[Pw_forum] problems with converge of geometric optimization
-0.09160.00016756 > atom 47 type 2 force = 0.00368611 -0.27950.00045003 > atom 48 type 1 force = 0.00031498 -0.00033410 -0.2164 > atom 49 type 1 force =-0.000720470.000634530.00013858 > atom 50 type 1 force = 0.000548500.000291280.00078326 > atom 51 type 1 force =-0.000189020.000619440.00105489 > atom 52 type 2 force =-0.000434750.28820.6407 > atom 53 type 2 force =-0.000567090.000120750.00041822 > atom 54 type 2 force =-0.000567410.000247150.00023894 > atom 55 type 2 force =-0.000412800.000596750.00160623 > atom 56 type 2 force = 0.000146510.000289600.00159954 > atom 57 type 2 force =-0.001714500.20960.00247953 > atom 58 type 2 force =-0.000171830.000857810.7596 > atom 59 type 2 force =-0.001026800.00054910 -0.00040500 > atom 60 type 1 force =-0.93790.00014378 -0.00127470 > atom 61 type 1 force = 0.00012531 -0.00025665 -0.00112033 > atom 62 type 1 force = 0.00080427 -0.000565580.00101772 > atom 63 type 1 force =-0.004982840.00033298 -0.00350540 > atom 64 type 2 force =-0.000868300.000663740.00055719 > atom 65 type 2 force =-0.00073070 -0.000576560.00125530 > atom 66 type 2 force =-0.00083096 -0.00034313 -0.00035115 > atom 67 type 2 force = 0.00106363 -0.00040306 -0.00077587 > atom 68 type 2 force = 0.000719540.00035647 -0.00132880 > atom 69 type 2 force = 0.000171700.00042807 -0.00121455 > atom 70 type 2 force =-0.00025769 -0.003956020.00174604 > atom 71 type 2 force =-0.00129773 -0.000719310.00039821 > atom 72 type 1 force =-0.00301732 -0.000652510.00435206 > atom 73 type 1 force = 0.003294640.000300460.00470849 > atom 74 type 1 force = 0.00484552 -0.00038429 -0.00313326 > atom 75 type 1 force = 0.00022645 -0.000125450.00117296 > atom 76 type 2 force =-0.000377190.00045763 -0.00319435 > atom 77 type 2 force =-0.000408140.88390.5440 > atom 78 type 2 force = 0.00033602 -0.19900.00046392 > atom 79 type 2 force = 0.00028389 -0.000673850.00205674 > atom 80 type 2 force =-0.40790.9821 -0.00339820 > atom 81 type 2 force =-0.00030699 -0.000387100.00080237 > atom 82 type 2 force = 0.8637 -0.00026351 -0.8527 > atom 83 type 2 force =-0.001563330.000234330.00055603 > atom 84 type 1 force =-0.00025828 -0.45070.00074608 > atom 85 type 1 force = 0.00057079 -0.00060726 -0.00014710 > atom 86 type 1 force =-0.00011808 -0.50270.00081874 > atom 87 type 1 force =-0.002822700.00131933 -0.00472331 > atom 88 type 2 force =-0.001587200.000675370.00032135 > atom 89 type 2 force =-0.00039389 -0.001117750.5539 > atom 90 type 2 force =-0.00056433 -0.5048 -0.00167901 > atom 91 type 2 force =-0.000148240.3539 -0.00627238 > atom 92 type 2 force =-0.002053340.00037334 -0.00044205 > atom 93 type 2 force =-0.000256930.004721850.00131091 > atom 94 type 2 force =-0.003663860.000125750.00058219 > atom 95 type 1 force =-0.00339605 -0.000779840.00213525 > atom 96 type 1 force = 0.004300430.000369770.00195456 > atom 97 type 1 force = 0.00347151 -0.00182346 -0.00493084 > Total force = 0.025508 Total SCF correction = 0.000128 > > Stephan Rix > Institute for Inorganic und Analytical Chemistry > Johannes Gutenberg-University Mainz, Germany > > > Date: Wed, 15 Sep 2010 17:10:07 +0200 > From: Stefano de Gironcoli > Subject: Re: [Pw_forum] problems with converge of geometric > optimization > To: PWSCF Forum > Message-ID: <4C90E1CF.6040400 at sissa.it> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > This happens when the energy landscape become noisy compared with the > expected energy decrease in the relaxation steps. > This can happen when you are very close to the minimum. > How large are your forces ? How tight is your force/energy threshold ? > stefano > > Stephan Rix wrote: > >> I am trying to calculate
[Pw_forum] problems with converge of geometric optimization
