[Pw_forum] slow relaxing
Nicola Marzari
marzari at MIT.EDU
Sun May 11 10:22:14 CEST 2008
hi Rudra,
ideally, if you have 18 atoms, i.e. 56 degrees of freedom (or 53, if you
remove traslations) a conjugate gradient algorithms should bring you to
the minimum in 53 steps. You are not far away from that - you are using
bfgs (not sure if there is an analytic results on the minimum number of
moves). Also, restart probably remembers the approximate hessian (second
derivative matrix) from the previous run, and that might not be that
helpful, at the end, if you started away from equlibrium.
Unless you are interested in empty states, or in ice, it's unlikely
that using k-point sampling would be very interesting - I'd use gamma,
and increase the sice of the cell.
Last - you might want to play around with the parameters - it's a bit of
a tricky issue, but basically what you need is that each ionic step the
hellman-feynman forces need to be converged well enough for the ions to
move in the right direction, but not too much.
Why don't you take a smaller system, like a water trimer, in a supercell
with gamma sampling, and try to work with etot_conv_thr and
force_conv_thr and upscale until you converge to the tolerance you want
in the least cpu time ?
nicola
Rudra Banerjee wrote:
> i am trying to do this PW calculation(water once again), but as you can
> see, the convergence ratio is very slow...is there any way to speed up
> the relaxation?
>
> 1 &CONTROL
> 2 calculation = 'relax' ,
> 3 restart_mode = 'restart' ,
> 4 outdir = '.' ,
> 5 pseudo_dir = '/home/Rudra/espresso-3.2.3
> /pseudo/' ,
> 6 prefix = 'ring' ,
> 7 etot_conv_thr = 1.0d-5 ,
> 8 forc_conv_thr = 1.0d-4 ,
> 9 nstep = 100 ,
> 10 tstress = .true. ,
> 11 tprnfor = .true. ,
> 12 /
> 13 &SYSTEM
> 14 ibrav = 14,
> 15 celldm(1) = 35,
> 16 celldm(2) = 1,
> 17 celldm(3) = 1,
> 18 celldm(4) = 0,
> 19 celldm(5) = 0,
> 20 celldm(6) = 0,
> 21 nat = 18,
> 22 ntyp = 2,
> 23 ecutwfc = 60 ,
> 24 /
> 25 &ELECTRONS
> 26 conv_thr = 1.0d-10 ,
> 27 /
> 28 &IONS
> 29 ion_dynamics = 'bfgs' ,
> 30 upscale = 30.D0 ,
> 31 trust_radius_max = 5.0D-1 ,
> 32 trust_radius_min = 1.0D-10,
> 33 /
> 34 ATOMIC_SPECIES
> 35 O 15.99000 O.BLYP.UPF
> 36 H 1.00000 H.fpmd.UPF
> 37 ATOMIC_POSITIONS bohr
> 38 O 0.452598E+01 0.199770E+01 -0.384251E+00
> 39 O -0.456645E+00 -0.494031E+01 0.587632E+00
> 40 O 0.417483E+01 -0.297334E+01 0.375756E+00
> 41 O -0.450000E+01 -0.199189E+01 0.492865E-01
> 42 O -0.413521E+01 0.302669E+01 -0.255022E-01
> 43 O 0.460540E+00 0.495314E+01 -0.751339E+00
> 44 H 0.565520E+01 0.257057E+01 0.950862E+00
> 45 H 0.433997E+01 0.115123E+00 -0.108584E+00
> 46 H -0.780642E+00 -0.583785E+01 0.216070E+01
> 47 H -0.195104E+01 -0.377112E+01 0.361755E+00
> 48 H 0.240252E+01 -0.368407E+01 0.466936E+00
> 49 H 0.500948E+01 -0.393069E+01 -0.955826E+00
> 50 H -0.432075E+01 -0.893828E-01 0.405553E-01
> 51 H -0.550895E+01 -0.239361E+01 -0.143549E+01
> 52 H -0.477693E+01 0.385892E+01 0.148439E+01
> 53 H -0.238065E+01 0.373511E+01 -0.293647E+00
> 54 H 0.196062E+01 0.377921E+01 -0.601195E+00
> 55 H 0.644296E+00 0.576927E+01 -0.238986E+01
> 56
> 57 K_POINTS automatic
> 58 3 3 3 1 1 1
>
>
> ENERGY
> -----------
>
> -205.1846032445 Ry
> -205.1846032445 Ry
> -205.1851434168 Ry
> -205.1859610477 Ry
> -205.1869410014 Ry
> -205.1873885307 Ry
> -205.1875891135 Ry
> -205.1878601381 Ry
> -205.1884629910 Ry
> -205.1886928104 Ry
> -205.1887148135 Ry
> -205.1889751236 Ry
> -205.1890151305 Ry
> -205..1890589222 Ry
> -205.1891438343 Ry
> -205.1892290624 Ry
> -205.1892951055 Ry
> -205.1894417283 Ry
> -205.1894831934 Ry
> -205.1895085510 Ry
> -205.1895218967 Ry
> -205.1895516832 Ry
> -205.1895640658 Ry
> -205.1895716262 Ry
> -205.1895746715 Ry
> -205.1895786741 Ry
> -205.1895832218 Ry
> -205.1895880554 Ry
> -205.1895927379 Ry
> -205.1895971238 Ry
> -205.1896015209 Ry
> -205.1896056880 Ry
>
> --
> Rudra
> JRF; SNBNCBS
> http://www.bose.res.in/~rudra
> <http://www.bose.res..in/%7Erudra%20%20%20%20%20%20%20>
>
> A bus station is where a bus stops. A train station is where a train
> stops. On my desk I have a work station.
>
> Please, if possible, don't send me MS Word or PowerPoint attachments
> Why?See: http://www.gnu.org/philosophy/no-word-attachments.html
> <http://www.gnu.org/philosophy/no-word-attachments.html%20%20%20%20%20%20%20%20%20%20%20>
>
>
>
> ------------------------------------------------------------------------
> Meet people who discuss and share your passions. Join them now.
> <http://in.rd.yahoo.com/tagline_groups_7/*http://in.promos.yahoo.com/groups/bestofyahoo/>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
--
---------------------------------------------------------------------
Prof Nicola Marzari Department of Materials Science and Engineering
13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA
tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu
More information about the Pw_forum
mailing list