Ceilidh ... Re^3: The first SCF step is a disaster

Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

We are NATO-free zone

Re^3: The first SCF step is a disaster

Dear David,
Dear Pedro,

The problem with this input is as follows:
the extra option expects that new basis functions are added at
the end of the old basis set of a selected atom. Unfortunately this is
not necessary the case. Indeed, for

 $BASIS  GBASIS=DZV NPFUNC=1 NDFUNC=1 DIFFSP=1 DIFFS=1  $END

basis on the carbon atoms is:

 C_3       

   8   S    1    4233.000000    0.456306 (  0.001220) 
   8   S    2     634.900000    0.842129 (  0.009342) 
   8   S    3     146.100000    1.361290 (  0.045452) 
   8   S    4      42.500000    1.834726 (  0.154657) 
   8   S    5      14.190000    1.869946 (  0.358866) 
   8   S    6       5.148000    1.068413 (  0.438632) 
   8   S    7       1.967000    0.172732 (  0.145918) 

   9   S    8       5.148000   -0.410106 ( -0.168367) 
   9   S    9       0.496200    0.446679 (  1.060091) 

  10   S   10       0.153300    0.174609 (  1.000000) 

  11   P   11      18.160000    0.990652 (  0.018539) 
  11   P   12       3.986000    0.926729 (  0.115436) 
  11   P   13       1.143000    0.650574 (  0.386188) 
  11   P   14       0.359400    0.253904 (  0.640114) 

  12   P   15       0.114600    0.095043 (  1.000000) 

  13   L   16       0.043800    0.068236 (  1.000000)     0.028562 (  1.000000) 

  14   D   17       0.750000    0.994871 (  1.000000)

i.e. the polarization D shell follows the diffuse L shell and
is the last in the list.

The workaround is to create a custom basis set. Attached please find
the modified input file. I have also changed SCF converger to DIIS
and tightened SCF convergence.

Set MKLNP and NP in $system group according to the number of physical
cores on your system and run this job sequentially.

You need to run this file as follows:

Firefly -ib hex3yne_2255tetraMe_mp4_gmoread2.inp -rf -p -stdext -o test.out

Hope this helps.

Kind regards,
Alex Granovsky

On Fri Mar 11 '16 1:19am, David Shobe wrote
-------------------------------------------
>I don't have the original input file that generated the orbitals, but it was something like this.

>--David
>
>
>On Thu Mar 10 '16 11:16pm, Alex Granovsky wrote
>-----------------------------------------------
>>Dear David,

>>could you please send me the complete input file (with orbitals),
>>as well as the input file that generates your input orbitals?

>>Kind regards,
>>Alex Granovsky
>>
>>
>>On Thu Mar 10 '16 2:21am, David Shobe wrote
>>-------------------------------------------
>>>I am having SCF convergence difficulty (and there aren't even transition metals in my molecule!). I am new at Firefly, and this is my first attempt to use GUESS=MOREAD.

>>>I copy the first 10 SCF iterations of my output below. The results (iteration 3 and beyond) vary according to which parameters I have selected in $SCF, but iterations 1 and 2 are always the same no matter what I put in $SCF! I assume iteration 1 is the wavefunction I read in from GUESS=MOREAD, but why does iteration 2 make that first, disastrous SCF step?

>>> DIRECT SCF CALCULATION, SCHWRZ=T FDIFF=T
>>> SCHWARZ INEQUALITY OVERHEAD: 41019 INTEGRALS, CPU TIME= 0.03

>>> NONZERO BLOCKS
>>> ITER EX DEM TOTAL ENERGY E CHANGE DENSITY CHANGE ORB. GRAD INTEGRALS SKIPPED
>>> 1 0 0 -388.903748740 -388.903748740************** 0.000000000 72600037 81761217
>>> 2 1 0 -276.111387353 112.792361387************** 0.869472191 80744208 81131630
>>> 3 2 0 -338.212891275 -62.101503922************** 1.285866356 80778736 81129727
>>> 4 3 0************************************************************ 77667278 81352497
>>> 5 4 0************************************************************ 81041056 81123783
>>> 6 0 0 64.831233214****************************** 1.965860844 81302263 81102913
>>> 7 1 0 104.155503738 39.324270524************** 0.895286223 80819938 81128165
>>> 8 2 0-21382.575588159-21486.731091897**************************** 80944456 81124231
>>> 9 3 0 -90.114643710 21292.460944449************** 0.732336451 80833808 81123917
>>> 10 4 0 -185.881527362 -95.766883652************** 3.334914872 80871551 81124031

>>>If it helps, the input file starts as below. Again, I have tried several options in $SCF, no change in iteration 1 and 2. The $VEC is taken from a punch file of a HF/DZV(d,p) job at the same geometry.

>>>!
>>> $CONTRL
>>> SCFTYP=RHF INTTYP=HONDO ICUT=11 ITOL=30
>>> RUNTYP=ENERGY MPLEVL=4 UNITS=ANGS
>>> $END
>>> $SYSTEM TIMLIM=3600 MEMORY=36000000 $END
>>> $SCF DIRSCF=.True. $END
>>> $GUESS GUESS=MOREAD NORB=240 EXTRA=.T. $END
>>> $EXTRAF
>>> NEXTRA(1)=4,4,4,4,4,4,4,4,4,4 !DIFFSP for 10 C
>>> NEXTRA(11)=1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 !DIFFS for 18 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-06
>>> 1 3 7.12360225E-06-1.27864732E-04 0.00000000E+00 0.00000000E+00 0.00000000E+00

>>>

This message contains the 925 kb attachment
[ hex3yne_2255tetraMe_mp4_gmoread2.inp ]

Mon Mar 14 '16 2:23pm

This message read 625 times