Talk:Loopholes in Bell tests/Archive 1

I've done it again -- gone and created a new page when I wasn't logged on! I'll try and get it properly attributed but so far my attempts with other pages have failed. I'm the author of Bell test loopholes (which rather overlaps with the current page), CHSH Bell test and Clauser and Horne's 1974 Bell test Caroline Thompson 11:27, 7 Aug 2004 (UTC)

Email: ch.thompson1@virgin.net

Web site: http://freespace.virgin.net/ch.thompson1/


 * This page should not be merged.  User:PRF Brown


 * Agreed. Instead this page should be linked from the Bell Tests page, and then the examples of that page merged with the ones on this page. Harald88 19:24, 6 December 2005 (UTC)

recently added loopholes, now parked here
(the following I copied from Bell test experiments and its Talk page; I only included comments that seem pertinent for this list):

Recently the following loopholes were added, making the article contents largely an article about loopholes (sorry dr.Chinese, I had not noticed that). I will cut them out and parked them temporarily here, for discussion about which ones are worth mentioning, and what to do with them. Most appropriate to me seems to make a subpage to this article with a detailed loopholes discussion, and that is linked to from the main article, as is common practice in other articles. Then we can shorten that paragraph in this article. Please comment on each loophole separately. Harald88 10:39, 3 December 2005 (UTC)

- (the first two now inserted in this list:)

Fair sampling The "fair sampling assumption" states that the sample of detected pairs is representative of the pairs emitted. The possibility of this not being true comprises the fair sampling, detection, efficiency or variable detection probability loophole (Pearle, 1970). In 2001 an experiment was conducted that used detection methods that were almost 100% efficient, thus avoiding this loophole (Rowe, 2001; Kielpinski, 2001). This experiment demonstrated violation of the CHSH inequality using two trapped ions. The setup, however, did not satisfy one of the essential criteria for Bell inequalities to apply: the two sides of the experiment were not sufficiently far separated (Vaidman, 2001). Many physicists, however, share Bell's opinion when he wrote that "it is hard for me to believe that quantum mechanics works so nicely for inefficient practical set-ups and is yet going to fail badly when sufficient refinements are made."

Subtraction of "Accidentals" No experiment done so far has ever had its equipment absolutely perfect. In a real situation it can be that for some reason entangled photons will not reach their respective detectors at the same time, or that 'noise' in the form of stray photons will interfere. These imperfections lead to so called 'accidentals', where coincidences are detected, even though they are not an entangled pair. To try and limit the effects of these accidentals an experimenter will define a 'coincidence window', a time in which they expect coincidences to occur. Only coincidences that occur inside this window are considered, those that occur outside it will be ignored as accidentals. While this seems good experimental conduct, it has been shown (Thompson, 2003) that this approach actually unfairly biases the data in favour of quantum mechanical predictions, giving proponents of local hidden variable theories cause to complain. -- (end of insertion from main article)

Enhancement

The CH74 inequality, and some others, are derived using the assumption that the presence of the analyzers never increases the probabilities of certain outcomes. If an experiment shows that these inequalities are violated then we do not have a general disproof for all local hidden variable theories, just those for which the analyzers cause no enhancement.

Rotational Invariance

It is assumed in Bell test experiments that it is only the angles between different detector orientations, not the actual angles themselves, that are significant in the data, meaning that rotating all detectors by the same angle will have no effect. If the photons involved in the experiments prefer some polarization direction then the test is not rotationally invariant, as a rotation of both detectors will have an effect. Rotational invariance is not an assumption required to derive Bell's inequalities, so this is not a theoretical problem, but the assumption is often used to analyze data in experiments and so could cause data to suggest violations of the inequalities when really they may not.

Double Detections

If local hidden variable theories are true then it is theoretically possible that in some situations a detector could have cause to measure both +1 and -1 simultaneously. This is not possible in quantum mechanics and is also not possible due to the electronics of the detectors used. This means that the physical constraints of the detectors may bias the data towards quantum physics.

The Memory Loophole

Local hidden variable theories could be constructed that violate the inequalities if the particles involved had memory, that is if the measurement of the nth pair was affected by the n-1 pairs that preceded it. A related loophole, the simultaneous measurement loophole, states that local hidden variable theories could be constructed that would violate the inequalities if all pairs are measured simultaneously. There has been some research in the area of these loopholes (Barrett, 2002; Gill, 2001) that show how the data can be analyzed so as not to be biased towards quantum mechanics. When analyzed in this way, however, the data still suggests that quantum mechanics is correct.

Topology change

Finally, there is a theoretical loophole, which holds that a Bell violation is not a superluminal influence if two entangled particles are viewed as colocated (for instance, if they are opposite ends of a wormhole, as proposed by Einstein and Rosen and more recently by Durand [1]). In other words, a local realistic theory remains possible if "local" is redefined to reflect the predictions of quantum mechanics. However, wormhole creation is a topology change (i.e. addition of a new "handle" to spacetime), which implies causal violations (see for instance the chapter on topology change in Visser's book Lorentzian Wormholes from Einstein to Hawking).'

.


 * "Enhancement" should be mentioned as it is considered important by Marshall et al, who are prominent members of the local realist camp. It could happen in actual experiments though I don't think personally that it does.


