When I prepare a state for use in a QNN I can use embeddings.AngleEmbedding and rotate using RX, RY or RZ.
Here, at Equation 14 it shows what it also describes as angle encoding. Am I correct in thinking this is something slightly different to what is being carried out by embeddings.AngleEmbedding? They go on to describe U_i, which is very similar to RY - is equation 14 showing the same thing as perform AngleEmbedding with RY?
The one thing Iām not sure on is the j vs i sub/superscripts - although my understanding is that x_j^(i) is the i-th component/feature of a feature vector j.
It may also be worth getting into contact with the authors if you are unsure.