Ask for BFD pseudopotential for Cu and pseudopotential conversion question

[yaoyi92]

Hi all,

I am looking for the BFD pseudopotential with upf format for Cu.

I checked the qmcpack package, where the Cu BFD pseudopotential is included with xml format but not the upf format. It will be great helpful is someone could provide the upf format pseudopotential for Cu. 

I also tried to convert the pseudopotential by ppconvert with such command I list here. However, it seems to take a very long time for the conversion (the conversion didn't finish in more than 1 hour). Is it normal for the conversion to take so long time? If so, do you have a sense how long it will take?

===

ppconvert --gamess_pot Cu.BFD.gamess --s_ref "1s(2)2p(6)3d(10)2s(1)" --p_ref "1s(2)2p(6)3d(10)2s(1)" --d_ref "1s(2)2p(6)3d(10)2s(1)" --xml Cu.BFD.xml

ppconvert --gamess_pot Cu.BFD.gamess --s_ref "1s(2)2p(6)3d(10)2s(1)" --p_ref "1s(2)2p(6)3d(10)2s(1)" --d_ref "1s(2)2p(6)3d(10)2s(1)" --log_grid --upf Cu.BFD.upf

===

Any suggestion and comment are welcome.

Thanks,

Yi

[Paul R. C. Kent]

ppconvert should run quickly - it is only recalculating the projectors, so should be similarly fast as an atomic pseudopotential code. Does is work for any other atoms?

You could also use the Cu potential from 

Pseudopotentials for quantum Monte Carlo studies of transition metal oxides, Jaron T. Krogel, Juan A. Santana, and Fernando A. Reboredo, Phys. Rev. B 93, 075143 (2016)

http://dx.doi.org/10.1103/PhysRevB.93.075143

The Supplemental Material with this paper includes Cu in UPF and QMCPACK xml formats.

We will add these potentials inside the QMCPACK distribution in the near future.

 

-- Paul

[yaoyi92]

On Tuesday, June 21, 2016 at 3:35:50 PM UTC-4, Paul R. C. Kent wrote:

ppconvert should run quickly - it is only recalculating the projectors, so should be similarly fast as an atomic pseudopotential code. Does is work for any other atoms?

The ppconvert works for O and H in my case very fast. But when I tried Cu, it took me more than a week without giving me a result.

 

 

You could also use the Cu potential from 

Pseudopotentials for quantum Monte Carlo studies of transition metal oxides, Jaron T. Krogel, Juan A. Santana, and Fernando A. Reboredo, Phys. Rev. B 93, 075143 (2016)

http://dx.doi.org/10.1103/PhysRevB.93.075143

The Supplemental Material with this paper includes Cu in UPF and QMCPACK xml formats.

These are really great information for me. Thank you very much.

Yi

[Younes Akabli]

Dear all,

I would like to ask if you could share the BFD pseudopotential for Cu in UPF format.

I tried to convert it myself using ppconvert, but I encountered the same issue reported in the QMCPACK discussion (very long runtime for Cu and no result).

Attached the GAMESS file used.

Cu-QMC GEN 2 1
3
19.00000000 1 6.25149628
118.77842940 3 6.72725326
-105.89982403 2 6.61024592
2
127.35069424 2 12.36568715
71.22984900 2 11.16072939 

Thank you very much for your help.

Best regards,
Younes

[Paul R. C. Kent]

You could try posting your ppconvert inputs and some of the output here in case anyone has a useful suggestion.

It is possible that the BFD potentials are simply not very friendly to the convergence schemes in ppconvert. They were not designed with plane-waves in mind. 

As should be clear from the rate of updates to ppconvert, the methods in ppconvert are not likely to change unless you or another contributor makes updates to them. Should we get the resources to do so, ppconvert would be replaced by a clean and modern python-based tool, where adding capabilities would be quicker. If you are set on converting BFD, perhaps another pseudopotential generating suite has recently gained a suitable tool, although I have to admit that I don't know of any that have.

If you can consider potentials other than BFD, I'll recommend the soft Cu ECP developed in  https://doi.org/10.1063/5.0109098 for plane-wave calculations. There is also the harder variant in https://doi.org/10.1063/1.5040472 . Both are available in UPF format for immediate download at https://pseudopotentiallibrary.org/

Regards, Paul

[Younes Akabli]

Dear Paul,

Thank you for your reply.

I have already tested the two pseudopotentials you mentioned, but I would like to compare them with BFD. Regarding ppconvert, I used the commands from the QMCPACK documentation and verified that they work well for O  atoms (as in the examples). However, for Cu I tried several reference configurations (see below), but the conversion does not work.

ppconvert --gamess_pot Cu.BFD.gamess --s_ref "3p(6)3d(10)4s(1)4p(0)" --p_ref "3s(2)3p(6)3d(10)4s(1)4(0)" --d_ref "3s(2)3p(6)3d(10)4s(1)4p(0)" --xml Cu.BFD.xml  
ppconvert --gamess_pot Cu.BFD.gamess --s_ref "3p(6)3d(9)4s(2)4p(0)" --p_ref "3s(2)3p(6)3d(9)4s(2)4(0)" --d_ref "3s(2)3p(6)3d(9)4s(2)4p(0)" --xml Cu.BFD.xml
ppconvert --gamess_pot Cu.BFD.gamess --s_ref "1s(2)2s(2)2p(6)3s(2)3p(6)3d(9)4s(2)4p(0)" --p_ref "1s(2)2s(2)2p(6)3s(2)3p(6)3d(9)4s(2)4(0)" --d_ref "1s(2)2s(2)2p(6)3s(2)3p(6)3d(9)4s(2)4p(0)" --xml Cu.BFD.xml

Thank you again for your help.

Best regards,
Younes

--
You received this message because you are subscribed to the Google Groups "qmcpack" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qmcpack+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/qmcpack/7a5ccd6b-d847-468d-9d34-ed1d3158d0bfn%40googlegroups.com.

[Jaron Krogel]

ppconvert is a bit heavy handed for this type of conversion.  It runs a DFT atomic solver even if you want a simple file format conversion.

Try this instead:

/your/path/to/qmcpack/utils/convert-gamess-ecp-to-qmcpack Cu.BFD.gamess

You will need to install Nexus first.  See, e.g. https://github.com/QMCPACK/qmcpack/blob/develop/nexus/README.md

~Jaron

To view this discussion visit https://groups.google.com/d/msgid/qmcpack/CAEzn3OhCCG67%2BsqYrO2eiGjUMYEROZyA4o2Cp6ehQ%2BL_sifMuA%40mail.gmail.com.

[Younes Akabli]

Dear Jaron,

Thank you for the information.

This works well to convert the GAMESS file to QMCPACK and CASINO formats, but not to Quantum ESPRESSO (UPF), so I still need to use ppconvert for that.

I was able to run ppconvert using the following command (based on QMCPACK tutorials: https://qmcpack.readthedocs.io/en/develop/lab_qmc_basics.html):

ppconvert --gamess_pot Cu.ccECP-soft.gamess --s_ref "1s(1)2p(0)3d(10)" --p_ref "1s(1)2p(0)3d(10)" --d_ref "1s(1)2p(0)3d(10)" --log_grid --upf Cu.ccECP-soft.upf

However, when I compare the generated UPF with the ones available on the pseudopotential library website (for ccECP-soft and ccECP), they are not identical, and the result changes depending on the chosen occupations.

Any guidance on this would be appreciated.

Best regards,
Younes