Re: [QE-users] pw.x Segmentation failed with libxc
It compiles and works normally, thanks again. -原始邮件- 发件人:"Paolo Giannozzi" 发送时间:2018-12-19 21:15:37 (星期三) 收件人: "Quantum Espresso users Forum" 抄送: 主题: Re: [QE-users] pw.x Segmentation failed with libxc That is? it "compiles and works" and "compiles but still does not work"? On Wed, Dec 19, 2018 at 2:14 PM 魏亚东 wrote: Thanks a lot, with reinstalling libxc-4.2.3, pw.x has been compiled normally. -Original Messages- From:"Paolo Giannozzi" Sent Time:2018-12-19 00:00:44 (Wednesday) To: "Quantum Espresso users Forum" Cc: Subject: Re: [QE-users] pw.x Segmentation failed with libxc On Tue, Dec 18, 2018 at 1:39 PM 魏亚东 wrote: forrtl: severe (174): SIGSEGV, segmentation fault occurred pw.x 009973E8 date_and_tim_ 23 date_and_tim.f90 It seems that libxcf90 were implemented in wrong way. maybe, but it is hard to see the link between compilation with libxc and a routine that just prints the date and time 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 ___ 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 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222 ___ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] SCF energy tolerance of GGA+U for Er2O3
Dear Lorenzo,My system is an output of optimized experimental primitive structure of Er2O3. Since I know it has a gap, I used a small smearing instead of occupation=fixed for better convergence of SCF. However, you know that using large smearing produces wrong results. I use your comment to change mixing mode. I know that SCF convergence in f-series with hubbard parameter is difficult. I will post new results. Regards David On Monday, December 17, 2018, 2:30:14 PM GMT+3:30, Lorenzo Paulatto wrote: You're using a very large value of U in a large system, with little symmetry, a very small smearing and no initial guess on the final state (starting_ns, mixing_fixed_eigenvalues, magnetization,..) it's going to take a lot of sweat, tears and possibly blood to converge it. I see no magic solution. Maybe try to use a different mixing_mode. Cheers -- Lorenzo Paulatto Written on a virtual keyboard with real fingers On Mon, 17 Dec 2018, 08:41 David Foster https://lists.quantum-espresso.org/mailman/listinfo/users ___ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] pw.x Segmentation failed with libxc
That is? it "compiles and works" and "compiles but still does not work"? On Wed, Dec 19, 2018 at 2:14 PM 魏亚东 wrote: > Thanks a lot, with reinstalling libxc-4.2.3, pw.x has been compiled > normally. > > > -Original Messages- > *From:*"Paolo Giannozzi" > *Sent Time:*2018-12-19 00:00:44 (Wednesday) > *To:* "Quantum Espresso users Forum" > *Cc:* > *Subject:* Re: [QE-users] pw.x Segmentation failed with libxc > > On Tue, Dec 18, 2018 at 1:39 PM 魏亚东 wrote: > > forrtl: severe (174): SIGSEGV, segmentation fault occurred >> >> pw.x 009973E8 date_and_tim_ 23 >> date_and_tim.f90 >> > It seems that libxcf90 were implemented in wrong way. >> > maybe, but it is hard to see the link between compilation with libxc and a > routine that just prints the date and time > > 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 > > ___ > 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 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222 ___ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] 答复: 答复: DFT+U can not converge
