## Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

** Re^2: An effect of $MCSCF iforb=.t. $END on SA-MCSCF **
Pavlo Solntsev

pavlo.solntsev@gmail.com

Sorry, Alex. I suddenly changed cvgtol value in $det (see below). I put a correct one and everything works fine.

I apologize for the false alarm.

Pavel.

On Fri May 30 '14 5:25pm, Alex Granovsky wrote

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

>Dear Pavel,

>this is indeed a bit strange. Check if the correct roots are

>properly captured by the CI's initial guess routine. If not,

>try to increase NSTGSS and NHGSS in $det.

>Kind regards,

>Alex Granovsky

>

>

>On Thu May 29 '14 8:38pm, Pavlo Solntsev wrote

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

>>Alex,

>>I faced a problem. I successfully got CASSCF(11,11) solution with two doublets and one quartet. I use SA-MCSCF for these states since they are sit very close to each other. The total number of states is 10. I did calculations without $MCSCF iforb=.t. sd=.t. $END. Then i added this keyword and reused optimized orbitals, but my solution was quite different. The state order before was D,D,Q, but with iforb=.t. it is Q,7... Is it normal effect of iforb or there is a pitfall somewhere here?

>>My input file looks

>> $CONTRL SCFTYP=mcscf RUNTYP=energy d5=.t. icharg=0 wide=.t. gencon=.t.

>> MAXIT=50 icut=12 itol=35 exetyp=run inttyp=hondo mult=2 $END

>> $det ncore=150 nact=11 nels=11 wstate(1)=1,1,1,-0 nstate=10

>> pures=.f. cvgtol=1d- $end

>> $mcscf cistep=aldet fors=.t. ntrack=5 numfo=5 iforb=1 canonc=1 sd=.t. $end

>> $TRANS DIRTRF=.t. mode=112 mptran=2 AOINTS=DIST altpar=.t. cuttrf=1d-13 $END

>> $SYSTEM MWORDS=250 nojac=1 mxbcst=-1 $END

>> $MPI mxgsum=0x40000 mnpdot=1d9 $end

>> $BASIS extfil=.t. $END

>> $p2p p2p=.t. dlb=.t. $end

>> $smp httfix(1)=.t. $end

>> $moorth nostf=1 nozero=1 syms=1 symden=1 symvec=1 tole=0 tolz=0 $end

>> $guess guess=moread norb=708 $end

>>

>>

>>I also tried to read orbitals with asis=.t. but it didn't help.

>>Sincerely,

>>Pavel.

>>

>>

>>

Fri May 30 '14 10:50pm

This message read **560** times