This is a simple attempt at writing a 2D FDTD simulation in R.
epsilon_0 <- 8.854187817e−12
mu_0 <- 1.2566370614e−6
Omega_0 <- sqrt(mu_0/epsilon_0)
dt <- 0.1
epsilon_r <- 1
mu_r <- 1
period <- 40*dt
dy <- dx <- dt/4*sqrt(2)
a <- -2
b <- 2
x <- seq(from = a, to = b+dx/2, by = dx)
y <- seq(from = a, to = b+dy/2, by = dy)
t <- seq(from = 0, to = 36 + dt/2, by = dt)
origin_x <- round(length(x)/2) + 1
origin_y <- round(length(y)/2) + 1
E_z <- array(data = 0, dim = c(length(x), length(y), length(t)))
H_y <- E_z # need ot be zeros arrays
H_x <- E_z
Example shows an air box surrounded by perfect conductor(fields must equal 0). A high refractive index box is set up right of center.
Epsilon_r <- array(data = 1, dim = c(length(x), length(y)))
Mu_r <- Epsilon_r
Epsilon_r[x > 1, ] <- 9
image(Re(Epsilon_r))
Source is a single sinusoidal pulse.
Source <- E_z # just zeros
Source[origin_x, origin_y, t < period] <- sin(2*pi*t[t < period]/4)
plot(Re(Source[origin_x,origin_y,]), type='l')
Solve the wave equation for magnetic (H) and electric(E) fields for each time step. Simulation is vectorised, need notes on this here.
nx <- length(x)
ny <- length(y)
for(t_i in seq(2, length(t))){
H_y[1:(nx-1), 1:(ny-1), t_i] <-
H_y[1:(nx-1), 1:(ny-1), t_i - 1] +
(1 / Mu_r[1:(nx-1), 1:(ny-1)] / Omega_0 *
(E_z[2:(nx), 1:(ny-1), t_i - 1] - E_z[1:(nx-1), 1:(ny-1), t_i - 1]) / sqrt(2))
H_x[1:(nx-1), 1:(nx-1), t_i] <-
H_x[1:(nx-1), 1:(nx-1), t_i - 1] -
(1 / Mu_r[1:(nx-1), 1:(ny-1)] / Omega_0 *
(E_z[1:(nx-1), 2:(ny), t_i - 1] - E_z[1:(nx-1), 1:(ny-1), t_i - 1]) / sqrt(2))
E_z[2:(nx-1), 2:(ny-1), t_i] <-
(E_z[2:(nx-1), 2:(ny-1), t_i - 1] +
1 / Epsilon_r[2:(nx-1),2:(ny-1)] * Omega_0 *
(
(H_y[2:(nx-1), 2:(ny-1), t_i] - H_y[1:(nx-2), 2:(ny-1), t_i]) -
(H_x[2:(nx-1), 2:(ny-1), t_i] - H_x[2:(nx-1),1:(ny-2) , t_i])
) / sqrt(2) + Source[2:(nx-1), 2:(ny-1), t_i])
}
for(i in 1:length(t)){
fields::image.plot(Re(E_z[,,i]), asp = 1,col = viridis::viridis(200))
}