Skip to content

Adding the TIP4P force field for water to ASE, compatible with the QM/MM module

username-removed-559845 requested to merge asod/ase:tip4p into master

Since some users are employing this force field for QM/MM simulations, I've implemented a version that works with the official, updated QM/MM scheme. Info on the FF: http://dx.doi.org/10.1063/1.445869

TIP4P uses a virtual interaction site on every water molecule that carries the charge, which makes the model reproduce liquid water quasistructure more accurately. We've tested that the previous implementation produced liquid TIP4P water, so for this version, I've tested that the energies and forces are consistent with the previous version, and with TIP4P forces from the NAMD MD program.

Having massless, virtual interaction sites while avoiding having to change all dynamics codes and constraints to be able to handle this means only adding the sites just before the energy/forces are calculated, and removing them again after the forces on the sites have been redistributed (according to 10.1002/(SICI)1096-987X(199906)20:8). However, this means that for the QM/MM electrostatic embedding to work, the virtual sites must also be added when the embedding takes place. Thus, I've made a new embedding class, EmbedTIP4P, that does this.

These tests are made on the water dimer.

MM Forces: forces This plot shows the force components of the left (blue) and right (red) water molecule in the dimer. The previous (old) implementation and the current (circles) give the same results. Both versions are consistent with NAMD.

MM dimer binding energy: epotMM

MM, QM/MM, and QM energies: epot

Merge request reports