DF-basis in sequence of RS2-F12 calculations

46 views
Skip to first unread message

var...@gmail.com

unread,
Dec 16, 2023, 12:05:29 PM12/16/23
to molpro-user
Dear Prof. Knowles,

   I have found that if I run a RS2-F12/cc-pVQZ-F12 calculation as a separate calculation, I get a different result from if I run it as a sequence with increasing basis set size (e.g., first cc-pVDZ-F12, then cc-pVTZ-F12, and then cc-pVQZ-F12). This appears to be a result of the DF-basis remaining fixed at the basis for the first of the calculations (e.g., CC-PVDZ-F12/MP2FIT for the sequence of basis sets or CC-PVQZ-F12/MP2FIT for the standalone calculation). 

   In contrast, for a sequence of CCSD(T)-F12 the DF-basis sets change consistently with the basis set and I do not notice similar behavior.

  Is there a simple way to make sure the DF-basis is consistent in RS2-F12 regardless of the sequence of calculations?

   For reference I repeat below two separate inputs that provide different outputs for the final RS2-F12/cc-pVQZ-F12 result (-55.652754823392 vs -55.653453366172) when running Molpro2023.2 from Sep. 22.

********************
memory,200,m
symmetry, nosym; orient, noorient
angstrom
geomtyp=xyz
geometry={
3
rotational dynamics
N, 0.227645, -0.0217272, -0.0760363
H, 1.02181, 0.204517, 0.562037
H, -4.18478, 0.0973675, 0.494439
}

nosym

basis=cc-pvqz-f12

{rhf;wf,9,1,3}

{multi; closed,3; occ,6; wf,9,1,1;orbprint,3}

rs2-f12,shift=0.2

---


**********

memory,200,m

symmetry, nosym; orient, noorient

angstrom

geomtyp=xyz

geometry={

3

rotational dynamics

N, 0.227645, -0.0217272, -0.0760363

H, 1.02181, 0.204517, 0.562037

H, -4.18478, 0.0973675, 0.494439

}

nosym

basis=cc-pvdz

{rhf;wf,9,1,3}

{multi; closed,3; occ,6; wf,9,1,1;orbprint,3}

 basis=cc-pvdz-f12

{multi; closed,3; occ,6; wf,9,1,1;orbprint,3}

rs2-f12,shift=0.2

basis=cc-pvtz-f12

{multi; closed,3; occ,6; wf,9,1,1;orbprint,3}

rs2-f12,shift=0.2

basis=cc-pvqz-f12

{multi; closed,3; occ,6; wf,9,1,1;orbprint,3}

rs2-f12,shift=0.2

---

Hans-Joachim Werner

unread,
Dec 17, 2023, 12:29:13 PM12/17/23
to var...@gmail.com, molpro-user
In fact is is not the desired behaviour and has been fixed. The fix will become available in the next Molpro release.

For the time being you can avoid the problem by specifying the fitting sets on the rs2-f12 command lines:

rs2-f12,shift=0.2,df_basis=CC-PVQZ-F12/MP2FIT,df_basis_exch=CC-PVQZ-F12/jkfit

Best regards
Joachim Werner
> --
> You received this message because you are subscribed to the Google Groups "molpro-user" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to molpro-user...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/molpro-user/dee9550c-c9cd-437c-9da0-fc917b7be84fn%40googlegroups.com.

Reply all
Reply to author
Forward
0 new messages