Weih's data: what ad hoc explanations do local and non-local models give?

[Gordon Watson]

That is surely a very specious definition of "non-locality" almost not worthy of any attention. Why not simply stick to "violating a Bell inequality", why use the term "non-locality" at all, rather than any other such as "imaginary" or "non-sense"?

Bill: Happily, in addition to Einstein-locality: something else that we agree about! Nicely put! GW

 

[gill1109]

Bill Schnieder, Gordon Watson: I don't like the terminology either. But: "what's in a name"? It is the *results* of Masanes et al. which are interesting. As for the terminology: this has nowadays become fairly standard in quantum information. Like it or not, it's what people say. It's become a technical term with an established technical meaning.

 

[Gordon Watson]

Bill Schnieder, Gordon Watson: I don't like the terminology either. But: "what's in a name"? It is the *results* of Masanes et al. which are interesting. As for the terminology: this has nowadays become fairly standard in quantum information. Like it or not, it's what people say. It's become a technical term with an established technical meaning.

Hi Richard, fair enough, and thanks for sharing your dislike. But when "non-locality" is so often "what people say" ... and when it is then so often taken or understood to negate or refute the well-established Einstein-locality ... one of my favourite classical principles ... and when the established technical meaning is not so clearly spelt out, as Bill suggests ... confusion tends to reign, imho. Witness the many on-going discussions on the subject. With thanks again, Gordon.

 

[harrylin]

[..] it will be good to clearly state the main competing explanations for high correlations with small time windows. [..]

common "local-realist" models: no "spooky" influence at a distance, nor wild science fiction.
Strangely enough I know only one such model, of De Raedt et al: influence of polarisation on detection time delay. This influence is inferred from QM predictions as well as from independent experiments, based on an assumed "local-realist" universe. [..]
For further discussion it will be particular interesting to compare De Raedt et al's model against QM (= all the other proposed models, I suppose) with Weihs' data. For that we should know more precisely:

- the basis for the 5%
- the basis for the 30 μs
- the exact formula that De Raedt et al used for their most successful simulations (see the thread on De Raedt's simulations)
- the possible delay time range according to the literature, insofar as this has been observed
- ?

Harald

I now came a little bit further with the possible variation in delay time as function of polarization according to the literature.

My first suspect in Weih's experiment was birefringence in his photonic crystal. However that was a dead-end road, for he writes in Arxiv9810080v1:

"we pump a BBO-crystal with 400 mW of 351 nm light
from an Argon-ion-laser. A telescope was used to nar-
row the UV-pump beam [12], in order to enhance the
coupling of the 702 nm photons into the two single-
mode glass fibers. On the way to the fibers, the pho-
tons passed a half-wave plate and the compensator crys-
tals necessary to compensate for in-crystal birefringence"

The next obvious suspect is the electro-optic modulator that he used, as these are commonly made of similar materials. About that he comments in that same paper which was published in Phys. Rev. Letters:

"Each of the observers switched the direction of local
polarization analysis with a transverse electro-optic modulator.
It’s optic axes was set at 45◦ with respect to the
subsequent polarizer. Applying a voltage causes a rotation
of the polarization of light passing through the modulator
by a certain angle proportional to the voltage [13].
For the measurements the modulators were switched fast
between a rotation of 0° and 45°. [..]
The total of the delays occurring in the electronics and
optics of our random number generator, sampling circuit,
amplifier, electro-optic modulator and avalanche photodiodes
was measured to be 75 ns. [..]

[13]Precisely speaking, the modulator introduces a phase
shift between the linearly polarized components parallel
and perpendicular to its optic axis (at 45°). Together
with two quarter-wave plates (at 0° or 90°) before and
after the modulator this results in a polarization rotation
in real space as usually seen in circularly birefringent
media. The latter quarter-wave plate can be abandoned
here because it is parallel to the axis of the subsequent
polarizer and thus introduces only a phase which cannot
be measured anyway. The quarter-wave plate in front of
the modulator is substituted by our fiber and the initial
polarization controllers."

To my regret, I do not understand this. What type of electro-optic modulator did he use, and how did he account for its birefringence?
Can someone else perhaps explain this to me?

 

[audioloop]

why not use only one pair of entangled electrons ?
just that.

 

[harrylin]

I now came a little bit further with the possible variation in delay time as function of polarization according to the literature.

My first suspect in Weih's experiment was birefringence in his photonic crystal. However that was a dead-end road [..] The next obvious suspect is the electro-optic modulator that he used, as these are commonly made of similar materials. [..] What type of electro-optic modulator did he use, and how did he account for its birefringence?
Can someone else perhaps explain this to me?

