Dear Frederico,
I really appreciate your advice.
As you mentioned, I checked my structure in electrode and scattering region.
The system I calculated is graphene nanoribbon structure and I used
two units in z-direction for the electrode (left and right) and four
units for the scattering region. So, I think this is the reason I got
nc= 0 when nc is calculated from NScatt - NL - NR.
As increasing the scattering region to 6 units, the tbtrans code works
fine and I got reasonable results.
But, still I'm confused about electrode region/scattering region.
Let say, in the z direction, I have the structure such as,
left electrode(atom 1-10) core region (11-40) right electrode (41-50)
Then, scattering region is atom 1-50 ? or 11-40?
If only 11-40, why nc=NScatt - NL - NR should be positive? (NScatt > NL+NR)
I'm looking forward to having any advice. ^^
Thanks.
Kyoung Kweon
Quoting Frederico Dutilh Novaes <[email protected]>:
Dear Kyoung,
As the manual says, "the number or nodes should not bigger than the number
of k-points".
It should not in the sense that you would have nodes doing nothing, because
this version of tbtrans parallelizes only over k-points.
I ran in 2 nodes, and single node, either of them is not working. And
always code stops at kgrid_Monkhorst_Pack block.
This is the error message :
<<<<Error>>
Transport : ERROR : nc <=0
ERROR STOP from Node: 0
[0] [MPI Abort by user] Aborting Program!
Abort signaled by rank 0: MPI Abort by user Aborting program !
Exit code -3 signaled from c45-202
Killing remote processes...DONE
<<End of output file>>
contour: 0.13809 0.00000 0.002969616 0.000000000 97
contour: 0.14106 0.00000 0.002969616 0.000000000 98
contour: 0.14403 0.00000 0.002969616 0.000000000 99
contour: 0.14700 0.00000 0.002969616 0.000000000 100
unit cell:
45.3534 0.0000 0.0000
0.0000 13.2281 0.0000
0.0000 0.0000 16.1168
unit cell:
45.3534 0.0000 0.0000
0.0000 13.2281 0.0000
0.0000 0.0000 16.1168
Transport : ERROR : nc = 0 !!
Transport : ERROR : nc <=0
ERROR STOP from Node: 0
This error is not related to the k-point sampling. Let's call NL the
dimension of you left electrode system (the number of orbitals used from you
left electrode calculation), NR the equivalent for the right relectrode, and
NScatt the dimension of your setup for the scattering region. nc is such
that,
nc = NScatt - NL - NR
So what you are having is that the sum of the dimensions of what is
attributed to left end right electrodes is actually greater than what you
have in your scattering region setup.
It would require more information (like your fdf files) to try to help you
further. For example, I can't see why your TranSIESTA calculation ran
(considering the problem you are having with tbtrans).
Best regards,
Frederico.
