Re: repeated bands while using MPI-MPB

I think I use the correct command, and for your infomation I attach the 
result and control file (line-defect.ctl provided in mailing list). I am 
wondering if there is something wrong with the installation of MPI-MPB. Hope 
someone can help me to find my problem. Thanks.

; A line-defect waveguide in a 2d triangular lattice of dielectric
; rods (c.f. tri-rods.ctl), formed by a row of missing rods along the
; "x" direction.  (Here, "x" and "y" refer to the first and second
; basis directions.)  This structure supports a single guided band
; within the band gap, much like the analogous waveguide in a square
; lattice of rods (see "Photonic Crystals" by Joannopoulos et al.).

(define-param supercell-y 7) ; the (odd) number of lateral supercell periods

(set! geometry-lattice (make lattice
    (basis1 (/ (sqrt 3) 2) 0.5)
    (basis2 (/ (sqrt 3) 2) -0.5)
    (size 1 supercell-y no-size)))

(define-param eps 12) ; the dielectric constant of the rods
(define-param r 0.2) ; the rod radius in the bulk crystal

(set! geometry (list (make cylinder
         (center 0 0 0) (radius r) (height infinity)
         (material (make dielectric (epsilon eps))))))

(set! geometry
      (append
       ; duplicate the bulk crystal rods over the supercell:
       (geometric-objects-lattice-duplicates geometry)

       ; add a rod of air, to erase a row of rods and form a waveguide:
       (list
 (make cylinder (center 0) (radius r) (height infinity)
       (material air)))))

(define Gamma (vector3 0 0 0))
(define K' (lattice->reciprocal (vector3 0.5 0 0))) ; edge of Brillouin 
zone.
(set! k-points (interpolate 4 (list Gamma K')))

; the bigger the supercell, the more bands you need to compute to get
; to the defect modes (the lowest band is "folded" supercell-y times):
(define-param extra-bands 5) ; number of extra bands to compute above the 
gap
(set! num-bands (+ supercell-y extra-bands))

(set-param! resolution 32)

; Compute the TM modes, outputting the Ez field in the *middle* of the
; band.  (In general, the guided mode in such an air defect may have
; exited the gap by the time it reaches the edge of the Brillouin
; zone at K'.)
(run-tm
 (output-at-kpoint (list-ref k-points (quotient (length k-points) 2))
     fix-efield-phase output-efield-z))


----- Original Message ----- 
From: "Steven G. Johnson" <[email protected]>
To: <[email protected]>
Sent: Sunday, April 30, 2006 12:00 PM
Subject: Re: [MPB-discuss] repeated bands while using MPI-MPB


> On Sun, 30 Apr 2006, Lingling Tang wrote:
>> I am using MPI-MPB (Linux 2.4.21-37.0.1.EL, gcc 3.2.3, mpich) to analyze 
>> 2D photonic crystal, such as supercell and line defect structure. It is 
>> strange that for the same control file, I get different band digram when 
>> I use one and two processors. The result got by one processor is correct, 
>> while repeated bands are obtained if two processors are used. For 
>> example, if num-bands is 16, 9th band to 16th is exactly the same as 1st 
>> to 8th. I can never go to higher frequency by increasing num-bands. Does 
>> anyone have any idea about this problem?
>
> That doesn't make any sense to me.  Perhaps you are running it 
> incorrectly?  You should use something like
>  mpirun -np 2 mpb-mpi foo.ctl
>
>> Another problem with MPI-MPB is the first two dimensions transposed as 
>> mentioned in user reference. Use mpb-data -r -n 32 epsilon.h5, I still 
>> cannot get correct image (I can if use non-mpi version mpb). Is there any 
>> thing wrong with my commands?
>
> Just pass the -T option to mpb-data, as described in the manual.
>
> _______________________________________________
> mpb-discuss mailing list
> [email protected]
> http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/mpb-discuss

_______________________________________________
mpb-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/mpb-discuss