[Debichem-devel] ITP: amp -- atomistic machine-learning potentials

[Graham Inggs]

retitle 790803 amp -- atomistic machine-learning potentials
owner 790803 ginggs at debian.org
thanks

Upstream have relaunched Neural as Amp.

* Package name    : amp
   Version         : 0.3
   Upstream Author : Andrew Peterson, Alireza Khorshidi
* URL             : https://bitbucket.org/andrewpeterson/amp
* License         : GPL-3.0+
   Programming Lang: Python
   Description     : Atomistic Machine-learning Potentials
Amp is an open-source package designed to easily bring machine-learning 
to atomistic calculations. This allows one to predict (or really, 
interpolate) calculations on the potential energy surface, by first 
building up a regression representation of a “train set” of atomic 
images. Amp calculator works by first learning from any other calculator 
(usually quantum mechanical calculations) that can provide energy and 
forces as a function of atomic coordinates. In theory, these predictions 
can take place with arbitrary accuracy approaching that of the original 
calculator.
.
Amp is designed to integrate closely with the Atomic Simulation 
Environment (ASE). As such, the interface is in pure python, although 
several compute-heavy parts of the underlying codes also have fortran 
versions to accelerate the calculations. The close integration with ASE 
means that any calculator that works with ASE - including EMT, GPAW, 
DACAPO, VASP, NWChem, and Gaussian - can easily be used as the parent 
method.

I intend maintaining this package as part of the DebiChem team.

I found there was a packaged named amp in Debian circa 2000; the Audio 
MPEG Player in non-free, but I don't believe this is a problem.