Thanks very much. Both cases I will try with and without saturation. 发件人: users 代表 Giuseppe Mattioli 发送时间: 2018年12月19日 21:11 收件人: users@lists.quantum-espresso.org 主题: Re: [QE-users] 答复: DFT+U can not converge Yes, I suppose so, but it is not easy to say exactly how it has to be done in this case (H or OH saturation? surface reconstruction?). If you are lucky, you may find something in literature... Best Giuseppe LEUNG Clarence ha scritto: > Dear Giuseppe, > > Thanks for your suggest. > > My structure is (3 1 0) surface of Co3(PO4)2 with 15 angstrom > vacuum. You mean that the unsaturated bond should be passivated by H > or something? > > LIANG Xiongyi > > 发件人: users 代表 Giuseppe > Mattioli > 发送时间: 2018年12月19日 19:30 > 收件人: users@lists.quantum-espresso.org > 主题: Re: [QE-users] DFT+U can not converge > > > Dear Liang Xiongyi > > In my opinion, your initial structure is possibly wrong or missing > necessary saturation (i.e. twofold-coordinated P atoms on the surface > and P=O free terminations, which are likely unstable in gas phase). > However, and generally speaking, if your calculation is going > somewhere not so far from convergence, you can overcome a difficult > initial step by using this very useful switch: > > scf_must_converge=.false. > > Coupled with a suitable electron_maxstep variable, it will calculate > forces and move atoms when scf ends. If the structure is meaningful > and not wrong (and this may not be the case) in a few steps you should > recover an acceptably smooth convergence. > HTH > Giuseppe > > LEUNG Clarence ha scritto: > >> Dear QEusers, >> >> When I use QE6.3 to perform DFT+U to relax a slab, it is very >> difficult to converge and even can not converge (more than 300 scf >> steps). Thanks for your suggestion in advances. >> >> The input file is: >> >> &CONTROL >> calculation='relax', >> nstep = 200 , >> wf_collect = .true. , >> etot_conv_thr = 3.5D-6 , >> forc_conv_thr = 4.0D-4, >> / >> >> &SYSTEM >> ibrav=0, >> celldm(1)=9.6431795847d0, >> nat=109, >> ntyp=3, >> ecutwfc=60, >> ecutrho=480, >> input_dft='PBE', >> occupations='smearing', >> smearing = 'gaussian' , >> degauss = 0.01 , >> nspin = 2 , >> starting_magnetization(1) = 0.5 , >> lda_plus_u = .true. , >> Hubbard_U(1) = 3.32 , >> vdw_corr = dft-d3 , >> / >> &ELECTRONS >> conv_thr=1d-05, >> mixing_beta=0.3d0, >> mixing_mode ='local-TF', >> electron_maxstep = 600 , >> / >> &IONS >> / >> &CELL >> / >> >> ATOMIC_SPECIES >> Co 58.933200d0 Co.pbe-spn-rrkjus_psl.0.3.1.UPF >> O 15.999400d0 O.pbe-n-rrkjus_psl.1.0.0.UPF >> P 30.973800d0 P.pbe-n-rrkjus_psl.1.0.0.UPF >> >> ATOMIC_POSITIONS {crystal} >> O0.175510868 0.823201091 0.0040419020 0 0 >> Co 0.752963028 0.655948399 0.0043135660 0 0 >> O0.301125723 0.416151961 0.0048799970 0 0 >> Co 0.846799898 0.775755648 0.0052751820 0 0 >> Co 0.346800351 0.275755641 0.0052752040 0 0 >> O0.392474471 0.135359341 0.0056702340 0 0 >> Co 0.940637166 0.895562903 0.006230 0 0 >> O0.518089326 0.728310212 0.0065083300 0 0 >> O0.902313840 0.059368885 0.0179593930 0 0 >> O0.807187857 0.964594997 0.0204575570 0 0 >> O0.036878250 0.705581722 0.0238905990 0 0 >> P0.893919682 0.538810250 0.0262428490 0 0 >> Co 0.451343782 0.481124665 0.0337500410 0 0 >> O0.028914218 0.314809876 0.0343233470 0 0 >> P0.323165643 0.700374526 0.0446558860 0 0 >> O0.412931350 0.644218586 0.0452437970 0 0 >> O0.180081232 0.532598515 0.0466851490 0 0 >> O0.551818853 0.324973894 0.0622826950 0 0 >> O0.318968869 0.213353805 0.0637240070 0 0 >> P0.408734429 0.157197863 0.0643119200 0 0 >> O0.702985853 0.542762514 0.0746444590 0 0 >> Co 0.280556437 0.376447725 0.0752177630 0 0 >> P0.837980537 0.318762141 0.0827249540 0 0 >> O0.695021832 0.151990660 0.0850772360 0 0 >> O0.924712372 0.892977386 0.0885102760 0 0 >> O0.829586379 0.798203506 0.0910084100 0 0 >> O0.213810893 0.129262179 0.1024594740 0 0 >> Co 0.791263053 0.962009487 0.1027311380 0 0 >> O0.339425748 0.722213050 0.1032975690 0 0 >> Co 0.885099923 0.081816736 0.1036927540 0 0 >> Co 0.385100377 0.581816730 0.1036927760 0 0 >> O0.430774496 0.441420429 0.1040878070 0 0 >> Co 0.978937191 0.201623991 0.1046542380 0 0 >> O0.556389351 0.034371300 0.1049259020 0 0 >> O0.940613865 0.365429972
