E = self._emle(
atomic_numbers, charges_mm, xyz_qm, xyz_mm, cell, charge
)
dE_dxyz_qm, dE_dxyz_mm = _torch.autograd.grad(
E.sum(), (xyz_qm, xyz_mm), allow_unused=allow_unused
)
-dE_dxyz_qm_bohr = dE_dxyz_qm.cpu().numpy()
+dE_dxyz_qm_bohr = dE_dxyz_qm.cpu().numpy() * _BOHR_TO_ANGSTROM
dE_dxyz_mm_bohr = (
dE_dxyz_mm.cpu().numpy() * _BOHR_TO_ANGSTROM
if dE_dxyz_mm is not None
else _np.zeros_like(xyz_mm.cpu().numpy())
)
+# Store the the MM and EMLE energies. The MM energy is an approximation.
+E_mm = E_mm_qm_vac + E.sum().detach().cpu().numpy()
+E_emle = E_tot

Intelligent Force Fields
Lab

Teaching molecular simulations to work smart, not hard

If you are interested in joining the group, see the contact page.

Recent

All news →

2026.03.10
New paper in Chemical Science: enzyme catalysis with EMLE/MM
Our work on simulating enzyme catalysis with electrostatically embedded machine learning potentials is published in Chemical Science.
2026.02.02
Rubén and Luis start their funded PhD periods
Rubén Montagud (IFFLab) and Luis Orta (TunonLab) have been awarded prestigious fellowships to support their doctoral research.

Intelligent Force FieldsLab

Recent

New paper in Chemical Science: enzyme catalysis with EMLE/MM

Rubén and Luis start their funded PhD periods

Intelligent Force Fields
Lab