MEMCHK ERROR

[cymantren]

Hello everybody,

I came across a problem, when I wanted to do a CAS(4,4) calculation.
-----------------------------------------------------------------------------------------------
Node 0: --- SEVERE ERROR, PROGRAM WILL BE ABORTED ---
MEMCHK ERROR, not a valid memget id in work(kalloc-1)

Program received signal SIGILL: Illegal instruction.
------------------------------------------------------------------------------------------------

The Input and the whole failure-messages are in the appendix.

It seems to be a memory problem , is'n' it ?

Please give me some tips, how I can solve the problem.

[magnus]

Give Dalton more memory by using the "-mb" or "-gb" option which is memory in MB or GB, respectively. For example:

Code: Select all

dalton -gb 2 -mol ScB3_631gdp_optim.mol -dal ScB3_optvib_c2v_cas44.dal

[magnus]

I noticed that you're using a development version. This is not recommended for production calculations.

[cymantren]

Dear Magus,

thank you very much for you quick reply.

I just want to mention that I have to use this syntax in order to start DALTON: $DALTON/dalton ScB3_optvib_c2v_cas44 ScB3_631gdp_optim
$DALTON, means where it is writen in the ".bashrc"-file:
export DALTON=/home/kernel20/chemie/DALTON/DALTON18_clc/dalton
-----------------------------------------------------------------------------------------------------
and got this message I mentioned before, which you told me, that I have to increase the memory.


However :$DALTON/dalton -gb 2 -mol ScB3_631gdp_optim.mol -dal ScB3_optvib_c2v_cas44.dal
does not allow this and expects that the order is dal-file mol-file , without extensions !
---------------------------------------------------------------------------------------------------
usage: dalton [-h] [-f F] [-d] [-ext EXT] [-noarch] [-nobackup] [-get GET]
[-put PUT] [-N N] [-exe EXE]
inputs [inputs ...]
dalton: error: argument -d: ignored explicit argument 'al'
-----------------------------------------------------------------------------------------------------------

This is not really a probelm, but I also usually use " -t option" and to circumvent this memory problem, I have to use the "-m "option.

Maybe that's really -as you told me- because of using the developement -version and it seems to be better to does'nt use it.


So I will switch to the normal - version.


Thank you again !

Best greetings

Alfred

[magnus]

Actually that looks weird. It looks like the launch script is from a really old version of Dalton.

[taylor]

This has nothing to do with your problem, I am sure, but I just wanted to point out that unless you are running a test calculation, or trying to reproduce other work, your basis set choice (6-31G**) is not a good one. To say the least. This set is poor for first- and second-row atoms (it is not even of split-valence quality because of the constraints on s and p exponents) and is very poor indeed for transition metals. Sets of the same size include Jensen's polarization-consistent or Dunning's correlation-consistent sets: these are higher quality and definitely more reliable. It is also difficult to get reliable transition-metal results without polarization functions (that is, an f-set on the metal), at least for orbital-based treatments. Proponents of DFT methods have been telling me for decades that f-sets are not needed for calculations on transition-metal systems --- I have no comment on this because I have no experience...

Best regards
Pete

[hjaaj]

I have looked into your problem. The "MEMCHK ERROR" is caused by your input of .SINGLY occupied with zeroes. I removed these two lines, and the calculation runs ok for me. (An input possibility we had not imagined, I will fix it in Dalton so it will work in future versions.)

PS. From the output of my test run I can see you have made a poor choice of active orbitals. I advice you to take a look at the MP2 occupation numbers and use them to guide you in the selection of the CAS space.