Re: [QE-users] pw.x Segmentation failed with libxc
Thanks a lot, with reinstalling libxc-4.2.3, pw.x has been compiled normally. -Original Messages- From:"Paolo Giannozzi" Sent Time:2018-12-19 00:00:44 (Wednesday) To: "Quantum Espresso users Forum" Cc: Subject: Re: [QE-users] pw.x Segmentation failed with libxc On Tue, Dec 18, 2018 at 1:39 PM 魏亚东 wrote: forrtl: severe (174): SIGSEGV, segmentation fault occurred pw.x 009973E8 date_and_tim_ 23 date_and_tim.f90 It seems that libxcf90 were implemented in wrong way. maybe, but it is hard to see the link between compilation with libxc and a routine that just prints the date and time 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 ___ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] 答复: DFT+U can not converge
Yes, I suppose so, but it is not easy to say exactly how it has to be done in this case (H or OH saturation? surface reconstruction?). If you are lucky, you may find something in literature... Best Giuseppe LEUNG Clarence ha scritto: Dear Giuseppe, Thanks for your suggest. My structure is (3 1 0) surface of Co3(PO4)2 with 15 angstrom vacuum. You mean that the unsaturated bond should be passivated by H or something? LIANG Xiongyi 发件人: users 代表 Giuseppe Mattioli 发送时间: 2018年12月19日 19:30 收件人: users@lists.quantum-espresso.org 主题: Re: [QE-users] DFT+U can not converge Dear Liang Xiongyi In my opinion, your initial structure is possibly wrong or missing necessary saturation (i.e. twofold-coordinated P atoms on the surface and P=O free terminations, which are likely unstable in gas phase). However, and generally speaking, if your calculation is going somewhere not so far from convergence, you can overcome a difficult initial step by using this very useful switch: scf_must_converge=.false. Coupled with a suitable electron_maxstep variable, it will calculate forces and move atoms when scf ends. If the structure is meaningful and not wrong (and this may not be the case) in a few steps you should recover an acceptably smooth convergence. HTH Giuseppe LEUNG Clarence ha scritto: Dear QEusers, When I use QE6.3 to perform DFT+U to relax a slab, it is very difficult to converge and even can not converge (more than 300 scf steps). Thanks for your suggestion in advances. The input file is: &CONTROL calculation='relax', nstep = 200 , wf_collect = .true. , etot_conv_thr = 3.5D-6 , forc_conv_thr = 4.0D-4, / &SYSTEM ibrav=0, celldm(1)=9.6431795847d0, nat=109, ntyp=3, ecutwfc=60, ecutrho=480, input_dft='PBE', occupations='smearing', smearing = 'gaussian' , degauss = 0.01 , nspin = 2 , starting_magnetization(1) = 0.5 , lda_plus_u = .true. , Hubbard_U(1) = 3.32 , vdw_corr = dft-d3 , / &ELECTRONS conv_thr=1d-05, mixing_beta=0.3d0, mixing_mode ='local-TF', electron_maxstep = 600 , / &IONS / &CELL / ATOMIC_SPECIES Co 58.933200d0 Co.pbe-spn-rrkjus_psl.0.3.1.UPF O 15.999400d0 O.pbe-n-rrkjus_psl.1.0.0.UPF P 30.973800d0 P.pbe-n-rrkjus_psl.1.0.0.UPF ATOMIC_POSITIONS {crystal} O0.175510868 0.823201091 0.0040419020 0 0 Co 0.752963028 0.655948399 0.0043135660 0 0 O0.301125723 0.416151961 0.0048799970 0 0 Co 0.846799898 0.775755648 0.0052751820 0 0 Co 0.346800351 0.275755641 0.0052752040 0 0 O0.392474471 0.135359341 0.0056702340 0 0 Co 0.940637166 0.895562903 0.006230 0 0 O0.518089326 0.728310212 0.0065083300 