Plane-wave DFT problem - unit cell optimization


Click here for full thread
Clicked A Few Times
I wanna self answer this question. Just convert the Cartesian coordinate to fractional coordinate. I use the website:

http://cci.lbl.gov/cctbx/frac_cart.html

to convert.

Quote:Chun-hung.wang Dec 15th 10:49 am
Hi all,

When I try to run partial unit cell optimization by plane-wave DFT calculation in NWChem 6.5 by Cascade supercomputer, the atom position is always rearranged. For example, the Zn position will "aggregate" with each other. My system has 272 atoms. In some cases the error message would be like: my atomic distance is too close. Here is my input file:




echo

title "ZIF-64"

memory stack 1800 mb heap 200 mb global 250 mb noverify

permanent_dir .

start zif64_PSPW

geometry noautoz
  system crystal
lat_a 21.110d0
lat_b 9.906d0
lat_c 21.110d0
alpha 90.0d0
beta 90.0d0
gamma 90.0d0
end
C 10.59088700 2.97180000 7.63759800
H 11.11019300 2.57556000 6.94730100
.........
Zn                 8.91200870    4.45472820   13.76794200
Zn 1.64299130 4.45472820 17.89705800
end

nwpw
  mult 1
xc lda
lmbfgs
end

driver
  maxiter 200
end

set geometry:actlist 1:2 9:10 13:14 17:18 23:24 27:28 35:36 41:42 69:74 93:98 105:110 117:122 135:140 147:152 171:176 189:194
set includestress .true.
task pspw optimize




The error message is like:
  atoms  219  and                    135  are similar
atom 219 coordinates 8.25280109501174 2.51446878737748 27.4179707412332
atom 135 coordinates 8.07364560376715 2.42962021218857 27.1928595327956
atoms 221 and 137 are similar
atom 221 coordinates 36.0274472419937 14.4455120450660 16.8622775957722
atom 137 coordinates 36.2066027332383 14.5303606202549 17.0873888042098
minimum distance 0.299951707644571


I've checked my original input file that no two atoms are too close to each other. Why does this happen here? How do I fix that error? Thank you.
Chun-Hung