Firefly and PC GAMESS-related discussion club



Learn how to ask questions correctly


Re^3: Convergence problems: Almost converged but oscillating

Alex Granovsky
gran@classic.chem.msu.su


Hi,

there is no easy way how to use converged orbitals as
the starting guess for the calculations using another,
different basis set. However, this is nevertheless
possible in most cases. If you need more details
please let me know.

Regards,
Alex Granovsky

On Sat Jun 26 '10 12:06pm, Miro Moman wrote
-------------------------------------------
>Now I am trying to MOREAD the orbitals from the converged Def2-TZVP run to be used with a larger basis set (aug-cc-pVTZ).

>I receive the following error:

> *** ERROR READING THE MO COEFFICIENTS AT   1  77
>     THE INPUT VALUES WERE   2   1

>Indeed,  1 77 does not exist :

> 1 75-2.45175653E-05 4.67451306E-07 8.18963875E-06-2.25179089E-05 1.05807521E-05
> 1 76 1.49291949E-05
> 2  1 7.09245510E-05 9.40896541E-06 5.16541291E-05-1.12328743E-04 1.24681158E-04

>Do I need to pad with zeroes the $VEC section?
>
>
>On Sat Jun 26 '10 11:35am, Miro Moman wrote
>-------------------------------------------
>>Ok. It has readily achieved convergence by switching to SOSCF. SOSCF appears to be more robust than DIIS in this particular case.
>>
>>
>>On Sat Jun 26 '10 10:34am, Miro Moman wrote
>>-------------------------------------------
>>>Hello,

>>>I am trying to optimize the geometry of a [Fe4S4]+ cluster.

>>>In fact, the entire structure is [(Fe4S4)(C2H2)(SCH3)3]-2.

>>>Having had problems with the full complex, I am now focusing on [Fe4S4]+

>>>This is my current input file:

>>> $CONTRL SCFTYP=UHF MULT=2 MAXIT=200 dfttyp=pbe0 d5=1 icharg=1 $END
>>>!$CONTRL RUNTYP=OPTIMIZE $END
>>> $CONTRL INTTYP=HONDO $END
>>> $SYSTEM TIMLIM=99999 MWORDS=99 $END
>>> $SCF DIIS=.T. $END
>>> $statpt nstep=1000 $end
>>> $scf dirscf=.t. fdiff=.f. $end
>>> $guess guess=moread norb=376 $end
>>> $smp httfix=.f. smppar=.t. $end

>>>I have tried many different options in a more or less combinatorial way, such as : ICUT=11, ITOL=30, rstrct=.t., fdiff=.t., etc.

>>>I have tried with larger and smaller basis sets (currently Def2-TZVP) and also I have used HF instead of DFT. In addition, thinking that the initial geometry could be the culprit, I have tried to optimize it.

>>>Often it seems that the density has nearly converged, but it oscillates quite a lot, for instance:

>>>  90 89 -6646.252407632     0.000030038   0.012915252   0.000146340     1661471091   9105267
>>>  91 90 -6646.252436008    -0.000028377   0.004901678   0.000065611     1661477904   9105225
>>>  92 91 -6646.252446605    -0.000010597   0.001334835   0.000051306     1661481388   9105176
>>>  93 92 -6646.252449524    -0.000002919   0.002216061   0.000043654     1661486660   9105169
>>>  94 93 -6646.252454263    -0.000004739   0.000922602   0.000045303     1661496334   9105149
>>>  95 94 -6646.252456240    -0.000001977   0.002316391   0.000044423     1661501828   9105141
>>>  96 95 -6646.252461196    -0.000004956   0.001103882   0.000043711     1661505726   9105136
>>>  97 96 -6646.252463602    -0.000002406   0.000039273   0.000042713     1661508744   9105140
>>>  98 97 -6646.252463639    -0.000000037   0.000755261   0.000041815     1661508666   9105145
>>>  99 98 -6646.252462045     0.000001594   0.000914736   0.000042353     1661506603   9105146
>>> 100 99 -6646.252463837    -0.000001792   0.004938129   0.000042491     1661508480   9105138

>>>Any suggestions?


[ Previous ] [ Next ] [ Index ]           Mon Jun 28 '10 2:05pm
[ Reply ] [ Edit ] [ Delete ]           This message read 897 times