As there were no comments I will clarify the above: apparently he did not consider the spread in time delays due to the EOM's birefringence. However, that should be done (and could be done!) in order to interpret the experimental findings.

 

[DrChinese]

Peter Morgan has been working with the raw Weihs data for several years. He has done some deep analysis of it, and today posted some results in the arxiv:

A graphical presentation of signal delays in the datasets of Weihs et al
Peter Morgan (2012)

http://arxiv.org/abs/1207.5775

"A graphical presentation of the timing of avalanche photodiode events in the datasets from the experiment of Weihs et al. [Phys. Rev. Lett. 81, 5039 (1998)] makes manifest the existence of two types of signal delay: (1) The introduction of rapid switching of the input to a pair of transverse electro-optical modulators causes a delay of approximately 20 nanoseconds for a proportion of coincident avalanche photodiode events; this effect has been previously noted, but a different cause is suggested by the data as considered here. (2) There are delays that depend on in which avalanche photodiode an event occurs; this effect has also been previously noted even though it is only strongly apparent when the relative time difference between avalanche photodiode events is near the stated 0.5 nanosecond accuracy of the timestamps (but it is identifiable because of 75 picosecond resolution). The cause of the second effect is a difference between signal delays for the four avalanche photodiodes, for which correction can be made by straightforward local adjustments (with almost no effect on the degree of violation of Bell-CHSH inequalities)."

 

[Peter Morgan]

Peter Morgan has been working with the raw Weihs data for several years. He has done some deep analysis of it

Hi DrC. "For several years" only in elapsed time. In committed time perhaps a few months. I have to disagree with "deep", though I think the paper I uploaded yesterday is kinda cute for its data visualization, which is different from anything I've seen other people do.

I was prompted to post this paper (the computation for which I did about two years ago) to the arXiv by Alejandro Hnilo, whose research group in Argentina has just finished a Bell-type experiment in which they use a pulsed laser. The distances are relatively short, but they record the timings of the laser pulses as well as the timings of Alice's and Bob's measurement events. They have posted their dataset privately, and it may become publicly available in time; they're currently working on their analysis.

 

[harrylin]

[..] A graphical presentation of signal delays in the datasets of Weihs et al Peter Morgan (2012)

http://arxiv.org/abs/1207.5775

Thanks that looks useful!

[..] I think the paper I uploaded yesterday is kinda cute for its data visualization, which is different from anything I've seen other people do.

I was prompted to post this paper (the computation for which I did about two years ago) to the arXiv by Alejandro Hnilo, whose research group in Argentina has just finished a Bell-type experiment in which they use a pulsed laser. The distances are relatively short, but they record the timings of the laser pulses as well as the timings of Alice's and Bob's measurement events. They have posted their dataset privately, and it may become publicly available in time; they're currently working on their analysis.

By any chance, did you also look into the time delay differences due to the EOM's birefringence?

 

[Peter Morgan]

By any chance, did you also look into the time delay differences due to the EOM's birefringence?

I basically did not much more than what you see in the paper on the arXiv. What looks pretty clear is that a local adjustment can be made to the timings that eliminates the timing features at the nanosecond scale that I at first identified in the longdist35 dataset. I didn't look quantitatively at what might be discovered by looking at multiple datasets (which can become a lot of work, so one wants a relatively strong feeling that it might be worthwhile).

I'm not sure whether the Weihs data contains enough information to characterize what parts of the various timing delays are caused by the electro-optical modulator.

BTW: let's get Gregor Weihs' name right. It's not Weih, nor Weih's. Of course I should feel especially sensitive about this, because there's one place in my arXiv paper where I use Wiehs.

 

[harrylin]

I basically did not much more than what you see in the paper on the arXiv. What looks pretty clear is that a local adjustment can be made to the timings that eliminates the timing features at the nanosecond scale that I at first identified in the longdist35 dataset. I didn't look quantitatively at what might be discovered by looking at multiple datasets (which can become a lot of work, so one wants a relatively strong feeling that it might be worthwhile).
I'm not sure whether the Weihs data contains enough information to characterize what parts of the various timing delays are caused by the electro-optical modulator.

You may have missed the concern that was perhaps first raised by De Raedt: a certain amount of unaccounted birefringence could explain the results instead of "non-locality". Regretfully nobody seems to know if this may have been caused by the EOM or not.

BTW: let's get Gregor Weihs' name right. It's not Weih, nor Weih's. Of course I should feel especially sensitive about this, because there's one place in my arXiv paper where I use Wiehs.

Yeah I know, regretfully my spelling error is in the title and I can't change it.