Re: [QE-users] vc-relax with PBE0 not converging
Should work now: see https://gitlab.com/QEF/q-e/-/merge_requests/1327, notably PW/src/hinit1.f90 Paolo On Wed, Feb 10, 2021 at 7:51 PM Paolo Giannozzi wrote: > On Wed, Feb 10, 2021 at 3:58 PM Bidault, Xavier wrote: > > So I really wonder what's wrong with QE >> > > So do I, but, you know, it is not always that simple to find what doesn't > work and why and how to fix it. > Anyway, please see the new comment at the link I sent previously ( > https://gitlab.com/QEF/q-e/-/issues/204 ) > > Paolo > > , concerning the same simulation using hybrid functionals? >> >> Best, >> Xavier >> -- >> *From:* Bidault, Xavier >> *Sent:* Tuesday, February 2, 2021 2:05 PM >> *To:* Quantum ESPRESSO users Forum >> *Subject:* Re: [QE-users] vc-relax with PBE0 not converging >> >> Hello, >> >> @Giovani: I have tested SCAN-rvv10 with the PP provided in the link. >> vc-relax works very well with it. Thank you! I just had to use a cutoff at >> least 200 Ry to get less than 2 meV/atom error. >> >> @All: I'm currently trying B3LYP-D3 (with mt PP files provided on QE >> website and cutoff larger than 140 Ry), and I get the same oscillations >> with no damping during a vc-relax. So it's not only PBE0. Something must be >> wrong with hybrid functionals and vc-relax. Does someone else get the same >> issue with hybrid functionals? Or am I missing a parameter setting? Below >> is the main part of the input script (without atom positions). Does someone >> see something obvious missing to make PBE0 run correctly with vc-relax for >> a molecular crystal (monoclinic insulator)? Maybe the bfgs algorithm to >> change? >> >> &control >> calculation='vc-relax', >> restart_mode='from_scratch', >> prefix='vc-relax_90_f4e5' >> pseudo_dir = './../../pseudo/', >> forc_conv_thr = 1.0d-4, >> etot_conv_thr = 1.0d-5, >> nstep = 1000 >> / >> &system >> ibrav=-12, >> celldm(1)= 12.331406938, >> celldm(2)= 1.691349322, >> celldm(3)= 1.128495901, >> celldm(5)= -0.2193360205, >> nat= 56, >> ntyp= 4, >> *input_dft = 'PBE0',* >> *vdw_corr = 'dft-d3',* >> >> *ecutwfc = 90 ! 2 meV/atom error * >> *ecutfock = 90* >> >> *nqx1 = 1 nqx2 = 1 nqx3 = 1 * >> / >> &electrons >> conv_thr =1.0d-8 >> / >> &ions >> *ion_dynamics = 'bfgs'* >> / >> &cell >> * cell_dynamics = 'bfgs',* >> cell_dofree = 'ibrav', >> press = 0.0, >> press_conv_thr = 0.1 >> / >> K_POINTS AUTOMATIC >> *3 2 3 1 1 1* >> ATOMIC_SPECIES >> C 12.011 C_ONCV_PBE-1.2.upf >> H 1.008 H_ONCV_PBE-1.2.upf >> O 15.999 O_ONCV_PBE-1.2.upf >> N 14.007 N_ONCV_PBE-1.2.upf >> -- >> *From:* users on behalf of >> Giovani Rech >> *Sent:* Saturday, January 30, 2021 12:28 PM >> *To:* Quantum ESPRESSO users Forum >> *Subject:* Re: [QE-users] vc-relax with PBE0 not converging >> >> Dear Xavier, >> I've faced the same issue a while back while using scan-rvv10. The >> pressure would not converge with the pseudo-potential that I was using >> (ONCV). For my case ( >> https://www.mail-archive.com/users@lists.quantum-espresso.org/msg36102.html >> <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.mail-archive.com%2Fusers%40lists.quantum-espresso.org%2Fmsg36102.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598609312%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=OsEnbXq4P4iScMir%2FNRrGlXWhHbpSGE%2BaporX1oPr7c%3D&reserved=0>) >> the issue was solved when I changed the PP to one that was built using SCAN. >> Please try to use one of the PP found here: >> https://yaoyi92.github.io/scan-tm-pseudopotentials.html >> <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyaoyi92.github.io%2Fscan-tm-pseudopotentials.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=XGWTblBQMpn92c3Cq9EuiNReSQAkxI3r5eZI97%2FN0sU%3D&reserved=0> >> or >> even look for a PP built usi