0 0 O0.902313840 0.059368885 0.0179593930 0 0 O0.807187857 0.964594997 0.0204575570 0 0 O0.036878250 0.705581722 0.0238905990 0 0 P0.893919682 0.538810250 0.0262428490 0 0 Co 0.451343782 0.481124665 0.0337500410 0 0 O0.028914218 0.314809876 0.0343233470 0 0 P0.323165643 0.700374526 0.0446558860 0 0 O0.412931350 0.644218586 0.0452437970 0 0 O0.180081232 0.532598515 0.0466851490 0 0 O0.551818853 0.324973894 0.0622826950 0 0 O0.318968869 0.213353805 0.0637240070 0 0 P0.408734429 0.157197863 0.0643119200 0 0 O0.702985853 0.542762514 0.0746444590 0 0 Co 0.280556437 0.376447725 0.0752177630 0 0 P0.837980537 0.318762141 0.0827249540 0 0 O0.695021832 0.151990660 0.0850772360 0 0 O0.924712372 0.892977386 0.0885102760 0 0 O0.829586379 0.798203506 0.0910084100 0 0 O0.213810893 0.129262179 0.1024594740 0 0 Co 0.791263053 0.962009487 0.1027311380 0 0 O0.339425748 0.722213050 0.1032975690 0 0 Co 0.885099923 0.081816736 0.1036927540 0 0 Co 0.385100377 0.581816730 0.1036927760 0 0 O0.430774496 0.441420429 0.1040878070 0 0 Co 0.978937191 0.201623991 0.1046542380 0 0 O0.556389351 0.034371300 0.1049259020 0 0 O0.940613865 0.365429972 0.1163769650 0 0 O0.845487883 0.270656085 0.1188751290 0 0 O0.075178275 0.011642810 0.1223081710 0 0 P0.932219708 0.844871338 0.1246604210 0 0 Co 0.489643808 0.787185753 0.1321676130 0 0 O0.067214244 0.620870964 0.1327409190 0 0 P0.361354279 0.006083789 0.142115621 O0.446802131 0.950135511 0.145110715 O0.214381915 0.837791409 0.146887224 O0.593110221 0.629859448 0.161855388 O0.
[QE-users] 答复: DFT+U can not converge
Dear Giuseppe, Thanks for your suggest. My structure is (3 1 0) surface of Co3(PO4)2 with 15 angstrom vacuum. You mean that the unsaturated bond should be passivated by H or something? LIANG Xiongyi 发件人: users 代表 Giuseppe Mattioli 发送时间: 2018年12月19日 19:30 收件人: users@lists.quantum-espresso.org 主题: Re: [QE-users] DFT+U can not converge Dear Liang Xiongyi In my opinion, your initial structure is possibly wrong or missing necessary saturation (i.e. twofold-coordinated P atoms on the surface and P=O free terminations, which are likely unstable in gas phase). However, and generally speaking, if your calculation is going somewhere not so far from convergence, you can overcome a difficult initial step by using this very useful switch: scf_must_converge=.false. Coupled with a suitable electron_maxstep variable, it will calculate forces and move atoms when scf ends. If the structure is meaningful and not wrong (and this may not be the case) in a few steps you should recover an acceptably smooth convergence. HTH Giuseppe LEUNG Clarence ha scritto: > Dear QEusers, > > When I use QE6.3 to perform DFT+U to relax a slab, it is very > difficult to converge and even can not converge (more than 300 scf > steps). Thanks for your suggestion in advances. > > The input file is: > > &CONTROL > calculation='relax', > nstep = 200 , > wf_collect = .true. , > etot_conv_thr = 3.5D-6 , > forc_conv_thr = 4.0D-4, > / > > &SYSTEM > ibrav=0, > celldm(1)=9.6431795847d0, > nat=109, > ntyp=3, > ecutwfc=60, > ecutrho=480, > input_dft='PBE', > occupations='smearing', > smearing = 'gaussian' , > degauss = 0.01 , > nspin = 2 , > starting_magnetization(1) = 0.5 , > lda_plus_u = .true. , > Hubbard_U(1) = 3.32 , > vdw_corr = dft-d3 , > / > &ELECTRONS > conv_thr=1d-05, > mixing_beta=0.3d0, > mixing_mode ='local-TF', > electron_maxstep = 600 , > / > &IONS > / > &CELL > / > > ATOMIC_SPECIES > Co 58.933200d0 Co.pbe-spn-rrkjus_psl.0.3.1.UPF > O 15.999400d0 O.pbe-n-rrkjus_psl.1.0.0.UPF > P 30.973800d0 P.pbe-n-rrkjus_psl.1.0.0.UPF > > ATOMIC_POSITIONS {crystal} > O0.175510868 