Method 
Features 
Storage 
CPU 
Availability 
Spectralelement 
Sparse matrix, direct solution 
n^{5} n^{3} 
n^{7} n^{4} 
(Philips/Signetics, 2D) 
Timedomain
finitedifference 
Massively parallel 
n^{3}
n^{2} 
n^{3} x
n^{2} x 
TEMPEST
(UC Berkeley, 2D&3D) 
Waveguide 
Small refractive index changes 
n^{4}
n^{2} 
n^{6} x
n^{3} x 
METROPOLE
(CarnegieMellon, 2D&3D) 
Differential 
Smooth topography 
n^{4}
n^{2} 
n^{6} x
n^{3} x 
iPHOTO (Intel, 2D)
LISI (TU Vienna, 3D) 
Physicaloptics &
fast multipole 
Piecewise linear topography 
n^{2}
n 
n^{3}
n^{1.5} 
(UC Berkeley, 3D) 
Integral &
fast multipole 
Iterative solution 
n^{2}
n 
n^{3.3} x
n^{1.5} x 
(UC Berkeley, 3D) 
Upper quantities characterize the storage and computational costs
for threedimensional simulations, whereas lower quantities
refer to twodimensions. The symbols
and
signify the number of iterations required for the
respective technique, and
denotes the accuracy
of the vertical discretization.