565580.00101772 atom 63 type 1 force =-0.004982840.00033298 -0.00350540 atom 64 type 2 force =-0.000868300.000663740.00055719 atom 65 type 2 force =-0.00073070 -0.000576560.00125530 atom 66 type 2 force =-0.00083096 -0.00034313 -0.00035115 atom 67 type 2 force = 0.00106363 -0.00040306 -0.00077587 atom 68 type 2 force = 0.000719540.00035647 -0.00132880 atom 69 type 2 force = 0.000171700.00042807 -0.00121455 atom 70 type 2 force =-0.00025769 -0.003956020.00174604 atom 71 type 2 force =-0.00129773 -0.000719310.00039821 atom 72 type 1 force =-0.00301732 -0.000652510.00435206 atom 73 type 1 force = 0.003294640.000300460.00470849 atom 74 type 1 force = 0.00484552 -0.00038429 -0.00313326 atom 75 type 1 force = 0.00022645 -0.000125450.00117296 atom 76 type 2 force =-0.000377190.00045763 -0.00319435 atom 77 type 2 force =-0.000408140.88390.5440 atom 78 type 2 force = 0.00033602 -0.19900.00046392 atom 79 type 2 force = 0.00028389 -0.000673850.00205674 atom 80 type 2 force =-0.40790.9821 -0.00339820 atom 81 type 2 force =-0.00030699 -0.000387100.00080237 atom 82 type 2 force = 0.8637 -0.00026351 -0.8527 atom 83 type 2 force =-0.001563330.000234330.00055603 atom 84 type 1 force =-0.00025828 -0.45070.00074608 atom 85 type 1 force = 0.00057079 -0.00060726 -0.00014710 atom 86 type 1 force =-0.00011808 -0.50270.00081874 atom 87 type 1 force =-0.002822700.00131933 -0.00472331 atom 88 type 2 force =-0.001587200.000675370.00032135 atom 89 type 2 force =-0.00039389 -0.001117750.5539 atom 90 type 2 force =-0.00056433 -0.5048 -0.00167901 atom 91 type 2 force =-0.000148240.3539 -0.00627238 atom 92 type 2 force =-0.002053340.00037334 -0.00044205 atom 93 type 2 force =-0.000256930.004721850.00131091 atom 94 type 2 force =-0.003663860.000125750.00058219 atom 95 type 1 force =-0.00339605 -0.000779840.00213525 atom 96 type 1 force = 0.004300430.000369770.00195456 atom 97 type 1 force = 0.00347151 -0.00182346 -0.00493084 Total force = 0.025508 Total SCF correction = 0.000128 Stephan Rix Institute for Inorganic und Analytical Chemistry Johannes Gutenberg-University Mainz, Germany Date: Wed, 15 Sep 2010 17:10:07 +0200 From: Stefano de Gironcoli Subject: Re: [Pw_forum] problems with converge of geometric optimization To: PWSCF Forum Message-ID: <4C90E1CF.6040400 at sissa.it> Content-Type: text/plain; charset=ISO-8859-1; format=flowed This happens when the energy landscape become noisy compared with the expected energy decrease in the relaxation steps. This can happen when you are very close to the minimum. How large are your forces ? How tight is your force/energy threshold ? stefano Stephan Rix wrote: > I am trying to calculate an H-center in CaF2, and I am having trouble with > the geometry optimization. > I have tried adjusting various parameters such as ion dynamics, trust > radius, convergence tresholds, but it seems that convergence cannot be > improved. > > Using bfgs I always get the following error: > > %% > from bfgs : error # 1 > bfgs history already reset at previous step > %% > > Alternatively the electonic convergence gets extremely slow. > > In addition, no matter what parameters I adjust the total forces in the > system never converge, e.g. > Total force = 0.026394 Total SCF correction = 0.000314 > Total force = 0.018379 Total SCF correction = 0.000282 > Total force = 0.018824 Total SCF correction = 0.000120 > Total force = 0.018857 Total SCF correction = 0.000155 > Total force = 0.020215 Total SCF correction = 0.000193 > Total force = 0.058765 Total SCF correction = 0.000241 > Total force = 0.016998 Total SCF correction = 0.000205 > Total force = 0.042967 Total SCF correction = 0.000238 > Total force = 0.106226 Total SCF correction = 0.73 > Total force = 0.137142 Total SCF correction = 0.71 > Total force = 0.109549 Total SCF correction = 0.000108 > Total force = 0.068723 Total SCF correction = 0.000103 > Total force =
[Pw_forum] problems with converge of geometric optimization
