Re: [QE-users] Problems with Fe ONCV pseudopotential
Related to this, we found a number of ghosts in the SG15, liaised with Francois, and that led to a number of updates in SG15.1 . Those are clearly visible in the band structure plots and matrix errors on the SSSP page. E.g. compare SG15 and SG15.1: https://www.materialscloud.org/discover/sssp/plot/precision/S nicola On 01/10/2020 10:48, Christoph Wolf wrote: There is a "known problem" (known at least to Don Hamann, the creator of the ONCV), I will paste part of a conversation I had with him a few years ago pertaining to the same problem: " The Fe SG15 input data must have been on the "hairy edge" of ghost behavior. A small change of the local potential in the input should make it go away with the corrected 3.3.1, although I have not tested this yet. It is unfortunate that some other SG15 input may have the same issue, and I'm confident that such a small local-potential change should not effect the quality of the psp, but the whole set apparently needs review. Deep ghost detection was an innovation with 3.3.0, although the SG15 automated input data refinement using real solid-state calculations would certainly have rejected data that produced ghosts with 2.2.1 " HTH! Chris -- Group Leader "Theory of Quantum Systems at Surfaces" IBS Center for Quantum Nanoscience Seoul, South Korea ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users -- -- Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL Director, National Centre for Competence in Research NCCR MARVEL, EPFL http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
There is a "known problem" (known at least to Don Hamann, the creator of the ONCV), I will paste part of a conversation I had with him a few years ago pertaining to the same problem: " The Fe SG15 input data must have been on the "hairy edge" of ghost behavior. A small change of the local potential in the input should make it go away with the corrected 3.3.1, although I have not tested this yet. It is unfortunate that some other SG15 input may have the same issue, and I'm confident that such a small local-potential change should not effect the quality of the psp, but the whole set apparently needs review. Deep ghost detection was an innovation with 3.3.0, although the SG15 automated input data refinement using real solid-state calculations would certainly have rejected data that produced ghosts with 2.2.1 " HTH! Chris -- Group Leader "Theory of Quantum Systems at Surfaces" IBS Center for Quantum Nanoscience Seoul, South Korea ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
On Wed, Sep 30, 2020 at 2:30 PM mkondrin wrote: May be "ONCV potentials" link in the QE site > (https://www.quantum-espresso.org/pseudopotentials) should be upgraded? > the previous link was selected because it provided PP files with pseudo-atomic orbitals. The last point of the README of https://github.com/pipidog/ONCVPSP contains this statement, that seems to explain what happened and why: * Use at your own risk! Although I used exactly the same input parameters as in the original ABINIT and SG15 database to regenerate ONCVPSPs in UPF format, the ONCVPSP code I used was v3.x where the original ABINIT and SG15 database were generate by v2.x. Therefore, the calculated results could be different. You should always double check your results carefully. Anyway, I have reverted the link to the original one (but won't answer complaints that "projwfc.x doesn't work" and the like) Paolo -- Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222 ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
On 29.09.2020 15:41, Lorenzo Paulatto wrote: Many thanks again, Lorenzo. I believe I have downloaded ONCV pseudopotentials from the link in the official QE site (https://github.com/pipidog/ONCVPSP). It might be outdated, at least psedopotential files in it have no version number. > The latest version of > SG15 pseudopotentials can be downloaded from the original site > http://www.quantum-simulation.org/potentials/sg15_oncv/ . > I don't know who "pipidog" is, but, yes, I would recommend downloading from the official website, especially if there is any problem. However, in this case, there is a problem because the official files do not include atomic wavefunctions (i.e. projwfc and exx with locla wavefunctions are not possible). Also, S&G used quite an old version of the ONCV code, newerv version produce substantially different pseudopotentials, but I have never done a study of this issue. cheers Hi Lorenzo and Nicola, Indeed downloading SG15 pseudopotentials from the original site http://www.quantum-simulation.org/potentials/sg15_oncv/ drastically changed the situation. Even Fe_ONCV_PBE-1.0.upf pseudopotential didn't produce ghost states. May be "ONCV potentials" link in the QE site (https://www.quantum-espresso.org/pseudopotentials) should be upgraded? Many thanks for help. Sincerely yours, M.V. Kondrin ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
Many thanks again, Lorenzo. I believe I have downloaded ONCV pseudopotentials from the link in the official QE site (https://github.com/pipidog/ONCVPSP). It might be outdated, at least psedopotential files in it have no version number. > The latest version of > SG15 pseudopotentials can be downloaded from the original site > http://www.quantum-simulation.org/potentials/sg15_oncv/ . > I don't know who "pipidog" is, but, yes, I would recommend downloading from the official website, especially if there is any problem. However, in this case, there is a problem because the official files do not include atomic wavefunctions (i.e. projwfc and exx with locla wavefunctions are not possible). Also, S&G used quite an old version of the ONCV code, newerv version produce substantially different pseudopotentials, but I have never done a study of this issue. cheers -- Lorenzo Paulatto - Paris ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