0.823201091 0.0040419020 0 0 > Co 0.752963028 0.655948399 0.0043135660 0 0 > O0.301125723 0.416151961 0.0048799970 0 0 > Co 0.846799898 0.775755648 0.0052751820 0 0 > Co 0.346800351 0.275755641 0.0052752040 0 0 > O0.392474471 0.135359341 0.0056702340 0 0 > Co 0.940637166 0.895562903 0.006230 0 0 > O0.518089326 0.728310212 0.0065083300 0 0 > O0.902313840 0.059368885 0.0179593930 0 0 > O0.807187857 0.964594997 0.0204575570 0 0 > O0.036878250 0.705581722 0.0238905990 0 0 > P0.893919682 0.538810250 0.0262428490 0 0 > Co 0.451343782 0.481124665 0.0337500410 0 0 > O0.028914218 0.314809876 0.0343233470 0 0 > P0.323165643 0.700374526 0.0446558860 0 0 > O0.412931350 0.644218586 0.0452437970 0 0 > O0.180081232 0.532598515 0.0466851490 0 0 > O0.551818853 0.324973894 0.0622826950 0 0 > O0.318968869 0.213353805 0.0637240070 0 0 > P0.408734429 0.157197863 0.0643119200 0 0 > O0.702985853 0.542762514 0.0746444590 0 0 > Co 0.280556437 0.376447725 0.0752177630 0 0 > P0.837980537 0.318762141 0.0827249540 0 0 > O0.695021832 0.151990660 0.0850772360 0 0 > O0.924712372 0.892977386 0.0885102760 0 0 > O0.829586379 0.798203506 0.0910084100 0 0 > O0.213810893 0.129262179 0.1024594740 0 0 > Co 0.791263053 0.962009487 0.1027311380 0 0 > O0.339425748 0.722213050 0.1032975690 0 0 > Co 0.885099923 0.081816736 0.1036927540 0 0 > Co 0.385100377 0.581816730 0.1036927760 0 0 > O0.430774496 0.441420429 0.1040878070 0 0 > Co 0.978937191 0.201623991 0.1046542380 0 0 > O0.556389351 0.034371300 0.1049259020 0 0 > O0.940613865 0.365429972 0.1163769650 0 0 > O0.845487883 0.270656085 0.1188751290 0 0 > O0.075178275 0.011642810 0.1223081710 0 0 > P0.932219708 0.844871338 0.1246604210 0 0 > Co 0.489643808 0.787185753 0.1321676130 0 0 > O0.067214244 0.620870964 0.1327409190 0 0 > P0.361354279 0.006083789 0.142115621 > O0.446802131 0.950135511 0.145110715 > O0.214381915 0.837791409 0.146887224 > O0.593110221 0.629859448 0.161855388 > O0.355569602 0.519337757 0.162613913 > P0.444333889 0.463
Re: [QE-users] DFT+U can not converge
Dear Liang Xiongyi In my opinion, your initial structure is possibly wrong or missing necessary saturation (i.e. twofold-coordinated P atoms on the surface and P=O free terminations, which are likely unstable in gas phase). However, and generally speaking, if your calculation is going somewhere not so far from convergence, you can overcome a difficult initial step by using this very useful switch: scf_must_converge=.false. Coupled with a suitable electron_maxstep variable, it will calculate forces and move atoms when scf ends. If the structure is meaningful and not wrong (and this may not be the case) in a few steps you should recover an acceptably smooth convergence. HTH Giuseppe LEUNG Clarence ha scritto: Dear QEusers, When I use QE6.3 to perform DFT+U to relax a slab, it is very difficult to converge and even can not converge (more than 300 scf steps). Thanks for your suggestion in advances. The input file is: &CONTROL calculation='relax', nstep = 200 , wf_collect = .true. , etot_conv_thr = 3.5D-6 , forc_conv_thr = 4.0D-4, / &SYSTEM ibrav=0, celldm(1)=9.6431795847d0, nat=109, ntyp=3, ecutwfc=60, ecutrho=480, input_dft='PBE', occupations='smearing', smearing = 'gaussian' , degauss = 0.01 , nspin = 2 , starting_magnetization(1) = 0.5 , lda_plus_u = .true. , Hubbard_U(1) = 3.32 , vdw_corr = dft-d3 , / &ELECTRONS conv_thr=1d-05, mixing_beta=0.3d0, mixing_mode ='local-TF', electron_maxstep = 600 , / &IONS / &CELL / ATOMIC_SPECIES Co 58.933200d0 Co.pbe-spn-rrkjus_psl.0.3.1.UPF O 15.999400d0 O.pbe-n-rrkjus_psl.1.0.0.UPF P 30.973800d0 P.pbe-n-rrkjus_psl.1.0.0.UPF