Talk:Delayed choice quantum eraser

From Wikipedia, the free encyclopedia

	Portal This article is within the scope of WikiProject Physics, which collaborates on articles related to physics.
???	This article has not yet received a rating on the assessment scale. [FAQ]
???	This article has not yet received an importance rating within physics.
Help with this template Please rate this article, and then leave comments to explain the ratings and/or to identify its strengths and weaknesses.

Please enter all new discussion at the bottom. Nobody who actively follows this discussion is likely to notice it if it goes up here with the much earlier stuff. Thanks.

Archive 01

Contents 1 Why this is not an ansible 2 Cramer ansible 3 Future ansible discussions 4 Wow! Great Job! 5 Associated questions 6 Back to basics

[edit] Why this is not an ansible

This particular device does not allow communication into the past or faster-than-light by the method of adding or removing a distantly-located quantum eraser and noting a change in the interference patterns as seen in the nearby D0 interference pattern.

You get an interference pattern (at D0) in the D1 case, and also in the D2 case, but not in the D3 case, nor in the D4 case. You could replace the D3 and D4 pick-off half-silvered mirrors with fully-silvered flip mirrors. Flip the mirrors into the beams and you get D3 or D4. Flip them out of the beams and you get D0 or D1. This is a delayed choice quantum eraser where the choice (of whether or not to do that erasure) is under your control. Attach the flip mirrors to a telegraph key and you send morse code.

I think your point is that at D0, while getting single photons, there is no pattern. One would need to get enough photons recorded that were all either self-interfering or all not self-interfering to be sure that whatever single photon was recorded was not just an incidental photon that either happened to fall in the center or happened to fall in an area that would be appropriate for the formation of a fringe. Right? P0M (talk) 17:00, 3 March 2008 (UTC)

That is not my point. My point is that at D0, if you take a million photons (which are coincident with a quarter million in each of D1,D2,D3,D4) then you will not see an interference pattern. If you select out the quarter million photons that are coincident with D1, then you will see an interference pattern at D0. Likewise the quarter-million coincident with D2 will produce an interference pattern on D0, but not the quarter million of D3 nor of D4. If you take the half-million photons that are in coincidence with either D1 or D2, then you do not get an interference pattern. (D1:yes, D2: yes, D1+D2:no). So even if you know that the second photon hit either D1 or D2, if you erase the knowledge of which one it hit, you erase the interference pattern. DMPalmer (talk) 00:23, 6 March 2008 (UTC)

I believe that what you say is valid. The apparatus is set up so that a photon may show up at D3 or D4, and in either of those cases there is no pathway by which it can interfere with itself. Or, a photon may show up at D1 or D2, and in both those cases it must interfere with itself because there are pathways open from both slit one and slit two to each of those detectors. I've worked out the reason that photons that show up at D1 and D2 will have opposing polarities. (See diagram above at the top of the section on Polarization.) I haven't had time to work out the phase changes, but it looks like a similar artifact of experimental design must be involved. P0M (talk) 04:48, 6 March 2008 (UTC)

However, flipping the mirrors in or out does not cause a detectable change in the pattern seen on D0. With the mirrors in or out, you still get a broad peak without any interference fringes. (This broad peak looks like what you would get from a single slit.) However, if you take the photons observed at D0 which occurred at the same time as photons observed at D1, you will see an interference pattern. (This can be done simultaneously, as in the Kim et al. experiment, or can be done after the fact, when floppy disks recorded from D1 on Wolf 359 are brought back by slowboat and compared to those recorded from D0.) Likewise, you see an interference pattern in coincidence with D2. But the recorded D0 data does not change at all once you have the D1 and D2 data.

However, the D1 and D2 coincidence interference patterns are 180 degrees out of phase. (See figs 3 & 4 of the paper on the arxiv.) So the D1 pattern looks like A_C_E_G_ and the D2 pattern looks like _B_D_F_H. If you don't have the D1 and D2 measurements, taking all the D0 data gives you ABCDEFGH, which is exactly the same as for the D3 and D4 cases.

Therefore, you don't know what message was sent until you get the D1/D2 data back via classical channels, and it is too late to play the lottery.DMPalmer (talk) 19:28, 1 March 2008 (UTC)

Rather than the complications of the original apparatus, which were designed to permit or prevent interference on a random basis, what would happen if the two paths in the lower part of the apparatus were either caused to diverge or were caused to converge on a single target? If they converge it would be equivalent to a simple double slit experiment that permits interference and produces fringes. If they were diverged it would be equivalent to a simple double slit experiment with an opaque wall separating the paths from the two slits. 152.17.115.79 (talk) 21:30, 3 March 2008 (UTC)

Since this idea that the experiment could be used for FTL or backwards-in-time signalling seems to be a fairly common confusion, I re-added a deleted section discussing the fact that the total pattern of signal photons at D0 never shows interference, and that the interference pattern can only be recovered by looking at the coincidence count between signal photons and idlers which went to one of the which-path-erasing detectors. Hypnosifl (talk) 21:51, 8 March 2008 (UTC)

I think it is worthwhile to add this part back in if it can be made entirely clear to the average well-informed reader.

Without the bottom (idler) part of the apparatus, interference would be seen at d0 because a part of the light wave emerges from each slit and an entangled wave-function emerges from the two regions (for the red path and the blue path) of the BBO. So it functions just as would a wave-function emerging from the two original slits. But the BBO sends a wave-function along the other two red and blue paths. If there is nothing in that path that "sequesters" a photon by making it show up in a place that only one path can reach (i.e. by collapsing the wave-function there), then it is unclear to me how it could influence what happens in the upper path. One would have to arrange to have something happen to keep the red and blue paths from merging, and then one would have to have a detector that would, with high efficiency, make a photon "show up" on it, i.e., making the wave-function collapse at that point. (Modified P0M (talk) 07:27, 9 March 2008 (UTC))

Once you have that lower (idler) apparatus, however, you have a 50/50 chance of a photon showing up at d3 or at d4, which makes the entwined photon show up on d0 as it would if it had simply been diffracted by a single slit, so there is no interference pattern 50% of the time. The other 50% of the time you should get photons showing up at d0 not as if they had been diffracted but as is they had been parts of an interference fringe. But the sum of two interference fringe patterns with a side-to-side displacement is not a black screen. It should be an evenly lighted screen since the two interference fringes are complementary. So the total result, when the lower (idler) apparatus is in place, should be a central bright patch (the diffraction pattern 50%) on the background of a very wide band of moderate illumination (the joint result of all the photons showing up in their appropriate diffraction fringes).(Modified P0M (talk) 07:27, 9 March 2008 (UTC))

So it looks to me as though a "telegraph" could indeed be sent, with "dit" being an ordinary diffraction pattern (when the lower two, idler, paths do not let the wave-function fall on the same detector) and a "dah" being a wide band of moderate background illumination (when the two idler paths are managed so that the wave-function falls, by way of two paths, on the same detector. (Modified P0M (talk) 07:27, 9 March 2008 (UTC))

One can also imagine doing things like putting single detectors in place of the two first-surface mirrors in the lower path. That way no matter where the wave-function collapsed and the photon showed up it would be unambiguously at the end of only the red path or only the blue path, and when the apparatus was put in the way of the lower red and blue paths the upper path would necessarily direct the two paths onto the same detector making the wave-functions collapse in such a way as to manifest a diffraction-type pattern. (modified P0M (talk) 07:27, 9 March 2008 (UTC))

Your argument is unclear to me. When you say "without the bottom part of the apparatus", what specific parts are you talking about removing? Are you talking about removing all the devices involved in measuring the idler photons--the prism, the beam splitters, the detectors D1-D4, all of it? And why do you think interference would be seen at D0? I don't understand what you mean by "ghost light wave" or by the phrase "nothing in that path that sequesters a photon by making it show up in a place that only one path can reach" (what specific path do you mean by 'that path'? Are you talking about the path of the signal photon or the path of the idler?) I think it would help me understand better if you rephrased your point in the language of signal photons and idlers, and referred specifically to particular detectors and paths.

I've tried to make it a little clearer. But to really work it out right I think it will be necessary to look at wave-functions, the collapse of wave-functions, and the results that are consistent for entangled wave-functions. P0M (talk) 07:27, 9 March 2008 (UTC)

Right.

In the part of the apparatus between the laser and d0 there is a double slit. With the exception of the BBO, the apparatus is the same as is seen in the Double-slit experiment. If the paths from BBO that lead into the Glen-Thompson prism were not treated so as to identify which-path information, what would prevent the light waves in the paths from the BBO to d0 from interfering?

As long as the BBO is present to create an entangled pair, the mere fact that the signal photons are entangled with idlers in such a way that there was the potential to measure the idlers in such a way as to determine the which-path information for the signal photons is enough to guarantee that the total pattern of signal photons will not show interference, regardless of what actually happens to the idlers. This is just one of the features of how entanglement works in QM, it's discussed in the physicsforums threads I linked to.

In any case, I can assure you that you're incorrect if you think it's possible to get an interference pattern in the total pattern of signal photons at D0, regardless of what you do to the idlers--when the signal photons are entangled with the idlers in such a way that there is even the potential to determine which path they went through by certain measurements on idlers, this alone is enough to destroy the interference pattern at D0, regardless of whether the appropriate measurements are in fact carried out on the idlers. This is just a consequence of the way entanglement works in QM. See the discussion here and here on physicsforums.com, for example (both threads contain a lot of links to papers by professional physicists to back up the points being made). Hypnosifl (talk) 01:46, 9 March 2008 (UTC)

If the signal photons' paths are combined, the which-path information is "erased," no? Once that is done, there is no way to get it back. If there were, we could look at a pattern of interference fringes and get the which-path information back from it.

Is there some unspoken assumption that the idler photons have some kind of logical priority over the signal photons? I don't see how either idler photons or signal photons can have priority. If that is the case, certain outcomes are possible or impossible for the entire apparatus, no?

What do you mean by "combining" the paths of the signal photons? You mean like running the signal photons through an apparatus similar to the one the idlers went through, where idler paths from different slits were combined if they go to detectors D1 or D2? For a pair of entangled particles, we should really just talk about whether the measurements made on both members of the pair preserve or erase their which-path information; in the standard DCQE setup, when the signal photon is measured at D0 and the idler is measured at D1 or D2 this means the which-path information for both is erased, and when the signal photon is measured at D0 and the idler is measured at D3 or D4 this means the which-path information for both is preserved, but in theory you could certainly alter the setup so that even if the idler was measured at D1 or D2, the signal photon could be measured in such a way as to preserve the which-path information for both photons. But I'm not sure if this answers your question--if not, could you elaborate why you think there might be an assumption about the idlers having "logical priority" over signal photons? Hypnosifl (talk) 03:31, 9 March 2008 (UTC)

Actually, I have a fair amount of trouble with the language and the logic that some of the writers on these subjects use. I have wanted for some time to work back through everything and try to distinguish "ways of talking about things" from operational definitions. Right now I do not have the time. Maybe in a few days...

Combining paths? It may be part of the problem that I just mentioned. Here is a starter: Consider the basic double-slit apparatus. Draw a line between the midpoints of each slit and (assuming they are arranged left and right rather than top and bottom) draw a long perpendicular line at the midpoint of that short line. Now construct a thin wall along the long vertical line perpendicular to the wall in which the slits have been made and make it reach all the way to the detector screen, the ceiling and the floor. Now when a laser is shone on the double slits one beam of light will go out the left slit and hit the detector screen in the left chamber just created,and one beam will come out of the right slit and hit the detector screen in the right chamber. Anything that travels through the left slit will be confined to the left chamber and anything that travels through the right slit will be confined to the right chamber. The "which-path information" is certain. Now remove the wall and the "which-path information" will be lost. Two diffraction patterns will be replaced by one interference pattern, the famous fringes, etc. The paths have been combined, or, more precisely, whatever travels along each path is not separated from whatever travels along its counterpart path.

Now consider:

Are not the paths in the upper diagram separated in an analogous way? And does not the introduction of a second beam splitter as shown in the lower diagram combine the paths, or, more precisely, whatever moves along those paths?

While we're looking at the simpler apparatus, When a single photon hits the first beam splitter, some people say that a given photon either gets reflected or gets transmitted. If the photon goes one way, what do you prefer to call the what-ever-it-is with which the photon may interfere if circumstances permit it? It is on one level merely a question of terminology, so hopefully you will have a preferred term and I can agree to use it. P0M (talk) 05:31, 9 March 2008 (UTC)

One source calls it a "presence," but I don't like that very much. [1] Another, [[2]]which I favor, calls it a "wave-function," and the general discussion there virtually prohibits (as does Wheeler) speaking of the photon as going by one path or the other. The wave-function goes by both paths, and the discovery of a photon in the detector screen is the result of the collapse of the wave-function. Nevertheless, at least in popularizations, people continue to speak of the photon going by one path or the other, seek to determine by which path the photon has really gone, etc. P0M (talk) 06:09, 9 March 2008 (UTC)

I'm not sure how familiar you are with the technical details of quantum mechanics, so apologies if I'm telling you things you already know here, but the wave function is a basic part of the mathematical machinery of the Schrödinger picture of quantum mechanics. The wave function is a mathematical function that assigns "amplitudes" (which are complex numbers) to every possible definite outcome that might be obtained from a measurent on the system; for example, for each possible position at which you might find the particle on measuring it, the wave function assigns some amplitude. The complete set of amplitudes at any given moment is the system's quantum state at that moment, and the quantum state changes over time, its evolution determined by the Schrödinger_equation. At the moment you measure some variable like position, the probability of finding any one definite value for that variable is given by the square of the complex amplitude which the wave function assigned to that value immediately before the measurement--this is the Born rule. This is also what is meant by "wave function collapse"--if you want to make further predictions about the behavior of the system after the measurement, you must assume that at the moment of measurement the wave function made a discontinous jump, so that immediately after the measurement all the amplitude for the variable you were measuring became concentrated on the value you obtained, while the amplitude for every other value of that variable dropped to zero. This special sort of state is known as an eigenstate of whatever variable you were measuring. Note that a quantum state which is an eigenstate for one variable may not be an eigenstate for another--for example, in the case of a position eigenstate, where the wave function assigns zero amplitude to all but one possible value of the position variable, this would not be an eigenstate of the momentum variable, this same quantum state would assign nonzero amplitudes to multiple possible values of momentum. Likewise, an eigenstate of the momentum variable cannot be a position eigenstate. This is one way of understanding the position-momentum uncertainty relation, which says that any measurement of position necessarily introduces some uncertainty into the value of the momentum, and vice versa.

Anyway, the upshot of all this is that the basic procedure for making predictions in the Schrodinger picture is to construct a wave function for the system (based on information from previous measurements), use the Schrodinger equation to find the evolution of the wave function/quantum state (the two terms are basically interchangeable) over time between measurements, then each time you make a measurement you assume that at the moment of measurement the wave function "collapsed" into the corresponding eigenstate of whatever variable you were measuring, which you can then evolve forward according to the Schrodinger equation until the next measurement.

I'd agree that all the talk of "paths" is hard to make sense of in the Schrodinger picture, since the wave function will assign different amplitudes to multiple possible positions at each moment between measurement, and one cannot really talk about the "probability" that the particle is at one position or another (along one path or another) except at the moment of measurement when the wave function behaves differently than it does in between measurements (discontinuously collapsing instead of evolving continuously according to the Schrodinger equation). But the Schrodinger picture is just one of several mathematically equivalent methods of making predictions in QM--there is also the Heisenberg picture, as well as the one that is probably most relevant to the talk of "which-path information", the path integral formulation.

In the path integral formulation, if you have some initial measurement of the system and you want to use that to calculate probabilities for the outcome of a later measurement, instead of using the initial measurement to construct a wave function and evolving it forward using the Schrodinger equation, you instead perform a type of quantum-mechanical summation over all possible paths from the initial state (the initial position a particle was emitted from an emitter, say) that end up at the state whose probability you want to calculate (the final position a particle is detected on the screen of a double-slit experiment, say), and this integral will give you the correct probability. In the double-slit experiment, you'd sum over paths which go through the first slit as well as paths which go through the second slit.

You can't ordinarily talk about the "probability" the particle took one path or another, since this would imply the total probability of the particle ending up at a given location would just be an ordinary sum of the probabilities of each path that ends up at that location, when in fact the quantum-mechanical sum over paths doesn't work this way, different paths can interfere with one another and cancel each other out. But when physicists say that certain measurements can reveal in retrospect which slit a particle went through, I think this can be understood in terms of the path integral--in a situation where you get the which-path information by measuring an idler, with the idler found to have gone through slit #1 for example, I think that this means you could correctly predict the probability the signal photon is measured at different positions by summing only over paths that went through slit #1, you'd no longer have to include paths that went through slit #2 in your path integral. I'm not absolutely sure this is correct since I haven't seen these calculations done for the type of signal-idler experiments we're discussing, but I know that in the case of the standard double slit experiment without entangled particles, if you measure which slit the particle went through then you only sum over paths that went through that slit in the path integral.

Also, although technically you're supposed to sum over all sorts of crazy and wiggly paths, paths which deviate from straight lines tend to mutually cancel each other out, which is the quantum-mechanical answer to why particles tend to move in straight lines, and which also means that for practical purposes you can get pretty accurate probabilities just by summing over the types of straight paths illustrated in the diagrams. In his initial discussion of the double-slit experiment in The Feynman Lectures on Physics, Feynman analyzes the double-slit experiment using just two paths which go in straight lines from the emitter to each slit and from each slit to the position on the second screen where he wants to calculate the probability the particle will be measured, for example.

Does this help at all? Hypnosifl (talk) 18:49, 9 March 2008 (UTC)

I'm familiar but not extremely familiar with the points you mentioned above. And what you say indicates that you are not the writer who will say thing in a casual way that will misrepresent the complexities of the situation. The way some writers discuss these experiments, the photon is still a little bullet and it either goes by one path or the other, and the existence of a second path not taken tells us that as it was approaching the split between paths the photon "decided" to take one path or the other. If somebody has no way of figuring out which path it took, then it will interfere with itself (somehow). If somebody has a way of figuring out which path it took, then it will not interfere with itself (for some "reason" of its own). In the extreme case, a photon decides to go around a gravitational lens one way or the other, and it will be seen many years, decades, centuries, later either by a telescope pointed toward one point in the sky or by a telescope pointed toward the other point in the sky. It will not appear in photographic film used by both telescopes. On the other hand, if light from the two telescopes were to be directed onto the same sheet of photographic emulsion, a photon would be recorded at a position in accord with its interfering with itself. (Don't ask me how this thought experiment could be implemented in the real world!) The secondary sources tend to explain this by saying that the choice of which way to observe the photon in the present determines how the photon "decided" to travel. It sounds so wacky as I try to remember how those writers explain these experiments that I have to doubt that I am getting it right.

To me, the idea that some "thing" goes by all possible paths is far easier to accept. For one thing, it doesn't stress my suspension of disbelief so much. A few days ago, I ran across a discussion that handled my problem by speaking of the photon as what is emitted and what is detected. But in between, the author did not speak of the photon. I think he just called it the "light wave."

I just wanted to know (1) whether you would say that in the basic double-slit experiment some "thing" goes through both slits, and (2) if you do believe that some "thing" goes through both slits, then what do you favor as a term to use to speak of it to the average well-informed reader?

I'm out of time for the present. I hope I have managed to ask the question clearly. P0M (talk) 21:08, 9 March 2008 (UTC)

I guess what I'd say is that we can't answer any questions about what is "really" going on between measurements without getting into the issue of the interpretation of quantum mechanics, since different "interpretations", such as the Bohm interpretation or the transactional interpretation or the many-worlds interpretation, could give different answers to this question, and all the different "interpretations" are experimentally indistinguishable so this is not a question that empirical science can settle. But in the context of talking about "which-path information", I think we can just talk about allowable paths in the path interpretation, so that if you know which slit the particle went through, that means all the allowable paths are the ones that went through that slit. You can talk this way while remaining agnostic about whether there was really any "thing" that was following one or all of these allowable paths between measurements. Hypnosifl (talk) 07:01, 11 March 2008 (UTC)

Fair enough. (When I mentioned "ghosts" before I was teasing people a little bit. It's very important to not let language and thinking habits get in the way of understanding.) P0M (talk) 13:55, 11 March 2008 (UTC)

Not having a lab with the requisite apparatus, I am of course in danger of missing the obvious. Let's take things a step at a time.

I put in the missing reflected paths. Thanks for bringing that omission to my attention. P0M (talk) 03:07, 9 March 2008 (UTC)

Thanks. If you have the time, could you also modify the angle of the middle beam-splitter so that it looks more like the diagram on this page? As it is the new reflected lines you just added to the diagram don't seem to be obeying the law that "the angle of incidence equals the angle of reflection". Also, could you add the labels "BSA", "BSB" and "BS" to the three beamsplitters as in the diagram on that webpage, so that specific beamsplitters could be referred to in the article if needed? Hypnosifl (talk) 03:39, 9 March 2008 (UTC)

I'm not very good with Inkscape. Wikipedia is rather insistent that it be used. The letters will be easy to add. Getting the angles more realistic may be harder than it seems it would be. I'll see. P0M (talk) 05:31, 9 March 2008 (UTC)

The new diagram looks good to me--again, thanks for taking the time to draw this up, it definitely makes the article more understandable. Hypnosifl (talk) 18:49, 9 March 2008 (UTC)

Actually, I have one last suggestion which should be easy to add--I just remembered that the article refers to "slit A" and "slit B" as these terms are used in the diagram in the paper, and it says the D3 detector gets idlers from slit A while the D4 detector gets idlers from slit B. So, could you add the label A next to the blue slit on the black double-slit screen, and B next to the red one? Hypnosifl (talk) 20:24, 9 March 2008 (UTC)

Done. (I should be elsewhere, sigh. But who can I blame but myself?) P0M (talk) 01:44, 10 March 2008 (UTC)

[edit] Cramer ansible

The particular device described in this article may have some limitations or special characteristics that obscure the non-local factors. Dr. John Cramer has produced an idea for an experiment that could test the idea of "retrocausal" activity.

I think I read one of his postings that said he was trying to get funds together to do the experiment. Maybe twenty km. of fiberglass cable is expensive. His take on the experiment is that he cannot see any theoretical problems with it, so he gives it at least a slim chance of actually working.

The way of determining whether the conditions are right for interference fringes in the top path is to either put the detector where an interference fringe would "focus," or else pull it far enough back that interference is negligible. I haven't put in the distances between various sets of components.

This experiment is designed so that it could operate with a continuous strong stream of photons being provided by the laser. So there is no need to count coincidences to eliminate the odd photon from outside the lab. P0M (talk) 04:17, 5 March 2008 (UTC)

One possibly more straightforward way of gaining or destroying which-path information would be to modify the Cramer experiment as indicated in the second diagram. With the mirror-shutter closed (in place), there is which-path information available. With it open the two paths are open and the lens converges them onto the detector screen where they will interfere. Once that happens, which-path information is lost. P0M (talk) 03:19, 9 March 2008 (UTC)

I recently got into an extended discussion about the Dopfer experiment which Cramer is using as a basis for his own experiment. It started when I read the page here which discusses the Dopfer experiment and speculates how it could be modified to become an ansible, or a device for sending information back in time.(a) Then I came across these posts (b) by someone named "Ben" who seems to have some familiarity with the Dopfer thesis, and claims that there are some misunderstandings in the previous page I linked to which had been arguing the experiment could be turned into an ansible. I posted Ben's comments on this thread (c) from physicsforums.com, and then got into a long discussion with another poster about the thesis and whether Ben's analysis makes sense; if you want to follow the discussion we had (there were a lot of posts on other subjects on that thread too), the relevant posts are #44, #50, #53, #57, #62, #66, #68, #69, #70, #71, #72, #74, #78, #80, #81, #82, and #83 (I was posting as 'JesseM'). I think we managed to come up with some sensible ideas about why, in orthodox QM, Dopfer's experiment could have the results it did and yet this would not open up the possibility of FTL or backwards-in-time communication. Hypnosifl (talk) 20:06, 16 March 2008 (UTC)

I just had a look at (b) above. It sounds like Ben believes that any "interference fringes" seen by means of the movable detectors are simply artifacts of the experimental design. I'm not entirely sure what either one of them actually means, but the other guy seems to be clearer in his thinking. Think of it this way: You have normal vision and can see an interference pattern. I have poor vision, so I use the equivalent of a photographic scanner to scroll over the surface of the detection screen (perhaps using a little telescope so my scanner won't cast a shadow on the screen it is trying to scan). I will get the same information you do, but it will come to me line by line (or maybe musical tone by musical tone if I am totally blind) as the stepper motor moves the scanning head in tiny increments. A significant detail of the experimental set-up is that the scanning head has to pause for some time at each station in order to collect a fair sample. Since the arrivals of photons are on a probabalistic basis, a short sample might be entirely too high or entirely too low to be representative. If 32 people flipped a coin 5 times, one of them probably would get all heads or all tails. Somebody who watched only his performance would probably suspect that the coin was loaded. The reason that they do not use a large screen electronic detector is simply because they don't have it. It would have to be not only very sensitive but also very "noise free." A large screen with a little intrinsic noise (or with lots of stray photons coming in from the general environment) would mean that it would be very hard to be sure what was relevant data and what was distraction. Even coincidence timing would be problematical if there was enough noise in the system to make multiple hits occur between beats of the timing clock or the recharge time of the photon emitter. P0M (talk) 04:10, 19 March 2008 (UTC)

I think Ben just meant that they intentionally made the detectors narrow, because the only way to see an interference pattern at one detector is to do a coincidence count between photon hits there and photon hits at the other detector which lie in a narrow range of positions; if you don't do a coincidence count, the total pattern of photons at a detector will never intereference, just like in the DCQE. Why do you think this is unlikely? After all, to suggest the total pattern of photons at one detector would vary depending on what happened at the other detector would clearly violate Eberhard's theorem. Hypnosifl (talk) 04:41, 19 March 2008 (UTC)

I clicked on the wrong link earlier, so I missed what I have now labeled as (a) and went to what I have now labeled as (b). The description at (a) makes it clear that if one eliminates the noise in the experiment (it indicates that some entangled photons do not have equal momenta, etc.) what one does with the idler (i.e., the part of the experiment with the detector that can be placed at two differences from the Heisenberg lens) determines what happens at the detector that is at a fixed distance from its lens. Using the coincidence counter does not, as far as I can see, do anything to the experiment except removing noise. So it appears that, as far as communication opportunities go, the problem would be how to "purify" the stream of photons coming into the Heisenberg lens. Since some of the photons coming out of the BBO have different energies, a narrow band-pass optical filter could absorb part of them. It is also possible that the BBO crystal could be better constructed or replaced with a more efficient converter. It is possible that the detection screen could be tuned to the frequency of the entangled photons. It is also possible that what looks like an evenly illuminated screen to human vision could still be processed to show that certain areas of the screen are more highly illuminated than random variance could account for.

However, I will keep going through the links to discussion. It's always a little difficult to avoid getting sidetracked by wishing that you could get clarifications as you are reading through. P0M (talk) 16:45, 21 March 2008 (UTC)

I just checked out (c) above. I do not see what Ben's diagram has to do with the Dopfer experiment. Here is how I see things:

In the lower part of the apparatus, entangled photons are sent through a double slit. They then form a or Young's interference pattern on their detector. Presumably a signal is generated by this detector and goes into the coincidence counter. Meanwhile, lots of photons have shown up on the wall adjacent to the double slits.

On the other side of the apparatus, a relatively large number of photons is coming through, since there is nothing corresponding to the double-slit barrier. They will hit the detector screen. Why? Their entangled mates have had their positions determined by landing on the wall around the double slits. So the ones on this side will not interfere with themselves since their paths have been determined. Moreover, they would have landed at random positions even if there were nothing specific set up for their entangled mates.

In order to get a high sensitivity reading, an appropriate detector is mounted on a trolley that is moved by a stepper motor. At each position along its course it would pick up a large number of photons that are paired to photons that don't get through the double slits. So the experimental strategy is to filter out the false positives by counting only those photons that match with the reception of a photon that has gone through the double slits. At some positions along its course there will be virtually no hits, corresponding to the dark fringes in the detection screen associated with the double slits. At other positions there will be very many hits, corresponding to the bright fringes in the interference pattern. If the experimenters are being truthful, the results of these readings should be analogous to what you would get by photographing the traditional interference and then scanning it with an ordinary desktop scanner and counting pixels in vertical bands (corresponding to the fringes or to fractional parts of them). P0M (talk) 22:10, 21 March 2008 (UTC)

(Unindenting) I just finished going through the long set of links provided above. I think I may have a different position than either of the main participants in that discussion.

I've downloaded Dopfer's dissertation. The original article shows, in Abbildung 4.3, an electron source at the bottom of the illustration, a double slit, a screen that would pick up an electron aimed obliquely through the slits, a lens, and two positions for a detector, one at the focal length of the lens, which would bring light waves into registration at a closer distance than would otherwise be the case. In the original Young double-slit experiment, light from the sun was brought into the lab from the outside because the distance between the sun and the photons that are detected at various points on the detection screen are virtually identical. Passing through a single slit, a much nearer point light source would tend to create a wider band because the angle between the source and the two sides of the slit is greater. (It'd the same geometry that makes nearer things look bigger.) Since two slits work the same way, if the light source is not at infinity it is helpful to use a lens to "pull" the divergent light paths together. So the center of the light wave that goes through the left slit would be pulled to the right, and the center of the light wave that goes through the right slit would be pulled to the left, and the center of each light wave will then meet at the center of the detection screen. In that case you will get a good set of interference fringes, and you will not be able to look at a single photon and tell which slit it has come through. If, however, you move the detection screen back and trace the "rays" through the apparatus, you will discover that they no longer meet on the detection screen. Instead, one ray will terminate at one point on the detection screen, and the second ray will terminate at another point on the detection screen. And if the experimenter discovers a photon at one or the other of these two points then it will be clear which path the photon can be associated with. The reason the experimenters do this is that they want to be able to choose either a case where they can see the wave character of light (in the nearer position where the light waves are neatly superimposed) or a case where they can see the particle character of light (in the farther position where they can say that if a photon is detected at position g then it must have gone through slit a).

The next diagram (p. 33, Abbildung 4.4) shows a related experimental set-up in which the electron is replaced by photons generated by passing laser output of 351.1 nm through a crystal that down-shifts them to create two entangled photons of 702.2 nm. The double-split wall interrupts one half of the entangled photons, and many of them terminate on that wall. The ones that find their ways through the double slits then pass through a "helper" lens which is intended to make sure that the light waves are nicely superimposed on the "double-slit detector." In order to avoid having the partner photons that reach the Heisenberg detector, only those that coincide in time with the ones that reach the double-slit detector are counted. If a large-area photon detector screen were used as the Heisenberg Detector, then there would often be more than one photon hitting the screen at the same time, so they have elected to place a narrow detector there. Then they have to make it scan across the width of the expected interference pattern, and recording how many hits per equal length of time are recorded for each increment of its journey. (Essentially, it's acting just like the detector in a desktop scanner. The original Macintosh scanners fit onto the printer head of a Mac printer, and the stepper motor of the printer was made to inch the photosensitive head across the screen, recording the light level for each pixel it could resolve, and then turning the whole thing into a low-resolution scan.)

If the Heisenber Detector is racked to its more remote position, it actually shouldn't have to move very far because it should (I think) just see two bright spots at about the separation of the double slits. But there are still going to be lots of entangled photons that reach the Heisenberg detector even though their partners have come to an early end on the first side of the double slit wall. So the coincidence counting is still going to be important.

In Diagram 4.5, it appears that they are holding what we have been calling the Heisenberg detector in its near position and not having it scan from side to side. Instead, they are having the "double-slit detector" move from side to side. When they do that and plot out the intensities (number of hits per "station" as the detector inches across the space where a traditional detector screen would have been), they get a standard Young interference pattern.

In Diagram 4.6 they appear to be holding what we have been calling the Heisenberg detector in its far position, and scanning from side to side with the double-slit detector. The diagram appears to be trying to express some understanding for why the interference fringes will disappear from the double-slit detector. The crystal appears to act as a reflector, transferring something from the Heisenberg detector to the double-slit detector. The result they diagram is that the double-slit detector sees no interference pattern. It sees a sort of bell curve of intensities.

In Diagram 4.7 they have put the Heisenberg detector back in its near position and they scan with it, recording the hits it gets when the double-slit detector gets a hit The result is an interference pattern.

In Diagram 4.8 they have put the Heisenberg detector into its far position again, and they check its hits that coincide with hits at the double-slit detector as the former racks back and forth. In this case, the detector remotely sees the two peaks that correspond to non-interfering beams of light from the two slits on the other side of the apparatus.

Why aren't 4.6 and 4.8 showing the same results? Apparently because there is no double-slit apparatus at the other end of the experimental setup. I am not sure that I have absorbed all of the implications of this experiment yet. The fact that a detector is racking across an incoming stream of light waves cannot determine whether the light waves are coming in or not, can it? I guess one experiment I'd like to see is to just insert a sheet of white paper in front of the double-slit detector for a while during the running of the experiment. (Or a nice wide digital camera CCD would work too.) Don't have it connected to any coincidence detector.

I think I'll look for some other secondary sources on this experiment. It seems to be worthy of an article of its own. P0M (talk) 04:07, 22 March 2008 (UTC)

I just re-read the article that evidently started this discussion (http://www.paulfriedlander.com/text/timetravel/experiment.htm) and it seems quite good to me. Can anyone identify inaccuracies in that article? P0M (talk) 04:54, 22 March 2008 (UTC)

[edit] Future ansible discussions

I think it would be good to find another place to discuss the general issues that gather around delayed choice quantum erasers and the possibilities of instantaneous communications. The apparatus described in this experiment could not be used for such purposes without substantial modification, and the design having been built for automatic production of randomly selected paths it may not be the best example to choose to try to elucidate these questions for the readers of Wikipedia.

Could we find one or two authoritative discussions of whether this apparatus could be used for instantaneous communication? While getting clear on the issues is very important for crafting articles that do not mislead readers, we cannot go farther and put forth our own conclusions. To do so would break the rule against publishing original research in Wikipedia.

Another apparatus (see Media:Walborn_EtAl_QuantumEraser.svg) is discussed in the article entitled Quantum eraser experiment. The authors of the article that reports the experiment indicate that they have done the experiment in a delayed erasure version, and the presence or absence of the polarizing element before detector P determines whether an interference pattern is detected at detector S. P0M (talk) 07:44, 12 March 2008 (UTC)

Well, I found this source which says in section 2.3 that backwards-in-time communication is absolutely ruled out in standard quantum field theory by "Eberhard's theorem", although the paper is mainly about looking for an alternative to quantum field theory which is similar enough that it can't be ruled out by past experimental results and which would allow backwards-in-time communication (but they don't produce a finished theory, only something that can replicate predictions of a simple 'toy' version of QFT). Cramer also mentions Eberhard's theorem in this article (in the paragraph beginning with 'At the AQRTP Workshop ...'), and also says that Eberhard's theorem rules out FTL or backwards-in-time communication in orthodox QM, although he mentions the possible that QM might be incorrect and that the true theory might look like QM with a small nonlinear element introduced into the equations, which would possibly mean FTL (or backwards-in-time) communication might be possible in the new theory.

Is there a link to the Walborn article you're talking about? I wonder if the interference pattern at detector S is claimed to be the total pattern of particles at that detector, or just a subset seen through coincidence-counting as in the delayed choice quantum eraser. Hypnosifl (talk) 15:59, 12 March 2008 (UTC)

http://grad.physics.sunysb.edu/~amarch/Walborn.pdf P0M (talk) 16:54, 12 March 2008 (UTC)

I think those sources should be cited in the article. That would make it "official." I'll have to have a look at Eberhard's theorem. The reason I'm curious is that "backwards-in-time" would seem to imply reversing the flow of time, and probably at minimum involves processes that occur in time, i.e., processes on which one can use the ordinary operational definition of time (watch the second hand go around as you watch the cannon ball fall from the Tower of Pisa), whereas the processes involved in these experiments are non-local and do not involve a process of transmission that one could time. P0M (talk) 16:54, 12 March 2008 (UTC)

But the two sources I linked to aren't talking about whether processes can involve backwards or FTL causality, but rather the more narrow question of whether observers could actually communicate information FTL or backwards in time (FTL and backwards-in-time communication would be equivalent in relativity, see Time travel#The equivalence of time travel and faster-than-light travel)--the first paper refers to the "backwards-time flows of information" and the Cramer article says that Eberhard's theorem shows QM "cannot be used for FTL observer-to-observer communication". So I'm pretty sure all they're talking about is the results of measurements made by each observer, and whether the results of one observer's measurement can instantly give him any information about what type of measurement was made by another distant observer, not interpretational questions of what is going on "behind the scenes" to explain the process that determined each observer's results.

And the Walborn article does seem to be just talking about interference patterns seen in coincidence counts, not in the total pattern of photons at any detector--all their graphs have the vertical axes labelled as coincidence counts. Hypnosifl (talk) 18:48, 12 March 2008 (UTC)

By the way, if we're going to cite sources in the article for the impossibility of FTL or backwards-in-time communication in orthodox QM, it might be a good idea to cite the original Eberhard article. The first paper I linked to lists two papers by him:

P.H. Eberhard, Bell’s Theorem and the different concepts of locality, Nuovo Cimento, Vol. 46B, No. 2, August 11, 1978, p.392-419. Also see P.H. Eberhard and R.Ross, Quantum theory cannot provide faster-than-light communication, Foundations of Physics Letters, Vol. 2, No. 2, 1989, p.127-149.

I assume the second paper would be the more relevant one. Hypnosifl (talk) 20:15, 12 March 2008 (UTC)

I don't have access to those articles.

One of the social problems seen in recent years seems to have been made worse, if not caused, by mass media sources that make statements about what "science proves" that are later refuted by scientists when new research bring forth new information. Experience in the real world will tell whether people like Cramer can find a way to exploit entanglement to communicate information outside the kind of processes that take time. In the meantime what we can report is that FTL or instantaneous communication is incompatible with theory x. P0M (talk) 15:15, 13 March 2008 (UTC)

Ironically, just after I wrote the above I walked out into the yard where a construction company is putting a new roof on an outbuilding. I mentioned something about global warming to the owner of the construction company. "It's happened before and it'll happen again. Mother Nature will take care of us." He's now got half of the message, but he won't listen to what scientists tell us about how many years it will/would take to get the carbon dioxide content of the atmosphere back down.P0M (talk) 16:54, 13 March 2008 (UTC)

Right, what we can say definitively is that FTL communication is incompatible with the current mathematical theory of quantum mechanics. But that doesn't rule out the idea that the current theory is wrong and that it is possible in the real world (personally I doubt it since FTL is the same as time travel in relativity, but you never know), and Cramer actually suggested a way it might be wrong in his article (nonlinear terms added to the equations). Anyway, it would be interesting to know what the current theory predicts about the results of Cramer's experiment--it must give some prediction about it. Perhaps, as with the delayed choice quantum eraser, it would predict the total pattern of photons at each detector would show no interference, but by looking at particular subsets one could recover an interference pattern if the top lens was at the right distance. Hypnosifl (talk) 18:43, 13 March 2008 (UTC)\

So what happens if one isolates the two paths that would have merged at either d1 or d2? Presumably d0 would not get an interference pattern from that "side" of the apparatus, no? P0M (talk) 01:34, 14 March 2008 (UTC)

You could ensure that all the signal photons end up at D3 or D4 by replacing the beam splitters BSa and BSb with mirrors, for example. In this case the total pattern of signal photons at D0 will just be the sum of the D0/D3 coincidence count and the D0/D4 coincidence count (since every signal photon at D0 is associated with an idler that went to one of those detectors), and neither coincidence count shows interference, so the total pattern of signal photons won't either. On the other hand, you could ensure that all the signal photons end up at D1 or D0 by removing the beam splitters BSa and BSb altogether, in which case the total pattern of signal photons at D0 will be the sum of the D0/D1 coincidence count and the D0/D2 coincidence count. Each of these coincidence counts does individually show interference, but as noted in the last paragraph of the "The experiment" section of the article, the two interference patterns are out-of-phase so the high points of one line up with the low points of the other, meaning that their sum will look like a non-interference pattern. So, no matter what you do to the idlers, the total pattern of signal photons at D0 always looks like a non-interference pattern. Hypnosifl (talk) 05:53, 14 March 2008 (UTC)

That's not what I had in mind. My original idea might have been difficult to make happen in the real world. Here is another one: Suppose that we remove BSc and redirect Ma and Mb so that the red and blue beams concide at a single detector. They would be of crossed polarizations, a factor introduced by the BBO, so a polarizer would have to be introduced into one or the other path to correct for that factor. P0M (talk) 06:55, 14 March 2008 (UTC)

Even if you removed BSc and redirected the mirrors so the blue path and red path would converge on a single detector, wouldn't they be converging on it at different angles so you could in principle tell which path they came from by measuring their momentum, or else wouldn't the path lengths be different so you could tell which path they came from by measuring the delay between the signal photon detection and the idler detection? Basically I don't know if there's any setup that will cause all the idlers to converge on a single which-path erasing detector in such a way that there's absolutely no way to distinguish idlers that came on the blue path from idlers that came on the red path. Hypnosifl (talk) 07:39, 14 March 2008 (UTC)

Ideally, one could arrange things so that the path lengths would all be equal, and that would seem (but only seem) to take care of time differentials. But if you do the Young experiment and use a detection screen formed by removing one surface sheet from a piece of corrugated cardboard you will still get an interference pattern. You can also cut a hole in the middle of the detection screen and put a secondary detection screen at some distance behind it without causing any problems with the interference pattern.

In the original double-slit experiment, the probability waves converge on the detector screen from different angles. Nevertheless, an interference pattern appears. Even in that simple situation, the times at which photons appear on the detection screen will vary with distances from the slits. But the time and place of positive interference still work out. At pairs of distances along the detector screen the arrival of a high probability part of the wave from the A slit will meet the high probability part of the wave from the B slit and at a random one of those distances a photon will be detected. One might think that the photon would always show up where and when the first high probability parts of the wave hit the screen, but even in this simplest of situations things do not work out in the time sequence that our experience in the macro world would indicate. It seems that one has to say that from the drop in orbital of an electron in the light source to the rise in orbital of an electron in the detection screen is one event. The probability wave that our model(s) say passes between them behaves in such a way that t=d/c elapses during the course of that event, but the photon's "choice" of where to pop up on the detection screen is, paradoxically, time independent. I've seen discussions of momentum measurement too, but I can't remember where. It may be implicit in the original argument where Einstein proposed to remove indeterminancy by physical measurements but Bohr (See:N. Bohr, in Albert Einstein: Philosopher-Scientist, edited by P. A. Schilpp ~Library of Living Philosophers, Evanston, 1949!;

reprinted in Quantum Theory and Measurement, edited by J.A. Wheeler and W. H. Zurek ~Princeton University Press, Princeton, NJ, 1983.) proved by Einstein's own relativity theory that it couldn't be done the way Einstein had assumed. (talk) 22:19, 14 March 2008 (UTC)

Position and momentum cannot be determined at the same time and information gained about one is information lost about the other. When a photon is manifested on a photographic emulsion, you get a chemical change at molecular size, and you lose momentum information. Gaining momentum information destroys the interference fringes.

http://prola.aps.org/abstract/PRA/v57/i3/p1519_1

The reason the entangled particles were used in the first place was to try to get around these strictures by measuring momentum of the idler while measuring position of the signal, or vice-versa. Basically all the ways I have seen that attempt to pin down the path of the idler consist of various ways of isolating (marking) the paths. The apparatus in the experiment under discussion was cleverly designed to automatically and randomly swap in a path isolating apparatus (the parts leading to d1 and d2) while swapping out the path merging apparatus (the parts leading to d3 and d4). The only reason that the two interference fringe patterns are complements of each other is that one path reflects from the first-surface side of BSc and the other path reflects from the opposite side of that silvered surface after having passed through a pane of glass. Actually, replacing Ma or Mb with a second-surface mirror might well fix the phase change problem.

But let's confront the issue directly. Let's just ask what will happen if part of the time we use the top front-end, and the rest of the time we use the bottom front-end on the idler portion.

Two different front ends

P0M (talk) 02:53, 15 March 2008 (UTC)

(responding to above comments but un-indenting to make more readable) You do raise some good points about it apparently not being possible to determine which slit a particle in the double-slit experiment went through just based on its momentum or on the timing of when it reaches a spot on the screen. It's possible there's something different about the modified DCQE where you rearrange the mirrors to make all the photons go to a single which-path-erasing detector, but I'm not sure.(1) I do have one other speculation, and it involves the fact that idler photons would not really be detected exclusively on the neat paths in the diagram, the paths just show where most of them would be detected; but if you placed a detector slightly to the left or right of one of the detectors in the DCQE, you should get some photons there too, though less (this is definitely true of the detectors in the Dopfer experiment, as you can see by looking at the diagrams on pp. 36-38 of Dopfer's thesis where they always keep one detector at a fixed position while varying the position of the other in order to graph the spatial distribution of hits on the position-varying detector that coincide with hits on the fixed detector; also see the discussion I just added to the "Cramer ansible" section earlier on this Talk page). So, my alternate speculation about what might happen in the modified DCQE where you aim all the photons at one which-path-erasing detector is this: perhaps if you placed an array of detectors side-by-side, or a wider CCD detector that could detect photons at a range of horizontal positions in the plane perpendicular to the path, then you would indeed find photons at a range of horizontal positions. Then if you took a coincidence count between idler photons at some very narrow range of horizontal positions at this detector and signal photons at the D0 detector, you'd see an interference pattern; but the total pattern of signal photons at D0 would still not show interference (as it shouldn't, by Eberhard's theorem), since the total pattern is really the sum of all the coincidence counts for each narrow range of horizontal positions at the idler detector.(2) Hypnosifl (talk) 20:07, 16 March 2008 (UTC)

I think one thing that is going on with these experiments is that when they are done with single photons the detector must be a very sensitive device, and technical issues (or maybe just the expense) limits the experimenters to devices with very small input ports. So what they do is to put the detector on a movable platform on tracks that is pulled across the "line of fire" by a stepper motor or something of that sort. They collect hits in this meter for equal lengths of time at each position as it moves from one side to the other, and then they assemble all of the data. That way they can build up an interference pattern on a computer screen or some similar device.

I guess they don't do them in total darkness either, and maybe there are some incidental photons that get into the detectors from who knows where, so one of the functions of the Co-incidence detector is to match the arrival of a photon in the detector with the departure of one from the laser. That way if something come into the detector at some other time the assumption is that it wasn't one of their photons and should be disregarded.

I had a go at locating the actual experiment reports on the simplest one of these Wheeler-like experiments after I had tried it out with some simple beam splitters and first-surface mirrors and got nowhere. One kind of thing to watch out for is, e.g., that the actual article may mention a very specific type of beam splitter (one that would be out of my financial reach, perhaps) and people who write secondary source materials omit that little detail. So I get the $10 kind and nothing happens.

Not having a cheap source of entangled photons at hand, I am very reluctant to depend at all on my own predictions of what ought to happen. If I were to forget to take a single factor into consideration I could get an entirely mistaken picture. For the same reason, I'm not very trustful of the experts who say what will happen without really having tried it. If you can't trust Einstein to get it right then whom could you trust?

A single photosensitive charge-coupled device (CCD) can pick up photons from a range of horizontal positions without having to move it back and forth--see here for example. According to the comments by "Ben" that I quoted in my most recent post in the "Cramer ansible" section above, the reason they use photodetectors with a narrow range in experiments like Dopfer's (and in the delayed choice quantum eraser too, presumably) is simply because they want to do a coincidence count rather than register the total pattern of photons--you can only recover an interference pattern by doing a coincidence count where you look at the subset of photons at one detector whose entangled twins were seen at the other detector with its position held fixed. Hypnosifl (talk) 03:46, 17 March 2008 (UTC)

(1) above:"It's possible there's something different about the modified DCQE where you rearrange the mirrors to make all the photons go to a single which-path-erasing detector, but I'm not sure." The reason that d3 and d4 pick up interference patterns that are complementary to each other, and that would produce an evenly illuminated screen if they were combined, is that bs3 is constructed so that there is a partially silvered first surface mirror if you look at it from one side, and a partially silvered second surface mirror (like a regular mirror except for not being fully silvered).

There is a phase shift when light goes from air to glass and/or glass to air, so the light that bounces off the first surface does not undergo a phase change, but the light that bounces off the other surface is 180 degrees out of phase. When the two beams form interference patterns in the same detector they are flipped, left to right.

(2) I need to work through the interferometer versions of the double-slit experiment to get clearer on what is going on. One of the problems with my little physics lab, I just realized, is that the beam that comes out of a laser pointer forms a spot on any detection screen. But what you want is the approximation of a geometrical line of photons, not a geometrical cylinder of photons.

Let's follow a single photon, and then keep in mind that in the real world there would be a bunch of other photons being fired along parallel courses. The single photon comes out of the BBO and we could argue about which port it "really" exits from. It is more useful to talk about the probability wave.The probability wave has to follow two paths and emerge from them in such a way that the maxima and minima do not exactly coincide, or otherwise we will not notice any interference. In the Young experiment this arrangement is guaranteed because of the lateral separation of slit a and slit b. In the interference version you must have to "aim" the two probability waves correctly at the same detector screen. If you use a single photographic emulsion you can fire single photon after single photon, and their probability waves (assuming the experimenter has gotten things lined up right) will interfere. When they interfere, the probability of a photon's appearing at some points will drop to 0 and the probabillity of a photon's appearing at other points will be enhanced. That means that instead of all of the hits occuring near the center (as would happen when a single slit yields a diffraction pattern in Young's apparatus with one slit blocked), you get very wide dispersion. If you haven't done the experiment, you ought to try it. It's very impressive how much the pattern flares out sidewise when the second slit is opened. (You can see the double-slit apparatus I made with the thinest brads I could buy, some household glue, plastic railway track, etc., here:http://www.wfu.edu/~moran/Physics/) From a distance of about 12 feet the diffraction pattern that appeared on the side of my refrigerator was about the size of my hand. When I opened the second slit it covered all of the side of the refrigerator and probably beyond. Since the bands farther from the center are dimmer, it would need some fancy apparatus to tell where the yield of photons is so low that it is impractical to wait around for one. If you are going to use a high-priced photon detector with a narrow input port, then you have to fire a large number of photons one at a time, and count how many out of some quantity (several hundred perhaps) show up at each place where the little trolley holding the detector has a stop on its route.

Back to something I may have said earlier. If the Kim device could be rigged so that temporarily it would permit photons only to show up at d1 or d2, then those would be "particle" finds. (If you measure the output of some apparatus looking for particles you will find particles. If you measure the output of some apparatus looking for waves, you will find waves.) The result on single-path transmissions, the result of something analogous to the double-slit apparatus with one slit blocked, would be a diffraction pattern. As the photograph shows, you are likely to see not just a simple spot in the middle if light is going through a single slit, but a couple of side spots that are not clearly isolated from the central spot. That's because of the diffraction that occurs when light goes through a slit. Probably with an interferometer there would be no diffraction, and then you would get just a spot.

If, on the other hand, you rigged the Kim device so that light could not reach d1 or d2, but could reach d3 or d4, then the detector screen would not be black and would not be lit up by a single spot or a diffraction pattern. Instead, it would be evenly illuminated. So the difference between the two set-ups would be center spot vs. evenly illuminated screen. Would d0 reflect the same thing? That seems to be the natural conclusion to draw from the experiment as it has been described.

The way Kim et al. did the experiment was governed by the desire to see whether the location of a photon appearing at d0 would match the location of an entangled photon appearing at d1, d2, d3, or d4. There may have been noise in the experiment. (I think I have read something about random quantum effects producing the occasional photon, and maybe the lab the experimenters were operating in was not totally black.) For whatever reason they decided that the practical way of being sure what they were seeing on the various detectors was to detect "coincidences". (If there were path length differences then they would have had to account for a lag between detection times.) I can't remember for sure, but they may not have needed tractors on detectors d1 and d2 because the photons would have shown up at or near the center in all cases. But they probably needed tractors on d0, d3, and d4, and then they would have had to run the experiment for some time to accumulate representative numbers of hits at all the "stations" of the trolleys.P0M (talk) 00:43, 19 March 2008 (UTC)

Discussion at http://www.sciencemag.org/cgi/content/full/315/5814/966 and elsewhere seems to me to indicate that at least at the current level of experimental device sophistication there must be problems with interference effects being washed out simply by the presence of unwanted unentangled photons getting through from the laser. I've had a go at reproducing the results of that much simpler experiment just using a continuous laser beam. So far I have not had any success whatsoever. Alignment is not easy to achieve, for one thing. Over long distances it would be extremely difficult to maintain alignment, and probably impossible to screen out all extraneous photons. Probably the best thing for the article is to point out that experts in the field maintain that instantaneous communication would contradict quantum mechanics and/or other highly successful theories. P0M (talk) 07:19, 16 March 2008 (UTC)

[edit] Wow! Great Job!

I added this article a few years ago and when I look at all the contributions people have made and all the improvements I am very pleased. I love the diagrams, which I was too lazy to do myself, and much improved text and lots of lively discussion on the talk page. Hooray! Thanks all for all the wonderful contributions. PS I DO think this can be used as the basis of a device that can transmit signals backwards in time. I do not believe that QM has any respect for the arrow of time nor that any impossible paradoxes would be created by this. Obviously that was my point when I added this articel a while back. I see lots of lively discussion, maybe some of the contributors are REAL physicists. If you are, back up your words with actions and DO THE EXPERIMENT.  ;-) Rich.lewis (talk) 19:34, 13 March 2008 (UTC)

Glad you like the changes, I too was happy to see that a diagram was added. But as to your point about sending signals back in time and the need to "do the experiment", the delayed choice quantum eraser experiment has already been done, and it's been seen that, as noted in the article, the D0/D1 coincidence count and the D0/D2 coincidence count are out-of-phase (the part in the article about the peaks of one matching up with the troughs of the other) so that their sum gives a non-interference pattern, which means that the total pattern of signal photons does not show any interference, and an interference pattern can only be recovered after the idlers have been measured and a coincidence count is done (at one point a while ago I emailed one of the authors of the paper to make sure I was understanding this correctly, and he confirmed it). Did you follow that section of the article? Of course that doesn't mean some new "DCQE-esque" experiment might not be done which does allow sending signals back in time (although as noted in the previous section, Eberhard's theorem shows that if the current theory of QM is correct, this should be impossible), but I think it's pretty unambiguous that the results of this experiment show it wouldn't be useful for that purpose. Hypnosifl (talk) 20:25, 13 March 2008 (UTC)

Then what happens to QM (or certain flavors of it) if some experiment such as the Cramer experiment works?

"In any case, our minds are designed to assume this [arrow of time], and cannot possibly learn to live with any other point of view.” Wrong, wrong, wrong. There is no more reason for the arrow of time to be a universal invariant at all levels of nature than there is for the direction “down” to be. And the human brain is perfectly capable of learning to make predictions based on a more powerful model.

Werbos http://www.werbos.com/reality.htm is looking for a way to ground the QM discoveries in something more basic.

Some of the URLs this discussion has turned up have been fascinating. Thanks. P0M (talk) 01:14, 14 March 2008 (UTC)

I've archived the earliest parts of this discussion. Nothing has disappeared, just click on the link at the top of this page. (And if you need anything, just copy it back.) P0M (talk) 02:01, 14 March 2008 (UTC)

[edit] Associated questions

Have a look at: http://www.quantumphil.org/history.htm P0M (talk) 11:07, 15 March 2008 (UTC)

oh cool! I think wikipedia now needs an article on Multisimultaneity Rich.lewis (talk) 17:42, 25 March 2008 (UTC)

quote from http://www.quantumphil.org/history.htm "The final results of the experiments with moving measuring devices (see experiments) rule out the possibility to describe the quantum correlations by means of real clocks, in terms of "before" and "after"; nonlocal quantum phenomena cannot be described with the notions of space and time. This means that there is no time ordering behind nonlocal correlations, so the causal order cannot be reduced to the temporal one. " Rich.lewis (talk) 17:49, 25 March 2008 (UTC)

I had been assuming causality and time were both illusions of our perception. turns out causality may be real, but time itself does seem to be an illusion. Rich.lewis (talk) 17:50, 25 March 2008 (UTC)

One way to look at things has been suggested by people in ancient China who tried to think about the logical consequences that follow from "all is one." If that statement is true, what does that imply with regard to all of the statements we make about "this photon" at "that time," etc.? They came to the belief, and it is a belief that is shared by modern Western philosophers of science and/or phenomenology, and the people like the Vienna Circle, that there is "something out there," but the way we try to understand it is to reach out with our minds and impose our structures on that external reality. Newton did it one way, and was extraordinarily successful. But there were some problems and so another system of "convenient fictions" came to the fore. Whenever we make a better set of fictions we tend to get excited by them and imagine that the real thing is no different from the models that we make to understand that real thing.

Some things that we believe turn out to be false in very palpable and obvious ways. Walking out in the yard at night I reach down for another length of the black polypropylene cord that the guy I bought my house from has scattered all over the place. It slithers away from my hand and I realize that it is a baby black snake. Other things turn out to be false, but in much harder to discriminate ways. Time is one of those "things" that seems perfectly obvious and undeniable. It only turns out to have problems when we think very carefully about extreme cases. And when we do that thinking we do not really destroy our old ways of thinking about time as much as refine them in unexpected ways.

Evolution has tuned our abilities to perceive accurately, and technology has multiplied our natural abilities. We have very good abilities to penetrate camouflage, for instance. We have innate abilities to perceive patterns even when they are mixed in deceptive ways with other patterns. To help us survive we also have good abilities to perceive patterns in moving systems (like the predator sneaking across the grass and brush toward us). Essentially what we do (even if it is only in our gel-ware) is to create a single enduring pattern that maps a series and sequence of images (in the most general sense that might include sounds and smells as well as sights). Some of the ways our environments change us leave records of what has befallen us that may persist for even millions of years. A giant predatory marsupial may have broken a bone a million years ago, but when its bones are recovered a close examination will show traces of that accident. There do not seem to be any serious problems with that kind of idea of time. But it is most dependable when it is possible to keep a single physical entity that shows the sequence of events in its own structures. A dependable record could be anything from a spool of motion picture film to a layer of sandstone that was laid down over thousands of years and then turned to stone under heat and pressure.

The idea of time appears to start to break down when high speeds are measured between two systems and sequences in time are attempted across inertial systems.

The idea of cause is not as simple as it ordinarily seen to be. We tend to think of the cause of an accidental shooting as something like the hound dog that puts its foot down on the trigger, but that's really only one facet of a complicated event. It's impossible (at least in most cases) to find an isolated or discrete event.

The importance of Einstein's work was to show how times as measured by different people in different inertial systems would work out. As long as any process involved is governed by the speed of light, we are on grounds that people over the last century have gradually accustomed themselves to accept. The idea of entanglement was rather an affront to Einstein, but it turns out that what he thought was the death knell of quantum mechanics turns out to be a reality that people have to explore and then will have to learn to accept. P0M (talk) 08:03, 26 March 2008 (UTC)

Take a look at http://www.signandsight.com/features/614.html Zeilinger has a clear statement about how much of what we discuss are actually constructions (or "convenient fictions," or "fish traps and rabbit snares," etc.) P0M (talk) 13:00, 27 March 2008 (UTC)

[edit] Back to basics

The discussion so far has been very helpful to me in trying to see what the real issues are. I have started to go back and look for the fundamental articles to see whether secondary sources have been leaving anything crucial out.

While starting with the first article I dug out this afternoon I started to wonder whether there might not be a way to take advantage of the Young experimental apparatus. One of the advantages it has is that it provides such a simple way of producing interference fringes. So examination of a related thought experiment might help us see the experiments being done today in a clearer way.

The crucial element of many of the quantum eraser experiments is the use of entangled photons. The hope is to be able to use one of the entangled photons to see which path the other entangled photon has taken. So suppose that we make two Young apparatuses. In the one that is in the top of the following diagram, there is no attempt to interfere with the development of the expected interference pattern. In the bottom of the two, which one might imagine first being tested as an exact duplicate of the top one, there are a couple of ways of providing "which path" information. The first way is to simply plug one of the two slits. The other way is to insert a wall as diagrammed. Either way, if anything comes out of slit b it cannot interact with anything coming out of slit a -- either because nothing is coming out or because whatever is coming out is coming out in the other room.

I've only drawn light beams from the BBO to the two double-slit devices. I have also included an idea from Cramer -- to insert a long glass cable between the BBO and the double slit apparatus.

If this experiment works as the others appear to work, whether the entangled photon in the lower part of the total apparatus is permitted to interfere with itself or not will have a potent influence on the the behavior of the "twin" in the upper part of the apparatus. It appears that making the bottom photon unable to interfere with itself will make the photon up top to also be unable to interfere with itself. The results on the detector screens should be unambiguous. There either will be an interference pattern or there will be a diffraction pattern. Without even considering whether changes are made in the bottom path after the photon is observed in the upper path, it is an obvious paradox just to have the choice of "one path or two paths" in the bottom part of the apparatus determine whether the photon in the upper part of the apparatus looks like it had one path or two paths. It obviously has two paths, so how could it be influenced to behave as though it onmly had one path open?P0M (talk) 03:30, 19 March 2008 (UTC)

An experiment sending a pair of entangled electrons through two different double-slits is discussed on p. 290 of this article by Zeilinger (p. 3 of the PDF). He says you won't observe an interference pattern behind either slit, but then points out that if you measure one of the photons in a way that erases any possibility of using it to determine which slit the other went through, then subsets of the hits at the the other double-slit can show interference; he uses the Dopfer experiment as an example of this. However, he says on the next page (p. 291) that the interference will never be seen in the total pattern of photons at the other slit, an interference pattern can only be seen by doing coincidence-counting. Hypnosifl (talk) 04:08, 19 March 2008 (UTC)

Why only one? Why not measure all of them so that their "which path" information is destroyed? Just wondering...P0M (talk) 05:01, 19 March 2008 (UTC)

I didn't really mean that only one would be measured (that isn't suggested by Zeilinger), I should have said something like "if you measure any one of the photons in a way that erases any possibility of using it...", i.e. you'd be free to do it for all of them. Hypnosifl (talk) 06:12, 19 March 2008 (UTC)

So if you measure all of them so that which path information is destroyed, which is what the apparatus I've dreamed up is meant to do, then there should be a simple interference pattern in the other detector, no?

Yet according to what you report, you would only get an interference pattern if you did matching? Why should matching be necessary if everything is constant in the experiment except for the probabilistic arrivals of photons at fringe maxima? (It might be interesting to see whether a photon arriving in the third fringe to the left of center in one apparatus is matched by one appearing in the third fringe to the left of center in the other apparatus, but what could possibly distort the interference patterns, and how would matching photon arrivals with photon arrivals unscramble anything if that were the case?) I'll have to work through the article more carefully.

To put this another way, suppose that we started with two identical Young apparatuses and fed them with individual lasers. Unless the world is coming to an end or our lasers are really only LEDs, we will probably get two expected interference fringe patterns. Now link the photons being fed to each apparatus by contriving that they be entangled somehow. What happens to destroy one of the two interference patterns? Or are both interference patterns destroyed? Why would that happen? It's not because we now have "which path" information. It might be because the BBO produces photons of complementary polarization, even though I can't intuit any basis for that. Maybe the math would show it. But then we could put in polarizing elements to fix things as is done in some of the quantum eraser experiments.

It might cost me as much as a couple thousand dollars to buy the appropriate laser, BBO, and protective glasses, but it would be almost worth it to see this thing work. Then I could answer interesting questions like, "What happens if polarizing light traps are put on the outputs of the BBO that would otherwise head to the lower (or upper) set of double slits?"

In the experiment with the five detectors and multiple beam splitters, there is a clear reason why the d3 and d4 detectors will pick up complementary diffraction patterns. It's because of a vanilla flavored optical phenomenon, the change of phase caused by going through the glass side of a beam splitter and then being reflected out the same side. But what is going on in this experiment to produce a similar effect? I'm puzzled, probably because I haven't digested the article yet.P0M (talk) 19:27, 19 March 2008 (UTC)

I just had another look at the Zeilinger article. He says:

If the Heisenberg detector is placed in the focal plane of the lens, it projects the state of the second photon into a momentum eigenstate which cannot reveal any position information and hence no information about slit passage. Therefore, in coincidence with a registration of photon 1 in the focal plane, photon 2 exhibits an interference pattern. On the other hand, if the Heisenberg detector is placed in the imaging plane at 2 f, it can reveal the path the second photon takes through the slit assembly which therefore cannot show the interference pattern (Dopfer, 1998).

The above statement should be true for all cases, i.e., for all photons originating in the experimental apparatus. As far as I can make out so far, he is substantiating, and even going beyond, what I have been saying. Can you give me an exact quotation of what you are referring to on page 291 of the article? P0M (talk) 21:21, 19 March 2008 (UTC)

"So if you measure all of them so that which path information is destroyed, which is what the apparatus I've dreamed up is meant to do, then there should be a simple interference pattern in the other detector, no?"

I don't think so, no. After all, you could easily do something like that with the DCQE, removing the beam-splitters BSa and BSb so that most of the idlers would end up at the which-path-erasing detectors D1 and D2. But since the D0/D1 interference patter is out-of-phase with the D0/D2 interference pattern, so that the sum of the two interference patterns is a non-interference pattern, then that suggests that even if all the idlers end up at D1 or D2, the total pattern of signal photons at D0 won't show interference. I would imagine that in any experiment where you erase the which-path information of one set of photons (let's keep calling these the 'idlers', even if the setup is different than the DCQE) so that there's no way to tell which slit their entangled twins (call them the 'signal photons') went through, something similar will be true; there will be different possible locations you might measure the idlers, and if you look at the coincidence graph between idlers at one of these locations and the corresponding signal photons, you'll see interference, but the sum of all these interference patterns in the coincidence graphs will not show interference, so the total pattern of signal photons behind the double-slit will be a non-interference-pattern.

Just to be clear, are we just talking about what would be predicted by orthodox QM if we actually did the math to find out its theoretical predictions for any given setup? If we drop the assumption that orthodox QM is correct, then of course pretty much anything could happen in any experiment that hasn't been performed. But as long as we stick to the predictions of QM, it seems to me that Eberhard's theorem demands that the total pattern of signal photons behind the double-slit can't vary depending on how the idlers are measured (whether their which-path info is preserved or erased), since if it did then just by looking at this total pattern you could gain information FTL. Do you agree? If so, I think you must agree that orthodox QM can't predict an interference pattern behind the double-slit even if the entangled idler photons are all measured in a way that erases their which-path info. And yet, as I said above, the DCQE suggests an elegant way that orthodox QM can both avoid the possibility of FTL information transfer but also avoid violating the principle of complementarity which leads us to think we should see an interference pattern when the which-path info is erased; the resolution would just be that you'd see an interference pattern in the subset of signal photons that correspond to idlers that were measured at one particular location, but that when you sum up all the signal photons whose idlers were detected at a range of locations, the interference patterns for each location are out-of-phase in a way that allows their sum to show no interference.

"To put this another way, suppose that we started with two identical Young apparatuses and fed them with individual lasers. Unless the world is coming to an end or our lasers are really only LEDs, we will probably get two expected interference fringe patterns. Now link the photons being fed to each apparatus by contriving that they be entangled somehow. What happens to destroy one of the two interference patterns? Or are both interference patterns destroyed? Why would that happen?"

Yes, I think both interference patterns would be destroyed, simply because entangled particles have different behavior from non-entangled ones; have a look at the discussion threads Does a beam of entangled photons create interference fringes? and entanglement and which-path from physicsforums.com, where various knowledgeable posters say that entangled photons don't show interference fringes, and give plenty of links to papers by scientists to support these assertions.

"Can you give me an exact quotation of what you are referring to on page 291 of the article?"

Yes, I was referring to the paragraph where Zeilinger wrote:

We note that the distribution of photons behind the double slit without registration of the other photon is just an incoherent sum of probabilities having passed through either slit and, as shown in the experiment, no interference pattern arises if one does not look at the other photon.

Does this not mean that if what he calls "photon 1" is not detected at the focus of the Heisenberg lens then what he calls "photon 2" will not register in the "double slit detector" at a position consistent with interference?P0M (talk) 03:19, 20 March 2008 (UTC)

When you say photon 1 is not detected at the focus (and here Zeilinger seems to be talking about the setup of the Dopfer experiment, although he doesn't use that name), do you mean it's not detected there because 1) the detector D1 is placed at some other position than the focus (say, at distance 2f in the image plane rather than distance f in the focal plane), or do you mean 2) that the detector D1 is placed at the focus, but we look at the subset of photons that registered at D2 where there was no corresponding photon hit at D1? Either way, I feel pretty confident that orthodox QM would predict that the total pattern of photon hits at D2 (when we don't do any coincidence-counting) will never show interference, again because of Eberhard's theorem. Hypnosifl (talk) 03:43, 20 March 2008 (UTC)

I primarily had in mind what happens when "photon 1" is not detected at the focus of the Heisenberg lens because the Heisenberg detector is moved to its more remote position. (I guess it is theoretically possible that the detector could be at the focus of the Heisenberg lens but a photon still would not be detected for some reason, but I don't know how anybody could be sure that a photon had gotten through or around the detector somehow.) What he appears to me to be affirming is that if the Heisenberg detector is at the focus of the Heisenberg lens, then "photon 2" must be found at some position consistent with an interference pattern. "A double-slit interference pattern for photon 2 is registered conditioned on registration of photon 1 in the focal plane of the lens." (p. 290) The converse would be that the lack of a double-slit interference pattern for photon 2 is registered conditioned on "placing the detector for photon 1 into the imaging plane of the lens" (p. 290) -- or simply removing the detector and letting the photon show up wherever it may. It would still be possible to determine its path, so that would imply which-path information for photon 2 and no interference pattern.

There would be no need for coincidence counting if the Heisenberg detector were left at one extreme (at the focal plane) or the other (at the imaging plane) unless there are non-entangled photons entering the experimental apparatus somehow.

He summarizes the two extremes as follows:

It is sufficient to destroy the interference pattern, if the path information is accessible in principle from the experiment or even if it is dispersed in the environment and beyond any technical possibility to be recovered, but in principle still ‘‘out there.’’ The absence of any such information is the essential criterion for quantum interference to appear.

P0M (talk) 07:37, 20 March 2008 (UTC)

On the other hand, in the section you quoted he was clearly just talking about an interference pattern in the coincidence count, which is why he said "Therefore, in coincidence with a registration of photon 1 in the focal plane, photon 2 exhibits an interference pattern." Also note on p. 290 where he talks about the experiment of sending two entangled photons through opposite slits, and says:

"Will we now observe an interference pattern for particle 1 behind its double slit? The answer has again to be negative because by simply placing detectors in the beams b and b' of particle 2 we can determine which path particle 1 took."

Note that he isn't saying you actually have to place those detectors in the beams of particle 2, the mere fact that you could do such a thing is sufficient to explain why you never see an interference pattern for particle 1 behind its slit (again, if the pattern for particle 1 depended on what you actually did to particle 2, this would give you an easy way to make an FTL telephone, forbidden by Eberhard's theorem). Hypnosifl (talk) 02:48, 20 March 2008 (UTC)

Theory does not rule over the universe, and physics makes progress by discovering places where theory is disproven. Right now it would be helpful to see what are the questions of real import and what are the questions that relate to artifacts of experimental design.P0M (talk) 16:56, 21 March 2008 (UTC)

But you are saying, correctly, that what the experimenter does not do (that is, the cases when the experimenter does not place any detector in the beams b and b') absolutely determines the lack of an interference in the other part of the apparatus.

"One problem in all experimental situations thus far is due to technical insufficiencies, namely that only a small fraction of all pairs emitted by the source is registered." (p.293) I think the only function of the coincidence counter, in this experiment, is to strain out the noise. If it is true, as he says, that "by simply placing detectors in the beams b and b' of particle 2 we can determine which path particle 1 took," then in that case the experimenter has definitely constrained the interference pattern from manifesting in photon 1. It's unclear to me what function the coincidence counters play other than to eliminate random photon hits. If some photon is registered in one half of the apparatus and is not registered in the other half, then the experimenters rule those appearances out since they cannot be caused by one of the entangled photons. I think that the BBO does not manage to convert every photon that is pumped into it. And I recall that some of the experiments use a filter on the detectors to eliminate photons that are not of the expected frequency.

The Kim experiment has a need for the coincidence counter that could not be reduced by improved experimental design intended to eliminate "rogue" photons -- in that experiment, entangled photons in the signal path are randomly related to photons detected and d1, d2, d3, and d4. It doesn't much matter whether the photons are detected at d1 or d2 because either way there will be no interference pattern manifested in d0. But photons detected in d3 and d4 are related to complementary interference patterns that would merge into an evenly illuminated screen if they were all added together. The Dopfer experiment doesn't have this complication. It's either "interference" or "non-interference" P0M (talk) 08:25, 20 March 2008 (UTC)

I just had a look at p. 3. It seems plausible, at least if you believe that photons are particles that have position even when the operation needed to determine position would also change the event (making the event emitter to point where the position was detected rather than emitter to detector screen). If we leave out the alleged particle, then we have paths to deal with. If a photon is detected somewhere on the screen beyond the double slit, the we know that the photon took path a or path b (or both). If the other photon is not observed we have no help in understanding whether the first photon took one path or the other. If it is observed, then that means that we have put a couple of detectors along paths a' and b', and one of the detectors picked up a photon.

On the other hand, suppose we put up a mirror double slit on the a' b' side? Then we lose "which path" information, no? Then it would be consistent to get interference on both detector screens. I'm out of time, so I haven't read what the author thinks/has discovered about this alternative. More later.

It seems that creating two paths is necessary to get interference, and creating two paths and then isolating them prevents interference as surely as if there were only one path. Or, to put it another way, now instead of one single path we have two single paths.

However, turn things around and they appear not to be symmetrical. If we had two paths for one entangled photon we would not have interference unless the two paths were merged. If the entangled twin photon is in a situation that enforces there being two paths for it, then attempting to merge the first pair of paths does not result in interference. In some ways the following argument seems persuasive: If the a'b' paths are separated by a wall, it would amount to "magic" to be able to merge a' and b' despite the wall. If a' and b' were heading off in opposite directions, it would also be "magical" to get them together just by operating on the a and b path events.

No more time. Maybe tomorrow...P0M (talk) 04:58, 19 March 2008 (UTC)