search

sustainable-processes / summit

Optimising chemical reactions using machine learning
https://gosummit.readthedocs.io/en/latest/
MIT License
118 stars 24 forks source link

Emulators models #248

Closed konkouz closed 1 year ago

konkouz commented 1 year ago

Hello, hope you are well. When you train an emulator do you always train a Gaussian Processor Regressor with spectral sampling ? Could you also please explain where ANNRegressor comes into play ? many thanks once again for the help :)

marcosfelt commented 1 year ago

Hi, Kostas!

1) We don't use GP's in the emulator; that is only for TSEMO 2) ANNRegressor is the model we train in the emulator; you can add custom models to the registry, but it's not very well documented currently:

https://github.com/sustainable-processes/summit/blob/49d6e80a3699d413a2f81e0477d86976f9b98099/summit/benchmarks/experimental_emulator.py#LL1475-L1476C32