Psi4 SAPT0 calculation was interrupted without error message

SAPT
1

The calculation of SAPT was interrupted. There was no SAPT Results and no error message. But replace Rb metal with Tl metal and it will finish normally. Why?

this is my input:

memory 1000MB
molecule dimer {
0 1 
C    -7.31946336   1.04157069   -0.10129091
H    -6.23303336   1.08651069   -0.10481091
H    -7.70284336   0.63852069   0.83301909
H    -7.70285336   0.49833069   -0.96167091
I    -8.03922336   3.07115069   -0.25989091
--
0 2
Rb    -6.599876640   -0.987520690   0.057270910
}
set {
  basis def2-svp 
  freeze_core True
}
set reference rohf
energy('sapt0')
  • The calculation stops at the following line of output:
DFHelper Memory: AOs need 0.015 GiB; user supplied 0.838 GiB. Using in-core AOs.

Output file:
CH3I-Rb.out.dat (29.2 KB)

Thank you for your help!

2

This might possibly be related to ECP problems, which IIRC are not well tested in SAPT calculations.
Our ECP technology is getting overhauled soon.

Do you have the stderr output? Which version of psi4 is this?

3

Dear Hokru,
Many thanks for your kind help.

I am running the latest psi4 version of 1.3.2.

The output I obtain is in the file below.
CH3I-Rb.out.dat (29.2 KB)

Under the same calculation method and basis set, I wonder why the metal Rb can be replaced by metal Tl to finish the calculation normally?

4

@hokru, can you try reproducing this? When I try on master, I get an error about DFHelper:add_transformation: second space (b_s), is not in space list! If you can reproduce this, one of us should put up an issue so we can bring this to the attention of the SAPT developers.

5

I can reproduce the error but I am not familiar with open-shell SAPT. Maybe @jgonthier can offer insight.

Though it seems there are no occupied orbitals to correlate in the metal fragment:

      Beta orbitals
    ------------------
    nmo       79    79
    nocc      17     4
    nvir      62    75
    nfocc      5     4
    naocc     12     0  <---
    navir     62    75
    nfvir      0     0
    ------------------
6

This looks like the bug described in seg faults in usapt0 solver · Issue #1830 · psi4/psi4 · GitHub that was solved last year… Is this bug resolution included in the version of Psi4 that you are using? Is there an error message or does the code just segfaults?