Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Entanglement and which-path

  1. Sep 22, 2007 #1
    I have a question that seems simpe: if you have two entangled photons, can you set up two pairs of slits (one pair for the signal photons, and one for the idlers) so that observing which slit the idler goes through tells you which slit the signal photon goes through?

    It seems that conservation of momentum should let you do this. However, the Walborn et. al. eraser experiment shows that there is interference when you do coincidence counting (in their base setup the signal photons go through a double slit, and the idlers are coincidence counted), so this implies that in a simple setup the idlers cannot reveal which-path information.
     
  2. jcsd
  3. Sep 23, 2007 #2

    JesseM

    User Avatar
    Science Advisor

    p. 3 of this PDF of a paper by Anton Zeilinger (the one with 'S290' in the upper left corner) shows a setup similar to the one you're describing, where one member of an entangled pair goes through a double slit, and the second member of the pair goes in the opposite direction, so you can measure the direction of the second photon to determine which slit the first photon went through. Zeilinger explains that the entangled photons going through the double slit will not show interference in the way non-entangled photons would:
     
    Last edited: Sep 23, 2007
  4. Sep 25, 2007 #3
    Jesse,
    I had not seen the earlier thread on this subject from last May. So, from the Zellinger paper, and also from the Shih "ghost" paper, and from the Cramer paper (http://faculty.washington.edu/jcramer/Nonlocal_2007.pdf) the idlers clearly do have the which path information.

    However, it is also true that Walborn et. al. obtained interference in their basic setup (no 1/4 wave plates) when they used coincidence counting.

    Were you ever convinced by Vanesch that "one arm of an entangled beam can be made to interfere" without coincidence counting?

    There's a book on entanglement by Amir Aczel who says that you do not get interference from one arm if you do not coincidence count, but that you do get interference if you do coincidence count, and so far everything I have seen agrees with that statement (though Zeilinger's version of Dopfer's experiment explores when coincidence counting works and when it does not). Unfortunately, I'm not sure why coincidence counting erases the which-path information in Walborn.

    I'd sure like to get a hold of Dopfer's thesis in English.
     
  5. Sep 25, 2007 #4
    Then you have not seen very much.
    This ridiculous idea that one arm of a entangled beam pair will not produce a interference while not testing the second arm at all is just plain wrong. You have seen this disputed in other threads, and just because PhD’s like Amir Aczel, Brian Greene and Zeilinger publish such a terrible mistake does not make the idea correct. The only thing that could demonstrate this right is the incredibly simple experiment it would take to test it. Just turn the coincidence counter OFF and detect for a pattern or no pattern for a beam going through a simple double slit. No one has ever referred to such an experiment showing no pattern because it has never happened and never will. Even without knowing German if you look at Dopfer's thesis you can see patterns come or go depending on the set up, but never is no interferance pattern behind a double slit appear unless coincidence counting is in use.
     
  6. Sep 25, 2007 #5

    JesseM

    User Avatar
    Science Advisor

    You keep saying this, but you never provide a single rational argument as to why you think it's wrong--not a theoretical argument based on QM, not an empirical argument based on experiments which have already been performed--it almost seems to be a matter of religious certainty with you. It's not even clear whether you think your predictions follow from mainstream QM, or whether you're saying that mainstream QM is wrong...I'm sure you know PF's policy on non-mainstream claims. You have also never addressed my question about how your belief that an interference pattern will show in situations with entangled particles could possibly avoid violating the principle of complementarity--if I have some entangled photons I send through a double-slit apparatus and I get an interference pattern as you predict, what's to stop me from measuring the entangled twins in such a way as to reveal the which-path information for each and every photon that formed the interference pattern? Alternatively, one might imagine that one will see or not see an interference pattern depending on whether or not the which-path information for the entangled twins is measured or erased, which would violate causality--John Cramer is doing an experiment to see if something like this is true, but he seems to admit that if it were true it would be a violation of standard QM, and would require modifying the theory with nonlinear effects (see this thread).
    There is already coincidence-counting in all the graphs shown in the Dopfer experiment, none show the raw data, so you could just as easily have said "never do you see an interference pattern unless coincidence counting is used"--read this page which has a description of it, which says:
    I guess Cramer's retrocausality experiment will by necessity be looking at the raw data rather than coincidence counts, so we'll see if the difference between the raw data and the coincidence counts is indeed "rather small" for the D1 detector. If there is any difference in the D1 detector depending on the setup of the D2 arm this would imply a possibility of violating causality, so I guess you'd probably agree with me that it's unlikely to make a difference. On the other hand, I'd predict the raw data for the D1 detector would look more like the lower right graph on the Fig. 4, or perhaps something in between the lower left and lower right graph if the detectors are picking up a fair amount of non-entangled photons, while you would presumably predict the raw data would look just like the graph on the lower left, is that correct?

