ERROR WHEN ADDING print debug


Clicked A Few Times
Hi there,

I am running NWChem in a very simple fashion. As I needed to get the Vxc matrix for my calculations I added "print debug" in the dft section of the input file.

It worked well for all the systems I needed so far (mostly atoms), but when I ran Neon it fails. I ran it without the "print debug" line and it finishes well. However, when I add the line the calculation Exits. I get the following lines at the end of the output file


0:0:nga_create_config:ga_set_data:wrong dimension specified:: 0
0:0:nga_create_config:ga_set_data:wrong dimension specified:: 0
Last System Error Message from Task 0:: Inappropriate ioctl for device
 0: ARMCI aborting 0 (0).
0: ARMCI aborting 0 (0).
system error message: Inappropriate ioctl for device


Your help on this will be greatly appreciated.

Forum Vet
Roycuevas,
Could you post the input file you are using?
Thanks, Edo

Clicked A Few Times
Quote:Edoapra Aug 23rd 5:56 pm
Roycuevas,
Could you post the input file you are using?
Thanks, Edo



Hi Edo,

Here is the input file



start

memory 700 MB noverify

echo


geometry units angstrom noautosym
Ne 0.0 0.0 0.0
end

basis "ao basis" spherical
* library cc-pvDz
end


dft
MULT 1
xc pbe
grid fine
odft
convergence energy 1d-6
iterations 150
print debug
end

task dft



The last line in the dft section (print debug) is that one that causes the problem. Without it the calculation is done but when adding it it exists. I have used this print debug for other calculations without a problem.

Thanks.

Roger

Forum Vet
Works fine for me in 6.1.1 (after changing the functional definition)
Roger,
I was not able to reproduce your failure in 6.1.1
However, I add to change the line "xc pbe" to "xc xpbe96 cpbe96" (since we do not have the pbe keyword).
One I have made this change, the input worked fine for me.

Cheers, Edo

Clicked A Few Times
Quote:Edoapra Aug 24th 9:14 am
Roger,
I was not able to reproduce your failure in 6.1.1
However, I add to change the line "xc pbe" to "xc xpbe96 cpbe96" (since we do not have the pbe keyword).
One I have made this change, the input worked fine for me.

Cheers, Edo



Hmmm...that is weird. It still does not work for me. Besides I have been using the pbe keyword with no problem whatsoever (with and without print debug).

In any event, I tried again, with debug print and xc b3lyp and still does not work.

Any other idea what may be going on?

Thanks Edo

Roger

Forum Vet
Roger,
Let's check what version of NWChem you are using.
What does the output say at the line that contains the text "nwchem branch" ?
Do you have access to the source code?
If you do, please send the output of the command

grep pbe $NWCHEM_TOP/src/nwdft/input_dft/xc_inp.F

Clicked A Few Times
Quote:Edoapra Aug 24th 9:48 am
Roger,
Let's check what version of NWChem you are using.
What does the output say at the line that contains the text "nwchem branch" ?
Do you have access to the source code?
If you do, please send the output of the command

grep pbe $NWCHEM_TOP/src/nwdft/input_dft/xc_inp.F


Edo,

I am running the 5.1 version.

Here is the output to the grep line:


c cpbe96 => 2100
c xpbe96
    &           'cpbe96','xpbe96','gill96','xperdew91',
, 'becke98','pbe0','becke97gga1','hcth407','hcthp14',
E 'op', 'bop', 'pbeop','rpbe','revpbe',
+ 'pbe', 'blyp', 'hartree'/
c cpbe96 Perdew-Burke-Ernzerhof functional for correlation
           write(LuOut,*)' cpbe96 is a nonlocal functional;',
c This is also done by default inside xc_pbe96.F
c xpbe96; Perdew-Burke-Ernzerhof functional for exchange
           write(LuOut,*)' xpbe96 "local" is equivalent to the',
c rpbe; Perdew-Burke-Ernzerhof + Norskov exchange
           write(LuOut,*)' rpbe "local" is equivalent to the',
c revpbe; Perdew-Burke-Ernzerhof + W Yang exchange
           write(LuOut,*)' rpbe "local" is equivalent to the',
c pbe
c xpbe96
c cpbe96
c Check nonlocal COR functionals: perdew86, perdew91, pbe96 to
           write(LuOut,*)' pbe96 is a nonlocal functional;',


Thanks

Roger

Forum Vet
Quote:Roycuevas Aug 24th 10:36 am

Edo,

I am running the 5.1 version.



Roger,
I strongly encourage you to use the 6.1.1 version.
It's pretty hard to go back and try to fix a possible bug in an older version of the code.

Edo


Forum >> NWChem's corner >> Running NWChem