Classical experiment violates Bell

[DirkMan]

"A unifying principle explaining the numerical bounds of quantum correlations remains elusive, despite the efforts devoted to identifying it. Here, we show that these bounds are indeed not exclusive to quantum theory: for any abstract correlation scenario with compatible measurements, models based on classical waves produce probability distributions indistinguishable from those of quantum theory and, therefore, share the same bounds. We demonstrate this finding by implementing classical microwaves that propagate along meter-size transmission-line circuits and reproduce the probabilities of three emblematic quantum experiments. Our results show that the “quantum” bounds would also occur in a classical universe without quanta. The implications of this observation are discussed."

http://journals.aps.org/prl/abstract/10.1103/PhysRevLett.116.250404

http://arxiv.org/abs/1511.08144

I found this interesting

"This shows that
quantum correlations can be
universally
recreated with
classical systems at the expense of some extra resources...extra memory...one bit of memory for each dichotomic decision ..."

 

[Heinera]

Their experiments are not Bell-type experiments. Among other things, their setup is manifestly non-local in the Bell sense, since results from the first measurement are passed on to the second measurement.

 

[Strilanc]

The paper didn't violate the Bell inequalities with a classical system. At least, not in the "we can pass Bell tests in real life" sense. Their pieces don't correspond to pieces from a Bell test, their pieces correspond to outcomes of a Bell test.

In the paper they say:

The classical equivalents to quantum states [that we use] are multichannel microwave signals propagating along independent waveguides with well-defined relative phases. Each classical microwave channel is identified with an element of the Hilbert space basis.

You know how, classically, three bits can be in the state 000, 001, 010, 011, 100, 101, 110, or 111 and so, quantumly, three qubits can be in any linear combination of those eight states? The authors didn't make a classical system with $n$ (3) pieces, one for each qubit. They made a system with $2^n$ (8) pieces, one for each basis state. Each basis state specifies all parts of the system, so their pieces are inherently non-local w.r.t. the system they are supposed to represent.

(To apply a local operation within the simulated system, they need to make all eight pieces do something. Only making one piece do something would apply a non-local operation within the system.)

In other words, they made an esoteric simulator and then avoided just saying that directly in the summary. They didn't violate the Bell inequalities anymore than I did by using Quirk to make this:

(Oh gosh, alert the presses! I s̶i̶m̶u̶l̶a̶t̶e̶d did something that's not possible classically!)

 

[Nugatory]

OP will not be back to answer follow-up questions, so this thread is closed.
PM any mentor if you want it reopened so that you can add to what Strilanc and Heinara have already said.