Are so-called non-local interactions really non-local?

MeJennifer

There is something I do not quite understand with regards to the so-called non-locality problem in EPR like experiments.

The wave function propagates at c, so even when two particles, that are part of the same quantum system, move in opposite directions they are still connected. They both travel along a null interval and thus there is a causal connection between the initial state and the measurement.

The same with a single photon, it gets emitted at one place and absorbed at another place but the distance is exactly zero.

What am I missing?

vanesch

I don't think one can in a reasonable way say that "the wavefunction propagates at c", given that it is not defined over normal euclidean (or minkowski) space, but over configuration space.


If a is a null vector, and b is a null vector, then a - b doesn't need to be a null vector. In the case of EPR, it is a spacelike vector.

Now, of course, you can consider that there is a null-curve linking the events A and B ; however, EVERY two events can be linked by a null curve ! However, you will have to walk this null curve sometimes with a dt > 0, and sometimes with a dt < 0.

So, yes, if you allow for "backward propagation of light signals" then every event is local to every other event. But that's not the idea, in general, although certain people look upon it that way (although I'm not an expert, the transactional view on QM does something of the kind I think).

MeJennifer

How do you demonstrate backpropagation on null intervals?
That seems to me a Lorentz variant assumption.

vanesch

Well, if Joe sends out a light pulse to Jack, I think that all observers can testify about the direction in which the lightpulse propagated. Nobody will see the lightpulse propagate from Jack to Joe.

This comes about because you cannot transform a lorentz vector (t,x,y,z) with t>0 into one with t<0 through a continuous lorentz transformation, EVEN when the vector is lightlike.