    Either way, I'd still like to know what your basis is for your claims, and whether you are arguing that your beliefs are in line with orthodox QM and that Aczel, Greene and Zeilinger have all made an error in their claims about the predictions of orthodox QM, or whether you are instead arguing that their claims about the predictions are correct theoretically but that orthodox QM is likely to give incorrect predictions about this sort of experiment.
     
    Last edited: Sep 25, 2007
  7. Sep 25, 2007 #6

    JesseM

    User Avatar
    Science Advisor

    As I understood it, Vanesch was talking about cases where the entanglement was not of a type that was relevant to determining the which-path information, so there'd be no possible way to determine which slit one member of the pair went through by measuring the other member of the pair. I have no problem with the idea that in this case you will see interference in the photons going through the slits, since there's no way to violate complementarity here.
     
  8. Sep 26, 2007 #7
    Actually, there's a paper by Shimizu et. al.
    ( "Quantum diffraction and interference of spatially correlated photon pairs generated by spontaneous parametric down-conversion") in which the authors get normal photons to interfere, and they get SPDC coincident pairs to interfere, but they cannot get the signal photons by themselves to interfere when they turn off the coincidence counting.
     
    Last edited: Sep 26, 2007
  9. Sep 26, 2007 #8
    Right. If I may rephrase it, you can get the signal photons to interfere by themselves only if the pairs are produced in a way that makes it theoretically and absolutely impossible to determine the which-path information from the idlers. So you cannot remotely turn the interference on and off in a FTL manner by restoring the which-path information to the idlers, since this information must be truly unavailable in order to achieve any interference absent a classical connection.
     
    Last edited: Sep 26, 2007
  10. Sep 26, 2007 #9
    All one needs use is a rational understanding of orthodox QM. It is you that seems to insist on dogmatically putting your faith in the miss-statements made by Aczel, Greene and Zeilinger. Apparently for no other reason than they produce a brand of Quantum Religious Science you want to have faith in with absolutely nothing to provided by anyone in the way of real data. Which you’ve had months to find and I note you have not referenced anyone as measuring no Young’s Double Slit pattern for a single beam no matter how that beam might be “entangled”.

    Also there is good old Commonsense, if this was true it would have been seen in the preparation and calibration data of thousands pattern-correlation experiments. There is no way an experimentalist would fail to be first to report such a dramatic finding. You can only eliminate the pattern by selecting photons of known “which way”. That cannot happen without selecting the appropriate photon of a known path by actually making the correlation counts. It does not get much simpler than that for mainstream fundamental QM science orthodox and other interpretations included.

    What on Earth can you be thinking
    - I, you, nobody can justify such a claim.
    Since Young using light in 1805 though 20th century single particle/photon beams going though a Young Double Slit have all saw an interference pattern. It was one of the key observations that lead to the HUP. Coincidence counting was needed to select “Which Way” info & and only with that info do we see the pattern “disappear” by selecting a group of photons from those that made the pattern. You know better than this, try to stay up to speed with what you already know.

    IMO it is a real shame that the ideas of a disappearing pattern as measured on a single beam, without using correlations to another beam have been presented to the public as credible mainstream science. NO proof of such thing has been presented to the scientific community and if were true such an experimental proof should be incredibly easy to perform.
    Why hasn't it? why haven't provided a link to such data by now?
    If you or anyone has a reference to such an observation without the use of correlation counting I’d like to see it.
     
