Firefly and PC GAMESS-related discussion club


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



Re^4: "Fatal error in MEMCI" while starting (16,16) or larger AS in CASSCF

Alex Granovsky
gran@classic.chem.msu.su


Dear Roman,
Dear Thom,

Indeed, this is out of memory error.
The required memory is computed in routine MEMCI which
generates this error message ("Fatal error in MEMCI")
and then aborts.

A (16e,16o) or similar CASSCF is currently a program's limit.
A sample (16,16) CASSCF input and output can be found here:

http://classic.chem.msu.su/gran/firefly/cnc16.tar.bz2

The program was improved since then so it is now faster.

Kind regards,
Alex

On Wed Mar 27 '13 3:08pm, Thomas wrote
--------------------------------------
>Dear Roman,

>The error you're receiving is most likely due to not enough memory being available. I'm not sure why Firefly gives an MEMCI fatal error instead of an out-of-memory error, but I do know that an active spaces of the size (24,20) is too memory demanding and therefore not possible.

>With Firefly 8.0.0, you can reduce the memory demands of the ALDET diagonalization procedure by specifying a non-zero value of DISTCI in $DET (when EXETYP=CHECK, the value of DISTCI should equal the number of FF threads). This activates a different diagonalization code that has a decreasing memory demand with an increasing amount of threads. Still, also with this code, anything larger than, say, (16,16) might not be possible.

>I would suggest to reduce the size of your active space. (16,14) should indeed be fine, provided you have the required computer power. Alternatively, you could choose an 'incomplete' active space approach by using the GUGA program and limiting the excitation number. For example:

>

 $MCSCF CISTEP=GUGA MAXIT=400 FORS=.F. $END
 $DRT GROUP=C1 NMCC=48 NDOC=12 NVAL=8 IEXCIT=2 $END
 $GUGDIA NSTATE=5 $END
 $GUGDM2 WSTATE(1)=1,1,1 $END 

>This specifies a (24,20) active space but limits the excitation number to 2. Such an active space can however not be used for a subsequent (X)MCQDPT2 calculation (as these always assume a complete active space). Also, it might not be able to describe the chemical problem you'd like to study.

>Hope this helps.
>
>
>Kind regards,
>Thom
>
>
>
>
>
>On Tue Mar 26 '13 7:26pm, Roman Kroik wrote
>-------------------------------------------
>>>Dear Roman,

>>>Could you post some more details on your calculation, preferably the input and output file? Without any details it is hard to say what is going wrong in your calculation.

>>Ok. I'm attaching input and output files for CASSCF(24,20) calculation. If I use CASSCF of (16,14) and less dimension, calculation proceeds without this error. Maybe, files will help in solving a problem, but I think there is a problem of some general nature.
>>
>>
>>


[ Previous ] [ Next ] [ Index ]           Thu Mar 28 '13 2:28am
[ Reply ] [ Edit ] [ Delete ]           This message read 1142 times