Hello, I am encountering a problem in my Gaussian16 calculation related to the convergency of the CPHF cycle: in short my calculation doesn't converge and I cannot understand why. That's the behavior I am observing :

LinEq1: Iter= 1 NonCon= 6 RMS=2.46D+01 Max=3.33D+04 NDo= 6

LinEq1: Iter= 2 NonCon= 6 RMS=2.08D+02 Max=2.02D+05 NDo= 6

LinEq1: Iter= 3 NonCon= 6 RMS=1.33D+03 Max=1.06D+06 NDo= 6

LinEq1: Iter= 4 NonCon= 6 RMS=1.08D+04 Max=1.12D+07 NDo= 6

LinEq1: Iter= 5 NonCon= 6 RMS=1.24D+05 Max=1.43D+08 NDo= 6

LinEq1: Iter= 6 NonCon= 6 RMS=7.02D+05 Max=5.40D+08 NDo= 6

LinEq1: Iter= 7 NonCon= 6 RMS=5.35D+06 Max=4.52D+09 NDo= 6

ecc.

LinEq1: Iter=247 NonCon= 3 RMS=3.29D+12 Max=1.21D+15 NDo= 3

LinEq1: Iter=248 NonCon= 3 RMS=1.97D+12 Max=7.35D+14 NDo= 3

LinEq1: Iter=249 NonCon= 3 RMS=1.15D+12 Max=4.08D+14 NDo= 3

LinEq1: Iter=250 NonCon= 3 RMS=6.81D+11 Max=2.22D+14 NDo= 3

and it stops due to wall time.

so it seems than more than converging is actually diverging

This is my input file (basis set is a defTZVPD , basically it's a small organic molecule on a gold sheet) :

%Mem=80GB

%NProc=64

%Oldchk=BPT.chk

%chk=BPT_32-0.01.chk

#p b3lyp/gen pseudo=read polar CPHF=RdFreq guess=read Nosymm Pop=Full IOP(6/7=3,9/40=45) int=ultrafine empiricaldispersion=GD3BJ

Title Card Required

0 1

785nm

I am wondering if this problem could be due to a non optimal geometry optimization in the previous step (that I took from another software), but the SCF cycle before the CPHF is converging, so I don't think it could be a structure optimization related problem.

Please, let me know if you have any suggestion to solve the issue.

Thanking you in advance

Best regards

Francesco Ciccarello - EPFL Lausanne