Mapping the phase distribution to the physical structure of the lens

[LumaFilter]

From your published paper, I can see the lens parameters that you've trained are as follows:

phase_initial = np.array([-0.3494864 , -0.00324192, -1., -1.,-1., -1., -1., -1.], dtype=np.float32)

Plugging these parameters into the phase distribution polynomial yields the phase distribution. However, I'm wondering how to map this phase distribution to the physical structure of the lens. Could you provide some guidance and references on how to proceed with this mapping?

[LumaFilter]

In supplementary you mentioned: "We restrict the parameter range for all phase coefficients to be within [−1000, 1000], and during optimization we normalize the range to be within [−1, 1].", that means the coefficients used to fabricate lens are [-349.4864, -3.24192, -1000.0, -1000.0, -1000.0, -1000.0, -1000.0, -1000.0]. But the phase profile will be very large and Far more than [0,2pi], how do we restrict the phase to [0,2pi]?

[0marMaher]

I have encountered difficulties when trying to optimize both components together and would appreciate any insights into this aspect of the training process.

Did you succeed in training the entire optimization cycle, with the metasurface coefficients and network simultaneously optimized?

[LumaFilter]

I have encountered difficulties when trying to optimize both components together and would appreciate any insights into this aspect of the training process.

Did you succeed in training the entire optimization cycle, with the metasurface coefficients and network simultaneously optimized?

No，I can't train normally either. I have described it here before.