DFT(M062X)-MD energy conservation problem

[Бранислав Миловановић]

Dear cp2k users/developers,

I want to perform NVT dynamics of the compound from the picture in the attachment.

However, when I try to run M062X DFT-MD calculation under present setup (input files in the attachment) I get a constant shift of the Cons Qty during my run for NVT runs (graph from my last attempt is in the attachment - last 2ps).

I've tested the DFT (M062X) part of the input file with the different geometry optimization protocols to match Gaussian09 results. I found excelent aggrement between the geometries so I think that M062X setup should not be a problem.

Could CSVR thermostat be a problem? Or something else in the MD section?

Also, I used this kind of input file earlier and encountered no problems.

Now I'm using cp2k 9.1 version.

Can anybody help me with this?

Best,
Branislav

[Matt Watkins]

Not sure.

If it was me, I'd run NVE and see if that conserves energy. If that does then you can focus on the thermostat.

[Krack Matthias (PSI)]

Hi Branislav

 

You are setting ANGVEL_TOL and COMVEL_TOL explicitly to zero and you &CENTER_COORDINATES in each MD step. All these options impact the energy conservation. Try to remove/comment all these options for an unbiased NVT MD run.

 

Matthias

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/86e033d9-53f6-4933-ba95-6baaba8502dan%40googlegroups.com.

[Sam Broderick]

Hi Everyone

Is there a library out there to extract this kind of information out of cp2k output? Or does everyone have to roll their own regex?

Thanks

Sam

[Jürg Hutter]

Hi

some additional changes to consider:

Cutoff: GAPW can be used with smaller cutoff, but you still should increase REL_CUTOFF

XC_GRID: Again GAPW is easier to handle and you don't need the smoothing

Both of these changes should make the calculation more stable.

regards

Juerg Hutter

________________________________________

From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Бранислав Миловановић <dr.bra...@gmail.com>

Sent: Wednesday, June 15, 2022 8:19 PM
To: cp2k

Subject: [CP2K:17176] DFT(M062X)-MD energy conservation problem

--
You received this message because you are subscribed to the Google Groups "cp2k" group.

To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/86e033d9-53f6-4933-ba95-6baaba8502dan%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/86e033d9-53f6-4933-ba95-6baaba8502dan%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Бранислав Миловановић]

Dear all,

Thanks for the advices.

As Matthias suggested removing:

#    COMVEL_TOL 0
 #   ANGVEL_TOL 0  

did the trick and no there is no Cons Qty shift anymore.

I will try what Juerg suggested considering cutoff and xc_grid as well.

Thanks once again!

Kind regards,

Branislav