Re: [QE-users] vc-relax with PBE0 not converging
On Wed, Feb 10, 2021 at 3:58 PM Bidault, Xavier wrote: So I really wonder what's wrong with QE > So do I, but, you know, it is not always that simple to find what doesn't work and why and how to fix it. Anyway, please see the new comment at the link I sent previously ( https://gitlab.com/QEF/q-e/-/issues/204 ) Paolo , concerning the same simulation using hybrid functionals? > > Best, > Xavier > -- > *From:* Bidault, Xavier > *Sent:* Tuesday, February 2, 2021 2:05 PM > *To:* Quantum ESPRESSO users Forum > *Subject:* Re: [QE-users] vc-relax with PBE0 not converging > > Hello, > > @Giovani: I have tested SCAN-rvv10 with the PP provided in the link. > vc-relax works very well with it. Thank you! I just had to use a cutoff at > least 200 Ry to get less than 2 meV/atom error. > > @All: I'm currently trying B3LYP-D3 (with mt PP files provided on QE > website and cutoff larger than 140 Ry), and I get the same oscillations > with no damping during a vc-relax. So it's not only PBE0. Something must be > wrong with hybrid functionals and vc-relax. Does someone else get the same > issue with hybrid functionals? Or am I missing a parameter setting? Below > is the main part of the input script (without atom positions). Does someone > see something obvious missing to make PBE0 run correctly with vc-relax for > a molecular crystal (monoclinic insulator)? Maybe the bfgs algorithm to > change? > > &control > calculation='vc-relax', > restart_mode='from_scratch', > prefix='vc-relax_90_f4e5' > pseudo_dir = './../../pseudo/', > forc_conv_thr = 1.0d-4, > etot_conv_thr = 1.0d-5, > nstep = 1000 > / > &system > ibrav=-12, > celldm(1)= 12.331406938, > celldm(2)= 1.691349322, > celldm(3)= 1.128495901, > celldm(5)= -0.2193360205, > nat= 56, > ntyp= 4, > *input_dft = 'PBE0',* > *vdw_corr = 'dft-d3',* > > *ecutwfc = 90 ! 2 meV/atom error * > *ecutfock = 90* > > *nqx1 = 1 nqx2 = 1 nqx3 = 1 * > / > &electrons > conv_thr =1.0d-8 > / > &ions > *ion_dynamics = 'bfgs'* > / > &cell > *cell_dynamics = 'bfgs',* > cell_dofree = 'ibrav', > press = 0.0, > press_conv_thr = 0.1 > / > K_POINTS AUTOMATIC > *3 2 3 1 1 1* > ATOMIC_SPECIES > C 12.011 C_ONCV_PBE-1.2.upf > H 1.008 H_ONCV_PBE-1.2.upf > O 15.999 O_ONCV_PBE-1.2.upf > N 14.007 N_ONCV_PBE-1.2.upf > -- > *From:* users on behalf of > Giovani Rech > *Sent:* Saturday, January 30, 2021 12:28 PM > *To:* Quantum ESPRESSO users Forum > *Subject:* Re: [QE-users] vc-relax with PBE0 not converging > > Dear Xavier, > I've faced the same issue a while back while using scan-rvv10. The > pressure would not converge with the pseudo-potential that I was using > (ONCV). For my case ( > https://www.mail-archive.com/users@lists.quantum-espresso.org/msg36102.html > <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.mail-archive.com%2Fusers%40lists.quantum-espresso.org%2Fmsg36102.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598609312%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=OsEnbXq4P4iScMir%2FNRrGlXWhHbpSGE%2BaporX1oPr7c%3D&reserved=0>) > the issue was solved when I changed the PP to one that was built using SCAN. > Please try to use one of the PP found here: > https://yaoyi92.github.io/scan-tm-pseudopotentials.html > <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyaoyi92.github.io%2Fscan-tm-pseudopotentials.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=XGWTblBQMpn92c3Cq9EuiNReSQAkxI3r5eZI97%2FN0sU%3D&reserved=0> > or > even look for a PP built using PBE0. > > Hope this helps, > > *Giovani Rech* > > Materials Physics and Advanced Ceramics > Universidade de Caxias do Sul > 95070-560 Caxias do Sul - RS, Brazil > orcid.org/-0003-0476-2056 > <https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Forcid.org%2F-0003-0476-2056&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sd