ATOMIC_POSITIONS {crystal} O0.175510868 0.823201091 0.0040419020 0 0 Co 0.752963028 0.655948399 0.0043135660 0 0 O0.301125723 0.416151961 0.0048799970 0 0 Co 0.846799898 0.775755648 0.0052751820 0 0 Co 0.346800351 0.275755641 0.0052752040 0 0 O0.392474471 0.135359341 0.0056702340 0 0 Co 0.940637166 0.895562903 0.006230 0 0 O0.518089326 0.728310212 0.0065083300 0 0 O0.902313840 0.059368885 0.0179593930 0 0 O0.807187857 0.964594997 0.0204575570 0 0 O0.036878250 0.705581722 0.0238905990 0 0 P0.893919682 0.538810250 0.0262428490 0 0 Co 0.451343782 0.481124665 0.0337500410 0 0 O0.028914218 0.314809876 0.0343233470 0 0 P0.323165643 0.700374526 0.0446558860 0 0 O0.412931350 0.644218586 0.0452437970 0 0 O0.180081232 0.532598515 0.0466851490 0 0 O0.551818853 0.324973894 0.0622826950 0 0 O0.318968869 0.213353805 0.0637240070 0 0 P0.408734429 0.157197863 0.0643119200 0 0 O0.702985853 0.542762514 0.0746444590 0 0 Co 0.280556437 0.376447725 0.0752177630 0 0 P0.837980537 0.318762141 0.0827249540 0 0 O0.695021832 0.151990660 0.0850772360 0 0 O0.924712372 0.892977386 0.0885102760 0 0 O0.829586379 0.798203506 0.0910084100 0 0 O0.213810893 0.129262179 0.1024594740 0 0 Co 0.791263053 0.962009487 0.1027311380 0 0 O0.339425748 0.722213050 0.1032975690 0 0 Co 0.885099923 0.081816736 0.1036927540 0 0 Co 0.385100377 0.581816730 0.1036927760 0 0 O0.430774496 0.441420429 0.1040878070 0 0 Co 0.978937191 0.201623991 0.1046542380 0 0 O0.556389351 0.034371300 0.1049259020 0 0 O0.940613865 0.365429972 0.1163769650 0 0 O0.845487883 0.270656085 0.1188751290 0 0 O0.075178275 0.011642810 0.1223081710 0 0 P0.932219708 0.844871338 0.1246604210 0 0 Co 0.489643808 0.787185753 0.1321676130 0 0 O0.067214244 0.620870964 0.1327409190 0 0 P0.361354279 0.006083789 0.142115621 O0.446802131 0.950135511 0.145110715 O0.214381915 0.837791409 0.146887224 O0.593110221 0.629859448 0.161855388 O0.355569602 0.519337757 0.162613913 P0.444333889 0.463046511 0.162481449 O0.733915313 0.849462282 0.171314557 Co 0.331397450 0.684149370 0.17454 P0.880835007 0.624048906 0.181575275 O0.729016571 0.457586972 0.184340848 O0.943694570 0.199413194 0.185888622 O0.842699539 0.103890146 0.188690045 O0.246376254 0.435621465 0.201025875 Co 0.819907469 0.267437250 0.200281311 O0.371654672 0.028624679 0.201597106 Co 0.919293748 0.386217687 0.198228118 Co 0.408991643 0.8
Re: [QE-users] Unexpected adsorption energy
Dear Mohamed I'm afraid that I cannot help you more than this without seeing your input files. If you want, you can send them in private. HTH Giuseppe Mohamed Safy ha scritto: Dear Giuseppe I took your advice but the E[cation+anion] system was very difficult to be converged. I tried to reduce the mixing_beta and conv_thr with no results. Thanks On Sun, 9 Dec 2018 at 14:37, Mohamed Safy wrote: Dear Giuseppe Many thanks for your response. I will try with your advice and give you the response Thanks On Thu, 6 Dec 2018 at 12:24, Giuseppe Mattioli < giuseppe.matti...@ism.cnr.it> wrote: Dear Mohamed charged ions are tricky in DFT for multiple reasons. The excess and well localized charge can suffer a very strong delocalization error which may lead to unbound electronic states. Moreover, charged ions are generally not stable in gas phase. They often require a polar solvent to exist. Finally, in periodic boundary conditions a distribution of charge (aka "jellium") is required to compensate the positive/negative charge in a supercell, and the reference potential is affected by the insertion of such charge, so that for example you cannot calculate the ionization energy of a molecule as E[q=+1]-E[q=0], as you do when you use GTO codes. This said, it is not impossible to calculate the adsorption energy of charged ions on a given substrate, provided that: 1) You use a hybrid EXX-GGA functional. This is not mandatory, but it is recommended because it generally avoids the accommodation of excess electrons in unbound states. 