How to extend to handle multiples eigenstates/eigenphases?

[SamDuffield]

For QPE, it is common for the input state to be contaminated with excited states which will have their own eigenphases.

Could potentially use a weighted mixture of Fourier/von Mises distributions to encode this, similar to the approach of van den Berg.