Plan for solving enhancement of cutoffs
This issue is about discussing plans for addressing different types of enhancement to the handling of cutoffs. Once a solution is agreed upon we can spawn new issues in regards to single tasks from the list below.
We have the following items to address
- Address the
MMCST
and how we should deal with smaller cells (see #45 (closed)). Potentially by taking advantage of theMMCST
we could reduce the DFT time a lot. This could be useful if we want to use long-ranged force-constants for a small supercell. Ificet
has a solution in place for this we can probably adapt the concept. - Introduce the
all
keyword for setting up cutoffs. This would be beneficial when comparing withphonopy
. If 1. is, however, properly addressed this might not be needed anymore. Specify cutoffs via a geometrical-order matrix. This would allow a user to fine tune cutoffs via a matrix with then,m
elements corresponding to the cutoffs for am
-th order cluster withn
bodies (points).Species dependent cutoffs. For example one might only want to include three-body interactions for a cluster containing carbon atoms.- Update the current
check_halfL
to actually check if given a cutoff this atoms object will span the clusters. Alternatively make this function simply return the maximum cutoff you can use for a given atoms object.
Note that 1 and 2 are deeply connected to the core of HiPhive
and fundamental principles of force constants, whereas 3 and 4 are more about extending the current functionality.
Edited by Erik Fransson