Brian Green's Beam Splitter Experiments

[selfAdjoint]

I'm at work so I cannot look at those "commercials". However, it seems to me that a physicist who posits an esoteric theory and then faces some (probably not all unjustified!) criticism in an attempt find an answer to this quandry is not entirely unadmirable.

Edit#2: Provided they understand the data and don't knowingly misrepresent it of course!

"What the bleep" was not produced by any kind of physicists; it is the product of a community of disciples of some "ascended master" and the quantum statements in it are new age pap. One physicist who was interviewed protested this on camera, but his protests were cut from the distributed film.

 

[MojaveJoe]

"What the bleep" was not produced by any kind of physicists; it is the product of a community of disciples of some "ascended master" and the quantum statements in it are new age pap. One physicist who was interviewed protested this on camera, but his protests were cut from the distributed film.

Ascended masters

Sounds pretty terrible. I think the idea is to posit as many non-metaphysical theories as possible because how the heck can you test anything else?

Cheers!

Edit: Your point about non-starter theories earlier in the thread strikes me as pretty important.

 

[Reagle]

Originally Posted by Reagle

1. It makes no difference what is at D3.

If the idler photon was sent to D3, and D3 had been a coffee cup (instead of a detector), then the interference pattern would be still destroyed.

Reply by Moving Finger

Agreed. That’s an important point (but we would have no way of detecting and displaying that destroyed pattern any more, because we’ve now lost our detector and substituted a cup of coffee instead).

Reply by Reagle

I don't find it so easy to understand what the author’s mean by "joint detection counting rates". Here is a quote for http://qopt.phys.msu.su/kulik/Papers/p1_1.pdf" on the subject.

It is easy to see that
these “joint detection” events must have resulted from the
same photon pair. It is predicted that the joint detection
counting rate R01 (joint detection rate between D0 and D1)
and R02 would show an interference pattern as a function
of D0’s position on its x axis. This reflects the wave
nature (both-path) of photon 1. However, no interference
fringes would be observed in the joint detection counting
events R03 and R04 during the same scan of detector D0
along its x axis. This is as would be expected because
we have now inferred the particle (which-path) property
of photon 1. It is important to emphasize that all four
joint detection rates R01, R02, R03, and R04 are recorded
at the same time during one scanning of D0. That is, in
the present experiment we “see” both wave (interference)
and which-path (particle like) with the same measurement
apparatus.
Different from

I thought the only thing that mattered, since the photon has traveled through a double slit, was that the position of arrival of the photon at D0 thus determining if the location coincides to an interference pattern.

Could someone who understands please explain what the "joint detection counting rates" are.

Thanks!
Reagle


http://qopt.phys.msu.su/kulik/Papers/p1_1.pdf"

 

[Farsight]

I don't know if it's relevant, but if you print two copies of page 10, figures 3 and 4 tally better if you turn one page over. They appear to be mirror images of one another. Then if you cut out figure 4 and align it carefully over figure 3, the peaks fill the troughs.

http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm

Selfadjoint, thanks for the info re:

Speakable and Unspeakable in Quantum Mechanics by Bell and Aspect and
Where Does the Weirdness Go? by David Lindley.

Much appreciated.

 

[selfAdjoint]

I don't know if it's relevant, but if you print two copies of page 10, figures 3 and 4 tally better if you turn one page over. They appear to be mirror images of one another. Then if you cut out figure 4 and align it carefully over figure 3, the peaks fill the troughs.

http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm

Selfadjoint, thanks for the info re:

Speakable and Unspeakable in Quantum Mechanics by Bell and Aspect and
Where Does the Weirdness Go? by David Lindley.

Much appreciated.

You're very welcome, and thank YOU for the suggestion re the diagram. Anything we can do to make this experiment as clear as possible will help everybody.

 

[MojaveJoe]

On the Beyond the Standard Model board, vanesch and I have discussed a paper detailing the RQM account of the entanglement problem. I am sure they can do double slit too, so it seems to me that if someone wanted to work it out, they could do DCQE as well. The basic principle is that quantum systems are only real in the process of interaction; that's where their observables take on real number values. And each interacting system sees the other's values relative to itself. This gets into the point that the photon doesn't have any spacetime real position between interactions, hence asking "which path did it go down" is a meaningless noise.

