Open maxscheurer opened 1 year ago
Overlap is defined as the product of two Gaussians. You should compare to
ref = integrate(
exp(-2.0 * (x ** 2 + y ** 2 + z ** 2)) ** 2, (x, -oo, oo), (y, -oo, oo), (z, -oo, oo)
)
for s-type orbitals, int1e_ovlp includes the normalization factor for angular part 1/4pi
Thanks, the 1/(4pi)
did the trick 👍 Regarding 1., I already had a product of two Gaussians, note the ** 2
after the exp
function.
I'll leave this issue open for follow-up questions during my implementation if that's okay.
Hi, I'm currently experimenting with some integrals for implementing GOSTSHYP in
pyscf
. For that, I need to build a custom "fakemol" and basis set, so I've been using pyscf'sfakemol_for_charges
to do the setup. The reason for this is that these special basis functions are not normalized.Along the way, I've been wondering which integrals
libcint
actually computes under the hood, so I've tried to "manually" compute the overlap of a primitive s-function with itself, with exponent of one, contraction coefficient of one, centered at the origin.I was wondering why the result from
libcint
is not identical to what I get withsympy
? What am I forgetting/missing here?Output:
Thanks for helping with this (probably stupid) question 😁