Learn how to ask questions correctly

Alex Granovsky

gran@classic.chem.msu.su

Dear Andrey,

in addition to what Thom has already suggested I'd like to share some points

specific to calculations on diatomic molecules.

1. Use coord=unique. Never use coord=cart!

2. Initially, it is better start with RHF orbitals of closed-shell

axially-symmetric ion or neutral molecule. This will give you

starting orbitals of proper symmetry. E.g., RHF orbitals of singlet

N_{2} in ground state are symmetric while RHF orbitals of singlet O_{2} are not!

For O_{2}, you would need RHF orbitals of O_{2}^{2+} system.

3. To preserve symmetry in subsequent MCSCF calculations and to give

physically reasonable solutions with no symmetry contaminants,

you need to use state-averaged MCSCF averaging over both components

of the given irrep of the full point group (the only exception are

sigma states). To do so, you actually need to lower symmetry so that

both components fall into the same irrep of some abelian subgroup of

the full symmetry group. E.g. one may use C2 (Cs) symmetry groups for

Pi (Delta) states, or even resort to C1 group.

4. When going to SOCI or XMCQDPT2, you need to get canonical MCSCF

orbitals first. Add $mcscf canonc=.t. $end to your input. This may

require further lowering symmetry. esp. with older versions of Firefly.

Kind regards,

Alex Granovsky

On Mon Feb 27 '12 5:51pm, Thomas wrote

--------------------------------------

>Dear Andrey,

>With respect to your first question, you can enable the use of symmetry during the MCSCF calculation with the GROUP and ISTSYM keywords in $DET. For example,

>$DET GROUP=D4H ISTSYM=1 NSTATE=10 NCORE=4 NELS=8 NACT=6 $END

>Here, ISTSYM chooses the irrep you would like to obtain states of. If you would like to obtain states of all irreps, you should give GROUP=C1 (which is the default). In this case, the symmetry will indeed be labelled as C1 in the output (though symmetry will still be exploited to speed up the calculation).

>As for your second question, I'm not sure if there is any option for that. Someone else should be able to answer that.

>Note that the $DRT group in your input currently doesn't do anything. Firefly can do MCSCF calculations with two different codes: the ALDET code and the GUGA code. The first is requested with CISTEP=ALDET and is controlled with keywords in the $DET group. The second is requested with CISTEP=GUGA and is controlled with keywords in the $DRT group (as well as a few other groups). Thus, when you give CISTEP=ALDET the program will use $DET, not $DRT, for MCSCF input.

>You also specified SOCI=1. This keywords requests a second order CI to be performed after the MCSCF calculation. Such a followup calculation is done by the GUGA code, but it is controlled by keywords in the $CIDRT group, not $DRT (so, $DRT SOCI=1 doesn't do anything). A followup CI calculation can be requested with $CONTRL SCFTYP=MCSCF CITYP=GUGA, after which the FOCI, SOCI, or IEXCIT keyword can be used to determine the CI type, i.e. which excitations will be done.

>Hope this helps. Please see the manual for some input examples.

>

>

>Kind regards,

>Thom

>

>

>

>

>

>

>On Mon Feb 27 '12 4:42pm, Andrey wrote

>--------------------------------------

>>Hi, collagues

>>I need to do some MCSCF calculations of electronic spectra of diatomic molecules, for instance, the O2 molecule.

>>I used the input file like that

>>! ================================================================

>>! Input file for FireFly

>>! ================================================================

>> $CONTRL SCFTYP=MCSCF RUNTYP=ENERGY EXETYP=RUN $END

>> $CONTRL COORD=CART UNITS=ANGS $END

>> $BASIS GBASIS=N311 NGAUSS=6 NDFUNC=1 $END

>> $MCSCF CISTEP=ALDET SOSCF=1 $END

>> $DET NSTATE=10 NCORE=4 NELS=8 NACT=6 $END

>> $DRT GROUP=D4H NPRT=1 SOCI=1 $END

>>

>> $DATA

>> Molecule specification

>>Dnh 4

>>

>>O 8.000000 0.000000 0.000000 -0.6035

>>O 8.000000 0.000000 0.000000 0.6035

>> $END

>>

>>$GUESS GUESS=MOREAD NORB=38 $END

>>$VEC

>>===

>>$END

>>In the output file I have the printout like

>>

>>

>>

>>CI EIGENVECTORS WILL BE LABELED IN GROUP=C1

>> PRINTING CI COEFFICIENTS LARGER THAN 0.050000

>> STATE 1 ENERGY= -149.7365263188 S= 1.00 SZ= 0.00 SPACE SYM=A

>> ALPHA | BETA | COEFFICIENT

>>--------|--------|------------

>> 111010 | 111100 | 0.6865861

>> 111100 | 111010 | -0.6865861

>> 101110 | 011110 | -0.1133390

>> 011110 | 101110 | 0.1133390

>> 110011 | 110101 | -0.0603978

>> 110101 | 110011 | 0.0603978

>> STATE 2 ENERGY= -149.7035781216 S= 0.00 SZ= 0.00 SPACE SYM=A

>> ALPHA | BETA | COEFFICIENT

>>--------|--------|------------

>> 111010 | 111010 | 0.6454186

>> 111100 | 111100 | -0.6454186

>> 111100 | 111010 | 0.2134040

>> 111010 | 111100 | 0.2134040

>> 011110 | 101110 | -0.1514058

>> 101110 | 011110 | -0.1514058

>> 110011 | 110011 | -0.0587975

>> 110101 | 110101 | 0.0587975

>>

>>

>>The questions are the following :

>>1) Why the symmetry is labelled as C1 for all the states, while

>>in the input I have used the D4H ?

>>2) Is there any option to print out the total angular momentum

>>projections of the states (and any other properties that

>>can be used to identify the states in the spectrum) ?

>>Thanks for any help in advance.

Mon Feb 27 '12 8:36pm

This message read