Re: [QE-users] vc-relax with PBE0 not converging
Hello, I finally turned to CP2K to use PBE0-D3. The equivalent of vc-relax worked for my molecular crystal and the structure converged. I did an all-electron simulation and have not yet used pseudopotentials. So I really wonder what's wrong with QE, concerning the same simulation using hybrid functionals? Best, Xavier From: Bidault, Xavier Sent: Tuesday, February 2, 2021 2:05 PM To: Quantum ESPRESSO users Forum Subject: Re: [QE-users] vc-relax with PBE0 not converging Hello, @Giovani: I have tested SCAN-rvv10 with the PP provided in the link. vc-relax works very well with it. Thank you! I just had to use a cutoff at least 200 Ry to get less than 2 meV/atom error. @All: I'm currently trying B3LYP-D3 (with mt PP files provided on QE website and cutoff larger than 140 Ry), and I get the same oscillations with no damping during a vc-relax. So it's not only PBE0. Something must be wrong with hybrid functionals and vc-relax. Does someone else get the same issue with hybrid functionals? Or am I missing a parameter setting? Below is the main part of the input script (without atom positions). Does someone see something obvious missing to make PBE0 run correctly with vc-relax for a molecular crystal (monoclinic insulator)? Maybe the bfgs algorithm to change? &control calculation='vc-relax', restart_mode='from_scratch', prefix='vc-relax_90_f4e5' pseudo_dir = './../../pseudo/', forc_conv_thr = 1.0d-4, etot_conv_thr = 1.0d-5, nstep = 1000 / &system ibrav=-12, celldm(1)= 12.331406938, celldm(2)= 1.691349322, celldm(3)= 1.128495901, celldm(5)= -0.2193360205, nat= 56, ntyp= 4, input_dft = 'PBE0', vdw_corr = 'dft-d3', ecutwfc = 90 ! 2 meV/atom error ecutfock = 90 nqx1 = 1 nqx2 = 1 nqx3 = 1 / &electrons conv_thr =1.0d-8 / &ions ion_dynamics = 'bfgs' / &cell cell_dynamics = 'bfgs', cell_dofree = 'ibrav', press = 0.0, press_conv_thr = 0.1 / K_POINTS AUTOMATIC 3 2 3 1 1 1 ATOMIC_SPECIES C 12.011 C_ONCV_PBE-1.2.upf H 1.008 H_ONCV_PBE-1.2.upf O 15.999 O_ONCV_PBE-1.2.upf N 14.007 N_ONCV_PBE-1.2.upf From: users on behalf of Giovani Rech Sent: Saturday, January 30, 2021 12:28 PM To: Quantum ESPRESSO users Forum Subject: Re: [QE-users] vc-relax with PBE0 not converging Dear Xavier, I've faced the same issue a while back while using scan-rvv10. The pressure would not converge with the pseudo-potential that I was using (ONCV). For my case (https://www.mail-archive.com/users@lists.quantum-espresso.org/msg36102.html<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.mail-archive.com%2Fusers%40lists.quantum-espresso.org%2Fmsg36102.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598609312%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=OsEnbXq4P4iScMir%2FNRrGlXWhHbpSGE%2BaporX1oPr7c%3D&reserved=0>) the issue was solved when I changed the PP to one that was built using SCAN. Please try to use one of the PP found here: https://yaoyi92.github.io/scan-tm-pseudopotentials.html<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyaoyi92.github.io%2Fscan-tm-pseudopotentials.