Re: [QE-users] vc-relax with PBE0 and QE6.7
Dear Lorenzo and Giuseppe, Thanks for the tips. Indeed ecutfock = ecuwfc doesn't add significant error in energy convergence for my system (+0.1 meV/atom). Then I changed the k-point grid from 3x2x3 to 2x1x2. 3x2x3 was perfectly converged for my system, and switching to 2x1x2 doesn't add significant error either (-0.1 meV/atom). As a result, the scf calculation (keeping the same convergence parameters) with PBE0 is now only 5 times as long as PBE. With a very little trade-off, I think. Thanks again! Xavier From: users on behalf of Giuseppe Mattioli Sent: Thursday, January 21, 2021 9:56 AM To: users@lists.quantum-espresso.org Subject: Re: [QE-users] vc-relax with PBE0 and QE6.7 Dear Xavier Lorenzo is totally right, but there are two further parameters that can improve your experience with exact-exchange functionals in QE: 1) ecutfock The cutoff for nonlocal integrals has a huge impact on computational resources. The default is ecutfock=ecutrho, but you rarely need that. In my experience, ecutfock=ecutwfc is quite satisfactorily for geometry optimizations, and ecutfock=1.5~2*ecutwfc is a sensible choice for properties calculation. You can test it on a small system. 2) adaptive_thr=.true. In systems having no particular scf convergence issues, this generally provides a speedup of inner scf loops. HTH Giuseppe Quoting Lorenzo Paulatto : > Yes. It mostly depends on the choice of nq1x, nq2x and nq3x. > Local-density is linear in the number of k-points, but exact > exchange is quadratic, and it can get slow very quickly > > Lorenzo Paulatto - Paris > On Jan 21 2021, at 4:14 pm, Bidault, Xavier wrote: >> >> Thank you, Giuseppe. I tried with ONCV SG15 PP library and it went >> fine. However I wasn't expecting PBE0 to be so much longer than PBE >> for the same calculation. I don't have yet the whole convergence >> curve w.r.t. cutoff, but with the same convergence parameters, same >> k-point grid... PBE0 is 300-400x slower than PBE for an SCF >> calculation. Is it the right order of magnitude? >> >> From: users on behalf of >> Giuseppe Mattioli >> Sent: Wednesday, January 20, 2021 6:22 AM >> To: users@lists.quantum-espresso.org >> Subject: Re: [QE-users] vc-relax with PBE0 and QE6.7 >> >> >> >> Dear Xavier >> Forces can be calculated in the case of PBE0 and other hybrid EXX >> functionals if you use norm-conserving (NC) pseudopotentials. No, >> there is no way to perform calculations without pseudopotentials in QE. >> HTH >> Giuseppe >> >> Quoting "Bidault, Xavier" : >> > Hello, >> > >> > I would like to run a geometry optimization (vc-relax) using PBE0 >> > and QE6.7. I got the following error message: >> > Error in routine setup (1): >> > forces for hybrid functionals + US/PAW not implemented >> > >> > >> > I guess this means I have to rely on another software, don't I? Or >> > is there a way to discard the need of PP with PBE0 and QE6.7? >> > >> > Thank you, >> > Xavier >> >> >> >> GIUSEPPE MATTIOLI >> CNR - ISTITUTO DI STRUTTURA DELLA MATERIA >> Via Salaria Km 29,300 - C.P. 10 >> I-00015 - Monterotondo Scalo (RM) >> Mob (*preferred*) +39 373 7305625 >> Tel + 39 06 90672342 - Fax +39 06 90672316 >> E-mail: >> >> ___ >> Quantum ESPRESSO is supported by MaX >> (https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2Fdata=04%7C01%7Cxavbdlt%40uic.edu%7C1dda01b591524bcb446808d8be255e4a%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637468414998966951%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=zzNw1Pbeo634CHxiN63AXdgtBw1bp95%2BzdqLkvds4hE%3Dreserved=0) >> users mailing list users@lists.quantum-espresso.org >> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusersdata=04%7C01%7Cxavbdlt%40uic.edu%7C1dda01b591524bcb446808d8be255e4a%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637468414998976910%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=u%2Fu%2BMLSKrN3MbKnOfhyi%2BB4%2BpVWfSScc3aQKgTPV8UU%3Dreserved=0 >> >> >> ___ >> Quantum ESPRESSO is supported by MaX >> (https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2Fdata=04%7C01%7Cxavbdlt%40uic.edu%7C1dda01b591524bcb446808d8be255e4a%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637468414998976910%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6
Re: [QE-users] vc-relax with PBE0 and QE6.7
