trust region dog leg algorithm
I wrote out Powell's 1968 Fortran code I refer to Ming2017 for the 4 equations I understand the logic flow:
- from d aka t2, and setting wx = wy, we can define weffx = wx +Īw ~= wx + 2d
- from weffx, pi (half wave) and choice of l=0,1,2,3.. (or m), we can define kx
- because of k_metal = root epsilon_metal x 2pi / lambda, and k_dielectric, and k_air, and x y z plasmonic wave direction definitions / momentum conservation, we can relate k_x squared + k_y squared = k_j squared + k_z,j squared, where j = metal, dielectric, air
- from the dispersion of the surface plasmons which is derived with a matrix method (Davis TJ (2009) Surface plasmon modes in multi-layer thin-films. Opt Commun 282:135ā140), we have a relation that relates z-direction waves in the dielectric and metal to each other and their attenuations due to layer thicknesses and material permittivities.
- And finally we have an equation for resonant wavelength which depends on nothing but epsilon_dielectric; choice of l or m; k_z, dielectric; and Weffx or Weffy (ultimately d and Wx).
Choose wavelengths:
400,500,600,700;
800+100=900,1000,1100,1300-100=1200
as resonant wavelengths of modes (0,1), (1,0)
W1, W2, W3, W4

MgF2 = 1.3737^2 = epsilon 1.9 at 1000nm al2o3 = 1.76^2 = 3.1 (thought it was 9 odd?!) sio2 = 1.45^2 = 2.1 TiN = metal but non metal under 400nm
https://refractiveindex.info/?shelf=main&book=TiN&page=Pfluger
SrTiO3 = 2.4089^2 = 5.8 sio2 = 1.45 odd for n (square it and times by ε0 to get ε)

al = -88 - i22 at 1000nm https://refractiveindex.info/?shelf=main&book=Al&page=Rakic
gaAs ϵ1 = 12.183 ϵ2 = 0.089731
TiN 900nm epsilon -12.330 1000nm -13.94, 1100nm -15.6, 1200nm -17.4
TiN-
moS2 https://refractiveindex.info/?shelf=main&book=MoS2&page=Beal Very interesting over the range of interest -- no absorption aka -i*0. But depends on doping concentration?