search

JuliaMolSim / JuLIP.jl

Julia Library for Interatomic Potentials
Other
83 stars 23 forks source link

JuLIP and Molly #134

Open cortner opened 3 years ago

cortner commented 3 years ago

What is the future for JuLIP.jl when [Molly.jl]() is moving far ahead in terms of features? A possibility would be to further emphasis interatomic potential implementations in JuLIP while Molly might focus more on the "simulation" i.e. MD, optimisation etc. or materials vs molecules. Yet another option might be that JuLIP simply becomes the abstract core of ACE.jl.

CC @jgreener64

I think it would be worthwhile considering this. We are now starting to produce new potentials for molecules as well, which so far seem to outperform ANN and GP potentials.

jgreener64 commented 3 years ago

I haven't had much time to work on Molly recently, but I'm hoping to be able to dedicate more time to it in future. My personal focus will be on features used for macromolecular simulation, differentiable simulation, simulation algorithms and optimisation. Similar to something like OpenMM but in pure Julia and fully differentiable. Clearly there's a long way to go.

I still need to have a deep dive into JuLIP.jl, but it seems to have more of a focus on implementing atomic potentials with use cases in materials science. As you say they could end up being complementary, with potentials from JuLIP working with Molly simulators.

cortner commented 3 years ago

On Optimisation we have a lot of experience and can help if you are interested. But this would only make sense if we can commit to some strategy going forward...