2) You embed your system in an implicit dielectric medium (maybe "water", in your case). In QE this is easily provided by the QUANTUM-ENVIRON plug-in. Then you can calculate the adsorption energy in two ways: A) you can start from the interacting configuration of your system and progressively remove the ion in several snapshots (or a few snapshot, depending on the computational resources you can afford). Then you build an interaction potential curve and yiu try to estimate its asymptotic value. It requires also a large supercell, of course. B) you can use a little trick (which however requires 1 and 2 above). Put a cation in a part of your supercell where the interaction energy with you polymer+anion system is negligible. Then calculate the energy of your cation+anion system in a neutral supercell where their charge is exactly compensated. The energy difference between three neutral supercells E[polymer+anion+cation]-E[polymer]-E[cation+anion] should be a sensible estimate of the anion adsorption energy HTH Giuseppe Mohamed Safy ha scritto: > Thanks for your valuable information but I have experimental results which > indicate the presence of adsorption. is this can be considered a > conflict?. I tried to validate the method using a smaller system. I > studied the adsorption of H2 on Graphene. > The adsorption energy was 17.17 kcal/mol. > the systems are below > Complex > &CONTROL > calculation = "scf" > forc_conv_thr = 1.0e-03 > max_seconds = 1.72800e+05 > nstep = 1000 > verbosity='high' > restart_mode='from_scratch' > iprint=1 > tprnfor=.true. > pseudo_dir = '/lfs01/workdirs/val/Test/pseudo', > outdir='/lfs01/workdirs/val/Test/Out/C', > / > > &SYSTEM > a = 7.40525e+00 > c = 9.99906e+00 > ibrav = 4 > nat = 19 > ntyp = 2 > ecutwfc = 45.0 , > ecutrho = 450.0 , > input_DFT = 'PBE-D2' , > occupations = 'smearing' , > degauss = 1.0d-4 , > vdw_corr = 'Grimme-D2' > assume_isolated = 'mt' > smearing = 'marzari-vanderbilt' , > / > > &ELECTRONS > conv_thr = 1.0d-7 , > electron_maxstep = 1000 > mixing_mode = 'plain' , > mixing_beta = 0.3d0 , > / > > &IONS > ion_dynamics='bfgs' > upscale=20.0 > / > > &CELL > / > > K_POINTS {automatic} > 3 3 3 0 0 0 > > ATOMIC_SPECIES > C 12.01070 C.pbe-n-kjpaw_psl.1.0.0.UPF > H 1.00794 H.pbe-kjpaw_psl.1.0.0.UPF > ATOMIC_POSITIONS {angstrom} > C1.280642168 0.685951341 -0.000431048 > C -1.236653977 3.539880413 -0.001566184 > C -0.000377617 2.903279130 -0.002911997 > C -2.489554615 5.710262290 -0.000852594 > C -1.229721248 4.990709007 -0.000338911 > C2.449440629 1.438897112 0.002319254 > C3.702198081 0.707454065 -0.001265064 > C1.236237242 3.539760579 0.000958837 > C2.478517989 2.856386275 0.004841971 > C -0.000246038 5.700684987 -0.000997560 > C1.229347070 4.990770096 -0.000716604 > C4.955272233 1.438694069 -0.002138838 > C6.124721243 0.686321393 0.000987763 > C3.702044434 3.562937903 0.001926384 > C4.925831271 2.856536000 -0.00173 > C2.489209922 5.710445901 -0.000342579 > C3.702309214 4.976078918 -0.48704 > H3.360489134 2.350036356 -3.014528460 > H2.719672863 2.741584163 -3.037540110 > > > Graphen > &CONTROL > calculation = "scf" >
[QE-users] DFT+U can not converge
