List view
In order for this library to be at all useful for the scientific community, we need introductory documentation and tutorials that tell a new user how to get up and running with the library. Developer documentation must, of course, also be continuously added and kept up to date. But that kind of documentation isn't very useful for a new user.
No due date•5/6 issues closedBeing able to use librascal to run MD simulations using machine learning models computed with librascal representations. This is also a good opportunity to benchmark the performance of these representations against existing implementations in a realistic application. Will require interfaces to common MD programs, such as LAMMPS, i-PI, and n2p2.
Overdue by 6 year(s)•Due by February 28, 2020Being able to use both SOAP (with KRR) and B-P (with ANNs) to fit a machine learning model of a potential energy surface. This will probably require an "external" fitting tool that will live in librascal for now.
Overdue by 6 year(s)•Due by January 17, 2020•14/16 issues closedPython bindings, access to both the full descriptor matrices and semantic access to individual components (indexing by atom, species, expansion component...)
No due date•1/1 issues closedThings that aren't necessarily critical to getting machine learning models with SOAP running, but significantly enhance the flexibility and speed of the representation.
Overdue by 6 year(s)•Due by November 29, 2019•6/7 issues closedWith gradients -- get it on the same footing as SOAP, ready to use to train and run ANN potentials Currently waiting on representation calculator update
No due date•1/1 issues closedSOAP (including tensor order > 0, "body order" > 2), with gradients, cell gradients, and everything else needed to use it for building and running ML models
Overdue by 6 year(s)•Due by November 29, 2019•6/7 issues closed