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=XGWTblBQMpn92c3Cq9EuiNReSQAkxI3r5eZI97%2FN0sU%3D&reserved=0> or even look for a PP built using PBE0. Hope this helps, Giovani Rech Materials Physics and Advanced Ceramics Universidade de Caxias do Sul 95070-560 Caxias do Sul - RS, Brazil orcid.org/-0003-0476-2056<https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Forcid.org%2F-0003-0476-2056&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=617XlpxPnbTJLbgJhYF7NAE7zb%2FSMyJIEQF9Fbw6IU0%3D&reserved=0> On Mon, Jan 25, 2021 at 12:19 PM Bidault, Xavier mailto:xavb...@uic.edu>> wrote: Hello, I have a molecular crystal (1 unit cell = 56 atoms). I use SG15 ONCV 1.2 pseudopotential and Grimme D2 correction. Using vc-relax with PBE, the structure converges, no problem. But with PBE0, density and pressure oscillate without dampening (see figure below - Pressure remains negative, and not close to zero). Is there any particular strategy with PBE0? * vc-relax with PBE and then starting from the res
Re: [QE-users] vc-relax with PBE0 not converging
Hello, @Giovani: I have tested SCAN-rvv10 with the PP provided in the link. vc-relax works very well with it. Thank you! I just had to use a cutoff at least 200 Ry to get less than 2 meV/atom error. @All: I'm currently trying B3LYP-D3 (with mt PP files provided on QE website and cutoff larger than 140 Ry), and I get the same oscillations with no damping during a vc-relax. So it's not only PBE0. Something must be wrong with hybrid functionals and vc-relax. Does someone else get the same issue with hybrid functionals? Or am I missing a parameter setting? Below is the main part of the input script (without atom positions). Does someone see something obvious missing to make PBE0 run correctly with vc-relax for a molecular crystal (monoclinic insulator)? Maybe the bfgs algorithm to change? &control calculation='vc-relax', restart_mode='from_scratch', prefix='vc-relax_90_f4e5' pseudo_dir = './../../pseudo/', forc_conv_thr = 1.0d-4, etot_conv_thr = 1.0d-5, nstep = 1000 / &system ibrav=-12, celldm(1)= 12.331406938, celldm(2)= 1.691349322, celldm(3)= 1.128495901, celldm(5)= -0.2193360205, nat= 56, ntyp= 4, input_dft = 'PBE0', vdw_corr = 'dft-d3', ecutwfc = 90 ! 2 meV/atom error ecutfock = 90 nqx1 = 1 nqx2 = 1 nqx3 = 1 / &electrons conv_thr =1.0d-8 / &ions ion_dynamics = 'bfgs' / &cell cell_dynamics = 'bfgs', cell_dofree = 'ibrav', press = 0.0, press_conv_thr = 0.1 / K_POINTS AUTOMATIC 3 2 3 1 1 1 ATOMIC_SPECIES C 12.011 C_ONCV_PBE-1.2.upf H 1.008 H_ONCV_PBE-1.2.upf O 15.999 O_ONCV_PBE-1.2.upf N 14.007 N_ONCV_PBE-1.2.upf From: users on behalf of Giovani Rech Sent: Saturday, January 30, 2021 12:28 PM To: Quantum ESPRESSO users Forum Subject: Re: [QE-users] vc-relax with PBE0 not converging Dear Xavier, I've faced the same issue a while back while using scan-rvv10. The pressure would not converge with the pseudo-potential that I was using (ONCV). For my case (https://www.mail-archive.com/users@lists.quantum-espresso.org/msg36102.html<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.mail-archive.com%2Fusers%40lists.quantum-espresso.org%2Fmsg36102.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598609312%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=OsEnbXq4P4iScMir%2FNRrGlXWhHbpSGE%2BaporX1oPr7c%3D&reserved=0>) the issue was solved when I changed the PP to one that was built using SCAN. Please try to use one of the PP found here: https://yaoyi92.github.io/scan-tm-pseudopotentials.html<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyaoyi92.github.io%2Fscan-tm-pseudopotentials.html&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=XGWTblBQMpn92c3Cq9EuiNReSQAkxI3r5eZI97%2FN0sU%3D&reserved=0> or