Ceilidh ... Re: state-averaged restricted active space SCF

Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

We are NATO-free zone

Re: state-averaged restricted active space SCF

Michal Krompiec
michal.krompiec@merckgroup.com

It seems that it is possible with GUGA. SO this is my attempt so far:
My molecule has 408 electrons. I want to freeze the lowest 200 orbitals, and do a SA4-RASSCF(8,20) with singlet excitations only (step B from the Truhlar group paper). But the calculation crashes with an error which doesn’t explain much (see below).
I tried SA4-RASSCF(8,16) with the same effect. I would be very grateful for any advice.
This is my input:
$contrl icut=12
SCFTYP=MCSCF RUNTYP=ENERGY MAXIT=600
COORD=UNIQUE D5=.TRUE.
UNITS=ANGS wide=.t. $END
$SYSTEM MWORDS=200 TIMLIM=60000 aoints=dist $END
$BASIS EXTFIL=.t. GBASIS=cc-pvtz $END
$SCF SOSCF=.f. DIIS=.t. DIRSCF=.t. MAXDII=600 NPUNCH=2
FDIFF=.f. NCONV=7 $END
$TRANS MPTRAN=2 DIRTRF=.T. AOINTS=DIST ALTPAR=.T. MODE=112 $END
$ciinp castrf=.t. $end
$INTGRL QFMM=.f. SCHWRZ=.TRUE. $END
$DRT group=C1 fors=.f.
foci=.f. soci=.f. iexcit=1
nmcc=200 ndoc=4 nval=20 next=0
naos=0 nbos=0 nalp=0 $END
$GUGDIA nstate=4 $END
$GUGDM2 wstate(1)=0.25,0.25,0.25,0.25 $end
$GUGEM CUTOFF=1.0D-20 dirci=.f. pack2=.t. $END
$GUGDIA CVGTOL=1.0D-7 $END
$GUGDM2 CUTOFF=1.0D-15 $END
$MCSCF cistep=guga ACURCY=1D-7 ENGTOL=1.0D-12 FORS=.FALSE. $END
$guess guess=moread norb=2200 $end

And this is the end of the output file:
ACTIVE= A A A A A A A A A A
DOC DOC DOC DOC VAL VAL VAL VAL VAL VAL
ACTIVE= A A A A A A A A A A
VAL VAL VAL VAL VAL VAL VAL VAL VAL VAL

MOLECULAR CHARGE = 0
NUMBER OF ALPHA ELECTRONS = 204
NUMBER OF BETA ELECTRONS = 204

THE ELECTRONIC STATE IS 1-A

THE DISTINCT ROW TABLE HAS 40 ROWS.
THE WAVEFUNCTION CONTAINS 65 WALKS (CSF-S).

TOTAL NUMBER OF INTEGRALS = 24645
NUMBER OF INTEGRALS/GROUP = 20000
NUMBER OF INTEGRAL GROUPS = 2
MAXIMUM RECORD SIZES ARE 8225 FOR UNIT 11 65 FOR UNIT 12
20000 FOR UNIT 15 22501 FOR UNIT 16
...... END OF -DRT- GENERATION ......

TIMING STATISTICS ON RANK 0:
CPU TIME: STEP = 0.72 , TOTAL = 49.6 SECONDS ( 0.8 MIN)
WALL CLOCK TIME: STEP = 0.73 , TOTAL = 45.6 SECONDS ( 0.8 MIN)
CPU UTILIZATION: STEP = 98.78%, TOTAL = 108.87%
OPENING FILE DAFL30 WITH 81 LOGICAL RECORDS OF 5456896 WORDS
DAFL30 WILL CONTAIN A MAXIMUM OF 6906384 PHYSICAL RECORDS
EACH PHYSICAL RECORD WILL CONTAIN 64 WORDS

EVALUATING EXCHANGE INTEGRALS FOR DIRECT INTEGRAL TRANSFORMATION
SCHWARZ INEQUALITY OVERHEAD: 937970 INTEGRALS, CPU TIME= 0.18

------------------------------
CASSCF INTEGRAL TRANSFORMATION
------------------------------
AO INTEGRALS WILL BE CALCULATED IN DIRECT MODE...

Fatal error in CASTRF, error code = -1

TIMING STATISTICS ON RANK 0:
CPU TIME: STEP = 4.70 , TOTAL = 54.3 SECONDS ( 0.9 MIN)
WALL CLOCK TIME: STEP = 4.71 , TOTAL = 50.3 SECONDS ( 0.8 MIN)
CPU UTILIZATION: STEP = 99.80%, TOTAL = 108.02%

24564883 WORDS OF DYNAMIC MEMORY USED
87883324 BYTES OF HEAP MEMORY USED, 414110 BYTES REMAIN IN USE

WARNING! THIS VERSION OF FIREFLY IS PROBABLY OUTDATED!
PLEASE CHECK FIREFLY HOMEPAGE FOR INFORMATION ON UPDATES!

EXECUTION OF FIREFLY TERMINATED ABNORMALLY AT 19:34:14 15-MAR-2018

On Thu Mar 15 '18 2:16pm, Michal Krompiec wrote
-----------------------------------------------
>Hello,
>is it possible to do a SA RAS-SCF MCSCF in Firefly? I would like to restrict the wavefunction to single excitations only (so esstentially, it would be a CIS MCSCF).
>I'd like to reproduce results of this paper from Truhlar's group: DOI: 10.1021/acs.jctc.8b00032 (Automatic Selection of an Active Space for Calculating Electronic Excitation Spectra by MS-CASPT2 or MC-PDFT), step B is (for a closed-shell ground state) a state-averaged CIS-MCSCF.

>Thanks,

>Michal

Fri Mar 16 '18 11:22am

This message read 371 times