After some more thinking, I’m still having trouble following this logic:
Let’s say I have a quantum search operator on a quantum computer and it turns out that 0000 are not the bits I’m looking for. Within that branch, 0000 splits again into an up-branch that doesn’t change into a cat, and a down-branch that changes into a cat, rotates a bit faster or slower, and then changes back out of a cat. Now we have two amplitudes in opposite phase so the whole quantum branch has deliberately decided to cancel itself out.
Specifically, I am not sure in what sense he uses the word “branches”. If this is an MWI concept, then different branches do not cancel, since they do not interact. Maybe it means different additive terms in the wave function of some subsystem? But those correspond to different orthogonal eigenstates and so they don’t cancel out, either. Maybe it is meant in terms of constructive/destructive interference, only with the destructive part in one place not being compensated by the constructive part elsewhere? This interpretation at least makes sense if you associate branches with propagation paths, but I still have no clue how to use time-dependent energy states to terminate rather than displace (in a perfectly sub-light way) interference maxima and minima.
Maybe someone else can speculate more successfully.
After some more thinking, I’m still having trouble following this logic:
Specifically, I am not sure in what sense he uses the word “branches”. If this is an MWI concept, then different branches do not cancel, since they do not interact. Maybe it means different additive terms in the wave function of some subsystem? But those correspond to different orthogonal eigenstates and so they don’t cancel out, either. Maybe it is meant in terms of constructive/destructive interference, only with the destructive part in one place not being compensated by the constructive part elsewhere? This interpretation at least makes sense if you associate branches with propagation paths, but I still have no clue how to use time-dependent energy states to terminate rather than displace (in a perfectly sub-light way) interference maxima and minima.
Maybe someone else can speculate more successfully.