Error in run_cioverlap_g09.pl

[Qunfei Zhou]

Hi all,

I am trying to use NX-1.4.0-2 to perform NAD with Gaussian09. when running moldyn.pl, I got the runnx.error as:

rm: cannot remove `link': No such file or directory

LA:ERROR - void LA::NRMat<C>::resize(int, int) [with T = double]: illegal dimensions

terminate called after throwing an instance of 'LA::LAerror'

sh: line 1: 25773 Aborted                 /home/qzh237/source/NX-1.4.0-2/bin/cioverlap-64/cis_casida -o -c < cis_casida.inp > cis_casida.out 2>> /scratch/qzh237/polymer/newtonx/dyn/TRAJECTORIES/TRAJ1/TEMP/../DEBUG/runnx.error

run_cioverlap: is dying now (cis_casida)

run-g09.pl: Error in run_cioverlap_g09.pl

And file run_cioverlap.out ended with:  Finished cis_casida -o -c   with ERROR

I am not sure what the LA error is. Can anyone help me? How to solve the cis_casida problem?

Thanks for your help,

Faye

[Rachel Crespo-Otero]

Hi Faye, 

Check whether you have the basis file in the JOB_AD/ folder, if not add it and send the calculation again.

Best Regards, 

Rachel 

[Qunfei Zhou]

Hi Rachel,

Thank you very much for your suggestion. I have the basis file in the JOB_AD/ folder. I checked and sent the calculation again. The errors are the same.

Best,

Faye

[Rachel Crespo-Otero]

Hi Faye, 

Could you send me your input files?

Best Regards, 

Rachel 

--
You received this message because you are subscribed to the Google Groups "Newton-X" group.
To unsubscribe from this group and stop receiving emails from it, send an email to newtonx+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Rachel Crespo Otero

[Zhou, Qunfei]

Hi, Rachel,

Here is my input file in the TRAJ1 directory. The last two files 'basis' and 'gaussian'(gaussian.com cannot be attached) are in JOB_AD/ folder.

In the NewtonX tutorial, I find that there should be the file 'intcfl' but I didn't have it after running nxinp. Could this be the problem? But I don't know why I don't have this file because I said yes to generate internal coordinate file.

Thanks for your help.

Faye

--
You received this message because you are subscribed to a topic in the Google Groups "Newton-X" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/newtonx/P_FubNfgt5E/unsubscribe.
To unsubscribe from this group and all its topics, send an email to newtonx+u...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Qunfei Zhou
Department of Materials Science and Engineering
College of Engineering
University of Kentucky
email: f...@uky.edu

[Rachel Crespo-Otero]

Hi Faye, 

Could you run some steps? Please check the gaussian.log in the TEMP/, does it finish normally? What about the

the sgaussian.log in TEMP/tmp.old/ directory? I can run some steps without problems (please check if you need to use 16 processors). 

Best Regards, 

Rachel    

[Qunfei Zhou]

Hi, Rachel,

The gaussian.log finished normally. there are no sgaussian.log in TEMP/tmp.old directory because the job stops when running cis_casida. I found that the grad file is void. The cis_casida.out file is 

core, occ, virt, freeze, discard -18 6 18 0 0

Using casida factors ? 1

Performing reorthonormalization ? 1

LA:ERROR - void LA::NRMat<C>::resize(int, int) [with T = double]: illegal dimensions

the cis_casida.inp file is attached. Is it the right format?

Thanks,

Faye

[Zhou, Qunfei]

This time I tried the calculation using the input in MD-SICH4-G09-TDDFT-NAD folder under text-nx directory.

Thanks,

Faye

[Felix Plasser]

For some reason the file believes that there are a negative number of core orbitals (-18). This seems to be the error. Did you change any settings in the "jiri.inp" file, by any chance?

-Felix

-- 
Felix Plasser
Institute for Theoretical Chemistry
University of Vienna
Währingerstr. 17/403
1090 Wien
felix....@univie.ac.at
http://homepage.univie.ac.at/felix.plasser/
Tel.: +43-1-4277-52761

[Rachel Crespo-Otero]

Hi Faye, 

Have you defined the path for gaussian in your submission script. Have you sourced $g09root/g09/bsd/g09.login? 

Take a look at the manual (page 17). It's likely that the internal gaussian executables are not found. If your 

TMP/tmp.old/MO_coefs is empty, that should be the reason of the problem.

If it doesn't work after checking these, please send me the whole directory. 

Best Regards, 

Rachel 

[Qunfei Zhou]

Thank you very much for your help. It works now. It is my mistake. Because NewtonX tries to read Gaussian outputs before the Gaussian job finishes and error occurs. So I run Gaussian job ahead using the gaussian.com in the JOB_AD file. But actually the input file has been changed.

Also for sgaussian.com I performed Gaussian09 separately after input files being generated and then copy the Gaussian output files to the directories running NewtonX. It finishes normally for the test job. But in this case, $g09root gaussian.com only run once but I see in moldyn the program has "initial run" "first run" and "second run". Are the input file for those runs different? 

How do you avoid the error produced when NewtonX tries to read Gaussian outputs before the Gaussian job finishes?

Sorry to bother you again and again. I am really appreciate your help.

Thanks,

Faye

[Qunfei Zhou]

I am sorry. It all works fine now. 

Thank you very much. I really appreciate your help.

Best Regards,

Faye