even look for a PP built using PBE0. Hope this helps, Giovani Rech Materials Physics and Advanced Ceramics Universidade de Caxias do Sul 95070-560 Caxias do Sul - RS, Brazil orcid.org/-0003-0476-2056<https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Forcid.org%2F-0003-0476-2056&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637476282598619268%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=617XlpxPnbTJLbgJhYF7NAE7zb%2FSMyJIEQF9Fbw6IU0%3D&reserved=0> On Mon, Jan 25, 2021 at 12:19 PM Bidault, Xavier mailto:xavb...@uic.edu>> wrote: Hello, I have a molecular crystal (1 unit cell = 56 atoms). I use SG15 ONCV 1.2 pseudopotential and Grimme D2 correction. Using vc-relax with PBE, the structure converges, no problem. But with PBE0, density and pressure oscillate without dampening (see figure below - Pressure remains negative, and not close to zero). Is there any particular strategy with PBE0? * vc-relax with PBE and then starting from the result and use PBE0? * relax with PBE0 and then vc-relax? * any simulation parameter to change (ion dynamics and/or cell dynamics - currently BFGS)? * any strategy with PBE0? [cid:1775486a5dacb971f161] Thank you, Xavier ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu<https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2F&data=04%7C01%7Cxavbdlt%40uic.edu%7C709e6f580f424bb29eaa08d8c54d02ce%7
Re: [QE-users] vc-relax with PBE0 not converging
Dear Xavier, I've faced the same issue a while back while using scan-rvv10. The pressure would not converge with the pseudo-potential that I was using (ONCV). For my case ( https://www.mail-archive.com/users@lists.quantum-espresso.org/msg36102.html) the issue was solved when I changed the PP to one that was built using SCAN. Please try to use one of the PP found here: https://yaoyi92.github.io/scan-tm-pseudopotentials.html or even look for a PP built using PBE0. Hope this helps, *Giovani Rech* Materials Physics and Advanced Ceramics Universidade de Caxias do Sul 95070-560 Caxias do Sul - RS, Brazil orcid.org/-0003-0476-2056 On Mon, Jan 25, 2021 at 12:19 PM Bidault, Xavier wrote: > Hello, > > I have a molecular crystal (1 unit cell = 56 atoms). I use SG15 ONCV 1.2 > pseudopotential and Grimme D2 correction. > Using vc-relax with PBE, the structure converges, no problem. > But with PBE0, density and pressure oscillate without dampening (see > figure below - Pressure remains negative, and not close to zero). Is there > any particular strategy with PBE0? > >- vc-relax with PBE and then starting from the result and use PBE0? >- relax with PBE0 and then vc-relax? >- any simulation parameter to change (ion dynamics and/or cell >dynamics - currently BFGS)? >- any strategy with PBE0? > > > Thank you, > Xavier > > ___ > 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] vc-relax with PBE0 not converging
Hello again, This is not really better with grid 3x2x3 (off-centered). See image below. Oscillations are visible on density, not much for pressure (but numbers end up periodic anyway). Pressure remains negative. However pressure oscillation decreases with higher number of k-point (Gamma -> 2x1x2 off centered -> 3x2x3 off-centered), but there is still a large offset. Note that SCF convergence of this organic crystal w.r.t. k-points is perfectly converged with 3x2x3 off-centered grid and PBE. Increasing the number of k-points for PBE0 may be better, but out of time for me. And my guess is it will not solve the pressure offset. Concerning adding smearing, I don't see the point. This molecular system (known as not conductor) does not need it with PBE, so I don't know why I should add smearing with PBE0. I gave a try to HSE instead of PBE0. Same oscillations. There is something wrong with vc-relax and hybrid xc functionals. [cid:61a1742b-37e7-41bb-81af-c20dd2d531a6] Best, Xavier From: Bidault, Xavier Sent: Tuesday, January 26, 2021 9:50 AM To: Quantum ESPRESSO users Forum Subject: Re: [QE-users] vc-relax with PBE0 not converging Hello, My system is an organic molecular crystal (atoms C, H, N, O). The cutoff is chosen so there is 2 meV/atom error for SCF energy. The k-point grid 3x2x3 (off-centered) yields perfectly converged SCF energy. But to get faster with the vc-relax, I lowered it to 2x1x2 (off-centered), which introduces an error of only -0.1 meV/atom. I don't know if this is the reason why vc-relax density and pressure oscillate without converging. I'm currently running v-relax with 3x2x3 grid... waiting for the result (10x slower)... Thank you, Xavier From: users on behalf of Paolo Giannozzi Sent: Monday, January 25, 2021 9:38 AM To: Quantum ESPRESSO users Forum Subject: Re: [QE-users] vc-relax with PBE0 not converging Maybe the same problem reported here? https://gitlab.com/QEF/q-e/-/issues/204<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitlab.com%2FQEF%2Fq-e%2F-%2Fissues%2F204&data=04%7C01%7Cxavbdlt%40uic.edu%7C3fdf0cc0f8bd4cd43e8308d8c14769ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637471859758045446%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=QCUVoTzncfJfau7FM88IUZSHHmCHJf9wUl50miJJ4W4%3D&reserved=0> Paolo On Mon, Jan 25, 2021 at 4:19 PM Bidault, Xavier mailto:xavb...@uic.edu>> wrote: Hello, I have a molecular crystal (1 unit cell = 56 atoms). I use SG15 ONCV 1.2 pseudopotential and Grimme D2 correction. Using vc-relax with PBE, the structure converges, no problem. But with PBE0, density and pressure oscillate without dampening (see figure below - Pressure remains negative, and not close to zero). Is there any particular strategy with PBE0? * vc-relax with PBE and then starting from the result and use PBE0? * relax with PBE0 and then vc-relax? * any simulation parameter to change (ion dynamics and/or cell dynamics - currently BFGS)? * any strategy with PBE0? [cid:1773a31d564cb971f161] Thank you, Xavier ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu<https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2F&data=04%7C01%7Cxavbdlt%40uic.edu%7C3fdf0cc0f8bd4cd43e8308d8c14769ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637471859758045446%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=sd3E5%2FhytwHgry%2Bvhoi4xrI%2F2AwuoSy5%2FOnqdAivFhs%3D&reserved=0>) users mailing list users@lists.quantum-espresso.org<mailto:users@lists.quantum-espresso.org> https://lists.quantum-espresso.org/mailman/listinfo/users<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusers&data=04%7C01%7Cxavbdlt%40uic.edu%7C3fdf0cc0f8bd4cd43e8308d8c14769ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637471859758055404%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=rLgz1kcGcloPgRTtOcAQJcRXdnlFNRtxNyn5cLZt9vE%3D&reserved=0> -- Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, Univ. Udine, via delle Scienze 206, 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] vc-relax with PBE0 not converging
