When I write a *.dll such as n(x,y,z) = a*x → where I expect a linear GRIN (non radially symmetric), instead what happens is that the rays do not bend at all, and the result is modeled as wavefront error with no ray deflection. I made sure to mark it as non-radially symmetric.

However, I do know that GRIN4 exists and works. But, there is no example of GRIN4 in sequential mode under surfaces. Any tips for what I might be doing wrong? I did double check all the derivatives, and an analogous version of my code works in non-sequential mode.