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

Cramer's Backward Causality Experiment

  1. Jul 18, 2007 #1
    Anyone read about this?


    I'm not sure I agree with that part. I'd say that the effect in signal B would at best show up simultaneously as Cramer tampers with signal A.

    Creating that longer circuitous route for signal A could be a way to simulate non-locality. After all, if signal A has to travel a longer route, it's as if it ends up at some distant location. So I suppose, in a technical way, simultaneity across non-local distances is a facsimile of time travel, but it isn't really genuine time travel.

    This is like when Stephen Hawking said, "I think we can look into the past, but we can't alter it."

    The Cramer experiment would work along the lines of Hawking's statement. Agree or disagree? Comments?

    (Gee, how come nobody thought to do this experiment before, given all the investigation into these issues?)
  2. jcsd
  3. Jul 18, 2007 #2


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    IMHO, the most accurate statement is that the effect doesn't show up until after the experiment is performed; no effect is visible until you have access to both data sets and can analyze correlations.
  4. Jul 19, 2007 #3
    I actually wrote to Cramer, after another thread linked to specific details he has given for the new experiment. I basically said that his apparatus looks exactly the same as the Delayed Choice Quantum Eraser (an experiment already done, which he did not reference). He replied to me, saying the DCQE is irrelevant because of the coincidence correlation requirement: a requirement which, importantly, his experiment removes. He emphasised that the new experiment is a synthesis of a previous experiment by the DCQE group and an experiment from a particular German PhD thesis (which, strangely, seems not to have led to any journal publications).

    I replied back to ask about a specific detail* of his apparatus, but also noted that the "ghost fringes" in the experiments he referenced were only observed in coincidence** measurements. He didn't respond again.

    Anyway, his apparatus is conceptually clear (even if it fails as I expect), so once published the experiment should have pedagogical value (especially when students misinterpret the DCQE). It's good news if it will be finished soon this year.

    *If I knew what he was using as the optical combiner, I think I could show where the path-information goes, and hence that it doesn't disappear. His one reply sort-of gave the impression he thought the coincidence requirement basically came only from "noise" photon pairs that don't interact with the slits.
    **A colleague thought Cramer may have mistakenly overlooked this aspect.
  5. Jul 20, 2007 #4
    Anyway, do you think that "long circuitous route" can be used to simulate "non-local scales"?

    For instance, if you have a beam of light travel straight out for a distance of 1-light-year, then you know that's obviously a non-local scale of distance. But if you have your beam of light travel the same amount of distance along some long circuitous route which merely ends up taking it only a few inches away from the original starting point, then is it still roughly the same thing?

    It seems to me that the Cramer experiment could merely serve as a partial disproof. If you tamper with signal A, and then your signal B which has taken the long circuitous route exhibits the same tampering effect much later, then you have disproven that "spooky action" is instantaneous. But if signal B demonstrates a tampering effect immediately, that result may not be conclusive, because it might merely be due to the long circuitous route not truly faking the non-locality.
  6. Jul 22, 2007 #5
    How does this differ from what was done in Double Slit Quantum Eraser by Walborn? They used a coincidence circuit in that setup, but in principle it's not necessary, right?

    I'm curious if Cramer's setup can collect enough photons, even with a 300 mW pump, to generate an interference pattern within the short period of time between s1 and s2 to demonstrate "backwards causality".
  7. Jul 23, 2007 #6
    Not right.

    (Please let me know if you think there is any published record of entangled particles producing interference patterns outside of coincidence.)
  8. Jul 23, 2007 #7
    In virtually all of these sorts of experiments bell inequality tests are performed to demonstrate that the photons are in fact entangled, and it's not possible to do that without coincidence counting. Also, using coincidence circuits is the easiest way to avoid noise. Especially when Type I downconversion is used, where the signal to noise ratio is exceptionally low. But I don't see how the coincidence counting itself somehow retroactively changes the Bell state after both photons have already been registered. That seems even spookier than what Cramer is trying to demonstrate.

    I do think that the "backwards causality" leg of the experiment will probably fail, since IMO they are misinterpreting the delayed erasure phenomenon; ie the wavefunction collapses at the interference pattern and coincidence counting is necessary to "pick and choose" the corresponding idler photons. But if the wavefunction collapses with the detection of the idler photons first then the signal photons should form the appropriate dispersion pattern irrespective of coincidence counting. And this positively demonstrates the "nonlocal communication" leg of the experiment.

