def run_TMM_simulation(wavelengths, polarization_amplitudes, k_inc, theta, phi, ER, UR, layer_thicknesses,\ transmission_medium, incident_medium): """ :param wavelengths: :param k_inc: holds kx, ky of the incident wave-vector (which is enough to specify kz since k0 is specified by wavelength :param theta: :param phi: :param ER: relative dielectric constants of each layer :param UR: relative permeability of each layer :param layer_thicknesses: :param transmission_medium: :param incident_medium: :return: """ assert len(layer_thicknesses) == len(ER) == len(UR); "number of layer parameters not the same" ref = []; trans = []; I = np.matrix(np.eye(2, 2)); # unit 2x2 matrix [e_r, m_r] = incident_medium; [e_t, m_t] = transmission_medium; n_i = np.sqrt(e_r*m_r); [kx, ky] = k_inc; normal_vector = np.array([0, 0, -1]) # positive z points down; ate_vector = np.matrix([0, 1, 0]); # vector for the out of plane E-field ## ================= specify gap media ========================## e_h = 1; m_h = 1; Pg, Qg, kzg = pq.P_Q_kz(kx, ky, e_h, m_h) Wg = I; # Wg should be the eigenmodes of the E field, which paparently is the identity, yes for a homogeneous medium sqrt_lambda = cmath.sqrt(-1) * Wg; # remember Vg is really Qg*(Omg)^-1; Vg is the eigenmodes of the H fields Vg = Qg * Wg * (sqrt_lambda) ** -1; ## ========================================== ## [pte, ptm] = polarization_amplitudes; for i in range(len(wavelengths)): # in SI units ## initialize global scattering matrix: should be a 4x4 identity so when we start the redheffer star, we get I*SR Sg11 = np.matrix(np.zeros((2, 2))); Sg12 = np.matrix(np.eye(2, 2)); Sg21 = np.matrix(np.eye(2, 2)); Sg22 = np.matrix(np.zeros((2, 2))); # matrices Sg = np.block( [[Sg11, Sg12], [Sg21, Sg22]]); # initialization is equivelant as that for S_reflection side matrix ### ================= Working on the Reflection Side =========== ## Pr, Qr, kzr = pq.P_Q_kz(kx, ky, e_r, m_r) ## ============== values to keep track of =======================## S_matrices = list(); kz_storage = [kzr]; X_storage = list(); ## ==============================================================## # define vacuum wavevector k0 lam0 = wavelengths[i]; # k0 and lam0 are related by 2*pi/lam0 = k0 k0 = 2*np.pi/lam0; ## modes of the layer Om_r = np.matrix(cmath.sqrt(-1) * kzr * I); X_storage.append(Om_r); W_ref = I; V_ref = Qr * Om_r.I; # can't play games with V like with W because matrices for V are complex ## calculating A and B matrices for scattering matrix Ar, Br = sm.A_B_matrices(Wg, W_ref, Vg, V_ref); S_ref, Sr_dict = sm.S_R(Ar, Br); # scatter matrix for the reflection region S_matrices.append(S_ref); Sg, D_r, F_r = rs.RedhefferStar(Sg, S_ref); ## go through the layers for i in range(len(ER)): # ith layer material parameters e = ER[i]; m = UR[i]; # longitudinal k_vector P, Q, kzl = pq.P_Q_kz(kx, ky, e, m) kz_storage.append(kzl) ## E-field modes that can propagate in the medium W_i = I; ## corresponding H-field modes. Om = cmath.sqrt(-1) * kzl * I; X_storage.append(Om) V_i = Q * np.linalg.inv(Om); # now defIne A and B A, B = sm.A_B_matrices(Wg, W_i, Vg, V_i); # calculate scattering matrix S_layer, Sl_dict = sm.S_layer(A, B, layer_thicknesses[i], k0, Om) S_matrices.append(S_layer); ## update global scattering matrix using redheffer star Sg, D_i, F_i = rs.RedhefferStar(Sg, S_layer); ##========= Working on the Transmission Side==============## Pt, Qt, kz_trans = pq.P_Q_kz(kx, ky, e_t, m_t); kz_storage.append(kz_trans); Om = cmath.sqrt(-1) * kz_trans * I; Vt = Qt * np.linalg.inv(Om); # get At, Bt At, Bt = sm.A_B_matrices(Wg, I, Vg, Vt) ST, ST_dict = sm.S_T(At, Bt) S_matrices.append(ST); # update global scattering matrix Sg, D_t, F_t = rs.RedhefferStar(Sg, ST); K_inc_vector = n_i * k0 * np.matrix([np.sin(theta) * np.cos(phi), \ np.sin(theta) * np.sin(phi), np.cos(theta)]); # cinc is the c1+ E_inc, cinc, Polarization = ic.initial_conditions(K_inc_vector, theta, normal_vector, pte, ptm) ## COMPUTE FIELDS Er = Sg[0:2, 0:2] * cinc; # S11; #(cinc = initial mode amplitudes), cout = Sg*cinc; #2d because Ex, Ey... Et = Sg[2:, 0:2] * cinc; # S21 Er = np.squeeze(np.asarray(Er)); Et = np.squeeze(np.asarray(Et)); Erx = Er[0]; Ery = Er[1]; Etx = Et[0]; Ety = Et[1]; # apply the grad(E) = 0 equation to get z components Erz = -(kx * Erx + ky * Ery) / kzr; Etz = -(kx * Etx + ky * Ety) / kz_trans; ## using divergence of E equation here # add in the Erz component to vectors Er = np.matrix([Erx, Ery, Erz]); # a vector Et = np.matrix([Etx, Ety, Etz]); R = np.linalg.norm(Er) ** 2; T = np.linalg.norm(Et) ** 2; ref.append(R); trans.append(T); return ref, trans
def run_TMM_anisotropic(wavelengths, polarization_amplitudes, theta, phi, ER, UR, layer_thicknesses,\ transmission_medium, incident_medium): ref = []; trans = [] wvlen_scan = np.linspace(0.5, 4, 1000); thickness = 0.5; # 1 layer for wvlen in wvlen_scan: k0 = 2*np.pi/wvlen; kx = np.sin(theta)*np.cos(phi); ky = np.sin(theta)*np.sin(phi); # we will build the system by rows? a11 = -1j*(ky*mu_tensor[1,2]/mu_tensor[2,2] + kx*(epsilon_tensor[2,0]/epsilon_tensor[2,2])) a12 = 1j*kx*(mu_tensor[1,2]/mu_tensor[2,2] - epsilon_tensor[2,1]/epsilon_tensor[2,2]); a13 = kx*ky/epsilon_tensor[2,2] + mu_tensor[1,0] - mu_tensor[1,2]*mu_tensor[2,0]/mu_tensor[2,2]; a14 = -kx**2/epsilon_tensor[2,2] + mu_tensor[1,1]- mu_tensor[1,2]*mu_tensor[2,1]/mu_tensor[2,2]; a21 = 1j* ky *(mu_tensor[0,2]/mu_tensor[2,2] - epsilon_tensor[2,0]/epsilon_tensor[2,2]); a22 = -1j * kx*(mu_tensor[0,2]/mu_tensor[2,2]) +ky *(epsilon_tensor[2,1]/epsilon_tensor[2,2]); a23 = ky**2/epsilon_tensor[2,2] - mu_tensor[0,0] + mu_tensor[0,2]*mu_tensor[2,0]/mu_tensor[2,2]; a24 = -kx*ky/epsilon_tensor[2,2] - mu_tensor[0,1] + mu_tensor[0,2]*mu_tensor[2,1]/mu_tensor[2,2]; a31 = (kx*ky/mu_tensor[2,2] + epsilon_tensor[1,0] - epsilon_tensor[1,2]*epsilon_tensor[2,0]/epsilon_tensor[2,2]) a32 = (-kx**2/mu_tensor[2,2] +epsilon_tensor[1,1] - epsilon_tensor[1,2]*epsilon_tensor[2,1]/epsilon_tensor[2,2]); a33 = -1j*(ky*(epsilon_tensor[1,2]/epsilon_tensor[2,2])+kx*(mu_tensor[2,0]/mu_tensor[2,2])); a34 = 1j*kx*(epsilon_tensor[1,2]/epsilon_tensor[2,2]-mu_tensor[2,1]/mu_tensor[2,2] ) a41 = ky**2/mu_tensor[2,2] - epsilon_tensor[0,0] + epsilon_tensor[0,2]*epsilon_tensor[2,0]/epsilon_tensor[2,2]; a42 = -kx*ky/mu_tensor[2,2] - epsilon_tensor[0,1] + epsilon_tensor[0,2]*epsilon_tensor[2,1]/epsilon_tensor[2,2]; a43 = 1j*ky*(epsilon_tensor[0,2]/epsilon_tensor[2,2]-mu_tensor[2,0]/mu_tensor[2,2] ); a44 = -1j*(kx*(epsilon_tensor[0,2]/epsilon_tensor[2,2])+ky*(mu_tensor[2,1]/mu_tensor[2,2])); A = np.matrix([[a11, a12, a13, a14], [a21, a22, a23, a24], [a31, a32, a33, a34], [a41, a42, a43, a44]]); #print(np.linalg.cond(A)) eigenvals, eigenmodes = np.linalg.eig(A); rounded_eigenvals = np.round(eigenvals, 3) sorted_eigs, sorted_inds = nonHermitianEigenSorter(np.round(eigenvals,10)); ## ======================================================== W_i = eigenmodes[0:2, sorted_inds]; V_i = eigenmodes[2:, sorted_inds]; Om = np.matrix(np.diag(sorted_eigs) ); #print(np.round(W_i,3), np.round(V_i,3), np.round(Om,3)) #then what... match boundary conditions... try using the gaylord formulation. # or use the scattering matrix formalism, where we still, technically deal with all field components... Sg11 = np.matrix(np.zeros((2, 2))); Sg12 = np.matrix(np.eye(2, 2)); Sg21 = np.matrix(np.eye(2, 2)); Sg22 = np.matrix(np.zeros((2, 2))); # matrices Sg = np.block([[Sg11, Sg12], [Sg21, Sg22]]); # initialization is equivelant as that for S_reflection side matrix ### ================= Working on the Reflection Side =========== ## Pr, Qr, kzr = pq.P_Q_kz(kx, ky, e_r, m_r) ## ============== values to keep track of =======================## S_matrices = list(); kz_storage = [kzr]; X_storage = list(); ## ==============================================================## # define vacuum wavevector k0 lam0 = wvlen; # k0 and lam0 are related by 2*pi/lam0 = k0 k0 = 2*np.pi/lam0; ## modes of the layer Om_r = np.matrix(cmath.sqrt(-1) * kzr * I); X_storage.append(Om_r); W_ref = I; V_ref = Qr * Om_r.I; # can't play games with V like with W because matrices for V are complex #print(Om_r) ## calculating A and B matrices for scattering matrix Ar, Br = sm.A_B_matrices(Wg, W_ref, Vg, V_ref); S_ref, Sr_dict = sm.S_R(Ar, Br); # scatter matrix for the reflection region S_matrices.append(S_ref); Sg, D_r, F_r = rs.RedhefferStar(Sg, S_ref); # longitudinal k_vector ## ============ WORKING INSIDE ANISOTROPIC LAYER ================# # now defIne A and B Al, Bl = sm.A_B_matrices(Wg, W_i, Vg, V_i); # calculate scattering matrix S_layer, Sl_dict = sm.S_layer(Al, Bl, thickness, k0, Om) S_matrices.append(S_layer); ## update global scattering matrix using redheffer star Sg, D_i, F_i = rs.RedhefferStar(Sg, S_layer); ##========= Working on the Transmission Side==============## Pt, Qt, kz_trans = pq.P_Q_kz(kx, ky, e_t, m_t); kz_storage.append(kz_trans); Omt = cmath.sqrt(-1) * kz_trans * I; Vt = Qt * np.linalg.inv(Omt); # get At, Bt At, Bt = sm.A_B_matrices(Wg, I, Vg, Vt) ST, ST_dict = sm.S_T(At, Bt) S_matrices.append(ST); # update global scattering matrix Sg, D_t, F_t = rs.RedhefferStar(Sg, ST); K_inc_vector = n_i * k0 * np.matrix([np.sin(theta) * np.cos(phi), \ np.sin(theta) * np.sin(phi), np.cos(theta)]); # cinc is the c1+ E_inc, cinc, Polarization = ic.initial_conditions(K_inc_vector, theta, normal_vector, pte, ptm) ## COMPUTE FIELDS Er = Sg[0:2, 0:2] * cinc; # S11; #(cinc = initial mode amplitudes), cout = Sg*cinc; #2d because Ex, Ey... Et = Sg[2:, 0:2] * cinc; # S21 Er = np.squeeze(np.asarray(Er)); Et = np.squeeze(np.asarray(Et)); Erx = Er[0]; Ery = Er[1]; Etx = Et[0]; Ety = Et[1]; # apply the grad(E) = 0 equation to get z components, this equation comes out of the longitudinal equation # or the divergence equation and is valid since the transmission region is VACUUM (as is the reflection region) Erz = -(kx * Erx + ky * Ery) / kzr; #uses the divergence law Etz = -(kx * Etx + ky * Ety) / kz_trans; ## using divergence of E equation here # add in the Erz component to vectors Er = np.matrix([Erx, Ery, Erz]); # a vector Et = np.matrix([Etx, Ety, Etz]); R = np.linalg.norm(Er) ** 2; T = np.linalg.norm(Et) ** 2; ref.append(R); trans.append(T); return ref, trans;