-1, 1) instead. `from sklearn.preprocessing import MinMaxScaler`
Update the main page of the online doc with the new additions in the README.
Substitute GPyOpt dependency with [BayesianOptimization](https://github.com/fmfn/BayesianOptimization) or just remove the bso flag for the projection matrix tuning. GPyOpt results in some errors for numpy 1.24+ and it is not maintained anymore.
Add the missing docstrings in the local active subspaces classes.
Add links and description to the readme in the tutorials folder.
Momentum parameter for nll train does not exist anymore.
This PR resolves issue #130 by implementing two changes to the ATHENA package: 1. Replacing GPyOpt with BayesianOptimization - [x] Migrated from the less maintained GPyOpt library to the more...
