LNO-MP2 calculations from LNO-CCSD(T)

More
4 months 5 hours ago #1100 by cnaim94
Dear MRCC developers,

I am making some LNO-CCSD(T) calculations and I want to compare the results also with LNO-MP2 ones. In my input, I am using the keywords calc=LNO-CCSD(T) and lcorthr=tight. From the output I can read that I have the values LMP2, are they actually proper LNO-MP2 calculations? I am saying so because if I directly run an LNO-MP2 calculation changing the corresponding parameters of the LNO decomposition I don't have exactly the same energies. Could you please let me know if I am not understanding properly what are the LMP2 energies or if I am making some mistakes in the keywords of my LNO-MP2 calculations?
These are the keywords that I change in the LNO-MP2 inputs in order to reproduce the LMP2 values:
calc=LMP2
localcc=2018
lcorthr=tight
lccoporder=lccfirst
lnoepso=3e-6
lnoepsv=3e-7
locintrf=direct
talg=lapl
naf_cor=1e-2

Best Regards,
Carmelo Naim
 

Please Log in or Create an account to join the conversation.

More
3 months 4 weeks ago #1101 by nagypeter
Dear Carmelo,

There is indeed a proper LMP2 correlation energy computed as an intermediate result of LNO-CCSD(T) calculations, which could be used by itself.

With the localcc=2018 version there is a small difference between the LMP2 correlation energy obtained via calc=lmp2 or as an intermediate of LNO-CCSD(T).
The "LMP2 correlation energy [au]:" line of the output of a calc=LNO-CCSD(T) job gives an LMP2 correlation energy as discussed in Ref [30] of the manual with minor changes reported in Ref [37]. The difference compared to calc=lmp2 is that the NAF approximation is not employed in the extended domains of an LNO-CCSD(T) calculation, while by default, the NAF approach is still turned on for LMP2 to stay consistent with Ref [30]. (This will be unified in a future release and a newer localcc option.)

So you should be able to get the "LMP2 correlation energy [au]:" value of an LNO-CCSD(T) calculation by setting
calc=LMP2
naf_cor=off

(Please, also note that natural orbitals are not employed in our method at the MP2 level, so there is no "LNO-MP2" method and some thresholds governing the LNO-CC calculations (e.g., lnoepso, lnoepsv) do not take effect for LMP2.)

Thanks for the question and let me know if you need more help.
Best wishes,
Peter

Please Log in or Create an account to join the conversation.

Time to create page: 0.015 seconds
Powered by Kunena Forum