    Without the coincidence circuit they'll need to subtract the noise out after the fact, and this will surely require more delicate and expensive equipment, not to mention longer collection periods. But I don't see why it shouldn't work.

    Or am I missing something?
  9. Jul 23, 2007 #8
    It seems you would agree with Cramer that coincidence counting is there just to reduce noise, whereas the mainstream community thinks coincidence counting is there to pick a fringe pattern out of a Gaussian distribution (leaving an anti-fringe pattern behind).

    What you're missing is that regardless of how long the collection period is, and even with theoretically zero noise, there is no existing reason to expect the gross signal to show an interference pattern, ever. As far as we know, no such thing has ever been demonstrated, despite that many such experiments have already been performed.
    Last edited: Jul 23, 2007
  10. Jul 23, 2007 #9
    I think we're talking about two different things here. With the 3 or 4 detector scheme outlined in Kim's "Delayed Choice Quantum Eraser" I agree that the fringes and antifringes are distinguished only when the which-way path has been determined after the fact. A classical communication channel is obviously necessary there. But with the setup I referenced above, the interference pattern forms no matter what the final state of the idler photon turns out to be. The only thing that matters is the simple fact that the idler has been detected -- accordingly, no additional information is needed to be exchanged classically to determine the dispersion pattern.

    If coincidence counting were necessary in this particular design, the coincidence circuit itself would be determining the Bell state that had already collapsed. How would that be possible?
  11. Jul 24, 2007 #10
    In the experiment that you mention (though it has not yet passed peer review, and noting it is preferable for several reasons to please link to the abstract only), the which path information for the signal photon can in principle be determined from its initial polarisation along a special axis (or equivalently, from the polarisation of its entangled idler photon "p" along that axis). Consequently, there is no gross interference (regardless of whether the signal photon is measured before or after the idler) as shown in figures 3 and 7.

    If a polariser is placed in the idler beam path, only half of the idler photons will be detected (but by rotating the polariser, you can choose which half). By appropriate choice of rotation, you will only detect idler photons for which the corresponding signal partner-photon's initial polarisation (along the important axis) can not be determined, and this half of the signal photons will produce fringes (fig 4 and 8). However, the opposite half of the idler photons (which are blocked by the polariser) would have corresponded to signal partner-photons that produced *antifringes* (for proof see figures 5 and 9). There is no way to distinguish these two sets of signal photons without using the measurement results from the idler photons (ie. that's why correlations/coincidence-counting *is* necessary). So either the idler photons are not measured for polarisation (and you get fig. 7), or the idler photons are measured for polarisation (and you still get fig. 7, which is just the sum of a fringe pattern superimposed with an anti-fringe pattern), either way you can't tell the difference by only looking at the signal photons.

    In summary, I'm saying you've not understood your experiment, because it only lends further evidence to my argument. uh.. "Next!" :smile:
  12. Dec 20, 2007 #11
    Sorry to re-open this old thread but it's the most recent I could find on the subject.

    What's the bottom line here? Referring to Cramer's (apparently dead) experiment:


    Will the photons passing through S1 always display an interference pattern to the naked eye / camera? The answer would seem to have to be yes, because despite the fact that members of the same pair will go through the same slit, which slit they go through is still governed by wave mechanics, and so as soon as more than one photon is used the interference pattern emerges.

    Subsequently, when the switch is set to "0", it becomes impossible to correlate the VLP photon to the HLP photon, similar to the signal photon in the DCQE choosing the "eraser." If the switch is set to "1", however, it would be possible to correlate the photons, and thus extract the particle-like pattern from the observed interference pattern, just as DCQE does, _if_ there was a coincidence circuit.

    In fact, the only difference between this and DCQE is that there are actually two erasers - the first, setting the switch to "0", is akin to the DCQE signal photon striking the eraser; but the second erasure occurs as soon as the photons at S1 strike the CCD without being recorded. Thus, even when the switch is set to "1", it is impossible to extract the particle-like image from the camera image because there is no record of photons from which to make the correlations. And consequently the camera always sees an interference pattern.

    Is this not the flaw?

    The only way to make this experiment work would be to force the photons at S2 to pass through one slit or the other without breaking the entanglement - not possible AFAIK.
  13. Jan 9, 2008 #12
    So does anyone know the final word on this? Has Cramer officially abandoned the project?

    Also, doesn't the interference pattern ALWAYS show up without coincidence counting? That would make a helluva lot more sense than A. Zellinger's statement that the pattern disappears when the entangled twin's slit is determined. Should be the other way around.
  14. Jan 9, 2008 #13
    Exactly what statement are you talking about?
    Your describing a statement about a condition that uses correlation counts, and the pattern disappearing is just what is expected ala standard Quantum Erasure.

    If you don’t quote or reference what your talking about how can anyone follow what your saying?
  15. Jan 9, 2008 #14
    Gee, no need to be snippy about it.

    A. Zellinger - Experiment and the foundations of quantum physics. Part III. A statement which you had quite a lot to say about in another thread!

    My point is that if you believe Zellinger, Cramer's detector/camera will always register a Gaussian pattern. But I'm not clear as to why.
  16. Jan 9, 2008 #15
    Don’t intend to be snippy,
    just brief in asking you to make your posts more complete for the benefit of other readers, rather than assume everyone is current with what you are reading.
    Those that wish to look won’t find anything under “Zellinger” if you mean Anton Zeilinger.

    If you are referring to my thread https://www.physicsforums.com/showthread.php?t=144792 that was about a published statement by Anton Zeilinger assuming results without correlation test actually being made, only that they could be made even though they are not.

    Your comment was about a statement on actual correlations being made “the pattern disappears when the entangled twin's slit is determined”.

    If you don’t have the quote available now I wait till you can find it. You are in a much better position to find it the complete statement your talking about.
  17. Jan 9, 2008 #16
    This (Cramer's experiment) seems to be pretty much the "double-double-slit" experiment which I was talking about in the other thread. A.Zeilinger describes it in the book "Einstein's Schleier" (currently available only in german, even though it was written in english, I think). I'll try once more to present his explanation, since I don't know whether it is available in english, anywhere. If you want to keep looking for alternative takes on this qustion, do let me know if you find any.

    The reason such experiments will always register an at first random looking set of data, according to AZ, is because, as I've said in the other thread, for such an experiment the source of light requires a minimum size (in order to produce impulse-entangled particles where the impulse of one is precisely enough the opposite of the other, so that one can conclude which-way-info of one from the other particle). And this minimum size will always create overlapping patterns, in so far as one can say it creates patterns.

    Then depending on whether both particles are allowed to act wave-like, or not, there will be interference pattern hidden in the random looking data set (or not), but available for analysis only with information from both measurement sites. This way it comes to the same situation as usually with entanglement: There is an instant effect, but it can't be used to send arbitrary messages since the measurement results are randomized, and the effect can only be discovered with the help of classically transmitted additional data which allows discovering the relationships between the two random looking data sets (when there is any).

    You might want to look into "quantum teleportation" (and perhaps "entanglement swapping"), since they go slightly further than "plain" entanglement, in transmitting qubits and such, and it becomes somewhat more obvious that there is 'something' being transmitted, which otherwise becomes more apparent only if one has an experiment using variable measurement angles, or such.
  18. Jan 9, 2008 #17
    NO its not.
    Cramer only uses a single double-slit in the path to the “Camera” area.

    The “Image Slits” are locations where he intends to capture photons for his switching system. He is not subjecting those photons to a real double slit.
  19. Jan 9, 2008 #18
    Actually, given the nature of the discussions surrounding this topic, I should have been more cautious when saying the double-double-slit experiment would be like Cramer's.

    I can only say that the intent textually described in the first link in the first post in this thread sounds very much like what I understand to be the purpose of A.Zeilinger's double-double-slit experiment (which has actually been performed).
  20. Jan 9, 2008 #19
    Yes, I just posted a response saying this (before reading yours).

    Otherwise my longer previous post still applies, IMHO.
  21. Jan 9, 2008 #20
    Ok - so - will the camera always see a Gaussian pattern or not!?!?! :) Now I'm even more confused. :/
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?