I guess this bothers me because isn't quantum theory supposed to explain the universe on a micro level? It seems to me that this is simply saying Quantum Theory is useless.

Cheers!

 

[MeJennifer]

I guess this bothers me because isn't quantum theory supposed to explain the universe on a micro level? It seems to me that this is simply saying Quantum Theory is useless.

Well a theory is supposed to give calculable results that are in accordance with experiments.

 

[Farsight]

Isn't that why we're all here? Because "Quantum Theory" predicts things correctly, but doesn't really explain what's going on under the covers in terms that the layman can understand?

 

[selfAdjoint]

Well a theory is supposed to give calculable results that are in accordance with experiments.

Exactly. In all the history of 20th century particle physics, people again and again started out hoping the latest thing "explained reality" and were disillusioned in one way or the other. But the ability to predict what experiments would see just got better and better. This is not lost on the physicists, hence the popularity of "SUAC".

Sometimes I wonder if all this marvelous ability to predict counterintuitive behavior rationally(as in this experiment!) is a sign of a deeper level of causality, and what we are seeing is the behavior of that level, which is counterintuitive but has these rational patterns to it. But what that deeper level might be like hasn't showed up yet. We do know (if the Aspect and other entanglement experiments hold up - again including this one) is that it isn't a naive classical theory. You have to break classicality some way or other to get the result to behave like QM. 't Hooft for example is experimenting with a huge gauge group; he puts Hamiltonian mechanics through that and comes out with something that behaves like quantum mechanics. But he can't do fermions, and hasn't calculated beyond the simplest behavior. There are other workers in this field, including of course all those QG people, who are starting to get to "coupling matter", which means deriving QM from their various ansatzen.

 

[selfAdjoint]

I just had a thought. Vanesch, in a post on another thread you said that Aspect-type tests of quantum entanglement had some irreducible uncertainty because detectors with fast enought switching time did not exist.

Isn't the DCQE interpretable as a test of quantum entanglement? And doesn't it get around the switching time problem?

 

[Farsight]

I bumped into this post from vanesch, third one down. Maybe it's relevant. Calling vanesch!

https://www.physicsforums.com/archive/index.php/t-59501.html

"Although it is a nice experiment and so on, I think it is a bit "oversold". You can give it the interpretation given by the authors if you want to, but in fact, something much simpler is going on. When you look at figure 3, you get an interference pattern, because you SELECT A SUBSAMPLE from all impacts at D0, which are coincident with a hit at D1. When you look at ANOTHER SUBSAMPLE, namely the impacts at D0 which are coincident with D2 (figure 4) you get a shifted interference pattern. This shift comes about, if you trace it back, to a difference in optical pathlengths in the polarizing beamsplitter, and this shift is utterly important..."

 

[moving finger]

And does Cramer have a detailed account of his approach dealing with the delayed choice quantum eraser?

I’ve not seen any. Applying the Transactional Interpretation (TI) to entangled entities such as the signal & idler photon would entail that advanced waves for both the signal & idler are sent back in time (the advanced wave for the signal from D0, and for the idler from either D1, D2, D3 or D4), these two advanced waves then arriving at the “source” at the same point in time (I guess this is a problem for the TI – how do two initially uncorrelated advanced waves arrange to travel back in time and arrive at the source at the same moment?). If this problem can be overcome, then the advanced waves “explain” how the signal photon “knows in advance where to land” on D0.

You and vanesch both complain about that. Who says nature has to be fully explained in 2006? What justification does "needing a useful philosophy" give for forcing the QM formalism into doing things it was never intended to?

I’m not suggesting forcing the QM formalism into doing any kind of things it was never intended to do - I’m interested in trying to find out what meaning, if any, lies behind the empirical data from quantum mechanics experiments, and whether we can understand more about what is really going on than what the QM formalism is currently telling us. The Copenhagen approach is great – for physicists. It’s not so very useful for most philosophers.

