Molcas Forum

DanielR

Member

Registered: 2015-12-15

Posts: 6

"Orthogonality violated" error in UHF calculation

Hi everyone,
I am trying to obtain initial orbitals from a UHF calculation in a radical using the ANO-RCC-VTZP basis set, but the SCF process does not converge, eventually leading to an error printing "orthogonality violated". I tried to decrease to set the OVLdelete parameter to 1.0d-5, but the problem persists.
I would appreciate any suggestions addressing this point.

Here is the input:

> EXPORT MOLCAS_MOLDEN=ON
> EXPORT MOLCAS_PRINT=3

&Gateway
coord=hgoh.xyz
basis=ANO-RCC-VTZP
RICD
GROUP=nosym

&Seward

&SCF
UHF
SPIN=2
CHARGE=0
OVLDelete=1.d-5

Here is the final output (from the SCF module):

 177  -717.966169873 -5047.562614204  4149.261230877   0.25E+03*   0.97E+09*   0.13E+03*    0.10E+04   0.10E+04   QNRc2DIIS    1.
 178 -1057.275815718 -5460.243230322  4222.632201150  -0.34E+03*   0.12E+10*   0.23E+03*    0.85E+03   0.10E+04   QNRc2DIIS    1.
 179 -1091.775782062 -5463.543089454  4191.432093939  -0.34E+02*   0.30E+09*   0.20E+03*    0.84E+03   0.10E+04   QNRc2DIIS    1.
 180 -1100.046857861 -5496.856442344  4216.474371029  -0.83E+01*   0.20E+09*   0.12E+03*    0.97E+03   0.10E+04   QNRc2DIIS    1.
 181 -1233.258041273 -5647.862586604  4234.269331878  -0.13E+03*   0.12E+10*   0.14E+03*    0.66E+03   0.10E+04   QNRc2DIIS    1.
 182 -1039.893220448 -5435.171642834  4214.943208932   0.19E+03*   0.23E+09*   0.13E+03*    0.11E+04   0.10E+04   QNRc2DIIS    1.
  iSym =                     1
  largest off diag element: [                   128 ,                   127 ] =
  5.439238094668425E-007

  ***** Orthogonality violated *****
  iSym =                     1
  diag element most different from 1.0: [                   128 ,
                   128 ] =   0.999998346135725
--- Stop Module: scf at Tue Jul 10 10:57:51 2018 /rc=_RC_INTERNAL_ERROR_ ---
--- Module scf spent 4 minutes 16 seconds ---

    Timing: Wall=332.41 User=325.90 System=5.83

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: "Orthogonality violated" error in UHF calculation

Try the (still undocumented) RS-RFO keyword if you have a recent OpenMolcas, or different combinations of NoDIIS, DIISThr, C1DIIS, QNRThr, NoDamp.