I get no gradient file

[Lara]

Hi

I want a structure optimization of my molecule. I transform my cartesian coordinates with x2t in a file called coord. After that I use define. I choose ired, eht, dft (b3-lyp) and TZVP basis for all atoms.

After finishing define I just gave the command jobex and I got no file called gradient.

What should I do?

Do you now some other tutorials than the official tutorial? This tutorial and the manual sucks extremly :-(

Thanks for help
Lara

[antti_karttunen]

Hi,

If your geometry optimization failed, you should get a file called GEO_OPT_FAILED. You could start troubleshooting by checking the contents of this file.

Best,
Antti

[Lara]

Hi

first of all: Thanks for the response.

I explain my error in more detail:

After my define procedure I just gave the command jobex and I got this message:

convgrep will be taken out of the TURBODIR directory

This don't change so I canceled it with strg+c

Than I have the message

Code: [Select]

forrtl: error (69): process interrupted (SIGINT)
Image              PC                Routine            Line        Source            
dscf               00000000004E69B2  klshloop_                 237  klshloop.f
dscf               00000000004DB663  shloop_                   652  shloop.f
dscf               0000000000428219  scf_                      926  scf.f
dscf               00000000004130FE  MAIN__                   1577  dscf.f
dscf               000000000040A2BC  Unknown               Unknown  Unknown
libc.so.6          00002B23BFE35C4D  Unknown               Unknown  Unknown
dscf               000000000040A1EA  Unknown               Unknown  Unknown
I have not the file GEO_OPT_FAILED but I have GEO_OPT_RUNNING and in it is

Code: [Select]


 CAUTION: this file will not be removed if user or the system decides
          to kill the job !!!
          To let jobex stop smoothly, simple create
          an empty file called \'stop\' in this directory
          jobex will stop at the next possible step

job.start file has been created with details about the job type:

jobex WAS CALLED AS: jobex
AN OPTIMIZATION WITH MAX. 20 CYCLES WILL BE PERFORMED
CONVERGENCY CRITERION FOR TOTAL SCF-ENERGY IS 10**(-6)
CONVERGENCY CRITERION FOR MAXIMUM NORM OF SCF-ENERGY GRADIENT IS 10**(-3)
CONVERGENCY CRITERION FOR MAXIMUM NORM OF BASIS SET GRADIENT IS 10**(-3)
PROCEDURE WILL START WITH A dscf like STEP
LOAD MODULES WILL BE TAKEN FROM DIRECTORY /opt/TURBOMOLE/bin/em64t-unknown-linux-gnu
DSCF  = /opt/TURBOMOLE/bin/em64t-unknown-linux-gnu/dscf
GRAD  = /opt/TURBOMOLE/bin/em64t-unknown-linux-gnu/grad
STATPT = /opt/TURBOMOLE/bin/em64t-unknown-linux-gnu/statpt
What should I do?

Greetings

[amavr78]

Hi Lara,
Obviously, you kill your job before even finishing the first step. That's why you do not get anything but errors. Just wait a bit, till the first energy step is completed.
If you want to run your calculation interactively, just run them in the background, by e.g:  nohup jobex > log.out &
and check if new files are created, such as: job.last or job.1

I will not agree with you about the tutorial and the manual. They are really helpful and explanatory. If you follow the examples in the tutorial, you will be able to complete successfully your calculations.

all the best,
Andreas

[Lara]

Hi

thanks for answering.

Now I used nohup jobex > log.out &

Could you explain what nohup means?

It is still working. This leads me to the following question:

Is there a way to see how long TURBOMOLE would need for a calculation. Just approximately?

greetings
Lara

p.s.
The tutorial and manula is hard for me, because it is to short in explaining things. For me especially it is hard because I am new to the topic of quantum chemistry. It would be great, if the tutorial or manual also explains in very short way the quantum chemistry behind every command. If you say that this is not possible, than have a look at the great manual of the MD software gromacs. Sadly it is only classic and not ab initio.

[amavr78]

Hi,

Obviously, you do not submit your calculations on a queue , but you run them locally on your computer.
nohup , will prevent your calculation from stopping , if you exit the terminal window, for example.
Take a look at the new files that have been created,e.g in job.last .  you can see how many scf iterations have been performed, and how close convergence is.
You can also view the scf steps in graphical window (if you have xmgrace installed) by typing : cgnce job.last

It will be much better, if you run some examples first as shown in tutorial.

If your molecule is too big, then of course the calculation will take more time.
1) as a start, you can use a smaller basis set , e.g : def-SV(P)
2) take advantage of the RI approximation. (RIDFT is 10 times faster than normal DFT)
3) run in parallel.

Theory is explained  in the manual. But if you need deep understanding, you should read the relevant papers or books. I would say that the Turbomole manual is very well written compared with manuals from other QM packages.
Anyway, it is a matter of taste.

good luck,

Andreas

[Lara]

Thanks for hints. Sadly I have TURBOMOLE not installed on my computer. Our admin says that it needs to much space and it is only installed on a kind of  research group computer for all people in our group and this computer has no graphical windows.

greetings