Re: scf convergence problem of Pt(111) slab ( No.1 ) |
- Date: 2015/02/02 19:41
- Name: Artem Pulkin <artem.pulkin@epfl.ch>
- Increase scf.mixing.history, decrease scf.mixing.weights. I would also go for a smaller K-grid to speed up calculations when finding convergence.
Artem
|
Re: scf convergence problem of Pt(111) slab ( No.2 ) |
- Date: 2015/02/02 19:43
- Name: Artem Pulkin <artem.pulkin@epfl.ch>
- And, of course, set the z component of your kgrid to 1 since you have vacuum there anyway.
|
Re: scf convergence problem of Pt(111) slab ( No.3 ) |
- Date: 2015/02/03 10:41
- Name: Wang Yuanqing <yuanqing.wang@riken.jp>
- Thanks. I will try.
|
Re: scf convergence problem of Pt(111) slab ( No.4 ) |
- Date: 2015/02/03 20:46
- Name: Eike Schwier
- You may also try changing
scf.EigenvalueSolver krylov
to
scf.EigenvalueSolver band
I tried krylov on my slab (Fe(001)) and while band solver works perfectly well, krylov does not converge.
best, Eike
|
Re: scf convergence problem of Pt(111) slab ( No.5 ) |
- Date: 2015/02/03 21:11
- Name: Wang Yuanqing <yuanqing.wang@riken.jp>
- Dear Eike,
Thanks for your advice!!!
Best,
Yuanqing Wang
|
Re: scf convergence problem of Pt(111) slab ( No.6 ) |
- Date: 2015/02/04 00:14
- Name: T. Ozaki
- Hi,
> I tried krylov on my slab (Fe(001)) and while band solver works perfectly well, > krylov does not converge.
This is not a statement applicable to general cases. In our experiences, the SCF convergence with the Krylov subspace method is almost the same as that for the conventional diagonalization if the parameters for the Krylov subspace method are properly selected.
As for Dr. Wang's case, there is no benefit from the O(N) Krylov subspace method because of the small number of atoms (64 atoms). Therefore, the conventional diagonalization method should be used because of the system size rather than the issue of SCF convergence.
As for improving the SCF convergence, scf.Kerker.factor should also be controlled when charge sloshing seriously happens. The default value for your system is found at the beginning of the standard output, and you can increase it starting from the value to suppress the charge sloshing.
Regards,
TO
|
Re: scf convergence problem of Pt(111) slab ( No.7 ) |
- Date: 2015/02/09 23:01
- Name: Wang Yuanqing <yuanqing.wang@riken.jp>
- Hi, everyone, I have finished the optimization. The output file said normally complete. However, there is one problem that in the last step it shows (in *.md)
time= 54.000 (fs) Energy= nan (Hartree) Pt nan nan nan nan nan nan Pt nan nan nan nan nan nan Pt nan nan nan nan nan nan Pt nan nan nan nan nan nan Pt nan nan nan nan nan nan Pt nan nan nan nan nan nan Pt nan nan nan nan nan nan
I *********************************************************** *********************************************************** History of geometry optimization *********************************************************** *********************************************************** MD_iter SD_scaling |Maximum force| Maximum step Utot (Hartree/Bohr) (Ang) (Hartree)
