• From: "Salter-Duke, Brian James -" <brian.james.duke##gmail.com>
• Subject: CCL: (Firefly): GUESS=MOREAD errors
• Date: Wed, 9 Mar 2016 07:33:45 +1100

``` Sent to CCL by: "Salter-Duke, Brian James -"
[brian.james.duke:_:gmail.com]
GAMESS(US) users are unlikely to be able to help as in GAMESS(US) you can not
add basis functions using GUESS=MOREAD from the run with a smaller basis. Does
EXTRA=.T.  allow this in Firefly? I have not used Firefly much.
Have you tried GUESS=MOREAD for the same basis set? Your error message, if I
understand it right in the mess below, suggests it is throwing away some basis
functions and you might have to lower NORB even when using the same basis set.
Cheers, Brian.
On Tue, Mar 08, 2016 at 04:08:32PM +0000, David Shobe avidshobe()yahoo.com
wrote:
diffuse functions?  The manual has been helpful, but not entirely so.
How do I calculate the proper value for NORB?
> I'm a new user of Firefly (formerly known as PC GAMESS, so any GAMESS(US)
users are free to contribute).
> I'm trying to use GUESS=MOREAD while changing the basis set to include
diffuse functions:
> ! \$CONTRL SCFTYP=RHF  RUNTYP=ENERGY  MPLEVL=4 UNITS=ANGS
\$END
\$SYSTEM TIMLIM=3600 MEMORY=36000000 \$END

\$SCF
DIRSCF=.True. \$END

\$GUESS
\$EXTRAF     NEXTRA(1)=4,4,4,4,4,4,4,4,4,4
!DIFFSP for 10 C
NEXTRA(11)=1,1,1,1,1,1,1,1,1,1,1,1,1,1   !DIFFS  for 14 H \$END

\$BASIS  GBASIS=DZV NPFUNC=1 NDFUNC=1 DIFFSP=1 DIFFS=1
\$END
\$ELMOM
IEMOM=2 WHERE=COMASS \$END          \$MP4
SDTQ=.T. \$END

\$DATA

2,2,5,5-tetramethyl-3-hexyne
RMP4/DZV(d,p)++//RMP2/DZV(d,p)

DNH 3
>  C_3         6.0   0.0000000000
0.0000000000  -0.6171527518 C_2         6.0
0.0000000000   0.0000000000  -2.0934695855 C_1A
6.0   1.4521398958   0.0000000000
-2.6001055211 H_1AA       1.0   1.9835899449
0.8863051714  -2.2440032755 H_1AC       1.0
1.4664232611   0.0000000000  -3.6950540786 \$END

\$VEC 1
1-7.05806381E-03-2.34618262E-04 4.28989286E-04 0.00000000E+00
0.00000000E+00 1  2-3.13369076E-04 0.00000000E+00 0.00000000E+00
5.72390159E-04 7.12360225E-06etc.
> I'm not sure how to figure out what goes after NORB=; here I assumed it was
the number of orbitals listed in \$VEC, which appears to be a 240 by 240 matrix.
Anyway, I get the following error message.
>                     * * *
WARNING * *
* ------------------------------------------------------------------------------ THE
OVERLAP MATRIX HAS   12 EIGENVALUES BELOW 1.2E-03. THE SMALLEST OF
THESE IS 2.20132E-08. THIS INDICATES A PARTIAL LINEAR DEPENDENCE IN YOUR
ATOMIC BASIS.
>  TO OBTAIN SCF CONVERGENCE MAY REQUIRE MORE ACCURATE INTEGRAL
EVALUATION (INTTYP=HONDO, ICUT=11, ITOL=30 IN \$CONTRL), MORE ACCURATE
DIRECT SCF FOCK MATRIX FORMATION (FDIFF=.FALSE. IN \$SCF), OR CHANGING
CONVERGERS (DIIS=.T. SOSCF=.F. IN \$SCF).
>  EIGENVALUES BELOW 1.0D-07 PROBABLY WON'T CONVERGE. EIGENVALUES
BETWEEN 1.0D-07 AND 1.0D-06 MAY REQUIRE TIGHTENING OF -NCONV- DENSITY
CONVERGENCE IN \$SCF.
