fci_tc_bi_ortho incorrect calculation

[kevrap121]

Hello @scemama ! I am trying to run a tc fci for Li2 to benchmark and get familiar with the Transcorrelation tools of QuantumPackage, but I'm getting energies that are below the energy minimum. I know this should be the case for the tc_scf, but I'm not sure why it is happening when I run fci_tc_bi_ortho. I attached a tarbell of the coordinates and script I am using, and I was wondering if you knows why this is happening. I also included results for when I set read_wf to false (start with a ROHF-like Slater determinant guess) and for when I set read_wf to true (start with the EZFIO wave function).
qp2-2.2.2_archive.tar.gz

[eginer]

Hello @kevrap121 !
I had a look at your script, and I have fixed it.
Roughly speaking, the TC energy strongly depends on the shape of the Jastrow, and the default parameter you used are not the best (you could not know :)
I attach an autonomous script to compute the energy of Li2 in the aug-cc-pVDZ (it runs very fast) with all detailed explained and some numbers to compare with.
WARNING: a ) you have to put in the script the QP_ROOT variable (where your QP2 is installed).
b ) I fixed some bugs in the QP2 related to TC (small stuffs) so I strongly suggest you to git pull & ninja clean & ninja
The commit code is "4af118c4e62fd99413f0917efab7b16cf66529fd"
I will not be available before September to potentially reply to further questions, but I will answer when I come back.
All the best,

1. Emmanuel
   example_tc.zip

[kevrap121]

Hello @eginer!

Thank you so much for this feedback. It is super helpful and I was able to get some cool results for a singlet and triplet carbon dimer using the parameter source you recommended. I'm curious to use the three-body Jastrow tools in the plugins, but I can't seem to find documentation for how to use them / set parameters. Could you point me in the right direction?

Best,
Kevin