  11. Sep 26, 2007 #10

    JesseM

    User Avatar
    Science Advisor

    No, I just trust they are more likely to know what they are talking about then you, especially when you never give any rational arguments about why orthodox QM should lead to the prediction of interference in the total pattern of entangled photons in a double-slit experiment, and you also never address my point that if interference was seen in this case, you could determine the which-path info for every photon in the interference pattern and thus violate complementarity, which seems to be a valid principle in orthodox QM.
    I haven't been trying to find it, although I note that bruce2g seems to have found some evidence in post #7. But you also have never provided any evidence that such an experiment has been performed and that interference was seen in the total pattern of entangled photons, so I figured it would be more useful to get some idea of why you believe this so fervently.
    Why would it be dramatic? If it is what is already predicted by orthodox QM then it wouldn't be particularly interesting, maybe someone would have reported it somewhere but it wouldn't be widely publicized unless it was surprising somehow. And you're still not giving any argument as to why you feel so sure that it is not what would be predicted by orthodox QM, or addressed my point about complementarity.
    Here you are just making assertions again without giving any reason why you think anyone should believe them. Are you familiar with Monty Python's argument sketch?
    So you assert. But can you find a single example of a paper or textbook in "mainstream fundamental QM science orthodox" which supports this claim about entangled photons?
    I was talking specifically about the Dopfer experiment since you had brought it up and said "Even without knowing German if you look at Dopfer's thesis you can see patterns come or go depending on the set up, but never is no interferance pattern behind a double slit appear unless coincidence counting is in use." My point was that all the graphs in Dopfer's thesis involved coincidence counting, so that although you're right that none of the graphs in his thesis show non-interference without coincidence counting, it's also true that none of the graphs in his thesis show interference without coincidence counting. There are no graphs that show anything without coincidence counting, period, so in no way does this support your claim that we'd see interference in the total pattern of signal photons if we didn't use coincidence counting.
    Um, yes, for a non-entangled photon. But just because you find it weird that entangled photons might behave differently doesn't make it wrong, and you haven't provided a single mainstream source or experiment or rational argument to support your idea that the total pattern of entangled photons will look just like the total pattern of non-entangled photons. Once again, please give an ARGUMENT as to why anyone should believe they must be the same, instead of just asserting it over and over and over and over again.
    But coincidence counting is also used when you erase the which-way info. Experiments like the DCQE only show coincidence count graphs. So you have no basis for saying the coincidence count makes a preexisting interference pattern "disappear", because the papers don't even show what the total pattern of photons looked like before the coincidence count was done and certain photon hits were thrown out. I would expect the total pattern was a non-interference pattern (or some blend of interference and non-interference, if a significant number of hits were from stray non-entangled photons), and that a coincidence count of signal photons whose idlers went to a detector that erased their which-path info would pick out a subset that show interference, while a coincidence count of signal photons whose idlers went to a detector that preserved their which-path info would show non-interference (this is plausible since the two different interference patterns in the DCQE experiment add up to a non-interference pattern, while your version is implausible since if the total pattern of signal photons shows interference, it's hard to see how you could pick out any subsets that would show non-interference--what about the valleys of your imagined total interference pattern where no photons hit, wouldn't this mean every possible subset of the total pattern would have dark bands in the same places?)

    Can you give any argument as to why my version is inconsistent with the results shown in the DCQE paper, or why it is inconsistent with the principles of QM? Or are you just going to keep asserting that it is without giving the slightest explanation as to where you got this notion?
    Again, bruce2g has provided an example. But I still want to know WHY you believe that orthodox QM predicts entangled photons should show the same interference pattern as non-entangled ones, when you haven't got a single experiment of your own to support this claim, or any explanation of why you believe it in terms of the basic principles of orthodox QM, or any response to my argument about how your claim would be inconsistent with complementarity.
     
    Last edited: Sep 26, 2007
  12. Sep 26, 2007 #11
    No; bruce2g provided an example that used correlation counts.
    Tons of experiments include data without correlation counts on one of two beams from a PDC. As I pointed out in the prior post if no pattern was there, I have no doubt it would have been reported and published widely. You seem to think the information was there and they just ignored it not reporting it, how can you explain that. Also, just how do you get "complementarity" into play with just one photon, ie. without using correlation to connect with a second.

    Without real results from a real experiment to support your view I see no reason for use to continue this further. Without that this can only become an argument, not even a rational debate as should be conducted in the threads. I’ll just stand off from this thread and allow you own view of what “mainstream” and that you and I; TRUST different people to teach QM.

    Until I notice real results posted based on using a single beam from an entangled pair I am done with this thread.
    RB

    Added note:
    JesseM, you have long ago covered the point that the QM view of a single beam from “entanglement” is no different a normal beam of light not generated from something like PDC “entanglement” with vanesch see vanash post from Nov 2006.
    "Well, if you only look at ONE beam, you cannot distinguish "entangled" photons from a statistical mixture of "non-entangled" photons." ( vanash)

    The only way you can recognize a beam as being in entanglement at all is by using the beam it is entangled with. I.E. correlation counting is required.
    Your argument is not with me; it is with fundamental QM. If you are still having trouble with that, get vanesch to help you out again.
     
    Last edited: Sep 26, 2007
  13. Sep 26, 2007 #12
    Could you please cite one?
    I'm very keen to see your evidence for this statement. (For an example of the significance, that "wrong" idea happens to be the precise reason why the mainstream expects John Cramer 's FTL to fail.) I previously cited a relevant article here, which basically shows that the signal photons from spontaneous parametric down conversion are in a superposition of incoherent phases.
     