>  THE OVERALL DEGREES OF AOS LINEAR INDEPENDENCE ARE:
0.9427       -    C   1 XY,   SHELL
7,   AO   17   0.9427       -
C   2 XY,   SHELL  14,   AO   36[I left out a
long list of nearly linearly dependent orbitals.]   0.7648E-04
-    C   9  S,   SHELL  62,   AO  162
>    0.1098E-04   -    C   3  Z,
SHELL  20,   AO   51   0.1098E-04   -
C   4  Z,   SHELL  27,   AO   70
0.3167E-06   -    C   3  S,   SHELL
20,   AO   48   0.3167E-06   -    C
4  S,   SHELL  27,   AO   67
0.1494E-06   -    C   1  Z,   SHELL
6,   AO   13   0.1494E-06   -    C   2
Z,   SHELL  13,   AO   32   0.2764E-07
-    C   2  S,   SHELL  13,   AO
29   0.2764E-07   -    C   1  S,
SHELL   6,   AO   10
>  YOU MAY CONSIDER DROPPING ONE OR MORE AOS, STARTING FROM THE END OF
THIS
LIST. ------------------------------------------------------------------------------ ***
ERROR READING THE MO COEFFICIENTS AT   1  49     THE
INPUT VALUES WERE   2   1
>  ADDRESS 0x0053D0BA HAS INITIATED PROGRAM ABORT BECAUSE OF FATAL
ERROR(S)
>  CPU        TIME:   STEP =
0.03 ,  TOTAL =        0.4 SECONDS (
0.0 MIN) WALL CLOCK TIME:   STEP =      0.04 ,
TOTAL =        0.4 SECONDS (    0.0
MIN) CPU UTILIZATION:   STEP =     71.30%,  TOTAL =
107.14%
>       506417 WORDS OF    DYNAMIC MEMORY USED
1563406 BYTES OF    HEAP MEMORY    USED,
120095 BYTES REMAIN IN USE EXECUTION OF FIREFLY TERMINATED
ABNORMALLY AT  9:29:44  8-MAR-2016
> Anyway, if I drop the AOs as suggested, it won't really be
RMP4/DZV(d,p)++//RMP2/DZV(d,p), will it, but some unnameable subbasis thereof?
And what does Firefly expect at "1 49"?   Shouldn't there
be 48 lines per orbital (each requiring 240 coefficients, and there are 5
coefficients per line, 240/5 = 48).  I copied that part directly from the
\$VEC section of the PUNCH file for an HF/DZV(d,p) calculation, so there
shouldn't be any errors there.  But, as I've implied, NORB=240 could be
wrong.
>
> My input is based partially on the following example in the Firefly manual.
I've tried to adapt it to my situation (C10H14 with an already converged
DZV(d,p) wavefunction), but as you saw above, I'm not succeeding.
>
> \$CONTRL SCFTYP=RHF D5=.T. \$END\$SYSTEM TIMLIM=1 MEMORY=500000 \$END\$GUESS
\$END\$DATARHF/aug-cc-pVDZ run that uses converged RHF/cc-pVDZ orbitalsCNV 2O 8.0
0.0000000000 0.0000000000 0.7205815395S 81 11720.00000 0.710000000E-032
1759.00000 0.547000000E-023 400.80000 0.278370000E-014 113.70000 0.1048000005
37.03000 0.2830620006 13.27000 0.4487190007 5.02500 0.2709520008 1.01300
0.154580000E-01S 81 11720.0000 -0.160000000E-032 1759.0000 -0.126300000E-023
400.80000 -0.626700000E-024 113.70000 -0.257160000E-015 37.03000
-0.709240000E-016 13.27000 -0.1654110007 5.02500 -0.1169550008 1.01300
0.557368000S 1811 0.30230 1.000000000P 31 17.70000 0.430180000E-012 3.85400
0.2289130003 1.04600 0.508728000P 11 0.27530 1.000000000D 11 1.18500
1.000000000S 11 0.07896 1.000000000P 11 0.06856 1.000000000D 11 0.33200
1.000000000H 1.0 0.0000000000 0.7565140024 0.1397092302S 31 13.01000
0.196850000E-012 1.96200 0.1379770003 0.44460 0.478148000S 11 0.12200
1.000000000P 11 0.72700 1.000000000S 11 0.02974 1.000000000P 11 0.14100
1.000000000\$END\$VEC---orbitals from a converged RHF/cc-pVDZ run---\$END
>
>
--
Brian Salter-Duke (Brian Duke)   Brian.Salter-Duke^_^monash.edu