Non-adiabatic coupling calculations different in different versions
When calculating non-adiabatic couplings, there seems to be something different between the current version of OpenMolcas and other codes on the market.
I'm using version 22.10 (tag 941-g2ccd2587-dirty ?) - the most recent one with support for CASPT2 gradients.
Using a very simple example of NH3 with two near degenerate 3p states (input file molcas.input), calculating the NACME with MCLR gives
*****************************************************
* *
* CI derivative coupling *
* *
*****************************************************
Irreducible representation: a
------------------------------------------------------------------------------------------
X Y Z
------------------------------------------------------------------------------------------
N1 -1.71057560579134E-05 2.88826915218245E-02 -2.88797371648402E-02
H2 -2.78900216733274E-02 -1.51557247036142E-02 2.60816820015898E-02
H3 2.78298898722781E-02 -2.61101793780746E-02 1.52633474036251E-02
H4 7.72375571067348E-05 1.23832125598674E-02 -1.24652922403771E-02
------------------------------------------------------------------------------------------
******************************************************
* *
* CSF derivative coupling *
* *
******************************************************
Irreducible representation: a
------------------------------------------------------------------------------------------
X Y Z
------------------------------------------------------------------------------------------
N1 1.33224900668379E-04 -2.44718799543312E-02 2.43586897182336E-02
H2 -1.35465695976892E-01 1.13567708850823E-01 1.20832058093483E-02
H3 1.35355807673248E-01 -1.20932051803874E-02 -1.13475931080057E-01
H4 7.49829471092158E-05 -1.05474278750654E-01 1.05434486214726E-01
------------------------------------------------------------------------------------------
Energy difference: -4.562499E-03
*******************************************************
* *
* Total derivative coupling *
* *
*******************************************************
Irreducible representation: a
------------------------------------------------------------------------------------------
X Y Z
------------------------------------------------------------------------------------------
N1 -1.77135945030357E-05 2.89943444439587E-02 -2.89908736566616E-02
H2 -2.72719596023715E-02 -1.56738772352931E-02 2.60265523899751E-02
H3 2.72123291665721E-02 -2.60550041443419E-02 1.57810811993380E-02
H4 7.68954475029529E-05 1.28644388266880E-02 -1.29463369538016E-02
------------------------------------------------------------------------------------------
Also, the calculation crashes with the error
Invalid disk address -512
I/O error in DaFile_checkarg
Unit = 20
--- Stop Module: alaska at Mon Feb 20 17:16:27 2023 /rc=_RC_FLOATING_EXCEPTION_ ---
Other NACME calculations have worked, so I can't be sure it's not something I've done wrongly in this calculation.
An old version of OpenMolcas (compiled some time last year, let me know if there is some way of identifying which version) gives me with the same input
*****************************************************
* *
* CI derivative coupling *
* *
*****************************************************
Irreducible representation: a
------------------------------------------------------------------------------------------
X Y Z
------------------------------------------------------------------------------------------
N1 -1.67284105760050E-05 -2.88945510396170E-02 2.88956824136539E-02
H2 2.78408369375394E-02 1.52727581404319E-02 -2.61215041655739E-02
H3 -2.79037117510538E-02 2.60921862182864E-02 -1.51611161046997E-02
H4 7.96032240869565E-05 -1.24703933191032E-02 1.23869378566223E-02
------------------------------------------------------------------------------------------
******************************************************
* *
* CSF derivative coupling *
* *
******************************************************
Irreducible representation: a
------------------------------------------------------------------------------------------
X Y Z
------------------------------------------------------------------------------------------
N1 1.39085244485584E-04 2.43403532107883E-02 -2.44594053652009E-02
H2 1.35341258011813E-01 -1.13463665425712E-01 -1.20922264501709E-02
H3 -1.35465779391889E-01 1.20884099197381E-02 1.13567006026437E-01
H4 8.70211261722886E-05 1.05429903808260E-01 -1.05481483293805E-01
------------------------------------------------------------------------------------------
Energy difference: -4.562603E-03
*******************************************************
* *
* Total derivative coupling *
* *
*******************************************************
Irreducible representation: a
------------------------------------------------------------------------------------------
X Y Z
------------------------------------------------------------------------------------------
N1 3.80550339707323E-03 6.35724962520584E+00 -6.35761664415228E+00
H2 -5.96662219316290E+00 -3.46084191345129E+00 5.71303967477950E+00
H3 5.98027814163454E+00 -5.70661778495587E+00 3.43647631932922E+00
H4 -1.73598668773753E-02 2.83860507471483E+00 -2.82036545903975E+00
------------------------------------------------------------------------------------------
clearly very different from the newer version.
Further more, checking these results with MOLPRO gives a coupling of
N1 0.000000046 6.350396164 -6.350396235
H2 -5.967929195 -3.444953445 5.705657129
H3 5.967927601 -5.705657404 3.444955251
H4 0.000001539 2.828658499 -2.828659950
much larger than in first example, very close to the second example.
The difference is simply that the second is divided by the energy, but the first is not. Will the output continue to be the same?