Dear QEusers, When I use QE6.3 to perform DFT+U to relax a slab, it is very difficult to converge and even can not converge (more than 300 scf steps). Thanks for your suggestion in advances. The input file is: &CONTROL calculation='relax', nstep = 200 , wf_collect = .true. , etot_conv_thr = 3.5D-6 , forc_conv_thr = 4.0D-4, / &SYSTEM ibrav=0, celldm(1)=9.6431795847d0, nat=109, ntyp=3, ecutwfc=60, ecutrho=480, input_dft='PBE', occupations='smearing', smearing = 'gaussian' , degauss = 0.01 , nspin = 2 , starting_magnetization(1) = 0.5 , lda_plus_u = .true. , Hubbard_U(1) = 3.32 , vdw_corr = dft-d3 , / &ELECTRONS conv_thr=1d-05, mixing_beta=0.3d0, mixing_mode ='local-TF', electron_maxstep = 600 , / &IONS / &CELL / ATOMIC_SPECIES Co 58.933200d0 Co.pbe-spn-rrkjus_psl.0.3.1.UPF O 15.999400d0 O.pbe-n-rrkjus_psl.1.0.0.UPF P 30.973800d0 P.pbe-n-rrkjus_psl.1.0.0.UPF ATOMIC_POSITIONS {crystal} O0.175510868 0.823201091 0.0040419020 0 0 Co 0.752963028 0.655948399 0.0043135660 0 0 O0.301125723 0.416151961 0.0048799970 0 0 Co 0.846799898 0.775755648 0.0052751820 0 0 Co 0.346800351 0.275755641 0.0052752040 0 0 O0.392474471 0.135359341 0.0056702340 0 0 Co 0.940637166 0.895562903 0.006230 0 0 O0.518089326 0.728310212 0.0065083300 0 0 O0.902313840 0.059368885 0.0179593930 0 0 O0.807187857 0.964594997 0.0204575570 0 0 O0.036878250 0.705581722 0.0238905990 0 0 P0.893919682 0.538810250 0.0262428490 0 0 Co 0.451343782 0.481124665 0.0337500410 0 0 O0.028914218 0.314809876 0.0343233470 0 0 P0.323165643 0.700374526 0.0446558860 0 0 O0.412931350 0.644218586 0.0452437970 0 0 O0.180081232 0.532598515 0.0466851490 0 0 O0.551818853 0.324973894 0.0622826950 0 0 O0.318968869 0.213353805 0.0637240070 0 0 P0.408734429 0.157197863 0.0643119200 0 0 O0.702985853 0.542762514 0.0746444590 0 0 Co 0.280556437 0.376447725 0.0752177630 0 0 P0.837980537 0.318762141 0.0827249540 0 0 O0.695021832 0.151990660 0.0850772360 0 0 O0.924712372 0.892977386 0.0885102760 0 0 O0.829586379 0.798203506 0.0910084100 0 0 O0.213810893 0.129262179 0.1024594740 0 0 Co 0.791263053 0.962009487 0.1027311380 0 0 O0.339425748 0.722213050 0.1032975690 0 0 Co 0.885099923 0.081816736 0.1036927540 0 0 Co 0.385100377 0.581816730 0.1036927760 0 0 O0.430774496 0.441420429 0.1040878070 0 0 Co 0.978937191 0.201623991 0.1046542380 0 0 O0.556389351 0.034371300 0.1049259020 0 0 O0.940613865 0.365429972 0.1163769650 0 0 O0.845487883 0.270656085 0.1188751290 0 0 O0.075178275 0.011642810 0.1223081710 0 0 P0.932219708 0.844871338 0.1246604210 0 0 Co 0.489643808 0.787185753 0.1321676130 0 0 O0.067214244 0.620870964 0.1327409190 0 0 P0.361354279 0.006083789 0.142115621 O0.446802131 0.950135511 0.145110715 O0.214381915 0.837791409 0.146887224 O0.593110221 0.629859448 0.161855388 O0.355569602 0.519337757 0.162613913 P0.444333889 0.463046511 0.162481449 O0.733915313 0.849462282 0.171314557 Co 0.331397450 0.684149370 0.17454 P0.880835007 0.624048906 0.181575275 O0.729016571 0.457586972 0.184340848 O0.943694570 0.199413194 0.185888622 O0.842699539 0.103890146 0.188690045 O0.246376254 0.435621465 0.201025875 Co 0.819907469 0.267437250 0.200281311 O0.371654672 0.028624679 0.201597106 Co 0.919293748 0.386217687 0.198228118 Co 0.408991643 0.881622266 0.208931455 O0.502982285 0.747275057 0.201700410 Co 0.015942502 0.506736170 0.203798000 O0.581988259 0.338815157 0.205761317 O0.989082264 0.670263393 0.214923633 O0.891588821 0.575388161 0.217463006 O0.102116329 0.316952504 0.225311380 P0.946018990 0.150667428 0.220957196 Co 0.509889754 0.092572883 0.231027714 O0.087929910 0.910622245 0.254736182 P0.390071467 0.310008795 0.243522265 O0.475452219 0.253683339 0.241349488 O0.235229551 0.143749020 0.238654386 O0.612387032 0.940393783 0.254302766 O0.441611912 0.82770 0.264038723 P0.540200975 0.770461487 0.2603