dft optimize failed


Clicked A Few Times
Hi there,

I have got "dft optimize failed" after 200 steps running the following input with NWChem 6.1 and 6.0. Please can anybody advise what is wrong with the input as below. This case is running on Cray XE6.


echo
start co
title "co"
print medium

geometry
 C                     0.02130486    -0.67185363     0.00000000
O -0.07773307 0.48364394 0.00000000
end

basis
 o   library "DZVP (DFT Orbital)"
h library "DZVP (DFT Orbital)"
c library "DZVP (DFT Orbital)"
end

driver
 tight
maxiter 200
xyz final
end

dft
 xc beckehandh
mulliken
end

task dft optimize



Thanks,

Yudong

Clicked A Few Times
removed

Forum Vet
Did you want to remove the whole post or are you still looking for feedback.

Bert


Quote:Ysun Jul 11th 2:40 pm
removed

Clicked A Few Times
Quote:Bert Jul 11th 8:10 am
Did you want to remove the whole post or are you still looking for feedback.

Bert


Quote:Ysun Jul 11th 2:40 pm
removed

No, I am still looking for feedback. My second post should be a new thread so I removed it from here.

Yudong

Forum Regular
Hi Yudong,

I have tried your input on my Redhat Linux box. I got the geometry optimization to converge in 5 steps to -112.58608241 Hartree without changing your input at all. I managed to get the same answer using both nwchem-6.0 and nwchem-6.1.1. So it seems that I cannot reproduce the problems you are seeing. Could you send me your output (hubertus dot vandam at pnnl dot gov), please? Maybe I can spot something.

Huub

Clicked A Few Times
Quote:Huub Jul 11th 2:21 pm
Hi Yudong,

I have tried your input on my Redhat Linux box. I got the geometry optimization to converge in 5 steps to -112.58608241 Hartree without changing your input at all. I managed to get the same answer using both nwchem-6.0 and nwchem-6.1.1. So it seems that I cannot reproduce the problems you are seeing. Could you send me your output (hubertus dot vandam at pnnl dot gov), please? Maybe I can spot something.

Huub

Hi Huub,

I emailed the output file to you a few days ago. I think you may have received it.

Thanks,

Yudong

Forum Vet
Huub is in Europe right now, he should get back to you early next week.

Bert


Quote:Ysun Jul 17th 3:29 pm
Quote:Huub Jul 11th 2:21 pm
Hi Yudong,

I have tried your input on my Redhat Linux box. I got the geometry optimization to converge in 5 steps to -112.58608241 Hartree without changing your input at all. I managed to get the same answer using both nwchem-6.0 and nwchem-6.1.1. So it seems that I cannot reproduce the problems you are seeing. Could you send me your output (hubertus dot vandam at pnnl dot gov), please? Maybe I can spot something.

Huub

Hi Huub,

I emailed the output file to you a few days ago. I think you may have received it.

Thanks,

Yudong

Clicked A Few Times
Hi Huub,

Have you spotted anything wrong in my output?

Cheers,

Yudong

Clicked A Few Times
Hi NWChem guys,

Please can you help me check if the input in the first post can work on your Cray XE6. It is not converging on our Cray XE6 with the same build of NWChem 6.1 as in my another post of "half int file".

Thanks,

Ysun

Forum Vet
Nothing wrong in the input file
Ysun
I didn't have any convergence problem running your input file with NWChem 6.1 on the ORNL Cray XE6.

I would suggest you to try to complete your NWChem build (that I am trying to assist in the thread "half int file"),
and then try this input file again with that NWChem binary.

Cheers, Edo

Clicked A Few Times
Edo,

The dft optimize failed error still exists using the newly-compiled NWChem 6.1.1. The job is run on 32 cores. The last lines of the output are:

               Final and change from initial internal coordinates
--------------------------------------------------



                               Z-matrix (autoz)
--------

Units are Angstrom for bonds and degrees for angles

Type Name I J K L M Value Change
----------- -------- ----- ----- ----- ----- ----- ---------- ----------
1 Stretch 1 2 1.15973 -0.00001

============================================================================== internuclear distances
------------------------------------------------------------------------------
center one | center two | atomic units | angstroms
------------------------------------------------------------------------------
2 O | 1 C | 2.19157 | 1.15973
------------------------------------------------------------------------------
number of included internuclear distances: 1
==============================================================================



------------------------------------------------------------------------
dft optimize failed 0
------------------------------------------------------------------------
------------------------------------------------------------------------
current input line :
28: task dft optimize
------------------------------------------------------------------------
------------------------------------------------------------------------
This type of error is most commonly associated with calculations not reaching
convergence criteria
------------------------------------------------------------------------
For more information see the NWChem manual at
http://nwchemgit.github.io/index.php/NWChem_Documentation


For further details see manual section: 



0:0:dft optimize failed:: 0
Application 2581875 exit codes: 134
Application 2581875 resources: utime ~42s, stime ~32s

Forum Vet
Does it fail at small core counts too?
Yudong,
Does the jobs fail when you run the job with a single core, too?

Cheers, Edo

Clicked A Few Times
Edo,

It has got the same "dft optimize failed" error when running on a single core.

Thanks,

Yudong

Forum Vet
Please add "clear" to the driver section
Yudong,
Since Hessian that were left from previous runs might be causing your problem,
please add the clear keyword to the driver section, e.g.

driver
clear
tight
maxiter 200
xyz final
end

Clicked A Few Times
Edo,

The clear keyword has done the job. The simulation works fine now with NWChem 6.1.1.

Thanks,

Yudong


Forum >> NWChem's corner >> General Topics