[Q-e-developers] inconsistent x-component of printed forces (qe 5.0)

[Nicola Manini]

Hi,
		in a simple pw calculation, the printed forces disagree with the successive 
displacements to achieve structural optimization.

The attached input & output (isotropically strained bulk Si, 2 atoms 
primitive cell) at the first iteration gives the following

      Forces acting on atoms (Ry/au):

      atom    1 type  1   force =     0.07692026   -0.07692026   -0.07692026
      atom    2 type  1   force =    -0.07692026    0.07692026    0.07692026

Note the opposite sign of the x components components.
However, the positions at the successive minimization step change from

ATOMIC_POSITIONS crystal
Si      0.000000000    0.000000000   0.00000000
Si      0.200000000    0.200000000   0.20000000

to

Si      -0.006410022  -0.006410022  -0.006410022
Si       0.206410022   0.206410022   0.206410022

suggesting that the actual value for the x component of the force coincides 
with that of the y and z components.
See that successive relaxation steps confirm this finding: correctly x=y=z 
symmetrical forces used in minimization, but incorrect sign of printed x 
force component.

Please check the code where it prints out the force components.

Thank you!
Ciao,
		Nick
-------------- next part --------------
A non-text attachment was scrubbed...
Name: si_strain.tgz
Type: application/targizipped
Size: 190759 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20130308/8b3e2e2e/attachment.bin>