Dear Xavier Lorenzo is totally right, but there are two further parameters that can improve your experience with exact-exchange functionals in QE: 1) ecutfock The cutoff for nonlocal integrals has a huge impact on computational resources. The default is ecutfock=ecutrho, but you rarely need that. In my experience, ecutfock=ecutwfc is quite satisfactorily for geometry optimizations, and ecutfock=1.5~2*ecutwfc is a sensible choice for properties calculation. You can test it on a small system. 2) adaptive_thr=.true. In systems having no particular scf convergence issues, this generally provides a speedup of inner scf loops. HTH Giuseppe Quoting Lorenzo Paulatto : Yes. It mostly depends on the choice of nq1x, nq2x and nq3x. Local-density is linear in the number of k-points, but exact exchange is quadratic, and it can get slow very quickly Lorenzo Paulatto - Paris On Jan 21 2021, at 4:14 pm, Bidault, Xavier wrote: Thank you, Giuseppe. I tried with ONCV SG15 PP library and it went fine. However I wasn't expecting PBE0 to be so much longer than PBE for the same calculation. I don't have yet the whole convergence curve w.r.t. cutoff, but with the same convergence parameters, same k-point grid... PBE0 is 300-400x slower than PBE for an SCF calculation. Is it the right order of magnitude? From: users on behalf of Giuseppe Mattioli Sent: Wednesday, January 20, 2021 6:22 AM To: users@lists.quantum-espresso.org Subject: Re: [QE-users] vc-relax with PBE0 and QE6.7 Dear Xavier Forces can be calculated in the case of PBE0 and other hybrid EXX functionals if you use norm-conserving (NC) pseudopotentials. No, there is no way to perform calculations without pseudopotentials in QE. HTH Giuseppe Quoting "Bidault, Xavier" : > Hello, > > I would like to run a geometry optimization (vc-relax) using PBE0 > and QE6.7. I got the following error message: > Error in routine setup (1): > forces for hybrid functionals + US/PAW not implemented > > > I guess this means I have to rely on another software, don't I? Or > is there a way to discard the need of PP with PBE0 and QE6.7? > > Thank you, > Xavier GIUSEPPE MATTIOLI CNR - ISTITUTO DI STRUTTURA DELLA MATERIA Via Salaria Km 29,300 - C.P. 10 I-00015 - Monterotondo Scalo (RM) Mob (*preferred*) +39 373 7305625 Tel + 39 06 90672342 - Fax +39 06 90672316 E-mail: ___ Quantum ESPRESSO is supported by MaX (https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2Fdata=04%7C01%7Cxavbdlt%40uic.edu%7Cb283a1a22511428303b508d8bd3e127e%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637467421621058448%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=nSRscevBNWTxUvuucpoteZlKTo5Ug%2BAIDVIuFATokZo%3Dreserved=0) users mailing list users@lists.quantum-espresso.org https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusersdata=04%7C01%7Cxavbdlt%40uic.edu%7Cb283a1a22511428303b508d8bd3e127e%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637467421621058448%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=rX2etOAO2hyg7NB6ciuRHq0zJzJnPRIjQjTU4hyVGIs%3Dreserved=0 ___ 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 GIUSEPPE MATTIOLI CNR - ISTITUTO DI STRUTTURA DELLA MATERIA Via Salaria Km 29,300 - C.P. 10 I-00015 - Monterotondo Scalo (RM) Mob (*preferred*) +39 373 7305625 Tel + 39 06 90672342 - Fax +39 06 90672316 E-mail: ___ 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 and QE6.7
Yes. It mostly depends on the choice of nq1x, nq2x and nq3x. Local-density is linear in the number of k-points, but exact exchange is quadratic, and it can get slow very quickly Lorenzo Paulatto - Paris On Jan 21 2021, at 4:14 pm, Bidault, Xavier wrote: > > Thank you, Giuseppe. I tried with ONCV SG15 PP library and it went fine. > However I wasn't expecting PBE0 to be so much longer than PBE for the same > calculation. I don't have yet the whole convergence curve w.r.t. cutoff, but > with the same convergence parameters, same k-point grid... PBE0 is 300-400x > slower than PBE for an SCF calculation. Is it the right order of magnitude? > > From: users on behalf of Giuseppe > Mattioli > Sent: Wednesday, January 20, 2021 6:22 AM > To: users@lists.quantum-espresso.org > Subject: Re: [QE-users] vc-relax with PBE0 and QE6.7 > > > > Dear Xavier > Forces can be calculated in the case of PBE0 and other hybrid EXX > functionals if you use norm-conserving (NC) pseudopotentials. No, > there is no way to perform calculations without pseudopotentials in QE. > HTH > Giuseppe > > Quoting "Bidault, Xavier" : > > Hello, > > > > I would like to run a geometry optimization (vc-relax) using