- edited description
Enhancements of the next version v0.4
Issue #1
resolved
The following enhancements need to be implemented in the next version of Amp:
- Amp v0.3 does not make use of accompanying fmodules. Fortran part should also get modularized in the context of Amp.
- Cartesian scheme should be added to Amp.
- Amp cannot be directly initialized yet as what we had before, say,
calc = BPNeural(...)
and
calc = CartesianNeural(...)
- Parameters are not ASE Calculator parameter type, rather parameters are just regular dictionaries (That's why it currently breaks BP_test.py).
P.S. In this version calculator is initialized as
calc = AMP(fingerprints=Behler, regression=NeuralNetwork, Gs=..., weights=..., ...)
It might be preferable to initialize it in turn as
calc = AMP(fingerprints=Behler(Gs=..., ...), regression=NeuralNetwork(weights=..., ...))
but I could not figure out how that is possible!
Comments (3)
-
reporter -
reporter - edited description
-
reporter - changed status to resolved
Impelemented.
- Log in to comment