Open davkovacs opened 2 years ago
Turns out the problem was that I didn't pass pin=2
to the basis construction.
I think we should have a better behavior than this if someone specifies an inner cutoff, but doesn't specify pin
.
There is always an rin even if you don’t choose it because it specifies the domain of orthogonality for the polynomials
maybe it could be better documented but if don’t see how to change the behaviour
I did choose rin
, and that makes perfect sense that you always have it. The problem was the not specifying pin=2
lead to 10^11 or -10^11, so completelycrazy behaviour of ACE where it shouldn’t have been defined.
So we could consider making pin = nothing the current pin = 0 and pin=0 becomes a jump to zero. But this isn’t going to solve anything. You still have to specify pin to make sure you get the desired behaviour.
Maybe this could be resolved by REQUIRING pin once rin is specified.
Yes, requiring pin
might be the best solution. It shouldn’t happen to anyone ever again, that you have rin
Specified, but the ACE is nonzero outside it.
I am trying to fit ACE models using ACEatoms, and when I plot the dimer curves of the potentials I get the 1-body and 2-body parts behave as expected, but the ACE component of the basis does not go to 0 at the inner cutoff, but it explodes: Code used to define the basis: