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

Causality in EPR experiments

  1. May 22, 2005 #1
    I have a question about quantum entanglement experiments, such as the two-photon "delayed choice" experiment performed by Aspect et al. http://prola.aps.org/abstract/PRL/v49/i25/p1804_1 [Broken]. Can anyone estimate how much time elapses between the arrival of a single photon at the detector, and the initiation of a current or voltage that could be considered "macroscopic"? I know too little about photomultipliers and the like to have any idea.

    In other words, does the experimental setup rule out the possibility that detector states become entangled with detected photon states, so that the "wavefunction collapse" is actually much later than the arrival of the photon at the detector? I am wondering if the time it takes for the detector to settle is long enough that a timelike signal could pass from one detector to another.

    Thanks in advance if anyone can shed light on this ... er .. so to speak.
     
    Last edited by a moderator: May 2, 2017
  2. jcsd
  3. May 22, 2005 #2

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Photomultipliers usually have response times in the order of a few nanoseconds...

    Ha, that's a nice idea :-))) Especially if the "detector" is the person looking at the results of the correlations ;-)

    cheers,
    Patrick.
     
    Last edited by a moderator: May 2, 2017
  4. May 23, 2005 #3
    photomultiplier

    Hmm ... how much current does the photomultiplier produce at the end of that response time? Is it just a few electrons per millisecond or Avagadro's number of them?
     
  5. May 23, 2005 #4
    Depends how many initial photons go in and what the photosensitive bit's made of!
     
  6. May 23, 2005 #5
    Only one initial photon goes in. Let's say the photosensor is made of silicon, then how many excited electrons are flowing per second after the response time has elapsed?
     
  7. May 23, 2005 #6

    DrChinese

    User Avatar
    Science Advisor
    Gold Member

    The Aspect experiment was intended to compensate for this. Over time, many improvements have been made to the process. The current state of the art is much more rigorous. Using fiber optics, distances are much longer and the time varying elements are more sophisticated. Thus the locality issue you describe is ruled out. Please reference:

    The 1998 Innsbruck Experiment (EPR with 1 kilometer of separation):
    http://arxiv.org/PS_cache/quant-ph/pdf/9810/9810080.pdf [Broken]
    by Weihs, Jennewein, Simon, Weinfurter and Zeilinger
     
    Last edited by a moderator: May 2, 2017
  8. May 23, 2005 #7

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    You get a short current pulse which integrates, to say, a few tens of femtocoulomb (say, 100000 electrons). That's good enough to be seen with a charge-sensitive amplifier. The pulse itself takes a few nanoseconds, and during that time, currents of the order of a few microamperes flow from the last anode.

    cheers,
    Patrick.
     
  9. May 23, 2005 #8
    Thanks for the reference. Yes, that does seem to settle the locality issue.
     
    Last edited by a moderator: May 2, 2017
  10. May 23, 2005 #9
    It solves part of the issue. That is, *if* A and B are causally
    affecting each other, then these causal influences must be
    travelling faster than light. We can be pretty sure of that.

    What we can't be sure of yet is whether or not A and B
    are causally affecting each other.

    Bell-type analyses show that the current state of the
    art of descriptive physics is quantitatively inadequate.
    Quantum theory isn't descriptive physics. So, there's
    no qualitative understanding of how the correlations are
    produced. They might be due to local interactions or
    they might be due to superluminal interactions. Nobody
    knows.

    The question of whether or not nonlocal causality
    is a fact of nature remains unanswered.
     
    Last edited: May 23, 2005
  11. May 23, 2005 #10

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member


    This is correct. The problem with these situations is that they are in a "twilight zone": on one hand the Bell conditions are violated. But on the other hand, there is no way to have an information transfer about the *choice* of polarizer of A by B. If there were such a transfer (that means, that B, by purely looking at his data, can find out what was the polarizer setting of A) then for sure there was a faster-than-light causal influence. But there is no such information transfer (B cannot find out what was the polarizer setting at A), and it can be shown that such transfer is impossible in quantum theory. You can only find out that there was a peculiar correlation by *bringing together* the data from both sides. And that leaves the possibility for locality to be still valid, depending on what is your view on quantum theory.

    As Sherlock said, the safest attitude is to say that nobody knows if locality holds or not as a fundamental principle.

    cheers,
    Patrick.
     
  12. May 23, 2005 #11
    Has there been any attempt to measure the speed of such faster-than-light influences? It should at least be possible to find a lower bound for that speed, one would think.
     
  13. May 24, 2005 #12

    DrChinese

    User Avatar
    Science Advisor
    Gold Member

    As I read the reference (Weihs et al), Figure 1, the lower bound was about 10c. That is assuming there is a non-local effect. The usual interpretation is that distance is not a factor, it is always "instantaneous".
     
  14. May 24, 2005 #13

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    You can't, really. As I said, the reason is that you only see the correlations, once you've brought together the data from both sides, using classical, slower-than-light communication.

    If you mean, can you measure "the speed with which one can do measurements at both sides", then this leaves me wondering what you are talking about. For instance, let us place ourselves in a non-Bell context, with classical correlations. Imagine the usual game: I have a white ball and a red ball, and put randomly one in a grey bag, and the other one in a green bag. The grey bag is sent to Tokyo, the green bag is sent to London. We decide that at 12 AM GMT, both bags will be opened and looked at. So now we have a "correlation" between both results. What is the speed at which this correlation is established ?
    In a similar way, what does it mean for the "correlation to propagate between both measurements" ? I can do (in a certain reference frame) the measurement at A slightly before, or slightly after, the measurement at B. In another reference frame (using relativity), I can inverse the order in which the measurements occur. In all these cases, the results are the same. So how are you going to attach a "speed" to this "propagation of correlation" ??

    cheers,
    Patrick.
     
  15. May 24, 2005 #14

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    What exactly was that ? I wonder what it can mean...

    cheers,
    Patrick.
     
  16. May 24, 2005 #15
    In these two-photon experiments, the polarizers and detectors are at rest with respect to one another, or nearly so. That is the reference frame I'm referring to when talking about the speed at which correlation hypothetically propagates. Of course, that speed will be measured differently by observers moving with respect to the apparatus, as you indicated. If this "propagation" point of view is valid, correlations of the EPR type would decrease or disappear as the distance between measurements increases.
     
  17. May 24, 2005 #16

    DrChinese

    User Avatar
    Science Advisor
    Gold Member

    I think the idea is: IF there were a causal effect that simply transmitted from A to B telling the polarization to comply with... then at what speed does that causal effect travel?

    In Bohmian mechanics, which attempts to insert non-locality explicitly, I don't think there is any limit to the speed of propagation of the correlation effects. If there WERE some such effect, we know that it must be able to propagate at 10 times the speed of light or more. That is per my reading of Weihs, since they specify that the Einstein light cone could have been a tenth the actual size and locality would have still been respected.

    Again, this is not standard interpretation of what is happening. You could just as easily say the purported causal correlation effect travels backward in time too - what speed is that?
     
  18. May 24, 2005 #17

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Yes, that is what I had in mind ! It is sufficient to cut 50 cm off one optical fiber or another, and you CHANGE THE ORDER in which things happen, so how can you reasonably define a speed ? Or do they simply take the duration of the two measurements (say, 3 ns) and the distance between the two measurements, and calculate a "speed" from the ratio ?

    cheers,
    Patrick.
     
  19. May 24, 2005 #18
    I may be wrong but I think the idea is to assume there is a FTL signal causeing the correlation. Then you look at the timeing of the measurements and calculate how fast such a signal would have to be to cause the correlation. It does not change the fact that you still have to look at both ends to see the correlation in the first place. And as far as I know most physicists don't believe in any such signal anyway.
     
  20. May 24, 2005 #19
    It's true that spacelike-separated events don't have a definite time ordering. However, it may be that EPR-type correlations are only allowed across certain spacelike intervals, and not others. Suppose there exists a reference frame in which all allowed correlations appear to propagate forward in time. In that case, the width of the cone enclosing all the corresponding vectors in Minkowski space can be viewed as the "speed" of the correlation signals.

    Of course, that implies the existence of a preferred reference frame, at least for quantum correlation phenomena, which is not the current philisophical fashion.
     
  21. May 24, 2005 #20

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Yes, but even without relativistic considerations, I have the following problem with trying to define a speed of propagation of any influence. Imagine an EPR setup in which the two particles are sent over long optical fibers, one end arrives at Alice, and the other at Bob. Now, Bob's fiber is slightly longer, so Alice measures "first" and Bob measures on average say 0.5 ns later. So we could then define a "speed" of the distance D between Bob and Alice divided by 0.5 ns. But now Bob shifts his photomultiplier 10 cm (0.3 ns) closer, by removing some piece of optical fiber. So now the speed will be something like D / 0.2 ns. Bob moves again his photomultiplier 10 cm closer: this time, Bob clicks first on average... so the speed is then D / (-0.1 ns) ?? No, because now suddenly the influence goes from Bob to Alice... So the speed is D/0.1ns but in the other direction. Given the detection time in a PM (if that's considered the "measurement process" whatever that may mean), some events will be going Alice-> Bob, others will be going Bob-> Alice and some will be damn close to equal times. How do you define a speed in this situation ?

    cheers,
    Patrick.
     
