I was working on a porphyrin structure with B3LYP/6-31+G* level of theory. At this stage the structure has already been optimized, and then I move on a frequency calculation. However, I didn't expect that such a frequency calculation would take such a surprisingly long time that it was terminated due to the time limitation of the cluster. Now I would like to restart the frequency calculation, but the task is killed upon submission. So could you please help me with it?
The input file starts as follows:
%nprocshared=16
%mem=32GB
%rwf=CoP4bs_fre_m.rwf
%NoSave
%chk=CoP4bs_fre_m.chk
# freq=restart B3LYP/6-31+g* int=grid=ultrafine
scf=xqc scrf=(cpcm,solvent=water)
While the log file ends as follows:
******************************************
%nprocshared=16
Will use up to 16 processors via shared memory.
%mem=32GB
%rwf=CoP4bs_fre_m.rwf
%NoSave
%chk=CoP4bs_fre_m.chk
----------------------------------------------------------------------
# freq=restart B3LYP/6-31+g* int=grid=ultrafine scf=xqc scrf=(cpcm,sol
vent=water)
----------------------------------------------------------------------
Use # Restart to restart analytic frequencies.
Error termination via Lnk1e in /software/gaussian-09RevA.02-el6-x86_64/g09/l1.exe at Mon Sep 7 12:19:01 2015.
I will appreciate your help with it, thanks!
Dear Xin,
Keep the Link-0 commands (i.e. the lines starting with %) but use the route specification
# Restart
(no other keywords) instead of # Freq=Restart B3LYP/... .
Best regards,
Jürgen
Dear Xin,
Keep the Link-0 commands (i.e. the lines starting with %) but use the route specification
# Restart
(no other keywords) instead of # Freq=Restart B3LYP/... .
Best regards,
Jürgen
Dear Xin,
For frequency calculation, keep your keyword same but before basis set just write "Freq=Raman"
Hi,
Take optimized checkpoint file and give "Freq" keyword in the route section.
Thanks for your suggestions.
Hi Jürgen, I have tried your suggestion but it prompted that No route information found on RWF. I am not sure where could possibly go wrong with the RWF.
******************************************
%nprocshared=16
Will use up to 16 processors via shared memory.
%mem=32GB
%chk=CoP4bs_fre_m.chk
---------
# Restart
---------
No route information found on RWF.
Error termination via Lnk1e in /software/gaussian-09RevA.02-el6-x86_64/g09/l1.exe at Thu Sep 10 11:06:10 2015.
******************************************
I have also tried to delete the "%rwf=CoP4s_fre.rwf %NoSave" lines and restart from the chk files, but it won't work either.
At this stage, I will try to run this freq calculation on another cluster without time limitation. And meanwhile, I will also try other method to restart the calculation on the original cluster. I will let you know if it works.
I appreciate your kindly help!
So there are in general three ways to do it:
1. "# Restart" (as Jurgen suggested). The problem here is that in order to restart freq calculations you need the rwf file, which was deleted due to the use of %nosave.
2. "SAME INPUT AS IN THE ORIGINAL FILE +
# Geom=checkpoint Guess=read"
This will actually restart the optimization from the last geometry from the checkpoint file, but since it was already optimized, it will finish promptly and then start the frequency calcualtions from the beginning.
3. You can also download the output/log file, open it in any viewer, save the last geometry and then run gaussian again without any OPT, but just do single-point energy + frequency calculations. This will also start the frequency calcualtions from the beginning.
To summarize: you can restart (analytic) frequency calcualtions only if you saved the .rwf file, otherwise you have to start it from the beginning.
Thank you Dr. Trzaskowski,
From Gaussian website, "%NoSave Causes Link 0 to delete scratch files at the end of a run". And in my case, the job is automatically killed by the cluster because of the time limitation. And since the job is not "at the end of a run", I have a rwf file up to 34 GB. However, # Restart cannot find route information in the rwf file for some reason.
Also, because of the time limitation (36 hours with 16 cores) and my relatively large porphyrin system, redo the freq calculation from the very beginning will not help.
So at last, I decided that I dislike this cluster, and turned to another slower cluster with 8-core/node for a single run but without time limitation. And currently, the job is going on and on...
Still, I appreciate your suggestions. And I believe these suggestions will help other guys who want to restart their calculation in their cases.
G09 can do the restart an analytic freq work if it can find out the rwf file.
Maybe you need rename the rwf file to, for example, restart.rwf
Then in the input file add the line %rwf=restart.rwf
Use a small molecule to test.
Best wishes,
Ding
Open the output (.log) with ChemCraft, save (extract) the optimized geometry structure, open with Avogadro and output for freq calc. as Gaussian (menu Extensions/Gaussian), the do the calc.
Use this if helps,
*************************************Method-A******************************
%nprocshared=16
%mem=32GB
%rwf=CoP4bs_fre_m.rwf
%chk=CoP4bs_fre_m.chk
# freq=(restart) B3LYP/6-31+g* int=(grid=ultrafine) scf=(xqc) scrf=(cpcm,solvent=water) geom=allcheck guess=read
Title
Charge Multiplicity
*************************************Method-B******************************
%nprocshared=16
%mem=32GB
%rwf=CoP4bs_fre_m.rwf
%chk=CoP4bs_fre_m.chk
# Restart
*************************************END***********************************
Hope This Helps,
Regards
Ajay Khanna
- Badges
- Science topic