Firefly and PC GAMESS-related discussion club

 
Learn how to ask questions correctly  
 
 

 

Panwang Zhou
pwzhou@gmail.com

Thanks for you reply.

First I reduce the MXBASE from the default 300 to 90, the "ROUTINE MQCACI" passed, however, it exited again at "ROUTINE MQLMB2" with the following information:

********** ROUTINE MQLMB2 (   1) ***************************************
***** ERROR: MEMORY REQUEST EXCEEDS MAXIMUM.
  494278798 WORDS REQUIRED    400000000 AVAILABLE.

It is still the memory problem and I can't set to MWORDS=500 to satisfy the required memory as the fifefly cannot allocate so much memory.

Then I try to redude the MXBASE to 70, and the required memory by "ROUTINE MQLMB2" is still 494278798 WORDS. Thus, the reduce of MXBASE could not reduce the memory request of MQLMB2. So, is there an option to reduce the memory request of MQLMB2?

For the EXETYP=CHECK, the job stops at "THE DETERMINANT FULL CI"

THE NUMBER OF DETERMINANTS HAVING SPACE SYMMETRY A  
IN POINT GROUP C1   WITH SZ=  0.0 IS        9018009
WHICH INCLUDES        2147145 CSFS WITH S=  0.0
WHICH INCLUDES        3864861 CSFS WITH S=  1.0
WHICH INCLUDES        2277275 CSFS WITH S=  2.0
WHICH INCLUDES         637637 CSFS WITH S=  3.0
WHICH INCLUDES          85995 CSFS WITH S=  4.0
WHICH INCLUDES           5005 CSFS WITH S=  5.0
WHICH INCLUDES             91 CSFS WITH S=  6.0
THE DETERMINANT FULL CI REQUIRES   514284263 WORDS

Fatal error, routine DETFCI needs at least 114284563 more words of memory!

This is not reasonable as the actual memory request of "THE DETERMINANT FULL CI" is "109225382 WORDS" (From the output of exetyp=run). Due to this error, I could not use exetyp=check to find a suitable value to fit into memory.

In addition, I'm intrested in "CSFs selection" and I'm not clear on how to do this, could you comment on this in detail? Thanks.



On Mon Jan 7 '13 4:41pm, Alex Granovsky wrote
---------------------------------------------
>Dear Panwang,

>normally, XMCQDPT code calls MQCACI two or three times depending
>on input options. The use of

>

 $MCSCF IFORB=.T. $END
 $XMCQDPT IFORB(1)=-1,1,1 $END

>allows one to skip the first call only.

>The best way to reduce the number of words required by MQCACI
>is to decrease the value of MXBASE variable.
>It defines the number of subspace vectors used in Davidson
>diagonalization. The default is rather large and is printed
>in output. Use exetyp=check to find a suitable value to fit
>into memory (say ca. 20 or so). You may also need to use CSFs
>selection (please drop a post here if you have questions on this).

>Note, aug-mcqdpt2 is not the same as xmcqdpt2. As I mentioned in my paper on XMCQDPT:

>

The systematic overestimation of off-diagonals within MCQDPT2 has been 
documented previously (2005-2006). The development of XMCQDPT was 
completed in 2008, and initially led to development of the so-called 
augmented effective Hamiltonian approach to MCQDPT2 (aug-MCQDPT2, or, 
more precisely, aug-MCQDPT2’) by Bochenkova. This procedure can be 
considered as the attempt to apply a variation of the stabilization 
graph technique to the MCQDPT2’ effective Hamiltonians of sequentially 
enlarged dimensions aimed to minimize effects caused by overestimated 
off-diagonal elements of the effective Hamiltonian.

>Kind regards,
>Alex Granovsky
>
>
>
>
>On Sun Jan 6 '13 10:41am, Panwang Zhou wrote
>--------------------------------------------
>>Dear all,

>>When performing the XMCQDPT2 calculation I encountered the following problem：
>>the job exited at the "ROUTINE MQCACI" and the error information is related to the memory usage:

>> ********** ROUTINE MQCACI **********************************************
>> ***** ERROR: MEMORY REQUEST EXCEEDS MAXIMUM.
>>  1298573444 WORDS REQUIRED    300000000 AVAILABLE.

>>Thus, it seems that there are not enough memory to carry this job. However, in my input file I have set

>>$MCSCF IFORB=.T. $END
>>$XMCQDPT IFORB(1)=-1,1,1 $END

>>as described in:
>>http://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C35e9ea902bHW-8295-1009+00.htmhttp://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C35e9ea902bHW-8295-1009+00.htm

>>these settings will avoid the first integral transformation and CASCI procedure (performed by the routine MQCACI) by overlapping part of calculations with those performed at the MCSCF stage and reusing them at XMCQDPT2 stage.

>>So, why this error occurred and is there a way to resolve this problem? Thanks.

>>The system I studied is the same as ref:
>>http://link.springer.com/article/10.1134%2FS1990793108050023?LI=truehttp://link.springer.com/article/10.1134%2FS1990793108050023?LI=true
>>The NACT and NELS I used is also same to the above ref, so this job should be carried out by firefly.

>>The part of my input file:
>> $CONTRL SCFTYP=MCSCF RUNTYP=ENERGY EXETYP=RUN MAXIT=100 ICHARG=-1
>>    MULT=1 FSTINT=.T. GENCON=.T. ITOL=30 ICUT=11 INTTYP=HONDO
>>    NOSYM=1 WIDE=.T. MPLEVL=2 $END
>> $SYSTEM MWORDS=300 KDIAG=0 NOJAC=100 TIMLIM=60000.0 $END
>> $SYSTEM MKLNP=1 NP=12 $END
>> $P2P P2P=.T. DLB=.T. $END
>> $SMP SMPPAR=.T. $END
>> $SCF DIRSCF=.T. FDIFF=.F. NCONV=8 $END
>> $TRANS MPTRAN=2 DIRTRF=.T. AOINTS=DIST ALTPAR=.T. MODE=112 $END
>> $MCSCF CISTEP=ALDET FULLNR=.F. SOSCF=.T. MAXIT=150 $END
>> $MCSCF IFORB=.T. $END
>> $DET NCORE=49 NACT=14 NELS=16 NSTATE=6 WSTATE(1)=1,1 $END
>> $XMCQDPT NSTATE=2 EDSHFT=0.02 THRGEN=1D-12 $END
>> $XMCQDPT IFORB(1)=-1,1,1 WSTATE(1)=1,1,-0 AVECOE(1)=1,1,-0 $END
>> $BASIS GBASIS=N31 NGAUSS=6 NDFUNC=1 NPFUNC=1 DIFFSP=.TRUE. $END
>> $GUESS GUESS=MOREAD NORB=359 $END
>> $MCQFIT $END