  14. Sep 26, 2007 #13

    JesseM

    User Avatar
    Science Advisor

    I'm pretty sure the paper shows the result of both coincidence counts and the "raw data" of all photons at a particular detector--look at the more recent version of the paper here which goes into a bit more detail. The last paragraph of the introduction says:
    Someone can correct me if I'm wrong, but I take a "one-photon detection scheme" to mean a scheme that looks at the total pattern of photons at a single detector without doing any correlations with the other member of the pair. If you look at the equations in the subsequent section, they define the "two-photon amplitude" in terms of a function of the form F(x,x') [equation 2], where x is a possible position for the first photon and x' is a position for the second photon, so what they're talking about is an amplitude (and hence a probability based on the square of this amplitude, which in equation 3 is written as R(x,x') and which they call the 'two-photon counting rate') for both members of the entangled pair to be found at a particular pair of locations. They then go on to talk about measuring the "transverse wave number" of each member of the pair rather than the position, for which they use the variable q (from the graphs and the definition of wave number I gather that q is just a function of position with dimensions of 1/position, and after eq. 19 they mention that the transverse wave numbers are normalized by q = 2pi/d), so that in equation 4 the two-photon counting rate for two values q and q' is written as R(q,q'). Then in equation 17 they talk about the one-photon counting rate, which they write as R(q), so it seems they're just talking about the total probability a single member of the pair will be measured to have the value q, without any correlations to the other member of the pair being used.

    Then after equation (18) they consider "an uncorrelated case", and derive the "two-photon counting rate of two independent photons" (i.e. no entanglement), and in Fig 2b they show two graphs of the predicted results, the top for the one-photon counting rate and the bottom for the two-photon counting rate; both show interference. But then they go on to consider the case of correlated photons, and in Fig. 4a they consider "perfect spatial correlation", with the corresponding graphs in Fig. 4b showing interference for the two-photon counting rate but no interference for the one-counting rate. What do you think this means, if not that their theoretical predictions is that the total pattern of photons at one detector will not show interference in the case where they are entangled with the photons at the other detector in such a way that they have "perfect spatial correlation"?

    Then if you look at the experimental results in section III, they agree with the theoretical predictions--Fig. 7 shows no interference observed in one-photon detection, with interference observed in two-photon detection. Also note that they ran the same experiment with non-entangled using "thermal light generated from a tungsten-halogen lamp", with the results in Fig. 9--in this case interference was observed in the one-photon detection case as well as the two-photon detection case.

    Please look over the sections of the paper I mention, and if you disagree with my understanding of what was going on and think that the "one-photon detection" involved correlation too, can you explain your reasons? What do you think was being correlated with what?
    But you're using circular reasoning here--you assume that the theoretical prediction is that there should be interference, and thus that if no interference was found this would be considered surprising and "published widely". You do not hold your own ideas to the same standard, since you do not expect that if interference was observed this would be "published widely" as well. So I think we'd both agree that results which were completely expected will not be "published widely" unless physicists find something interesting or nontrivial about them, we just disagree on which results are in fact expected in orthodox QM. But you haven't given a single argument as to why you are sure that the theoretical prediction is that interference should be seen in the total pattern of entangled photons, you just assume that because it's your gut feeling or something, in spite of the fact that all the experts are telling you something different.
    Just as you seem to think that if interference was observed, that information would not have been widely reported. Again, I think we'd both agree that ho-hum results which everyone expected and which have no new applications or theoretical interest will not be "widely" reported (which is not to say they'll be 'ignored', just that you'll have to do some digging to find papers that bother to report it), we just disagree over which result physicists would expect in the case of entangled photons.
    I don't think that word means what you think it means. "Complementarity" has nothing specifically to do with correlations, it's basically just the notion of "wave-particle duality", that quantum objects behave like particles when you measure them while they behave like waves when you don't (see here or here or here for more info). In the case of the double-slit experiment with ordinary non-entangled particle, "complementarity" is used to sum up the notion that if you are able to determine which slit the particle went through, you won't see a wavelike interference pattern, but if you don't determine which slit it went through, you will see interference. The interference pattern disappears in the same way regardless of what physical method you use to determine what slit it went through--bouncing photons off the slits (in the case of the double slit experiment with electrons), putting polarization filters on the slits (in the case of the experiment with photons), putting shutters on the slits which open and close at different times, etc. So, if complementarity is to work the same way in the case of entangled particles, you should not be able to simultaneously see an interference pattern and yet determine which slit each particle went through by measuring its twin; this is what I meant when I said your idea would violate complementarity. This is a conceptual argument and so not really rigorous, but it's better than no argument at all, which is what you've given so far for your position.

    edit: I just want to note an example of a paper which says that the fact that you get an interference pattern when you erase the which-path info in DCQE can be understood in terms of "complementarity":

    http://arxiv.org/pdf/quant-ph/0512207
    Also, note that I provided a second argument in my last post which doesn't depend on complementarity and which I think is even more clear:
    Do you disagree that in any interference pattern, there will be some positions where the destructive interference is complete so that you see no photon hits at all in these positions? Do you disagree that in the corresponding-non interference pattern, the probability of photon hits at those locations is not zero? If so, how could you possibly get a correct non-interference pattern by taking a subset of all the photon hits in an interference pattern? (i.e. throwing away some of the hits and keeping the rest, but not adding any new hits)

