[Wannier] Difficult to converge for large supercell

[Yun Yang]

Dear wannier90 users/authors,

I have some convergence problems for one kinds of antiferromagnetic
structures.

1. For some structures, it converge to different total spread for different
spin (spin up: 204.960; spin down: 218.967 ). I am not sure it is right.
2. And for some other structures, they have very slow convergence. ( the
Delta Spread is still -0.102E+01 at 6000 Iteration. )

The relevant portion of input/output is  below.
Could someone give me some idea about it? Any response will be highly
appreciated.

Thanks a lot,
Yun Yang

--------------------------------input -------------------------------------
*num_wann=273
num_bands=288

write_xyz = T
use_bloch_phases = false
iprint= 0
kmesh_tol=0.000001
search_shells=12

dis_win_max       = 6.6d0
dis_froz_max      = 6.6d0
dis_num_iter      =  200
dis_conv_tol     = 1.0d-9
dis_conv_window  = 20

num_iter         =  3000
num_dump_cycles  =  50
num_print_cycles =  50
conv_tol         = 1.0d-9
conv_window      = 20

begin unit_cell_cart
     3.7845100     3.7845100     0.0000000
    -3.7845100     3.7845100     0.0000000
     0.0000000     0.0000000    24.5299569
end unit_cell_cart

begin atoms_cart
O       -1.0921385     2.4088291     1.8252586
O       -2.7756156     4.5809594     2.1393497
O        0.7082736     2.7943602     3.9900185
...
O        0.9577591     2.4313320     0.0062152
O        1.0055154     4.5917031    22.8282041
O       -1.0807797     6.1075258    22.3838324
Ti       0.8626886     2.7391920     1.8671318
Ti       0.8437991     2.7496945     5.8787678
Ti       0.8605979     2.7599011     9.9301707
Ti       0.8469372     6.5394526     1.8671320
Ti       0.8574395     6.5205642     5.8787681
Ti       0.8676460     6.5373628     9.9301705
Fe1       0.8658522     2.7836692    13.9401049
Fe1       0.8407141     6.5252260    18.1598293
Fe1       0.8340616     2.7450507    22.2712639
Fe2       0.8914140     6.5426166    13.9401044
Fe2       0.8484616     2.7329699    18.1598279
Fe2       0.8527955     6.5108253    22.2712648
Pb       2.8009749     4.6932647    11.9173422
Pb       2.7784213     4.6590583     3.8623337
Pb       2.7694058     4.6497561     7.8921138
Pb      -0.9835019     4.6932299    11.9173430
Pb      -1.0177070     4.6706755     3.8623326
Pb      -1.0270087     4.6616619     7.8921147
Bi      -0.8992322     0.9947882    24.1623899
Bi      -0.8788219     4.8469538    15.8531172
Bi      -0.8847488     1.0075145    20.0257920
Bi      -0.8974663     4.7775330    24.1623905
Bi      -0.8298123     1.0134344    15.8531163
Bi      -0.8847394     4.7920174    20.0257938
end atoms_cart*

--------------------------output of slow convergence
---------------------------------------------------------
* +--------------------------------------------------------------------+<--
CONV
 | Iter  Delta Spread     RMS Gradient      Spread (Ang^2)      Time  |<--
CONV
 +--------------------------------------------------------------------+<--
CONV
      0     0.302E+04     0.0000000000     3018.8073804503       1.35  <--
CONV
      1    -0.211E+03   593.3584753528     2808.0050842612      24.37  <--
CONV
     60    -0.216E+02    40.5745863477      559.0956333879    1423.02  <--
CONV
    120    -0.408E+01    25.5247645416      364.6667507769    2817.40  <--
CONV
    180    -0.271E+02    25.0479925342      334.9080840874    4197.15  <--
CONV
    240    -0.958E-02     3.9459500641      257.0796991899    5577.72  <--
CONV
......

   2340    -0.883E+01    10.6302809602      461.6904093376   54906.15  <--
CONV
   2400    -0.131E+01    11.6467913061      360.4940697012   56256.80  <--
CONV
   2460    -0.310E+00    11.2637242429      350.7045045057   57661.03  <--
CONV
   2520    -0.141E+00     5.0814249412      343.5058238333   59067.39  <--
CONV
   2580     0.145E+04    11.8963919634     1720.9205589367   60475.19  <--
CONV
   2640    -0.252E+00    17.7236351947      687.1808548555   61912.97  <--
CONV
   2700    -0.980E+02    69.7570987576      713.8856599995   63347.75  <--
CONV
   2760    -0.146E+02    35.0834174784      317.7924512781   64778.08  <--
CONV
   2820    -0.126E+02    16.1100999806      429.5499556053   66224.50  <--
CONV
   2880    -0.806E+00     7.9652306641      340.6791136736   67659.84  <--
CONV
   2940    -0.102E+01    31.1404131228      319.7016524658   69099.76  <--
CONV*
*   3000    -0.291E+02    51.7853102534      408.5370262188   70484.98  <--
CONV*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20110914/f4100ad9/attachment.html>