This happens when the energy landscape become noisy compared with the expected energy decrease in the relaxation steps. This can happen when you are very close to the minimum. How large are your forces ? How tight is your force/energy threshold ? stefano Stephan Rix wrote: > I am trying to calculate an H-center in CaF2, and I am having trouble with > the geometry optimization. > I have tried adjusting various parameters such as ion dynamics, trust > radius, convergence tresholds, but it seems that convergence cannot be > improved. > > Using bfgs I always get the following error: > > %% > from bfgs : error # 1 > bfgs history already reset at previous step > %% > > Alternatively the electonic convergence gets extremely slow. > > In addition, no matter what parameters I adjust the total forces in the > system never converge, e.g. > Total force = 0.026394 Total SCF correction = 0.000314 > Total force = 0.018379 Total SCF correction = 0.000282 > Total force = 0.018824 Total SCF correction = 0.000120 > Total force = 0.018857 Total SCF correction = 0.000155 > Total force = 0.020215 Total SCF correction = 0.000193 > Total force = 0.058765 Total SCF correction = 0.000241 > Total force = 0.016998 Total SCF correction = 0.000205 > Total force = 0.042967 Total SCF correction = 0.000238 > Total force = 0.106226 Total SCF correction = 0.73 > Total force = 0.137142 Total SCF correction = 0.71 > Total force = 0.109549 Total SCF correction = 0.000108 > Total force = 0.068723 Total SCF correction = 0.000103 > Total force = 0.076599 Total SCF correction = 0.000168 > Total force = 0.033233 Total SCF correction = 0.000342 > Total force = 0.026883 Total SCF correction = 0.000248 > Total force = 0.025846 Total SCF correction = 0.000378 > Total force = 0.025596 Total SCF correction = 0.000225 > Total force = 0.025482 Total SCF correction = 0.000275 > Total force = 0.025508 Total SCF correction = 0.000128 > > I am thankful for any suggestions. > > Stephan Rix > Institute for Inorganic und Analytical Chemistry > Johannes Gutenberg-University Mainz, Germany > > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum >
[Pw_forum] problems with converge of geometric optimization
I am trying to calculate an H-center in CaF2, and I am having trouble with the geometry optimization. I have tried adjusting various parameters such as ion dynamics, trust radius, convergence tresholds, but it seems that convergence cannot be improved. Using bfgs I always get the following error: %% from bfgs : error # 1 bfgs history already reset at previous step %% Alternatively the electonic convergence gets extremely slow. In addition, no matter what parameters I adjust the total forces in the system never converge, e.g. Total force = 0.026394 Total SCF correction = 0.000314 Total force = 0.018379 Total SCF correction = 0.000282 Total force = 0.018824 Total SCF correction = 0.000120 Total force = 0.018857 Total SCF correction = 0.000155 Total force = 0.020215 Total SCF correction = 0.000193 Total force = 0.058765 Total SCF correction = 0.000241 Total force = 0.016998 Total SCF correction = 0.000205 Total force = 0.042967 Total SCF correction = 0.000238 Total force = 0.106226 Total SCF correction = 0.73 Total force = 0.137142 Total SCF correction = 0.71 Total force = 0.109549 Total SCF correction = 0.000108 Total force = 0.068723 Total SCF correction = 0.000103 Total force = 0.076599 Total SCF correction = 0.000168 Total force = 0.033233 Total SCF correction = 0.000342 Total force = 0.026883 Total SCF correction = 0.000248 Total force = 0.025846 Total SCF correction = 0.000378 Total force = 0.025596 Total SCF correction = 0.000225 Total force = 0.025482 Total SCF correction = 0.000275 Total force = 0.025508 Total SCF correction = 0.000128 I am thankful for any suggestions. Stephan Rix Institute for Inorganic und Analytical Chemistry Johannes Gutenberg-University Mainz, Germany -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20100915/b78f2dd0/attachment.htm
[Pw_forum] problems with converge of geometric optimization
Hi Stephan, If the only problem is bfgs, you can change ion_dynamics to damp. Regards -- Jia Chen Dept of Chemistry Princeton University Princeton, NJ 08544