Hi Sonu
Some possibilities:
- You need MD.VariableCell set up as true (sorry, just in case)
- In these cases, it is always good practice to check the file fdf.log
(out.fdf in older versions) to make sure that siesta is using what
you think
it is. (possible typos in input, for instance). Standard output
would also
show that in many cases, but not always very directly.
- You may have too low a mesh cutoff, especially If you have basis
orbitals with
derivative discontinuities at their cutoff radii. If, say,
expanding the lattice,
and the mesh (slightly) expanding with it, you can get many mesh
points getting
out of atoms at once, which can produce a jump in the P(V) curve.
If your target
pressure is in the gap, siesta oscillates never getting there.
Solution: increase mesh cutoff. You can also add soft confinement
to your
orbitals. And/or filter them.
In general, different tolerances relate to each other: a finer
relaxation
(small force and stress tolerance) would demand correspondingly fine
mesh cutoff and scf tolerance, for example.
Emilio
On 24 May 2011, at 05:32, sonu kumar wrote:
Dear all siesta users,
i am trying to relax cell parameters in rhombohedral unit cell.
But stress tensor components are not becoming less than 1.0
GPa. However pressure is ~zero.
part of output is:-
Stress-tensor-Voigt (kbar): -61.58 29.02 32.56
-23.03 -3.31 -13.80
siesta: Pressure (static):
siesta: 0.00120848 -0.00280164 kBar
I used these options:-
MD.MaxStressTol 1.0 GPa
MD.TargetPressure 0.0 GPa
%block MD.TargetStress
-1.0 -1.0 -1.0 0.0 0.0 0.0
%endblock MD.TargetStress
What should i do to reduce it?
Also, interesting thing is that off diagonal stress components are
non-zero,
but should be zero. Is this explainable?
Thank you.
regards,
Sonu
IITD
--
Emilio Artacho
Department of Earth Sciences, University of Cambridge
Downing Street, Cambridge CB2 3EQ, UK
Tel. (+44/0) 1223 333480, Fax (+44/0) 1223 333450
emi...@esc.cam.ac.uk, http://www.esc.cam.ac.uk/~emilio