On the Beyond the Standard Model board, vanesch and I have discussed a paper detailing the RQM account of the entanglement problem. I am sure they can do double slit too, so it seems to me that if someone wanted to work it out, they could do DCQE as well. The basic principle is that quantum systems are only real in the process of interaction; that's where their observables take on real number values. And each interacting system sees the other's values relative to itself. This gets into the point that the photon doesn't have any spacetime real position between interactions, hence asking "which path did it go down" is a meaningless noise.

I disagree. Asking “which path did it go down” has meaning if and only if one measures which path it goes down – as the DCQE itself demonstrates. This fits with your description of RQM – that quantum systems are real only in the process of interaction. In other words – whether a quantum entity is manifest as a wave or a particle depends on how you measure it.

Could someone who understands please explain what the "joint detection counting rates" are.

For any given entangled (signal plus idler) state, both “parts” of that state leave the source (the slit) at the same time. We know the distances to the various detectors, therefore for any signal photon detected at D0, we can calculate the exact time that its partner idle photon should reach each of D1, D2, D3 and D4. But that particular idler only reaches one of these detectors – therefore by “looking” for an idler arriving at either of D1, D2 D3 or D4 at just the right time (the effective coincidence time as calculated from D0) we can determine which detector (either D1, D2, D3 or D4) the idler photon went to.

Repeat the above for each entangled pair leaving the source – that way we build up a correlation of signal photons at D0 (with x-position information) with corresponding idler photons at either D1, D2, D3 or D4.

Sometimes I wonder if all this marvelous ability to predict counterintuitive behavior rationally(as in this experiment!) is a sign of a deeper level of causality, and what we are seeing is the behavior of that level, which is counterintuitive but has these rational patterns to it.

I like it! Sounds like super-determinism to me (but I’m not sure I agree it’s counterintuitive). Having been forced (thank you, selfAdjoint) to think about Cramer’s TI in more detail and especially to think about its application to entangled states, I’m leaning more towards hidden variables and the world being super-deterministic. It seems to me that this is the only way to explain the temporal sequence of how the signal photon in the DCQE “knows where to land” on detector D0, in advance of the idler photons “deciding which detector to go to” – the information telling the signal and idler where to go is already built-in at the start..

As far as I can see, no other “interpretation” can coherently “explain” the temporally asymmetric information inherent in the DCQE experiment.

Best Regards

 

[moving finger]

I bumped into this post from vanesch, third one down. Maybe it's relevant. Calling vanesch!

https://www.physicsforums.com/archive...p/t-59501.html

"Although it is a nice experiment and so on, I think it is a bit "oversold". You can give it the interpretation given by the authors if you want to, but in fact, something much simpler is going on. When you look at figure 3, you get an interference pattern, because you SELECT A SUBSAMPLE from all impacts at D0, which are coincident with a hit at D1. When you look at ANOTHER SUBSAMPLE, namely the impacts at D0 which are coincident with D2 (figure 4) you get a shifted interference pattern. This shift comes about, if you trace it back, to a difference in optical pathlengths in the polarizing beamsplitter, and this shift is utterly important..."

Agreed. Which is why I asked earlier in the thread what happens if one adds together the D1 and D2 data, and is this important?

Vanesch’s observation, however, does not explain what is going on. He claims that “something much simpler is going on”, but does not go on to explain exactly what is going on, or how the signal photon at D0 “knows” exactly where to land (the D0 data is spatial) in advance of the idler photon arriving at either of D1, D2, D3 or D4.

The phase shift is due to differences in optical path lengths, yes, but that does not explain why there is an interference pattern in the first place. It also does not explain why we see no interference pattern at all in the D3, D4 data.

If the idler (after the signal has landed on D0) goes to either D3 or D4 then there is no interference pattern observed in the coincidence data; whereas if the idler (after the signal has landed on D0) goes to either D1 or D2 then there is an interference pattern observed in the coincidence data. The interference pattern is encoded within the spatial distribution of the D0 data – ie where the signal photon lands on D0.

Since the interference pattern depends on the spatial information encoded within D0, and “idler arriving at D1 or D2” entails that there be an interference pattern, whereas “idler arriving at D3 or D4” entails no interference pattern, and the signal photon arrives at D0 BEFORE the idler “chooses” which detector to go to, how does the signal photon know exactly where it is to land on D0?

As selfAdjoint has pointed out, the QM formalism tells us how to calculate the numbers in a self-consistent fashion, but it seems that is all that the QM formalism can do for us – it does not provide any insight into what is “really happening” in this experiment. As far as I can see, nobody has so far come up with a coherent interpretation of what is going on which explains the time-dependence of the process.

Best Regards

 

[cesiumfrog]

Does anyone know of any similar experiment, except where the experimenter can actively choose whether or not to erase, and where (if the information is erased) the interference fringes are immediately observable without examining correlations?

I guess I'm basically wondering if there exist conditions where it would seem as though the choice could be delayed until after observation of whether interference occured.

In the paper referenced above, even if the first beamsplitters are replaced with mirrors (to always erase the which-path information), the aggregate data from the scanning detector (D0) would still appear featureless (until time-correlated with either D1 or D2). Can the phase difference between R01 and R02 be removed with a different experimental layout?

 

[gptejms]

I started two threads 'Prob. density in spacetime' and 'Earlier or later' in 2004 which I think are relevant to the present discussion.I suggest that you all read these and let me have your comments.The links are:-

https://www.physicsforums.com/showthread.php?t=45264
https://www.physicsforums.com/showthread.php?t=45267

 

[moving finger]

I started two threads 'Prob. density in spacetime' and 'Earlier or later' in 2004 which I think are relevant to the present discussion.I suggest that you all read these and let me have your comments.The links are:-

https://www.physicsforums.com/showthread.php?t=45264
https://www.physicsforums.com/showthread.php?t=45267

sorry, both these threads seem unreadable - is anyone else having the same problem?

Best Regards

 

[Brinx]

Yup, the first part of those threads look messy, like they use some script that isn't supported by the site anymore.

But anyway, I was also struggling with the matter of the delayed-choice quantum eraser after having read http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm . In particular, if the setup described on that page would be changed by firing, say, a thousand photons from the laser of which the signal photons would all be detected before any of the idler photons arrived at the other detector. Suppose the 'far' detector used for detecting the idler photons can be any of two detectors, namely one which does detect which-path information and one which doesn't.

Further, suppose the person controlling that far-away detector position was given the choice to install one of those two detectors after the time the thousand signal photons were registered, but before the idler photons would arrive at his location. No FTL communication would be necessary to enable the person manning the far-away detector station to know when the signal photons are registered in this setup of course, this way of working could have been planned in advance. As the fate of the idler photons is unknown (whether they will have their which-path information extracted or not, in other words) at the time of detection of all signal photons, I'm not sure what pattern will be visible at the signal photon detection screen. Either all the idler photons will have their which-path information determined (in which case the signal photon detector screen would have to show no interference pattern but instead a classical lumped distribution), or they will all be detected without their which-path information being known, in which case the signal photons should show an interference pattern - if my reasoning is correct.

The problem with this situation is that the person observing the signal photon detector after all signal photons are registered would know in advance which type of detector the person manning the far-away detector is going to choose! This sounds quite bizarre to me as it implies sending information back in time, but I can't find out where the fault lies within this line of reasoning.

 

[cesiumfrog]

... if the setup described on that page would be changed by firing, say, a thousand photons from the laser of which the signal photons would all be detected before any of the idler photons arrived at the other detector.

I don't think that change would work, because (see the PRL paper for some detail) the pattern only exists in the part of the signal data coincident with D1 or D2 (singularly):

If you're just looking at D0, there is no interference pattern

Nonetheless, this seems almost like some mere artifact of the final beamsplitter. Perhaps I can propose a different variation (entanglement + Wheeler's delayed choice):

Let the thousand entangled photons be produced (by pumping through a double slit, as in this DCQE experiment). Let all the signal photons be (almost immediately) observed on a nearby screen (or the scanning D0). Let all the idler photons propogate some (much greater) distance, to where an observer is located. The observer either places a screen in the path of all the idler photons (in which case, "surely" an interference pattern *will* be seen in the raw data, at both individual ends of the apparatus), or alternatively, the observer instead places a telescope focussed on just one of the slits (in which case "surely" potential time correlation would expose the which-path information for each photon, so there should be no reason to expect interference in the signal data).

