Learning Forum

[iqnlo]

Hi all

I'm doing the mode convergence test following this instruction:

Convergence testing process for EME simulations – Lumerical Support

But there is something wrong in the codes:

pol_converter (1);

Error: C:/Users/Downloads/pol_converter (1).lsf line 63: Can not find specified member 's52' in dataset 'S_mode_convergence_sweep'

Below is my question:

Why the transmission efficiency is called S52? The transmission should be s12 (forward propagation) or s21 (backward propagation). And also there is only s12, s21, s22, s11 in the matrix S.

[iqnlo]

By the way, the first 65 lines are:
# convergence with number of cells in tapered region
if(1){
switchtolayout;

# reset values of other settings in case convergence testing was
# performed previously for other settings
setnamed("EME","number of modes for all cell groups",10);
setnamed("mesh","dy",0.01e-6);
setnamed("mesh","dz",0.01e-6);

npts = 10; # number of points to sweep
ncells = linspace(2,20,npts); # values to sweep

S52 = matrix(npts); # initialize result matrix

for(i = 1:npts){ # sweep
switchtolayout;
setnamed("EME","cells",[1;ncells(i);1]);
run;
emepropagate;
S = getresult("EME","user s matrix");
S52(i) = S(2,1);
}

plot(ncells,abs(S52)^2,"Number of cells in tapered region" "|S52|^2");

# delta_T
delta_T = (abs(S52(2:npts))^2-abs(S52(1:npts-1))^2)^2/abs(S52(2:npts))^4;

# delta_T_N
delta_T_N = (abs(S52)^2-abs(S52(npts))^2)^2/abs(S52(npts))^4;

# plot delta_T and delta_T_N
plotxy(ncells(2:npts),delta_T, ncells,delta_T_N "Number of cells in tapered region","delta_T");
legend("delta_T","delta_T_N");
}

# the convergence with number of eigenmodes
if(1){
switchtolayout;

# reset values of other settings in case convergence testing was
# performed previously for other settings

setnamed("mesh","dy",0.01e-6);
setnamed("mesh","dz",0.01e-6);

start_mode = 4; #set smaller number of modes for convergence test
mode_interval = 1; #set mode interval for convergence test

setnamed("EME","number of modes for all cell groups",25); #should be the largest number of modes desired for test

run;
setemeanalysis("mode convergence sweep", 1);
setemeanalysis("start mode", start_mode);
setemeanalysis("mode interval", mode_interval);
emesweep("mode convergence sweep");

S = getemesweep("S_mode_convergence_sweep");
nmodes = S.modes;
S52 = S.s52;
npts = length(S52);

plot(nmodes,abs(S52)^2,"Number of eigenmodes","|S52|^2");

[Guilin Sun]

You are right that the script does not fit the simulation file. I guess originally the simulation file has 5 modes in the two ports so it has 5by5 S matrix. Current simulation file has one mode at each port, so please modify
S52(i) = S(2,1);
This will peak S21 even the variable name is called S52. If you want, please modify the output variable name.

[iqnlo]

Thanks for you reply!
The problem was solved.
But I meet another problem when I'm applying this code on my own structure (EME solver). In my original project, the "allow custom eigensolver settings" box need to be checked. Because I need to select the mode manually (also define the refractive index manually) rather than let the software choose mode at the highest refractive index. However for applying your codes, I need to uncheck this box and let the codes select the default mode. This make my simulation give a wrong result.

[Guilin Sun]

Ideally each topic only contains one question so we can suggest one solution. In this way the forum is more convenient for others to find the solution if they have the same issue, which is the main purpose of the forum. So please write a new post next for any new questions.
The online script file has 3 different sweeps. So if you want manually set something, please do it before the sweep, and make sure the sweep does not need to modify what you set. For example, only when you sweep the number of cells, then the cell property needs to be specified. If you sweep the mesh size, or EME size (yz spans), it does not change the cell property. If the modes need to be specified manually, then you may not be able to use the script. In this case, you will do one simulation at a time and collect the results for later use.
Please note that EME has a built-in mode number sweep. probably you can try it.