  22. May 24, 2005 #21
    You have defined the speed already (D/T), though I would think of it as the absolute value |D/T|. Yes, you can change the direction of the supposed "signal propagation" by choosing whether Bob or Alice performs the measurement first; the fact that the Bob->Alice and Alice->Bob results are indistinguishable only shows the symmetry of Bob's and Alice's views of the experiment.

    The interesting result would be to find that for a large enough value of |D/T|, the correlation effect disappears, i.e. the Bell Inequality is no longer violated, or perhaps to find that the threshold value of |D/T| is anisotropic with respect to the direction of the Bob-Alice axis. This would tend to support a "superluminal signal" view of wavefunction collapse.

    No such effect has been seen yet, but has it been positively ruled out?
     
  23. May 24, 2005 #22

    DrChinese

    User Avatar
    Science Advisor
    Gold Member

    Technically, no. As noted, we are at 10c (or maybe -10c) and we can expect that to rise over time as distances increase. I don't expect us to ever see anything happen in this regard - i.e. a finite interval.

    This does allow for some interesting speculation as to how to describe what is happening. In my opinion, there is something fundamental about the act of observation: if there are no hidden variables, then the photon polarization was not determinate prior to the observation. So what happens at the time of observation to explain the results?

    a) We have Vanesch's favorite, Many Worlds. There is branching.
    b) Perhaps in some rolled up unseen spatial dimension, the photons are actually not separated by any distance at all. Thus there is no superluminal effect because there is no distance to traverse.
    c) Perhaps there is in fact a new superluminal force carrier - "chanceons".
    d) Perhaps (my favorite speculation) the observation of one photon causes an effect to propagate to the past at the speed of light, and then change direction so it travels forward in time to the other photon. This would exactly trace a light cone that is consistent with observed results. Of course, I have no idea how or why such behavior occurs sometimes but not others, and this is a completely ad hoc idea. The combo of traveling both forward and backward in time allows arbitarily large distances to be traversed in arbitrary time intervals. Which is more or less what appears to happen anyway :)
     
  24. May 24, 2005 #23
    From what little I know about the EPR problem, I would tend to agree with Vanesch that Many Worlds is correct, but only if no evidence of superluminal propagation can be found.

    I would think (b) implies a superluminal effect after all. It's the effective speed of the signal in the lab frame that matters, not the speed in the particle's own frame.

    This is my favorite explanation for EPR, except that the "chanceons" wouldn't carry any energy -- they can only select between energetically degenerate states. Otherwise you'd get superluminal transmission of energy and/or information, leading to causal paradoxes.

    Interesting idea ... it does have some attributes of "Many Worlds" though, since the past which receives the backward-propagated effect isn't the same past that you started out with. Would it be "Many Histories"?
     
  25. May 24, 2005 #24
    There is a group headed by Gisin who have performed these types of experiments, and I have one reference [1] (4 years old now) that sets a lower bound at 2/3 * 10^7 c (!!). Basically they did an EPR type experiment with entangled photons sent via optical fiber network to two villages near Geneva, with the source smack dab in the center. One detector is set spinning at some high angular velocity so that the frames of reference of each detector are not the same, and such that *each* detector, in its own frame, is the first to do the measurement! Pretty cool, eh?

    I agree with some of the other comments that there is no actual "speed" because there is nothing actually being transferred from A to B. And I agree with vanesch that the MWI is conceptually the simplest way to understand what's going on here.

    David


    PS Patrick - wanted to make sure you saw my post on the Born and MWI thread ;)


    [1] Zbinden et al. Experimental test of non-local quantum correlation in relativistic configurations. quant-ph/0007009
    See also Gisin, Scarini, Tittel, and Zbinden quant-ph/0009055
     
  26. May 24, 2005 #25

    vanesch

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Yes, I understand that. What I wanted to say is that certain events in the sample will be almost perfectly synchronized, making the speed arbitrary high.

    I have difficulties with that idea, in that brick walls, rivers, tunnels etc... don't seem to do anything.

    My personal theory is that beyond a spacelike separation of 70802 lightyears, the effect suddenly disappears :tongue2: But I can't get funding for an experimental verification... :redface:

    cheers,
    Patrick.
     
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook