My inputs have been running for 7 days on komodo and are only 1/3 complete.
I looked at the log files and almost the entire time is spent computing forces at displacements.
KCZ has the convenient property that every term in the force depends only on a single bond vector. (i.e. there's no bond angle or dihedral angle stuff) This means that for each displacement, we only need to compute the terms that involve the displaced atom, instead of all of them. If I can make this optimization, I think the rest of the computations could probably finish in one or two days.
My inputs have been running for 7 days on komodo and are only 1/3 complete.
I looked at the log files and almost the entire time is spent computing forces at displacements.
KCZ has the convenient property that every term in the force depends only on a single bond vector. (i.e. there's no bond angle or dihedral angle stuff) This means that for each displacement, we only need to compute the terms that involve the displaced atom, instead of all of them. If I can make this optimization, I think the rest of the computations could probably finish in one or two days.