1 0.94486299 0.02759150 0.01379575 -5856.91586861 2 0.94486299 0.01606989 0.00803495 -5857.14984447 3 0.94486299 0.01907598 0.00953799 -5856.93657253 4 0.94486299 0.00729197 0.00364599 -5857.21061504 5 0.94486299 0.01615617 0.00854948 -5856.93089749 6 0.94486299 0.02375217 0.03175063 -5857.27171959 7 0.94486299 0.01988136 0.00315597 -5856.84984541 8 0.94486299 0.02875813 0.02146401 -5857.09521834 9 0.94486299 0.01072951 0.00941771 -5856.78896521 10 0.94486299 0.01096619 0.01697957 -5856.78876624 11 0.94486299 0.01402977 0.00118832 -5856.78442288 12 0.94486299 0.01383315 0.03175063 -5856.78479801 13 0.94486299 0.02117774 0.03175063 -5856.77018401 14 0.94486299 0.02934267 0.02816966 -5856.74596006 15 0.94486299 0.03694607 0.03175063 -5856.71572710 16 0.94486299 0.02787128 0.02346278 -5856.74965165 17 0.94486299 0.02280401 0.02127977 -5856.76638343 18 0.94486299 0.01824109 0.00844446 -5856.77590884 19 0.94486299 0.01750925 0.03175063 -5856.77804686 20 0.94486299 0.01212411 0.03175063 -5856.78769485 21 0.94486299 0.01590588 0.02719466 -5856.78315328 22 0.94486299 0.01127205 0.03175063 -5856.79143028 23 0.94486299 0.01063169 0.02375255 -5856.79642091 24 0.94486299 0.01281693 0.03175063 -5856.79884203 25 0.94486299 0.01266900 0.03175063 -5856.80225287 26 0.94486299 0.01086150 0.03175063 -5856.80606466 27 0.94486299 0.00837924 0.03175063 -5856.80951555 28 0.94486299 0.00911700 0.03175063 -5856.81236912 29 0.94486299 0.00814020 0.03175063 -5856.81499534 30 0.94486299 0.00634491 0.02325574 -5856.81765437 31 0.94486299 0.00573265 0.01900838 -5856.81925946 32 0.94486299 0.00457758 0.01320247 -5856.82059333 33 0.94486299 0.00458463 0.02329327 -5856.82151099 34 0.94486299 0.00394223 0.01632725 -5856.82258070 35 0.94486299 0.00294108 0.01040009 -5856.82304907 36 0.94486299 0.00234652 0.01690421 -5856.82333041 37 0.94486299 0.00227785 0.01094532 -5856.82356428 38 0.94486299 0.00158086 0.00256572 -5856.82362277 39 0.94486299 0.00142641 0.00794105 -5856.82366014 40 0.94486299 0.00160206 0.00329423 -5856.82374936 41 0.94486299 0.00103458 0.00363371 -5856.82379305 42 0.94486299 0.00101118 0.00187525 -5856.82380781 43 0.94486299 0.00089659 0.00410220 -5856.82382358 44 0.94486299 0.00090536 0.00422437 -5856.82385138 45 0.94486299 0.00079277 0.00329734 -5856.82386156 46 0.94486299 0.00067582 0.00282583 -5856.82387177 47 0.94486299 0.00064635 0.00258873 -5856.82386955 48 0.94486299 0.00047301 0.00155999 -5856.82388068 49 0.94486299 0.00034814 0.00083058 -5856.82388685 50 0.94486299 0.00033708 0.00102269 -5856.82389209 51 0.94486299 0.00033250 0.00244315 -5856.82389823 52 0.94486299 0.00027979 0.00141702 -5856.82390876 53 0.94486299 0.00034983 0.00149838 -5856.82391768 54 0.94486299 0.00042407 0.00130811 -5856.82393016 55 0.94486299 0.00034016 0.00086016 -5856.82392574 56 0.94486299 0.00031982 0.00114946 -5856.82392631 57 0.94486299 0.00028552 0.00078077 -5856.82392729 58 0.94486299 0.00034575 0.00101126 -5856.82392953 59 0.94486299 0.00035150 0.00215855 -5856.82393137 60 0.94486299 0.00044210 0.00062501 -5856.82394280 61 0.94486299 0.00045145 0.00060282 -5856.82394213 62 0.94486299 0.00045259 0.00062623 -5856.82394293 63 0.94486299 0.00046286 0.00054825 -5856.82394323 64 0.94486299 0.00043714 0.00028397 -5856.82394177 65 0.94486299 0.00044944 