PBE0 > > and QE6.7. I got the following error message: > > Error in routine setup (1): > > forces for hybrid functionals + US/PAW not implemented > > > > > > I guess this means I have to rely on another software, don't I? Or > > is there a way to discard the need of PP with PBE0 and QE6.7? > > > > Thank you, > > Xavier > > > > GIUSEPPE MATTIOLI > CNR - ISTITUTO DI STRUTTURA DELLA MATERIA > Via Salaria Km 29,300 - C.P. 10 > I-00015 - Monterotondo Scalo (RM) > Mob (*preferred*) +39 373 7305625 > Tel + 39 06 90672342 - Fax +39 06 90672316 > E-mail: > > ___ > Quantum ESPRESSO is supported by MaX > (https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2Fdata=04%7C01%7Cxavbdlt%40uic.edu%7Cb283a1a22511428303b508d8bd3e127e%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637467421621058448%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=nSRscevBNWTxUvuucpoteZlKTo5Ug%2BAIDVIuFATokZo%3Dreserved=0) > users mailing list users@lists.quantum-espresso.org > https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusersdata=04%7C01%7Cxavbdlt%40uic.edu%7Cb283a1a22511428303b508d8bd3e127e%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637467421621058448%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=rX2etOAO2hyg7NB6ciuRHq0zJzJnPRIjQjTU4hyVGIs%3Dreserved=0 > > > ___ > 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 and QE6.7
Thank you, Giuseppe. I tried with ONCV SG15 PP library and it went fine. However I wasn't expecting PBE0 to be so much longer than PBE for the same calculation. I don't have yet the whole convergence curve w.r.t. cutoff, but with the same convergence parameters, same k-point grid... PBE0 is 300-400x slower than PBE for an SCF calculation. Is it the right order of magnitude? From: users on behalf of Giuseppe Mattioli Sent: Wednesday, January 20, 2021 6:22 AM To: users@lists.quantum-espresso.org Subject: Re: [QE-users] vc-relax with PBE0 and QE6.7 Dear Xavier Forces can be calculated in the case of PBE0 and other hybrid EXX functionals if you use norm-conserving (NC) pseudopotentials. No, there is no way to perform calculations without pseudopotentials in QE. HTH Giuseppe Quoting "Bidault, Xavier" : > Hello, > > I would like to run a geometry optimization (vc-relax) using PBE0 > and QE6.7. I got the following error message: > Error in routine setup (1): > forces for hybrid functionals + US/PAW not implemented > > > I guess this means I have to rely on another software, don't I? Or > is there a way to discard the need of PP with PBE0 and QE6.7? > > Thank you, > Xavier GIUSEPPE MATTIOLI CNR - ISTITUTO DI STRUTTURA DELLA MATERIA Via Salaria Km 29,300 - C.P. 10 I-00015 - Monterotondo Scalo (RM) Mob (*preferred*) +39 373 7305625 Tel + 39 06 90672342 - Fax +39 06 90672316 E-mail: ___ Quantum ESPRESSO is supported by MaX (https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2Fdata=04%7C01%7Cxavbdlt%40uic.edu%7Cb283a1a22511428303b508d8bd3e127e%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637467421621058448%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=nSRscevBNWTxUvuucpoteZlKTo5Ug%2BAIDVIuFATokZo%3Dreserved=0) users mailing list users@lists.quantum-espresso.org https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusersdata=04%7C01%7Cxavbdlt%40uic.edu%7Cb283a1a22511428303b508d8bd3e127e%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637467421621058448%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000sdata=rX2etOAO2hyg7NB6ciuRHq0zJzJnPRIjQjTU4hyVGIs%3Dreserved=0 ___ 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 and QE6.7
Dear Xavier Forces can be calculated in the case of PBE0 and other hybrid EXX functionals if you use norm-conserving (NC) pseudopotentials. No, there is no way to perform calculations without pseudopotentials in QE. HTH Giuseppe Quoting "Bidault, Xavier" : Hello, I would like to run a geometry optimization (vc-relax) using PBE0 and QE6.7. I got the following error message: Error in routine setup (1): forces for hybrid functionals + US/PAW not implemented I guess this means I have to rely on another software, don't I? Or is there a way to discard the need of PP with PBE0 and QE6.7? Thank you, Xavier GIUSEPPE MATTIOLI CNR - ISTITUTO DI STRUTTURA DELLA MATERIA Via Salaria Km 29,300 - C.P. 10 I-00015 - Monterotondo Scalo (RM) Mob (*preferred*) +39 373 7305625 Tel + 39 06 90672342 - Fax +39 06 90672316 E-mail: ___ 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