need your help above woodpile calcalation

Hi, Dear all:

    I¡¦m trying to use MPB to calculate the band structure of ¡§woodpile¡¨ 
photonic crystals recently. There is no problem of the result when I use a 
FCC lattice. But when I change the height of stacked logs to make a FCT 
structure, the calculated results is incorrect comparing to FDTD 
transmission spectrum. There is always bandgap at higher frequency range 
than the actual one. The geometry of my structure is w=0.25a, h=0.25a, 
eps=10.3684. My code is shown as the following:

(set! geometry-lattice (make lattice

                 (basis-size (sqrt 0.375) (sqrt 0.375) (sqrt 0.5))

                 (basis1 0 1 (sqrt 0.5))

                 (basis2 1 0 (sqrt 0.5))

                 (basis3 1 1 0)))



(define (kinterpolate-uniform n ks)

  (map cartesian->reciprocal

       (interpolate-uniform n (map reciprocal->cartesian ks))))

(define-param k-interp 9)



(define X (vector3 0 0.5 0.5))

(define U (vector3 0.25 0.625 0.625))

(define L (vector3 0.5 0.5 0.5))

(define Gamma (vector3 0 0 0))

(define W (vector3 0.25 0.5 0.75))

(define K (vector3 0.375 0.375 0.75))



(set! k-points (kinterpolate-uniform k-interp (list X U L Gamma X W K)))

(define-param eps 10.3684) ; the dielectric constant of the "log" material



(define diel (make dielectric (epsilon eps)))



(define-param w (/ 0.25 (sqrt 2))) ; width of the logs

(define-param h (/ 0.25 (sqrt 2))) ; height of logs



; shortcut for cartesian->lattice function:

(define (c->l . args) (cartesian->lattice (apply vector3 args)))



 (define (first-gap w)

  (set! geometry

     (list

      (make block (material diel)

            (center (c->l 0 0 0))

            (e1 (c->l 1 1 0))

            (e2 (c->l 1 -1 0))

            (e3 (c->l 0 0 1))

            (size infinity w h))

      (make block (material diel)

            (center (c->l 0.125 0.125 h))

            (e1 (c->l 1 1 0))

            (e2 (c->l 1 -1 0))

            (e3 (c->l 0 0 1))

            (size w infinity h))))

(set-param! resolution 32)

(set-param! mesh-size 5)

(set-param! num-bands 20)



I¡¦ve checked the geometry distribution by employing Vis5D plot. It¡¦s 
correct showing a FCT structure I want. So the problem may happen at other 
places I can¡¦t see. If any person have had the similar experience? I would 
be very grateful for any help. Thanks in advance,

                                                                             
                                                               Ji-Feng Chen