I am having difficulties understanding this paper (arXiv:1910.12390), so asking this question.
As far as what I see from mathematics of this paper, all improvements over Grover's search disappear when probability of obtaining a postselected state is accounted for. But this is not what this paper seems to claim. After all, if the improvement disappears after accounting for probability of the postselected state, and we already know that PostBQP contains NP, what the paper aims for seems vacuous.
So is this paper really vacuous, or am I understanding something wrongly?