 * "Rotational invariance" deserves a mention if only because most quantum theorists don't seem to stop to think whether or not it holds. They just take it for granted.  Experimenters are more careful, but all the same may find its failure biasing their presentations.


 * "Double detections" are an interesting real possibility but probably too technical to deserve mention.


 * As far as I'm concerned, the other two "loopholes" are not real possibilities and I would not mention them. I've always taken the word "loophole" to mean real weakness in the experimental design or analysis, not theoretical fantasy.


 * Despite this desire to stick to reality, perhaps one more loophole should be added, since it is generally referred to as such: the "light cone" or "locality" loophole -- the one that Aspect's and Weihs' time-switching experiments were designed to close.


 * Caroline Thompson 09:51, 4 December 2005 (UTC)


 * [...] By the way, most experiments no longer subtract accidentals so as to parry this objection; or at a minimum they report them and address the issue straight on so there is no question that accidentals are not an issue. See for example Dehlinger & Mitchell (2002) which I will be adding to the page. -DrChinese 02:09, 6 December 2005 (UTC)


 * [...] probably there is a way to solve this matter [...] in an agreeable way: starting with agreeing on a linked page about the subject of related non-hypothetical loopholes to which all relevant loophole examples are moved, and a short, unbiased paragraph about it on this page. BTW, I also discovered that already a loophole page exists that should be merged(?!) with this page, but I trust that that is not the concensus, at least it's obviously not what you want and I think that most editors will agree not to do so. Regards, Harald88 08:03, 6 December 2005 (UTC)


 * Dr Chinese: You may have a point re the subtraction of accidentals.  My original entry (see my user page) read simply:
 * ===Subtraction of “accidentals”===
 * Adjustment of the data by subtraction of “accidentals”, though standard practice in many applications, can bias Bell tests in favour of quantum theory. After a period in which this fact has been ignored by some experimenters, it is now once again accepted .  The reader should be aware, though, that it invalidates many published results.  Notable examples in which there were large numbers of accidentals are Aspect's experiments (Aspect, 1981, 1982a,b) and the early "long-distance" Bell tests conducted in Geneva (Tittel, 1997).  These experiments are discussed in (Thompson, 2003).


 * You might be happier with this. Someone else had expanded it and diverged from the original usage of the term as per Stuart Freedman and Alain Aspect's PhD theses.  I'm sorry I had not explained fully in the first place but the issue is all about possible mis-pairings when more than one atom emits a pair of photons at the same time.  Aspect had to assume that all atoms emitted independently so that there was nothing special about simultaneous emissions.  He justified this decision in a paper in 1985, but his arguments depended on a lot of quantum mechanics and I know of other physicists who dispute them.  Under a wave theory of light it is much more likely that the whole bunch of atoms will act in resonance and between them produce only one pulse of light.  If this is what happened, then the observations would have been made on two particular frequencies, filtered from a whole spectrum emitted, these two classical waves being designated the two experimental "photons".  The important thing is that at any one point of time only one pulse (of finite duration) is emitted so there can be no accidentals due to this cause at or near the zero point of the coincidence time spectrum.


 * See http://freespace.virgin.net/ch.thompson1/Aspect/ch5.pdf
 * (Translation of part of Aspect's thesis)
 * and:
 * http://arxiv.org/abs/quant-ph/9711044
 * (Timing, 'accidentals' and other artifacts in EPR experiments)


 * Caroline Thompson 11:00, 6 December 2005 (UTC)

It seems like this page needs considerable revision, which I am not in a position to undertake right now. CT seems to cite her own work exclusively for all the loophole discussion. An Inspec search reveals plenty of other articles on the same topics, e.g. Pavicic, Phys. Lett. A, 209, 255 (1995), discussing enhancement and detection loopholes and Berardi et al., J. Opt. B 2, 476 (2000) on the detection loophole. Discussing some of this work would put the article on a firmer footing. Also, the loopholes discussed here are not specific to optical Bell tests, and since Bell experiments are becoming more common, I think it would be better to have a single page covering Bell loopholes in a generic way. Dave Kielpinski 05:16, 14 December 2005 (UTC)


 * I agree that the name should be changed. Most loopholes apply to any Bell test setup.


 * Re other refs, yes of course these should be added if they are good. However, the almost universal way in which fair sampling is assumed without good reason suggests that no paper to date can have explained in sufficiently comprehensible terms just what happens in the detection loophole.  I have recently updated and submitted to J Phys B my Chaotic Ball paper on the subject, explaining things intuitively. Caroline Thompson 10:43, 14 December 2005 (UTC)

Hmmm. I thought perhaps you'd like to update the reference list to make it, shall we say, more NPOV, since you obviously wrote it in the first place. I'll get back to it when I have a chance. Dave Kielpinski 06:58, 15 December 2005 (UTC)

"reputable journal" does not mean mainstream!
"divided references into "mainstream" and "non-mainstream" according to publication in reputable journals" -> That doesn't work: Reputable journals sometimes accept excellent non-mainstream (happily, otherwise scince would be near-dead!). Thus a choice must be made. And what is supposed to be the "mainstream opinion, who can claim to know the statisitcs on that? I thus rephrase the headers to "reputable journals". Harald88 09:21, 18 December 2005 (UTC)

Fair enough. I felt that people should know the difference between work that has been vetted in some reasonable way and work that has not. Sorry for the confusion. Dave Kielpinski 18:34, 18 December 2005 (UTC)