    In comparison, if the total pattern of signal photons shows a non-interference pattern, as I (and Aczel and Greene and Zeilinger) am saying, then it's easy to take two subsets of this and get two interference patterns (with the peaks of one lining up with the valleys of another), or to take two different subsets and get two different non-interference patterns, which is what the different possible subsets look like in the delayed choice quantum eraser depending on whether you erase or preserve the idlers' which-path info.
    Again, I think the paper bruce2g references does indeed show what happens when you look at the total pattern of entangled photons. And I'll note again that you're showing something of a double standard here, since you demand others show results demonstrating the total pattern shows non-interference, but you don't feel any need to look for results that demonstrate your claim that the total pattern will show interference.
    Again, some experimental results have been provided, hopefully you'll address them. But in any case, people certainly have rational debates about theoretical predictions on physicsforums threads all the time--the only reason this has not been much of a "rational debate" so far is that you consistently refuse to address any of my arguments or provide any theoretical arguments of your own. Since you're unwilling to provide any reasons whatsoever to believe that mainstream QM predicts the results you claim it does, and you obviously haven't done any mathematical calculations to check, I take it you don't have any intellectual reasons at all, it's just a sort of unfounded gut feeling of yours, which makes your absolute confidence that you are right and all these professional physicists wrong all the more ridiculous.
    I don't think you understood the followup posts (#7 and #8 here), because in the next post I asked if vanesch was talking about not being able to distinguish entangled photons from a mixed state of photons which definitely came from either slit A or from slit B. In such a mixed state, there is no interference between photons that came from the first slit and photons that came from the second! (A mixed state is a statistical mixture of pure quantum states that reflects ordinary classical uncertainty, like if you know there's a 50% chance the system is in quantum state |A> and a 50% chance it's in quantum state |B>...in this case there's no interference between |A> and |B>, the expectation value for an operator O would just be 0.5*<A|O|A> + 0.5*<B|O|B>...see here, for example. So if |A> and |B> are states which evolved from position eigenstates of the photon at the moment it passed through one slit or the other, you won't get interference on the screen.) And vanesch confirmed that this is what he meant. Only when the non-entangled photons going through the slits are in a "pure state" (or a mixed state where none of the pure states it's a mixture of are ones that tell you which slit the photon went through) do you get interference between the two slits.
    Like I said, you misunderstood what vanesch meant (if you doubt this, just ask him). Anyway, it's odd that you would cite vanesch as an authority here, when in post #49 of this thread vanesch said:
    Also, if you look at the rest of that post, you see he talks a lot about pure states vs. statistical mixtures (i.e. mixed states), and how the beam A of entangled photons will behave like a statistical mixture rather than a pure state, which explains why they don't show interference.
     
    Last edited: Sep 27, 2007
  15. Sep 27, 2007 #14
    No, actually they used correlation counts in one run, and then they used the raw data from one arm in another run, and the correlation counts show interference but the raw data from one arm do not.

    Also, here is an abstract of an article (not available for free online, unfortunately), which also demonstrates this ('single-photon interference' means looking at just one arm of the SPDC output):

    Two-photon double-slit interference experiment
    C. K. Hong and T. G. Noh

    Abstract
    We report a two-photon double-slit interference experiment. Two correlated photons, generated by spontaneous parametric downconversion and traveling in different directions, are made to pass through a double slit in such a way that the path of one photon can be identified with that of the other photon. NO SINGLE-PHOTON INTERFERENCE PATTERN IS OBSERVED even if the path is not actually identified, but a spatial interference pattern appears in the joint detection counting rates of the photon pairs. It can easily be shown that these nonlocal phenomena cannot be exhibited by any kind of classical field [emphasis added].

    JOSA B, Vol. 15, Issue 3, pp. 1192-1197


    http://josab.osa.org/abstract.cfm?id=35389


    PS - I found this by doing a search on 'correlated photon interference' -- back in the early days of SPDC, they called them 'correlated photons' instead of 'entangled photons' so the really basic results, like this one, would be found using that search term. I don't think you can find a study where they achieved an interference pattern without coincidence counting, but if there is one this would be the way to find it.
     
    Last edited: Sep 28, 2007
  16. Sep 28, 2007 #15
    This is starting to make a lot of sense. The down conversion introduces a random delay in the photon's journey (when you look at the particle side), but when it does this the crystal also has to completely mess up the coherence of the wavefront as well (when you look at the wave side). So the interference disappears since the wave is no longer coherent (i.e., the + and - parts of the wave do not line up neatly so they can cancel each other).

    Next, in order for the coincidence counting to produce the interference, it has to somehow fix this. Maybe it somehow selects photon pairs whose wave is coherent, so that the selected subset can interfere. Maybe it does something else, though I'm not sure what. From one of the papers, we see the coincidence count is something like 1/1000 of the raw count. So, there is apparently something special about the photon pairs that make it to a coincidence count. Any ideas about what that might be?
     
