run_cioverlap_turbo.pl error (NX v.1.4, TURBOMOLE v.7.1)

[joanna.j...@student.uw.edu.pl]

Dear All,

I've just set up NewtonX 1.4 and, following the tutorial steps for the TD-DFT calculations with Turbomole (v.7.1), I encountered an error at the stage of NAMD calculation.

After initial successful termination of the dscf and egrad steps, I'm getting an error from 'run_cioverlap_turbo.pl' regarding the overlap integrals. 

The exact message from the runnx.error file reads:

>> dscf ended normally

>> Could not read overlap integrals!

>> Check TEMP/tmp.old/overlap/dscf.out

>> run-turbo.pl: Error in run_cioverlap_turbo.pl

I attach a compressed directory with the problematic trajectory calculation.

I already applied two related fixes found on the forum: Mario' s suggestion from topic "Interface to Turbomole 7.2" (modification of the run_cioverlap_turbo.pl file), and tip from topic "run_cioverlap_turbo.pl error" by Felix (I added a line 'nbf(AO)=92' to the control file in JOB_AD). Maybe it is worth to mention, that the latter modification was essential to get the calculation at least starting.

However, my problem remains :( Has anybody had such problem before? 

Thanks,

Joanna

[Felix Plasser]

Dear Joanna,

I am copying Mario's most recent fix below. Did you do this already: change cao to sao?

Cheers

-Felix

---------------

Dear All,

It was brought to our attention that Newton-X (all versions up to 2.0) is not calculating the nonadiabatic couplings when interfaced to Turbomole 7.2. 

When running dynamics with either TDDFT or ADC(2), run_cioverlap_turbo.pl ends in error during in step 1
and DEBUG/runnx.erorr shows the message:

Could not read overlap integrals!

 Check TEMP/tmp.old/overlap/dscf.out

run-turbo.pl: Error in run_cioverlap_turbo.pl

The problem can be easily solved by a quick change in run_cioverlap_turbo.pl in your $NX folder:

1. Open $NX/run_cioverlap_turbo.pl with a text editor
2. Search for "$intsdebug    cao"

3. Change this sentence to ""$intsdebug    sao"

4. Save $NX/run_cioverlap_turbo.pl

That's all. Note that this change will not affect the behavior of Newton-X with other versions of Turbomole.
It will still work normally.

Let me know if you need more information about this issue.

Best regards,

Mario

--
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.

[Joanna Jankowska]

Dear Felix,

Thank you very much for the reply..
Yes, I did this already, just in case turbo7.1 had the same problems
as turbo7.2.
This has to be something else.

Best,
Joanna

[Felix Plasser]

Hi Joanna, I just tried this out with NX-2.2-B05 and TM 7.21 . It works after I change in run_cioverlap_turbo.pl "cao" to "sao"

531   @control=change_control("intsdebug","  sao",@control);

Is it possible for you to upgrade NX and TM to these version numbers?

-Felix

---

From: new...@googlegroups.com <new...@googlegroups.com> on behalf of Joanna Jankowska <joanna.j...@student.uw.edu.pl>
Sent: 23 May 2018 11:35:29
To: new...@googlegroups.com
Subject: Re: [Newton-X] run_cioverlap_turbo.pl error (NX v.1.4, TURBOMOLE v.7.1)

 

[Joanna Jankowska]

Hi Felix,

Unfortunately, I don't have access to TM 7.21. Yesterday I tried
my current NX-1.4 with another, slightly older TM version (7.0.1)
and got the very same error with the negative root in the Cholesky
decomposition.

Joanna

[Christian...@mail.huji.ac.il]

Hi Joanna,

in my case, the following versions of the programs have worked:

Newton-X version 2.0 (build 18, 26/September/2017)

TURBOMOLE V7.0.1( 19978 ) 29 Sep 2015

As far as I remember, this was working for me without any updates. So using a more recent Newton-X version might be worth a try.

Best regards,

Christian

[Felix Plasser]

Hi, the negative root in the Cholesky decomposition is ok, that's
inevitable. The only question is whether or not it prints the overlap
matrix before it crashes.

You can try going to

TEMP/tmp.old/overlap
And then you can try the two $intsdebug settings in control. And just run dscf in that folder. Is there any way you can make it print the overlap matrix before it crashes?

-Felix

[Joanna Jankowska]

Hi Felix,

Thank you for the tip. Actually, it does print something which looks
like overlap; please have a look on the attached files.

Joanna

[Joanna Jankowska]

Hi Christian,

Thank you very much. I'll check this out.

Joanna