Learning Forum

[NFrench]

I have been running a simulation where I am interested in the power coupling between a gaussian beam and the modes within a fiber. However, the program returns different results every time I run it. Below are two images of the identical simulation. I believe the simulation should be returning the exact same results every time I run it, so I am confused why the power coupling and TE polarization fraction are different for these two runs.

[Guilin Sun]

As you can see the TE Polarization value changes, which is an unfortunate result from degeneracy. In order to fix this issue, you will need to break the degeneracy. A direct way doing this is to choose symmetry BCs: Symmetric and anti-symmetric BCs in FDTD and MODE
example: Photonic crystal cavity

[NFrench]

Gotcha. I will try that, thanks. Perhaps I am misinterpreting the document that you have linked, but I was under the impression that symmetry BC are used to increase computation time. Shouldn't using symmetric/anti-symmetric BC give identical results for the simulation?

[Guilin Sun]

I guess you mean "ymmetry BC are used to DECREASE computation time" which is correct.
yes, you are right that using symmetric/anti-symmetric BC give identical results for the simulation, as long as it gives non-rotated mode under normal PML BCs. Please compare the two cases.
Using symmetric/anti-symmetric BCs is to pick up the modes that satisfy the BCs. It will only give the modes that have such symmetry property.

[NFrench]

Okay thanks for the help! I'll have to read up on how to best use symmetric/anti-symmetric BC. If they return identical results though, shouldn't I still have this issue? Or does adding symmetric/anti-symmetric BC break the degeneracy and thereby remove this error while keeping power coupling calculations/everything else about the simulation the same. I suppose I am still a bit confused and curious why changing the boundary would fix this problem I have been having.

[Guilin Sun]

adding symmetric/anti-symmetric BC break the degeneracy and thereby remove this error while keeping power coupling calculations/everything else about the simulation the same.
Identical can have different understanding. I mean it is equivalent to rotate the modes to have a polarization along the axis, eg, the dominant polarization is along one axis according to the symmetry, since your original result is the combination of the two polarizations.
please try and compare the results under different BCs.