    Last edited: Sep 28, 2007
  17. Sep 28, 2007 #16
    This misconception of ‘no double slit pattern from a single entangled beam’ has been discussed enough times on these forums before that the mentors here should know and understand the issue to help prevent this misinformation from be promoted as fact.

    What has never been cited is an experiment or any data that supports the ‘no double slit pattern from a single entangled beam’ claim with real single beam data. Why aren’t you asking for the incredibly simple to test, experimental proof of that to be cited?

    Can I cite one experiment to support my case?
    Many more than that, I site any PDC – Double Slit experiment used to demonstrate the results of knowing QM “Which Way” information.

    By necessity they must review large amounts of raw data including what is needed here. But the QM issue they are dealing with only requires they discuss in detail the correlation count data points less than one thousandth of the large amount of uncorrelated data that was a natural part of conducting their experiments. Although they don’t report detailed information on that uncorrelated data in their publications, it does not mean they ignored it in their reports. Just follow along here and be sure you understand what those experiments say. Just take the time to actually think, without blindly applying unsupported external comments.

    The simplest objective for those tests was to determine Which Way information about signal photons with carefully testing idler photons that can be paired with an otherwise untested signal photons. Thus the only location testing is performed on the signal photons to determine what type of pattern is created by them. The achievement here is getting the Which Way information without ever disturbing any signal photon on it way to making it individual contribution to whatever pattern is found. The experiment demonstrates that once “Which Way” information is known by using the correlation information the “Inference Pattern Disappears” . They go to great lengths documenting those correlation detections in their results, concluding that it confirms a fundamental principle of QM; namely when you know ‘which way’ the interference pattern disappears.

    All these experiments necessarily include “data without correlation counts” as it is from that large data group that photons are selected by correlation with the idler detector to create the subset of photons to plot with which way information. They don’t re-report findings about that raw uncorrelated data as it is basically old news. Seen since the early 1900’s when they first began lowering photon or electron density in tests like these until they could be sure experiments could run tests one particle at a time. Without multiple particles interacting with each to establish a Young’s interference pattern, many expected that pattern should go away. But of course it did not, and contrary to JesseM’s comments; that was a big deal back then, it was widely discussed and published, and it was a very important contribution to creating the HUP / QM theory. That ultimately includes establishing the simple QM principle that “Which Way” information will cause double slit type patterns to stop and change into a typical single slit diffraction distribution pattern.

    But according to cesiumfrog, JesseM, bruce2g, ZapperZ and their favored Authors; all this documentation by the experimenters was unnecessary, because according to them there never was an interference pattern to disappear. Therefore those tests do not verify the QM rule of Which Way rule about removing the interference pattern. For them the signal beam gave no pattern before correlation counting because the idler beam was made of “entangled” photons, and not capable of producing a double slit pattern in the first place. Gaining “which way” info did nothing, in their “no double slit pattern from a single entangled beam” view to support those Experimentalist claims of confirming the QM “Which Way” rule. In effect they are they are saying those conclusions are flawed as the testers failed to recognize and report the raw data that would support the ‘no double slit pattern from a single entangled beam’ claim. And thus render the remaining part of their reported conclusions about the QM Which Way issue meaningless.

    For me; I am satisfied with those testers implication in statements like “the interference pattern disappeared” with which way information, and “the pattern returned” or the “pattern remained” when correlations with idler observations did not include which way detail. I accept those comments as confirmation that raw data from a single entangled signal beam did in fact produced typical double slit interference patterns, was observed by them, and these statements are reflective of those observations. Likely they also assumed it as an obvious fact for their intended readers. Assuming that ‘no double slit pattern from a single entangled beam’ is correct also assumes that these experiments are wrong and flawed to the core of their conclusions. I do not, and expect a lot more than I’ve seen before even thinking of rejecting the QM conclusions they have made as incomplete or wrong.

    Promoting the opinion that ‘no double slit pattern from a single entangled beam’ is an absolute fact is done by to many in this forum, often being abusive to any that question that opinion. And I challenge any supporter of the ‘no double slit pattern from a single entangled beam’ to put up or shut-up. Just produce real data from any real experiment using absolutely no correlation methods of any kind that demonstrates a double slit that fails to produce a young’s type pattern after the input beam is replaced by one of a pair of “entangled” beams. If you are not satisfied with the information available in reported experiments ask for the help of ZapperZ one of our experimentalist experts. He may know how to reach more of the Raw Background data, or help you suggest an experiment to someone that can produce the detail you would like.

    No one has the right to demand others accept something like ‘no double slit pattern from a single entangled beam’ as an established fact when they have been given several opportunities in the past to offer references to real data to support this claim. Responding with references to other unsupported opinions and argumentative irrational comments is not what these forums should be about. IMO the experiments on QM “which way” issues mentioned above have been done adequately to already to refute this claim and IMO is falsified by existing data. Anyone not satisfied with the current state of reporting on those experiments can easily have an even simpler test preformed to attempt to verify their claim. Based on existing evidence they will fail in such an attempt. And unless proof is found otherwise demanding that ‘no double slit pattern from a single entangled beam’ be accepted as fact is no less than promoting the spread of MISINFORMATION about that real facts as they stand experimental now. It borders on ‘crackpot’ assertions and should be addresses with them by the mentors.

    RB
     
  18. Sep 28, 2007 #17

    JesseM

    User Avatar
    Science Advisor

    Supporting your point? Your claim is that the raw data does show interference, so how does the result "the raw data from one arm do not" support your point?
    Here you ignore the context. In the paragraph after equation 9 of the paper, they distinguish between two completely different cases, one in which there is no correlation between the two photons and one in which there is perfect spatial correlation:
    Then equations 10, 11 and 12 continue to follow from "the former case", and the sentence after equation 12, saying "The one-photon counting rate [tex]R^{(1)}_{cl}[/tex] corresponds to the classical Fraunhofer diffraction pattern" (the sentence you quoted), is still referring to the former case.

    Only after equation 13 do they analyze the case where the photons are correlated, starting with the paragraph that says "On the other hand, in the case of G(x - x') = delta(x - x'), a pair of signal and idler photons passes together through the same point x at the object", and then they go on to say "In this case, the two-photon amplitude is not separable into any product of two independent amplitudes. In other words, it is a spatially entangled state." So only in this section are they talking about an entangled signal-idler pair, the sentence you quoted was about the "former case" where two photons are not entangled. They derive some more equations for the entangled case, finally getting to the one-photon counting rate in equation 17, where they say "the one-photon counting rate becomes constant", and go on to say "This result means that the one-photon diffraction-interference of the biphoton will exhibit no modulation." This case is illustrated in Fig. 4 later in the paper, where you can see that the upper graph predicts the one-photon counting rate will just be a flat line with no interference, and this is confirmed with experimental data in Fig. 7.
     
    Last edited: Sep 28, 2007
  19. Sep 28, 2007 #18
    Sorry I was focussing on the comments about the one photon pattern displayed in Fig 4 and misread the comment by Bruce. Since as I mentioned they are dealing with diffraction interference, This example does not speak to your issue anyway – so I will remove my post it. And stick with true double slit experiments to look for raw examples of data with no correlation counts, which so far I cannot pick any better than as in the prior post. If you find someone to actually run such a test to try and verify your claim with real data, you know which side I’ll lay my bet on.
     
  20. Sep 28, 2007 #19

    JesseM

    User Avatar
    Science Advisor

    In the second-to-last sentence of the paper's introduction they say they send the light through a "transmission grating", which is basically just a large series of slits instead of just two. For example, slide 24.6 of this powerpoint lecture used in a physics course says:
    So just to be clear, are you saying you accept the following:

    1. If you send non-entangled photons through a large number of slits, you'll get an interference pattern
    2. If you send signal photons from entangled pairs through a large number of slits, the interference pattern disappears
    3. If you select the subset of signal photons whose idlers had their which-path information erased, you recover an interference pattern

    ...but that you find the following totally implausible?

    1. If you send non-entangled photons through two slits, you'll get an interference pattern
    2. If you send signal photons from entangled pairs through two slits, the interference pattern disappears
    3. If you select the subset of signal photons whose idlers had their which-path information erased, you recover an interference pattern

    (the only difference between them is the words in bold)

    The math for calculating the predicted pattern in each version would be pretty similar. If you really agree with the first three but deny the possibility that the second three are correct, I guess I should go back and address the reasons you gave in the previous post:
    OK, why don't you cite a specific example of one of these "PDC - Double Slit experiment" papers so we can actually "follow along" and understand what they're saying as you suggest (just as I did with the last paper in showing you that they found no interference in the total pattern of signal photons when you originally thought they did).
    I would say that the objective was to show that depending on whether they determined the which-way information or erased it with the idlers, they would either get interference or non-interference in the signal photons.
    Do they actually use the words "the interference pattern disappears", or something equivalent? Again, can you name a specific paper so we can look at the context, or are you just going on vague memories? Even if they do use such language, it might be that they say it "disappears" because they were previously talking about the interference pattern that you see in the subset of signal photons whose idlers had their which-path information erased, not because they were referring to an interference pattern in the total pattern of signal photons.
    Of course that was a big deal, and I don't know what "comments" of mine would say otherwise--the reason it was a big deal was because it was a new result and quantum physics wasn't fully established. Nowadays everyone accepts that even QM's weirdest predictions are likely to be correct, so unless a prediction is particularly weird or has important practical applications, it's not necessarily going to be widely publicized (of course any new confirmation of a prediction is going to be published somewhere, but that doesn't mean it'd be trivial to find it on a google search or whatever). And assuming I (and everyone who's weighed in on this so far) am correct that orthodox QM predicts that the total pattern of signal photons through a double-slit will not show interference if they're entangled with idlers that could potentially reveal their which-path information, there's no reason to think this would be widely publicized in the media or that it'd be a simple matter to find papers mentioning this.
    Again, what experimenters describe the results in terms of getting an interference pattern to "disappear"? Are you sure that anyone has actually said this, or is this just your vague memory?

    Of course I don't claim these experiments were unnecessary, they are quite interesting, I just don't think the thing that makes them interesting is that they got an interference pattern to "disappear", and I don't think the experimenters claimed that that was the point of the experiments either.
    What nonsense! Of course I don't think the experiments "did nothing ... to support those Experimentalist claims of confirming the QM 'Which Way' rule". They do validate the which-way rule, specifically they verify the rule that if you have the which-way information you don't see an interference pattern, and if you don't have the which-way information you do see an interference pattern. I think the reason the experiment is referred to as the "delayed choice quantum eraser" is precisely because they consider that the most interesting feature of the experiment, the fact that when you "erase" the which-way info from a subset of idler photons that go to a particular detector, you recover an interference pattern in the corresponding subset of signal photons, which is not visible in the total pattern of signal photons. That's a great validation of the part of the which-way rule that says "no which-way = interference", which is just as important a component of the rule as the "which-way = no interference" part (which is also validated by looking at the subset of signal photons whose idlers had their which-way info preserved).
    Once again, please name some testers who make statements like this, so we can verify that you're not just misremembering, and see the specific context if you're not.
    Ditto for this.
    As always, the double standard here is ridiculous, since never once have you provided any "references to real data" to support your claim that interference is seen in an entangled beam, and it's also pretty ridiculous to treat a situation where photons are sent through a large series of slits in a transmission grating as having nothing whatsoever to do with the question of what happens when you send them through a double slit. Also, I don't think I've said the lack of an interference pattern is an "established fact", just that it's what is predicted by orthodox QM.
    When dealing with questions about what a theory predicts, references to statements by physicists who have actually done the calculations are perfectly valid. And since you've never even bothered to explain what you think is wrong with my arguments about complementarity or about the fact that it's impossible to get a non-interference pattern by taking a subset of data from an interference pattern, merely dismissing them as "argumentative irrational comments" rings pretty hollow.
    The experiments themselves are perfectly valid and interesting even if you accept that no interference is seen in the total pattern of signal photons, as I explained above; it's only your weird interpretation that says they'd somehow be worthless in this case, and your unsupported memories that say any of the experimenters have ever talked about the interference pattern "disappearing".
     
    Last edited: Sep 29, 2007
  21. Oct 1, 2007 #20
    gift: good article on entangled photons

    I have purchased this article and uploaded it as an attachment to this post. I have separately uploaded figure 2, coincidence vs signal photon.jpg which, for the benefit of RandallB, shows interference with coincidence counting, but absolutely no interference at all for the signal photons taken by themselves (I believe that you can see the straight-line at the top of the thumbnail, which is the count of the raw signal photons, though to a different scale).

    Also, RandallB, please note their equation (3) on p. 1194:
    The probability density Psignal(y) of detecting a signal photon at y is given by:
    P(signal at y) = ... = K, (3)
    where K is a normalization constant.
    That is, the probability of finding the signal photon at y, when it is not coincidence counted, is a constant; I certainly hope you understand what that means : they have derived the fact that the signal photons alone do not exhibit interference.

    Randall, I find this stuff very interesting, as I'm sure you do. Please, if you still maintain your former position, please argue with this post with everything you have so that we might settle this matter once and for all. This article has lots of great references to SPDC experiments, and if your position has any merit then it should be mentioned in at least one of them. Please note that although I purchased this article, I first went to the main San Francisco library and loaded up the microfiche and printed it out. When I saw how relevant it was, I decided to make a purchase on behalf of the forum. You can also go to a good library to track down the references which are not available online, and find out if your position has any merit. If it does not, then I hope you'll agree with the very insightful posts of JesseM and CesiumFrog and allow us to get on to other matters that hopefully will merit further attention.

    The other reason I purchased the article is because it actually deals with the exact question with which I opened this thread -- they use an interesting technique to show that the idler always goes to the opposite slit of the signal photon (basically, there are no same-slit coincidence counts (count(signal-j idler-j) = 0); therefore, all of the coincidence counts must be signal-j idler-k (where k ≠ j). They also appear to agree with post #15 when the authors say: "On the other hand, the interference pattern that appears in the coincidence counting rates implies that some kind of hidden coherence between them has existed and that it is uncovered by the detection of the idler photons in an appropriate arrangement."
     

    Attached Files:

    Last edited: Oct 1, 2007
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?