results of optimization have a incorrect size ,what does it means?


new2.fsp (5.3 MB)
i am simulating a uniform grating coupler to caculate coefficient and optimizate,please help me check what is wrong with my structure and optimization setup,thanks.


Hi @chenzl23,

You get this error because the figure of merit in an optimization can only be a number. In your case, you are trying to optimize “expansion for t” that contains 8 different results, all of them defined over the bandwidth of your source.

To use the optimization tool, you have to select one of these results (for instance, “T_forward” that gives the power coupled in the TM mode of the fiber) and extract a value from this result (for instance, the maximum transmission). to extract such result, you can use an analysis group.

I hope this helps.


put optimization aside. is it right to get coupling coefficient by T_forward? help me check my structure and source setup.can TE mode source transfer in fiber ? and so on.


The choice of T_Forward or T_Backward depends on what you want to get and also the direction of propagation. In your case, since the light will propagate from the grating coupler to the fibre in the positive y direction, T_Forward will give you the coupling into the mode you selected (fundamental TM mode) in that direction. If you want to see the coupling to the TE mode, you need to select the corresponding mode. You can find additional information about the mode expansion monitors and how to use them at

Also, in our last release, we introduced a new object, the ports, that is using a combination of a mode source, field monitors and a mode expansion monitor to facilitate the S-parameter extraction.


i think the TM in FDTD mode source is actually TE mode. because TE mode is defined as its most electric field in Z(propagate in x direction ),which is the same as TM in fdtd solutions . is it right? fiber only propagate the TE(TM in fdtd) mdoe , so i choice the TM to caculate the coupler coefficient.


The definition of TM and TE modes is very user dependent! In our case, you are correct, we defined the TE mode as the mode with most E field in the x direction (propagation in the y direction).


so, is my simulation about coupler coefficient right? because the grating structure have no paper to prove,and i have to ask you lumerical team to prove my thoughts. please give me a answer, is the coefficient what light couple from waveguide to fiber?thanks.
Best wishes.


Hi @chenzl23,

Sorry for the late reply, I missed your last post! Overall, the simulation seems correctly set, but I would make sure the mode source and monitors (including the mode expansion) are large enough to not truncate the fields. There’s probably some elements you could delete as they are redundant (for instance, there is a mode expansion monitor in the fibre group).

Also, you are using dielectric object only defined by the value of their index. This type of material should be using only in single wavelength/frequency simulation.

The mode source is defined at one wavelength. It means the source will inject the same mode profile at all wavelength. This is usually a good approximation if the mode profile doesn’t vary. If needed, you an used the multi-frequency mode calculation that was introduced in the latest version of FDTD Solutions. In the mode source, check the box as shown and set the number of points:

You can find additional information about this here.

Finally, once the simulation is checked and you think correctly set, I would do a convergence testing to see the effect of the mesh size on the result. I would used mesh override object to specify a smaller mesh over the grating.