Solntsev Pasha
solntsev@univ.kiev.ua
Pavel
On Fri Jul 27 '12 2:19pm, Alex Granovsky wrote
----------------------------------------------
>Dear Pavel,
>The NREAD is simply the number of passes over a file containing
>two-electron integrals which will be performed to complete disk-based
>two-electron integral transformation. It is computed based on the
>available memory, the overall number of AOs and the number of MOs of
>different types (i.e. core, inactive, active, virtual).
>Firefly aborts if NREAD is predicted to be too large (i.e. larger than
>MXTRFR).
>You can reduce NREAD by allocating more memory. Alternatively,
>you can increase MXTRFR threshold:
>
$xmcqdpt MXTRFR=
>However, I strongly suggest you to use semidirect transformation with
>XMCQDPT code as described here.
>Kind regards,
>Alex
>
>
>
>On Thu Jul 26 '12 7:10am, Solntsev Pasha wrote
>----------------------------------------------
>>Dear Alex.
>>I used not latest version but anyway.
>>I am working on XMCQDPT job and before i played with different active space and didn't see any problem. Now i have SA-CASSCF(2,3) over first and second states. Total number of states is 3. So with CASSCF - no problem. With XMCQDPT i setup kstate(1)=1,1,1 and wstate(1)=1,1,0 avecoe(1)=1,1,0. Below you will find last part of my output file.
>>********** ROUTINE MQSYM **********************************************
>> ### RESULTS ###
>> # OF SYMMETRY ADAPTED CSFS = 6
>> CPU TIME: STEP = 0.00 , TOTAL = 1335.1 SECONDS ( 22.3 MIN)
>> WALL CLOCK TIME: STEP = 0.00 , TOTAL = 1338.6 SECONDS ( 22.3 MIN)
>> CPU UTILIZATION: STEP = 0.00%, TOTAL = 99.74%
>> ***** CPU TIME FOR MQSYM = 0.000 SEC.
>> ********** ROUTINE MQTRF **********************************************
>> CPU TIME FOR MQHTRN = 0.320 SEC.
>> *** ERROR STOP IN SUB.MQTRMC ***
>> NREAD IS GREATER THAN MXTRFR.
>> NREAD = 1921 MXTRFR = 1000
>> ADDRESS 0xF1C2C901 HAS INITIATED PROGRAM ABORT BECAUSE OF FATAL ERROR(S)
>>
>>
>>
>>In my previous jobs NREAD wasn't so high. It was 60. Is this a problem with my input file and we can avoid it by some settings or it is something more serious? I didn't put input/output file(s) but i am ready to provide them.
>>
>>
>>I also have a question not to you but to any familiar with CASSCF person.
>>We have polyene chain with phenyl rings from both sides. Problem: construct active space to describe most intense transition (first excited state). From CIS we can see that most intense transition will be 1st state (CIS reference) which consist of:
>> 124 131 0.12189782
>> 124 138 -0.06298572
>> 127 130 -0.13577728
>> 127 134 0.08853389
>> 128 131 0.21288424
>> 129 130 0.93620478
>>From TD-DFT(PBE96) also first state
>> 127 130 0.21359097
>> 130 127 0.02713894
>> 129 130 1.00828966
>> 130 129 -0.26609958
>>MO#129 is HOMO
>>Can we make any deassign about active space? I mean, what would be the best strategy to use NATURAL CIS ORBITALS or consider DFT orbitals that have better agreement with experiment than CIS. I already tried to put all pi orbitals from the chain into active space( 122,123,124,127,128,129) after reordering. In all cases i used 2 virtual orbitals. If we want to get only HOMO->LUMO transition do we need to worry about another transitions, since they also contribute to formation of the 1st excited state. After unsuccessful attempts i decided to check smallest active space and got small problem ( see above).
>>Thank you so much for your time and help.
>>Pavel.
>>
>>
>>
>>
>>
>>
>>
>>
>>