DMFF
DMFF copied to clipboard
deepmodeling
DMFF (Differentiable Molecular Force Field) is a Jax-based python package that provides a full differentiable implementation of molecular force field models.
### Summary Expand the function of parameter optimization in DMFF for dynamic properties, such as diffusion coefficient, viscosity, etc. This feature requests, 1. An efficient algorithm for evaluating the gradient....
[Feature Request] Efficient neighborlist construction/usage plugin for nonbonded force calculation
### Summary Construct neighborlist using non-JITable code (for speed), and use it in downstream nonbonded force calculation. ### Motivation The construction of nblist is the most time-consuming part of DMFF....
### Summary Implement the communication of LAMMPS and DMFF trained model in c++ interface. ### Motivation Similar with #85 , supporting of popular MD engines would make cutting-edge forcefield easier...
- add `method` in API for QEq - set the original method as `root_finding` (default) - add `mat_inv` (matrix inversion) and `pgrad` (projected gradient) method
### Bug summary I think the use of switch function is wrong in DMFF. https://github.com/deepmodeling/DMFF/blob/dcc9fa5a4516c06b2fd3383372a68bd68b70ffb0/dmff/classical/inter.py#L49C1-L54C21 It should be: ```python if self.isSwitch: x = (dr_norm - self.r_switch) / (self.r_cut - self.r_switch)...
1.Matrix inversion method and projected gradient method were added to calculate the equilibrium charges，and step stride was added to control method 2.units of Chi and J parameters were unified. 3.qeq...
### Summary Thank you for the great library! DMFF sets JAX to use float64. Therefore any code using DMFF has to be refactored if one wants to keep using Float32....
Add two features to avoid short range failure: 1. Add hard core potential (A/Br)^14 to Slater function 2. Bound the induced dipole using soft truncation Both are not supposed to...
- fix bug in qeq gaussian damp function 2, and make eta as gaussian width - spcific constant `DIELECTRIC` based on scipy.constants
This PR introduces the following changes: - **New Directory**: Added the source code to the `dmff/dpnblist` directory. The package can be installed as a dependency using the command `pip install...
Metadata
Owner
Metadata
DMFF (Differentiable Molecular Force Field) is a Jax-based python package that provides a full differentiable implementation of molecular force field models.