local CCSD(T) does not converge

Dear all,

When I run the file below, the ccsd program does not converge at occupied orbital 11. I have tried all dfalg settings (InvSqrt, Cholesky, LinEq). They produce different correlation energies for this orbital but all fail. The same problem exists for the aug-cc-pv5z calculations and different mrcc versions. Is there any way to fix that?

Thank you in advance for your help,

-Eugene

basis=aug-cc-pvqz
calc=CCSD(T)
mem=250GB
#rest=1
localcc=2016
lcorthr=tight
ccmaxit=250

geom=xyz
14

C 0.936008352176 1.416520416155 0.000000000000
C -0.241637230572 0.668566370331 0.000000000000
C -0.243671543247 -0.750713210657 0.000000000000
C 0.985545414872 -1.427932310833 0.000000000000
C 2.154848517559 -0.689076832284 0.000000000000
C 2.131704987347 0.718382031556 0.000000000000
H 0.920120480363 2.498420120164 0.000000000000
H 1.017199520730 -2.509400275402 0.000000000000
H 3.108453028233 -1.198743244070 0.000000000000
H 3.064766270333 1.264631211555 0.000000000000
C -1.607944701754 -1.165127889476 0.000000000000
C -2.373083619219 -0.025346395704 0.000000000000
H -3.443296067907 0.090949278302 0.000000000000
N -1.551506866233 1.075088906648 0.000000000000
H -1.863670573596 2.029398364206 0.000000000000
H -1.985351920807 -2.173088085078 0.000000000000

Dear Eugene,

I started to run your input file, it will take some time.
Looking at your molecule it has quite strange structure. In that form I think it should have some biradical character. In that case a closed-shell CC model might not be the best choice.
Adding two more H atoms on C(11) and N(14) would yield indole, which seems a much more probable system to me.
If you are indeed interested in the provided molecule, please, also attach your output file and list the MRCC versions you have tried.

Thank you,
Peter

Dear Peter,

This is indole C8H7N. The last two lines of the input are the hydrogens you mentioned.

-Eugene

Dear Eugine,

In that case, please, correct the number of atoms specified after the geom=xyz line. That should be 16 instead of 14. I was looking at the COORD.xyz file produced by MRCC and that is why the two H atoms were missing for me.

This should probably solve your convergence problem. Let me know, if you have any remaining issue with the corrected calculation and I will look at it again.

Thank you for using our program,
Peter

Thank you very much!

Best regards,

-Eugene