NO SIGNALING

Bell inequality violates local realism by quantum measurements. However, it assumes lack of signaling between the parties. Each party completes its measurement before it receives the information about the choice of the measurement of the other party. It is normally forbidden for distances and times unavailable by the speed of light limit by generally accepted relativistic axioms. However, in practice, simple lack of interaction should suffice. Quantum computers, like IBM, have distances and timescales below relativistic limit, but still no reasonable interation should occur between the parties. Nevertheless, the tests violate no-signaling in the context of IBM architecture at a very high confidence level.

Before publishing in Advanced Quantum Technologies, our result underwent peer-review at Physical Review Applied. Their referee wrote:

The paper demonstrates violations of a CHSH inequality on IBM Q computers, and claims to demonstrate the violation of no-signaling. However, it is known that there are (unwanted) interactions between qubits when measuring multiple qubits in a superconducting computer -- i.e., the qubits and/or the measurements pulses controlling the qubits are interacting during the measurement (often called measurement-induced crosstalk). Seeing a violation of no-signaling in a CHSH test on an IBM Q computer therefore tells us nothing surprising at all, because the qubits (and/or readout lines) are simply interacting according to ordinary quantum (and/or classical) physics, and nothing about this is inconsistent with our knowledge of quantum theory (e.g., that there cannot be superluminal signaling). There are various papers studying measurement-induced crosstalk in superconducting systems, e.g., the following paper from IBM scientists: https://iopscience.iop.org/article/10.1088/1367-2630/ad0e19/pdf.

Comment:
Instead of asking for a second round at Physical Review Applied, we transferred the paper to AQT. Nevertheless we completely disagree with the reviewer.

• We have distance 2 and 4 at our test while existing tests, including the one in the mentioned paper, have only distance 1 case
• The observed measurement crosstalk in the mentioned paper is $<{10}^{-4}$ while our observation gives $\sim {10}^{-2}$
• ZZ crosstalk and the model in the mentioned paper is unlikely as the settings differ only by phases
• there are no mid-circuits measuremenets
• the mentioned paper (and any other) does not address signaling in Bell-type tests on IBMQ or other quantum computers

[1] Bell theorem
[2] Wightman axioms;