Can this signal data reveal what choice the idler observer is going to make?

 

[Brinx]

Cesiumfrog, your suggestion sounds the same as mine did to me - although your description of the actual detectors used is more accurate. The question you pose is the one I was also wondering about, in any case. :)

 

[gptejms]

sorry, both these threads seem unreadable - is anyone else having the same problem?

Yep,the threads seem unreadable--may be you could try reading s.p.r. if you are a member.

 

[moving finger]

But anyway, I was also struggling with the matter of the delayed-choice quantum eraser after having read http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm . In particular, if the setup described on that page would be changed by firing, say, a thousand photons from the laser of which the signal photons would all be detected before any of the idler photons arrived at the other detector. Suppose the 'far' detector used for detecting the idler photons can be any of two detectors, namely one which does detect which-path information and one which doesn't.

Which other one or two detectors are you talking about here? There are 4 in the paper... D1, D2, D3 and D4. It's not clear which ones you intend to keep and which ones throw away.

Best Regards

 

[moving finger]

I need some help please.

In the commentary paper http://www.bottomlayer.com/bottom/ki...scully-web.htm it says :

Without more, we would expect the pattern developing at detector D0 to be an interference pattern. QM predicts that without which-path information, photons arriving from either A or B should interfere and distribute themselves one-by-one according to the statistical distribution of interfering waves.

In other words, it actually says that in absence of which path information, the pattern of signal photons at D0 should be an interference pattern. But we can get that pattern simply by removing mirrors BSB and BSA (remove these mirrors, and all idlers go to D1 or D2 via BS, hence no which path information).

But the original paper at http://xxx.lanl.gov/PS_cache/quant-ph/pdf/9903/9903047.pdf also made it clear that the raw data at D0 does not contain any interference pattern, because the interference patterns from D1 and D2 are 180 degrees out of phase - thus we will only see the pattern when we deconvolve the D0 signal data by correlating it with the idler coincidence data. This 180 degree phase shift is not referred to at all in the commentary paper at http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm.

If we remove BSB and BSA, do we see interference at D0 even without coincidence correlations with D1 and D2? If not, why not?

Am I missing something here?

Best Regards

 

[Brinx]

Moving finger, I proposed to place only one detector in the path of the idler photons instead of the more complex setup used in the paper - namely either a detector which does not read which-path information (such as a lens with a screen behind it, similar to the detector for the signal photons in the paper setup), or a detector which does read which-path information (such as the telescope detector Cesiumfrog suggested).

I should make a sketch to clear things up. I'll do so later on.

 

[moving finger]

Moving finger, I proposed to place only one detector in the path of the idler photons instead of the more complex setup used in the paper - namely either a detector which does not read which-path information (such as a lens with a screen behind it, similar to the detector for the signal photons in the paper setup), or a detector which does read which-path information (such as the telescope detector Cesiumfrog suggested).

I should make a sketch to clear things up. I'll do so later on.

OK, my problem (see my last post above) is that the original paper points out that there is a 180 degree phase shift between the D1 and D2 interference data, which means (I believe) that the raw D0 data (ie the data you want to use) will show no interference at all - I don't understand this and need someone to explain what is going on!

Best Regards

 

[gptejms]

OK, my problem (see my last post above) is that the original paper points out that there is a 180 degree phase shift between the D1 and D2 interference data, which means (I believe) that the raw D0 data (ie the data you want to use) will show no interference at all - I don't understand this and need someone to explain what is going on!

Best Regards

I've just had a quick look at the paper.The paper discusses this point(see page 3 of the paper).The culprit seems to be the beam splitter BS-- reflected and transmitted waves have a phase difference.

 

[moving finger]

I've just had a quick look at the paper.The paper discusses this point(see page 3 of the paper).The culprit seems to be the beam splitter BS-- reflected and transmitted waves have a phase difference.

