Learning Forum

[partha.nakti]

Hello,

I am working on an inverse-designed 2x2 3-dB SWG-based power splitter using Lumopt, aiming for equal power split between the bar and cross ports. To achieve this, I set the figure of merits (FOMs) to 0.5 for each port and used two separate FOM monitors (fom1 and fom2) to track the power distribution. However, I consistently observe that the FOM at the bar port exceeds that at the cross port, leading to an imbalance.

Could you advise on whether it’s possible to introduce an additional FOM or constraint to minimize this mismatch? Any guidance on adjusting the optimization setup to achieve a more balanced output would be greatly appreciated.

Thank you very much for your assistance.

-Partha

[Guilin Sun]

The optimization algorithm is only for one target, not two targets foms. Therefore you will need to give them the weights:

https://optics.ansys.com/hc/en-us/articles/1500010507742-Inverse-Design-of-a-Grating-Coupler-with-co-opt-Weights

eg:

[partha.nakti]

Hi Guilin,

Thank you so much for your reply. For a 3dB splitter, the fom_bar and fom_cross portions of the code would be similar, and I would assign a weight of 0.5 to each. Is this what you're suggesting? Please refer to the attached picture for clarification.

Thank you.

.

[Guilin Sun]

seems good. Please try. Please remember that the actual FOM is olnly one  scalar quantity.

[partha.nakti]

Thank you for your reply. However, the configuration shown in the screenshot I provided in my last message isn’t working. As you mentioned, "the actual FOM is only one scalar quantity." Does this mean that in the part

"target_T_fwd = lambda wl: 0.5 * np.ones(wl.size)"

0.5 cannot be used here, and it must always be a single integer? Could you please elaborate on this?

Thank you.

[Guilin Sun]

Please note the weights are passed to proper variables. You do not need to modify other script. Please visit the example online for more details.