[QE-users] Neverending geometry optimization
Paolo Giannozzi
p.giannozzi at gmail.com
Wed May 12 12:58:46 CEST 2021
Try one of the following:
- use the development version
- use smearing='gauss'.
Explanation here: https://gitlab.com/QEF/q-e/-/issues/81. I am aware of
cases of oscillations between two configurations that were caused by this
(now solved) problem.
Paolo
On Wed, May 12, 2021 at 12:42 PM Mauro Sgroi <maurofrancesco.sgroi at gmail.com>
wrote:
> Dear users,
> I'm writing to ask for help with a geometry optimization of Li2TiS3.
> For a particular structure the BFGS algorithm gives the message CASE:
> energy _new > energy _old and the optimization starts oscillating between
> the same energy values.
>
> For example:
> number of scf cycles = 184
> number of bfgs steps = 4
>
> energy old = -3664.4342647062 Ry
> energy new = -3664.4341932047 Ry
>
> CASE: energy _new > energy _old
>
> new trust radius = 0.0000060248 bohr
>
> trust_radius < trust_radius_min
>
> resetting bfgs history
>
> new conv_thr = 1.0E-11 Ry
>
> number of scf cycles = 185
> number of bfgs steps = 4
>
> energy old = -3664.4342647062 Ry
> energy new = -3664.4341596224 Ry
>
> CASE: energy _new > energy _old
>
> new trust radius = 0.0002779227 bohr
> new conv_thr = 1.0E-11 Ry
>
> number of scf cycles = 186
> number of bfgs steps = 4
>
> energy old = -3664.4342647062 Ry
> energy new = -3664.4341932047 Ry
>
> CASE: energy _new > energy _old
>
> new trust radius = 0.0000060248 bohr
>
> trust_radius < trust_radius_min
>
> resetting bfgs history
>
> new conv_thr = 1.0E-11 Ry
>
>
> My input is below. I played a bit with the values of the trust_radius min
> and max and also tried the damped optimization with no results.
> Have you any suggestions to solve this issue?
> Thanks a lot and best regards,
> Mauro Sgroi.
> Centro Ricerche FIAT
>
> &CONTROL
> title = 'Li2TiS3'
> pseudo_dir =
> '/workhpc/FCA/FCA_CRF_STRUT/sgroi/DATABASE/ESPRESSO'
> prefix = 'POSCAR-1012.xyz'
> outdir = '/workhpc/FCA/FCA_CRF_STRUT/sgroi/tmp/LTS'
> calculation = 'relax'
> disk_io = 'none'
> nstep = 500
> /
> &SYSTEM
> ibrav = 0
> A = 10.736
> nat = 54
> ntyp = 3
> ecutwfc = 52
> ecutrho = 575
> occupations = 'smearing'
> degauss = 0.005
> smearing = 'mv'
> /
> &ELECTRONS
> electron_maxstep = 100
> conv_thr = 1.0D-9
> diago_thr_init = 1e-4
> startingpot = 'atomic'
> startingwfc = 'atomic'
> mixing_mode = 'plain'
> mixing_beta = 0.1
> diagonalization = 'david'
> diago_david_ndim = 4
> scf_must_converge = .false.
> /
> &IONS
> ion_dynamics = 'bfgs'
> trust_radius_max = 1.5
> trust_radius_min = 1.0d-4
> /
>
> ATOMIC_SPECIES
> Li 6.941 Li.pbesol-sl-kjpaw_psl.1.0.0.UPF
> Ti 47.867 Ti.pbesol-spn-kjpaw_psl.1.0.0.UPF
> S 32.06 S.pbesol-nl-kjpaw_psl.1.0.0.UPF
>
> K_POINTS {gamma}
>
> CELL_PARAMETERS {alat}
> 0.000000 0.707107 0.707107
> 0.707107 0.000000 0.707107
> 0.707107 0.707107 0.000000
>
> ATOMIC_POSITIONS {CRYSTAL}
> S 0.1633831501 0.1505176116 0.1670420977
> S 0.8343669301 0.8304670782 0.8418548293
> S 0.1790412476 0.1764413502 0.4935621000
> S 0.8247343817 0.8408510709 0.4917316687
> S 0.1694651936 0.5002232433 0.1698005825
> S 0.8212707269 0.5090769817 0.8266090665
> S 0.4974749523 0.1673871640 0.1804228455
> S 0.4927868323 0.8404319567 0.8296118314
> S 0.1692023862 0.8072918196 0.8335429994
> S 0.4890640593 0.1774332300 0.5033543118
> S 0.1641898391 0.8317001666 0.4917151005
> S 0.4912509798 0.4872129048 0.1840572305
> S 0.1641576299 0.4968050469 0.8418554032
> S 0.8474555693 0.1703269784 0.1606609978
> S 0.8507828797 0.1558342390 0.8390628857
> S 0.1743639033 0.4835156874 0.5118083355
> S 0.8447227515 0.1643048203 0.4777090676
> S 0.1590903063 0.8641836879 0.1479836531
> S 0.4727465739 0.1769467379 0.8347894587
> S 0.8594550184 0.8282327233 0.1509046931
> S 0.1615534801 0.1693308123 0.8476401199
> S 0.8542522557 0.4928254752 0.1478553155
> S 0.4891060282 0.8469926505 0.1578063062
> S 0.5072663898 0.5105322763 0.8198128542
> S 0.8244012240 0.5104213752 0.4987828905
> S 0.4982413076 0.8198464677 0.5160373591
> S 0.4783486409 0.4993651057 0.5240038320
> Ti -0.0265989094 0.0242519853 0.0081914789
> Ti 0.3631233375 -0.0215014503 -0.0102111071
> Ti 0.0266113267 0.3269523470 0.0099648008
> Ti -0.0075871154 -0.0172900760 0.3646703424
> Ti 0.3419155890 0.3288881722 0.6436175035
> Ti 0.3571094741 0.6656725341 0.6686450804
> Ti 0.6930348083 0.6408751632 0.6601932583
> Ti -0.0207406070 0.6850583681 -0.0154635777
> Ti 0.3285981104 0.3393287536 0.3118356500
> Li 0.6587856458 0.6678472981 0.3328700152
> Li 0.6556429250 0.3430241928 0.3388163527
> Li 0.3345541443 0.6611504368 0.3292467652
> Li 0.6715595113 0.3269333260 0.6678239753
> Li 0.0060226298 0.6605967591 0.3365283850
> Li 0.0126944052 0.3286927896 0.3395337969
> Li -0.0038974319 0.3373998592 0.6566343563
> Li 0.6576999667 0.0033507055 0.3346167570
> Li 0.3246726313 0.0045111369 0.6656645737
> Li 0.3330391120 0.0093228380 0.3367520122
> Li 0.3419873713 0.3234714033 0.0165197646
> Li 0.3395750477 0.6570272725 0.0135071994
> Li 0.6551579591 0.3336210050 0.0149205898
> Li 0.6653074562 0.0041400082 0.0044649389
> Li 0.0073038566 0.0092546868 0.6496021276
> Li 0.6635360145 0.0085920982 0.6662860686
> Li 0.6489610883 0.6668026630 0.0048670464
> Li -0.0102428937 0.6735271531 0.6598841003
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at 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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20210512/83c8e35d/attachment.html>
More information about the users
mailing list