[MPB-discuss] linear defect waveguide in tri-lattice mpi serial vs parallel
Hello all I'm currently investigating the band structures of these waveguides (W1) and using the serial version of MPB to accomplish this. However, I'd like to move the the parallel version to speed things up but I'm having difficulty with producing the same structure as the serial version. Using the line-defect.ctl example file in serial mpb, I can view the correct structure using mpb-data -r -x 3x epsilon.h5 then h5topng epsilon.h5:data-new (as demonstrated in the tutorial). However, the same technique does not produce the same triangular lattice when using parallel mpb. The unit cell appears rotated by 90 degrees relative to the serial output. Transposing using mpb-data -T does not produce the same lattice either. Attached are the outputs of the serial run, using the mpb-data command above and the parallel output, using mpb-data -r -T -y 3y epsilon.h5. Is there a subtlety I am missing or do I need to use a rectangular basis for the unit cell rather than triangular? <><>___ mpb-discuss mailing list mpb-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/mpb-discuss
Re: [MPB-discuss] linear defect waveguide in tri-lattice mpi serial vs parallel
OK, almost immediately after posting this I found this: http://ab-initio.mit.edu/wiki/index.php/MPB_User_Reference#MPB_with_MPI_parallelization which explains why x and y are transposed. However, I'm still confused as to why the resulting lattice is not the same after transposition i.e. it is still rotated by 30 degrees relative to the serial case. Many thanks in advance. Rikki On 17 May 2012 09:52, Rikki J Coles wrote: > Hello all > > I'm currently investigating the band structures of these waveguides (W1) > and using the serial version of MPB to accomplish this. However, I'd like > to move the the parallel version to speed things up but I'm having > difficulty with producing the same structure as the serial version. > > Using the line-defect.ctl example file in serial mpb, I can view the > correct structure using mpb-data -r -x 3x epsilon.h5 then h5topng > epsilon.h5:data-new (as demonstrated in the tutorial). > > However, the same technique does not produce the same triangular lattice > when using parallel mpb. The unit cell appears rotated by 90 degrees > relative to the serial output. Transposing using mpb-data -T does not > produce the same lattice either. > > Attached are the outputs of the serial run, using the mpb-data command > above and the parallel output, using mpb-data -r -T -y 3y epsilon.h5. > > Is there a subtlety I am missing or do I need to use a rectangular basis > for the unit cell rather than triangular? > ___ mpb-discuss mailing list mpb-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/mpb-discuss
[MPB-discuss] Output in-plane fields when using non-cartesian basis vectors
Dear MPB users I am trying to obtain an E-field profile for the first y-odd gap band in a W1 waveguide in a triangular lattice hole slab structure. As you know, this mode has non-zero Ex,Ey and Hz. I am using a modified version of the tri-holes.ctl file, which uses the basis vectors (basis1 (/ (sqrt 3) 2) 0.5) and (basis2 (/ (sqrt 3) 2) -0.5))). When I output the Hz field and run mpb-data to restore a rectangular cell, everything is OK. However, the fields appear skewed when I do the same for the E-fields and I'm not sure why this is the case. Can anyone shed some light on this issue please? Thanks Rikki Coles ___ mpb-discuss mailing list mpb-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/mpb-discuss