Hello, My system is an organic molecular crystal (atoms C, H, N, O). The cutoff is chosen so there is 2 meV/atom error for SCF energy. The k-point grid 3x2x3 (off-centered) yields perfectly converged SCF energy. But to get faster with the vc-relax, I lowered it to 2x1x2 (off-centered), which introduces an error of only -0.1 meV/atom. I don't know if this is the reason why vc-relax density and pressure oscillate without converging. I'm currently running v-relax with 3x2x3 grid... waiting for the result (10x slower)... Thank you, Xavier From: users on behalf of Paolo Giannozzi Sent: Monday, January 25, 2021 9:38 AM To: Quantum ESPRESSO users Forum Subject: Re: [QE-users] vc-relax with PBE0 not converging Maybe the same problem reported here? https://gitlab.com/QEF/q-e/-/issues/204<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitlab.com%2FQEF%2Fq-e%2F-%2Fissues%2F204&data=04%7C01%7Cxavbdlt%40uic.edu%7C3fdf0cc0f8bd4cd43e8308d8c14769ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637471859758045446%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=QCUVoTzncfJfau7FM88IUZSHHmCHJf9wUl50miJJ4W4%3D&reserved=0> Paolo On Mon, Jan 25, 2021 at 4:19 PM Bidault, Xavier mailto:xavb...@uic.edu>> wrote: Hello, I have a molecular crystal (1 unit cell = 56 atoms). I use SG15 ONCV 1.2 pseudopotential and Grimme D2 correction. Using vc-relax with PBE, the structure converges, no problem. But with PBE0, density and pressure oscillate without dampening (see figure below - Pressure remains negative, and not close to zero). Is there any particular strategy with PBE0? * vc-relax with PBE and then starting from the result and use PBE0? * relax with PBE0 and then vc-relax? * any simulation parameter to change (ion dynamics and/or cell dynamics - currently BFGS)? * any strategy with PBE0? [cid:1773a31d564cb971f161] Thank you, Xavier ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu<https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2F&data=04%7C01%7Cxavbdlt%40uic.edu%7C3fdf0cc0f8bd4cd43e8308d8c14769ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637471859758045446%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=sd3E5%2FhytwHgry%2Bvhoi4xrI%2F2AwuoSy5%2FOnqdAivFhs%3D&reserved=0>) users mailing list users@lists.quantum-espresso.org<mailto:users@lists.quantum-espresso.org> https://lists.quantum-espresso.org/mailman/listinfo/users<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusers&data=04%7C01%7Cxavbdlt%40uic.edu%7C3fdf0cc0f8bd4cd43e8308d8c14769ce%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637471859758055404%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=rLgz1kcGcloPgRTtOcAQJcRXdnlFNRtxNyn5cLZt9vE%3D&reserved=0> -- Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, Univ. Udine, via delle Scienze 206, 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] vc-relax with PBE0 not converging
Maybe the same problem reported here? https://gitlab.com/QEF/q-e/-/issues/204 Paolo On Mon, Jan 25, 2021 at 4:19 PM Bidault, Xavier wrote: > Hello, > > I have a molecular crystal (1 unit cell = 56 atoms). I use SG15 ONCV 1.2 > pseudopotential and Grimme D2 correction. > Using vc-relax with PBE, the structure converges, no problem. > But with PBE0, density and pressure oscillate without dampening (see > figure below - Pressure remains negative, and not close to zero). Is there > any particular strategy with PBE0? > >- vc-relax with PBE and then starting from the result and use PBE0? >- relax with PBE0 and then vc-relax? >- any simulation parameter to change (ion dynamics and/or cell >dynamics - currently BFGS)? >- any strategy with PBE0? > > > Thank you, > Xavier > > ___ > 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 -- Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche, Univ. Udine, via delle Scienze 206, 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