Hi everyone,

I am having trouble with the charge transfer methodology of SAPT implemented in Psi4.

I did calculations with an external decontracted basis set generated by ORCA4 (as discussed here) at SAPT0/TZVP. SAPT0 trends **beautifully agree** with EDA calculations performed with ADF 2016 (ZORA-BP86-D3/TZ2P).

But I would like to use the charge transfer methodology and it has just worked for one compound so far.

I keep getting “not a number” issues.

This is a typical input:

```
memory 29 gb
molecule dimer {
0 1
Te 0.00000 -0.00000 3.05800
C 1.47926 1.51846 3.26142
H 1.29403 2.28304 2.51234
H 2.44452 1.05067 3.08494
H 1.43690 1.92765 4.26757
C -1.66964 1.30621 3.26157
H -1.58882 2.08868 2.51242
H -1.68241 1.71746 4.26769
H -2.56342 0.71309 3.08529
--
0 1
I 0.00000 0.00000 0.00000
Cl -0.01030 0.15341 -2.65615
units angstrom
}
set globals {
maxiter 500
relativistic DKH
df_scf_guess false
scf_type direct
# guess gwh
}
basis {
assign oldDKHtzvp
}
df_basis_sapt {
assign oldDKHtzvp-ri
}
energy('sapt0-ct')
```

And the corresponding output (rather long, some parts suppressed):

```
...
//>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>//
// SAPT Charge Transfer //
//<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<//
//>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>//
// Dimer Basis SAPT //
//<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<//
*** tstart() called on tarsus.cdi.unifran.br
*** at Thu Apr 13 08:54:39 2017
SAPT0
Ed Hohenstein
6 June 2009
Orbital Information
--------------------------
NSO = 390
NMO = 390
NRI = 1992
NOCC A = 35
NOCC B = 35
FOCC A = 0
FOCC B = 0
NVIR A = 355
NVIR B = 355
Elst10,r = -0.071295970477 [Eh]
Exch10 = 0.101705542455 [Eh]
Exch10 (S^2) = 0.095167728553 [Eh]
Iter Energy [mEh] dE [mEh] Residual Time [s]
1 -339.31922137 339.319221372 86.019096848 1
2 -428.41275195 89.093530581 27.015253835 2
...
EVERYTHING GOES FINE AT THIS POINT
...
CHF Iterations converged
...
Total HF -22.66371892 [mEh] -14.22169893 [kcal/mol] -59.50359402 [kJ/mol]
Total SAPT0 -38.92147546 [mEh] -24.42359561 [kcal/mol] -102.18833383 [kJ/mol]
Special recipe for scaled SAPT0 (see Manual):
Electrostatics sSAPT0 -71.29597048 [mEh] -44.73889879 [kcal/mol] -187.18757049 [kJ/mol]
Exchange sSAPT0 101.70554245 [mEh] 63.82119409 [kcal/mol] 267.02790171 [kJ/mol]
Induction sSAPT0 35.08726412 [mEh] 22.01759156 [kcal/mol] 92.12161194 [kJ/mol]
Dispersion sSAPT0 -15.12940925 [mEh] -9.49384804 [kcal/mol] -39.72226399 [kJ/mol]
Total sSAPT0 50.36742684 [mEh] 31.60603883 [kcal/mol] 132.23967917 [kJ/mol]
--------------------------------------------------------------------------------------------------------
*** tstop() called on tarsus.cdi.unifran.br at Thu Apr 13 08:56:14 2017
Module time:
user time = 633.67 seconds = 10.56 minutes
system time = 37.93 seconds = 0.63 minutes
total time = 95 seconds = 1.58 minutes
Total time:
user time = 4432.58 seconds = 73.88 minutes
system time = 48.71 seconds = 0.81 minutes
total time = 578 seconds = 9.63 minutes
//>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>//
// Monomer Basis SAPT //
//<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<//
*** tstart() called on tarsus.cdi.unifran.br
*** at Thu Apr 13 08:56:14 2017
SAPT0
Ed Hohenstein
6 June 2009
Orbital Information
--------------------------
NSO = 390
NSO A = 230
NSO B = 160
NMO = 390
NMO A = 230
NMO B = 160
NRI = 1992
NOCC A = 35
NOCC B = 35
FOCC A = 0
FOCC B = 0
NVIR A = 195
NVIR B = 125
Elst10,r = -nan [Eh]
Exch10 = -nan [Eh]
Exch10 (S^2) = -nan [Eh]
Iter Energy [mEh] dE [mEh] Residual Time [s]
1 -nan -nan -nan 0
CHF Iterations did not converge
...
EVERY NUMBER NOW IS A NAN
...
Total HF -23.85721138 [mEh] -14.97062679 [kcal/mol] -62.63710849 [kJ/mol]
Total SAPT0 -nan [mEh] -nan [kcal/mol] -nan [kJ/mol]
Special recipe for scaled SAPT0 (see Manual):
Electrostatics sSAPT0 -nan [mEh] -nan [kcal/mol] -nan [kJ/mol]
Exchange sSAPT0 -nan [mEh] -nan [kcal/mol] -nan [kJ/mol]
Induction sSAPT0 -nan [mEh] -nan [kcal/mol] -nan [kJ/mol]
Dispersion sSAPT0 -nan [mEh] -nan [kcal/mol] -nan [kJ/mol]
Total sSAPT0 -nan [mEh] -nan [kcal/mol] -nan [kJ/mol]
--------------------------------------------------------------------------------------------------------
*** tstop() called on tarsus.cdi.unifran.br at Thu Apr 13 08:56:46 2017
Module time:
user time = 232.98 seconds = 3.88 minutes
system time = 9.27 seconds = 0.15 minutes
total time = 32 seconds = 0.53 minutes
Total time:
user time = 4665.56 seconds = 77.76 minutes
system time = 57.98 seconds = 0.97 minutes
total time = 610 seconds = 10.17 minutes
SAPT Charge Transfer Analysis
------------------------------------------------------------------------------------------------
SAPT Induction (Dimer Basis) -56.1102 [mEh] -35.2097 [kcal/mol] -147.3174 [kJ/mol]
SAPT Induction (Monomer Basis) nan [mEh] nan [kcal/mol] nan [kJ/mol]
SAPT Charge Transfer nan [mEh] nan [kcal/mol] nan [kJ/mol]
*** Psi4 exiting successfully. Buy a developer a beer!
```

But the following input (pretty much the same!) works (it gives me a -30.2 kcal/mol charge transfer term):

```
memory 29 gb
molecule dimer {
-1 1
I 0.000 0.00000 3.069
--
0 1
I 0.00000 0.00000 0.00000
I 0.000 0.0000 -3.069
units angstrom
}
set globals {
maxiter 500
relativistic DKH
df_scf_guess false
scf_type direct
guess gwh
}
basis {
assign oldDKHtzvp
}
df_basis_sapt {
assign oldDKHtzvp-ri
}
energy('sapt0-ct')
```

I don’t believe it has something to do with either the initial guess (I have tested both the default, which sometimes gives me convergence problems for the dimer calculation, and `gwh`

, which works fine overall) or the `scf_type`

keyword (I have tested both `pk`

, which sometimes gives me storage problems, and `direct`

).

And since the basis set is the same, I doubt it has something to it too.

It would be very nice if someone could give me advice on this.

Thanks in advance!

All the best,

Felipe