DF-CCSD(T) can't find my USERDEF basis set

[Rob Lavroff]

Hi there, I'm trying to do a df-ccsd(t) calculation with a manually-entered basis set to exactly compare results to Cryscor code I'm working on. For some reason hf has no problem using this basis but then df-ccsd(t) gives the error:

Enter DF-CCSD [ITF]

 Cannot find default basis USERDEF for atom O

 Default basis sets for atom=O  key= context=JKFIT
 Number of basis sets found:    0

 Type=FIT
 Context=JKFIT

 Please specify a default basis or define basis sets for all atoms!
 ? Error
 ? Cannot find USERDEF-JKFIT-FIT in basis catalogue
 ? The problem occurs in binput

Is there something I can do to remedy this? My input is below, and weirdly enough, the commented out osv-lmp2 with density fitting has no issue finding the basis set, just df-ccsd(t)

***,h2o
memory,500,m

symmetry,nosym
angstrom
geometry={
O,  0.000000,  0.000000,  0.000000
H,  0.758602,  0.000000,  0.504284
H,  0.758602,  0.000000,  -0.504284
}
basis={
s,o,2266.1767785,340.87010191,77.363135167,21.479644940,6.6589433124,0.8478937000,0.2790534900
c,1.5,-0.005343180993,-0.039890039230,-0.178539119850,-0.464276849590,-0.443097451720
c,6.6,1.
c,7.7,1.
p,o,17.721504317,3.8635505440,1.0480920883,0.2623818000
c,1.3,0.043394573193,0.230941207650,0.513753110640
c,4.4,1.
d,o,0.4790426200
c,1.1,1.
s,h,24.62344998,3.640664800,0.773177290,0.196065500
c,1.3,0.01857031,0.14010862,0.55558583
c,4.4,1.0
p,h,0.184819860
c,1.1,1.
}

hf
df-ccsd(t), DF_BASIS=avtz/mp2fit
!osv-lmp2, DF_BASIS=avtz/mp2fit

[Grant Hill]

Hi Rob,

Have you tried setting df_basis_exch as well as df_basis? The error message suggests to me that this may be what the program is looking for.

One of the “JKFit” auxiliary basis sets could be used for this purpose, although it’s hard to say how robust the density fitting will be with a custom orbital basis.

Best wishes,

Grant

--
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/d2c0e244-70cc-4dd9-a378-1bf8efa2141cn%40googlegroups.com.

[Rob Lavroff]

Hi Grant, thank you for your fast reply! I just tried adding that (df-ccsd(t), DF_BASIS=avtz/mp2fit,DF_BASIS_EXCH=avtz/jkfit) and unfortunately it didn't change anything

[Rob Lavroff]

Hi Grant and all, two updates I wanted to give are

1) If I do exactly the same input but with df-mp2 instead of df-ccsd(t), it works (even without specifying an exchange DF basis, just the mp2fit one). Since ccsd by default begins with a mp2 guess of the amplitudes, this leads me to believe this stop in the ccsd code should be looked at more closely code-wise, as the code halts before a MP2 guess is even attempted.

2) I'm using Molpro 2023 but have reached out to someone who has 2024 access to see if they can reproduce the issue. My input is in my initial email if anyone here is interested in trying it.

Thank you,

Rob