0.00058327 -5856.82394195 66 0.94486299 0.00040858 0.00175268 -5856.82394497 67 0.94486299 0.00039595 0.00210916 -5856.82394736 68 0.94486299 0.00036718 0.00181225 -5856.82395054 69 0.94486299 0.00035589 0.00048490 -5856.82394763 70 0.94486299 0.00041868 0.00149762 -5856.82394804 71 0.94486299 0.00033379 0.00078205 -5856.82394898 72 0.94486299 0.00032111 0.00062764 -5856.82394918 73 0.94486299 0.00035179 0.00035633 -5856.82394996 74 0.94486299 0.00031683 0.00055199 -5856.82395128 75 0.94486299 0.00030294 0.00030488 -5856.82395183 76 0.94486299 0.00027244 0.00032820 -5856.82395270 77 0.94486299 0.00027800 0.00078233 -5856.82395281 78 0.94486299 0.00034311 0.00174234 -5856.82395381 79 0.94486299 0.00033426 0.00030745 -5856.82395628 80 0.94486299 0.00031417 0.00036228 -5856.82395808 81 0.94486299 0.00032500 0.00012257 -5856.82395928 82 0.94486299 0.00031693 0.00101867 -5856.82395953 83 0.94486299 0.00035753 0.00095990 -5856.82395630 84 0.94486299 0.00031074 0.00036934 -5856.82395970 85 0.94486299 0.00032070 0.00222151 -5856.82395864 86 0.94486299 0.00034637 0.00124637 -5856.82396337 87 0.94486299 0.00031230 0.00039361 -5856.82396227 88 0.94486299 0.00030502 0.00044654 -5856.82396117 89 0.94486299 0.00030776 0.00017436 -5856.82395976 90 0.94486299 0.00030705 0.00057602 -5856.82396034 91 0.94486299 0.00031326 0.00021038 -5856.82396039 92 0.94486299 0.00032040 0.00025081 -5856.82396122 93 0.94486299 0.00034254 0.00184341 -5856.82396161 94 0.94486299 0.00048906 0.00059169 -5856.82396574 95 0.94486299 0.00054134 0.00056914 -5856.82396782 96 0.94486299 0.00046433 0.00008597 -5856.82396684 97 0.94486299 0.00045067 0.00047094 -5856.82396653 98 0.94486299 0.00040977 0.00029057 -5856.82396587 99 0.94486299 0.00037245 0.00135650 -5856.82396519 100 0.94486299 0.00032333 0.00030617 -5856.82396226 101 0.94486299 0.00033768 0.00060915 -5856.82396286 102 0.94486299 0.00034173 0.00062457 -5856.82396490 103 0.94486299 0.00032409 0.00059394 -5856.82396449 104 0.94486299 0.00035214 0.00000000 -5856.82396668 105 0.94486299 0.00035214 0.00212633 -5856.82396669 106 0.94486299 0.00049741 0.00210756 -5856.82396254 107 0.94486299 0.00054096 0.00257494 -5856.82388909 108 0.94486299 0.00058930 0.00000000 -5856.82390787 109 0.94486299 0.00000000 0.00000000 nan \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
Is this OK?
|
Re: scf convergence problem of Pt(111) slab ( No.8 ) |
- Date: 2015/02/10 21:23
- Name: T. Ozaki
- Hi,
The calculation seems to be reaching almost convergence, while I don't know what happened at the last step. To make sure whether the calculation is proper or not, you can restart your calculation using the structure stored in *.md.
Regards,
TO
|
Re: scf convergence problem of Pt(111) slab ( No.9 ) |
- Date: 2015/02/10 21:27
- Name: Wang Yuanqing <yuanqing.wang@riken.jp>
- Thank you very much!
BTW, why does "nan" occur? What is the meaning of this sign?
Best,
Yuanqing Wang
|
Re: scf convergence problem of Pt(111) slab ( No.10 ) |
- Date: 2015/02/10 21:45
- Name: T. Ozaki
- Hi,
"nan" is an error message in C, and means not a number. This appears when an improper calculation is performed such as division by zero.
Regards,
TO
|