Guilin Sun
Ansys Employee

I believe the analytical solution has some limitations and ignore some quantities.  Δn=In2 ~=0.69 will be too large to index perturbation, even for the simulation.

The simulation takes more terms and for such strong nonlinearity, anything can happen.

I suggest to check the first principle and make sure the analytical solution does not lose any thing, and does not voilate the original assumption.