Firefly and PC GAMESS-related discussion club


 
Learn how to ask questions correctly  
 
 
We are NATO-free zone
 



Re^5: density convergence

Slawomir Janicki
slawomir.janicki@comcast.net


David,

Actually, after a closer look at your convergence I would re-order the ideas to try:

- increase precision:

 $CONTRL INTTYP=HONDO ICUT=15 ITOL=35 LEXCUT=5 $END 

- modify DIIS parameters:
 $SCF MAXDII=5 DIITOL=(less than 100.0) $END 

- select different DIIS internals:
 $SCF DIIMOD=(0 to 5) DIIERR=(0 or 1) $END 

- try turning on the new FDIFF version:
 $SCF FDIFF=.T. XFDIFF=.T. XFDNR=(2 and up) $END 

Slawomir

On Tue Dec 31 '13 6:06pm, Slawomir Janicki wrote
------------------------------------------------
>David,

>Here are a few things you may try.

>- increase precision:

 $CONTRL INTTYP=HONDO ICUT=15 ITOL=35 LEXCUT=5 $END 

>- make sure that DIIS is on and starts right away:
 $SCF DIIS=.T. ETHRSH=1E4 SOSCF=.F. $END 

>- turn off extrapolation:
 $SCF EXTRAP=.F. $END 

>- turn on damping and shifting:
 $SCF DAMP=.T. FSHIFT=0.1 $END 

>- if all fail, restrict orbital interchanges:
 $SCF RSTRCT=.F. $END 

>The last option is risky as it may lead to an incorrect occupation of orbitals.
>You can also try to modify the DIIS parameters: MAXDII, FSTDII, DIIMOD, DIIERR, DIITOL.

>Finally, if you just run the same input file again you will find that in a multithreaded/parallel run the SCF iterations will be slightly different. Some may just converge.

>Slawomir
>
>
>
>On Tue Dec 31 '13 3:45pm, David G. wrote
>----------------------------------------
>>>
>>
>>I should have said that the B3LYP method was used for the calculation.

>>Sincerely, David


[ Previous ] [ Next ] [ Index ]           Tue Dec 31 '13 6:23pm
[ Reply ] [ Edit ] [ Delete ]           This message read 597 times