Dear MV, you can find the tests for the SG15 pseudos on the SSSP pseudopotentials page - I checked the Fe, and it doesn't have any ghosts, at least in the elemental bulk form. nicola On 29/09/2020 12:59, mkondrin wrote: On 29.09.2020 14:15, Lorenzo Paulatto wrote: Thank you for your exhaustive and enlightening comment. I would rather stick to the abinit version of ONCV pseudopotentials. I really don't know which version I am using (at least I haven't managed to find the version number in the ONCVPSP folders). The zip-archive was downloaded 4 December 2019 and I believe it was the latest version at the moment. Hello, the version number is in the file name, it can be 1.0 - January 2015 1.1 - May 2015, not for all elements 1.2 - (and not 1.5, as I incorrectly recalled) from June 2016, but only published ready-made in UPF version in February 2020 For Iron you have Fe_ONCV_PBE-1.0.UPF Fe_ONCV_PBE-1.2.UPF I would say, try the 1.2 if the 1.0 has ghosts. Many thanks again, Lorenzo. I believe I have downloaded ONCV pseudopotentials from the link in the official QE site (https://github.com/pipidog/ONCVPSP). It might be outdated, at least psedopotential files in it have no version number. The latest version of SG15 pseudopotentials can be downloaded from the original site http://www.quantum-simulation.org/potentials/sg15_oncv/ . Surely I give them a try. Sincerely yours, M.V. Kondrin cheers Sincerely yours, M.V. Kondrin ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users -- -- Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL Director, National Centre for Competence in Research NCCR MARVEL, EPFL http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
On 29.09.2020 14:15, Lorenzo Paulatto wrote: Thank you for your exhaustive and enlightening comment. I would rather stick to the abinit version of ONCV pseudopotentials. I really don't know which version I am using (at least I haven't managed to find the version number in the ONCVPSP folders). The zip-archive was downloaded 4 December 2019 and I believe it was the latest version at the moment. Hello, the version number is in the file name, it can be 1.0 - January 2015 1.1 - May 2015, not for all elements 1.2 - (and not 1.5, as I incorrectly recalled) from June 2016, but only published ready-made in UPF version in February 2020 For Iron you have Fe_ONCV_PBE-1.0.UPF Fe_ONCV_PBE-1.2.UPF I would say, try the 1.2 if the 1.0 has ghosts. Many thanks again, Lorenzo. I believe I have downloaded ONCV pseudopotentials from the link in the official QE site (https://github.com/pipidog/ONCVPSP). It might be outdated, at least psedopotential files in it have no version number. The latest version of SG15 pseudopotentials can be downloaded from the original site http://www.quantum-simulation.org/potentials/sg15_oncv/ . Surely I give them a try. Sincerely yours, M.V. Kondrin cheers Sincerely yours, M.V. Kondrin ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
Thank you for your exhaustive and enlightening comment. I would rather stick to the abinit version of ONCV pseudopotentials. I really don't know which version I am using (at least I haven't managed to find the version number in the ONCVPSP folders). The zip-archive was downloaded 4 December 2019 and I believe it was the latest version at the moment. Hello, the version number is in the file name, it can be 1.0 - January 2015 1.1 - May 2015, not for all elements 1.2 - (and not 1.5, as I incorrectly recalled) from June 2016, but only published ready-made in UPF version in February 2020 For Iron you have Fe_ONCV_PBE-1.0.UPF Fe_ONCV_PBE-1.2.UPF I would say, try the 1.2 if the 1.0 has ghosts. cheers Sincerely yours, M.V. Kondrin ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users -- Lorenzo Paulatto - Paris ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
On 29.09.2020 13:30, Lorenzo Paulatto wrote: *-123.5858-123.5848-123.5814-123.5812-123.5812-123.5812-123.5812 P.S. the "***" mean that there is an eigenvalue more negative than -999 eV, it is the ghost state which manages to make itself orthogonal to the repulsive non-local projectors, to snuggle in the attractive local potential. Hi, Lorenzo, Thank you for your exhaustive and enlightening comment. I would rather stick to the abinit version of ONCV pseudopotentials. I really don't know which version I am using (at least I haven't managed to find the version number in the ONCVPSP folders). The zip-archive was downloaded 4 December 2019 and I believe it was the latest version at the moment. Sincerely yours, M.V. Kondrin ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
*-123.5858-123.5848-123.5814-123.5812-123.5812-123.5812-123.5812 P.S. the "***" mean that there is an eigenvalue more negative than -999 eV, it is the ghost state which manages to make itself orthogonal to the repulsive non-local projectors, to snuggle in the attractive local potential. -- Lorenzo Paulatto - Paris ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Problems with Fe ONCV pseudopotential
*-123.5858-123.5848-123.5814-123.5812-123.5812-123.5812-123.5812 That is quite obviously a sign of a ghost state of the pseudopotential. I found a similar one in the sg16-ONCV pseudo of Mn some time ago. However, the problem was only present with pseudos v 1.1 or 1.2 of the library, while the new v 1.5 did not have it. If a new version does not fix it, you can either change pseudo, or contact S&G to signal the issue and ask for help. hth Increasing ecut value to 85 didn't help to get rid of these asterisks. I have also tried abinit flavour of Fe ONCV potential and this solve the problem with wrong number of electrons and the spin state is the one that would be expected from the Hund's rule (although calculations are significantly slower than compared to sg15 pseudopentials). Sincerely yours, M. V. Kondrin ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users -- Lorenzo Paulatto - Paris ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users