Yes, I know the original paper discusses this point :

http://xxx.lanl.gov/PS_cache/quant-ph/pdf/9903/9903047.pdf

But the explanation paper does not :

http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm

My questions are :

The explanation paper says that :

QM predicts that without which-path information, photons arriving from either A or B should interfere and distribute themselves one-by-one according to the statistical distribution of interfering waves.

But this is obviously not true, since the pattern at D0 is a combination of the two interference patterns which are 180 degrees phase-shifted, so the "raw data" hitting D0 (in absence of any coincidence correlations with the other detectors) shows NO interference pattern.

If the phase shift is due to the beamsplitter BS, what happens if we remove BS - does the raw data at D0 suddenly show an interference pattern?

If the beamsplitter at BS causes a 180 degree phase shift between transmitted & reflected photons, then 50% of all photons reaching D1 and 50% of all photons reaching D2 should be phase-shifted relative to the other 50% - because half of all photons reach each detector are transmitted and half are reflected! Thus the explanation for the phase shift between D1 and D2 data cannot be due to phase shift diferences between reflection and transmission.

Best Regards

 

[gptejms]

But this is obviously not true, since the pattern at D0 is a combination of the two interference patterns which are 180 degrees phase-shifted, so the "raw data" hitting D0 (in absence of any coincidence correlations with the other detectors) shows NO interference pattern.

The joint detections D0,D1 or D0,D2 show intereference patterns--that is the only claim in either of the papers.Besides,I don't see how the phase shift is 180 degrees--seems to be 90 degrees to me(one term contains cos,the other sin).

If the beamsplitter at BS causes a 180 degree phase shift between transmitted & reflected photons, then 50% of all photons reaching D1 and 50% of all photons reaching D2 should be phase-shifted relative to the other 50% - because half of all photons reach each detector are transmitted and half are reflected! Thus the explanation for the phase shift between D1 and D2 data cannot be due to phase shift diferences between reflection and transmission.

Best Regards

I don't really know how a beam splitter works,but the situation may not be the same on either side of the beam splitter.Imagine the following situation:-reflection coeff. R,transmission coeff. 1-R on one side; reflection coeff. -R,transmission coeff. 1+R on the other side.If you work with this, you get the signs as in the paper.

 

[gptejms]

But this is obviously not true, since the pattern at D0 is a combination of the two interference patterns which are 180 degrees phase-shifted, so the "raw data" hitting D0 (in absence of any coincidence correlations with the other detectors) shows NO interference pattern.

The joint detections D0,D1 or D0,D2 show intereference patterns--that is the only claim in either of the papers.Besides,I don't see how the phase shift is 180 degrees--seems to be 90 degrees to me(one term contains cos,the other sin).

If the beamsplitter at BS causes a 180 degree phase shift between transmitted & reflected photons, then 50% of all photons reaching D1 and 50% of all photons reaching D2 should be phase-shifted relative to the other 50% - because half of all photons reach each detector are transmitted and half are reflected! Thus the explanation for the phase shift between D1 and D2 data cannot be due to phase shift diferences between reflection and transmission.

Best Regards

I don't really know how a beam splitter works,but the situation may not be the same on both sides of the beam splitter.Imagine the following situation:-reflection coeff. R,transmission coeff. 1-R on one side; reflection coeff. -R,transmission coeff. 1+R on the other side.If you work with this, you get the signs as in the paper.

 

[moving finger]

The joint detections D0,D1 or D0,D2 show intereference patterns--that is the only claim in either of the papers.Besides,I don't see how the phase shift is 180 degrees--seems to be 90 degrees to me(one term contains cos,the other sin).

The original paper (page 3) clearly says Pi radians phase-shift - that's 180 degrees. A 180 degree phase shift would mean complete loss of fringes (destructive interference) when the data are superimposed - which explains why there is no structure in the D0 data alone. A 90 degree phase shift would not lead to complete loss of interference fringes when you superimpose the data.

I don't really know how a beam splitter works,but the situation may not be the same on both sides of the beam splitter.Imagine the following situation:-reflection coeff. R,transmission coeff. 1-R on one side; reflection coeff. -R,transmission coeff. 1+R on the other side.If you work with this, you get the signs as in the paper.

OK, thanks for that

Best Regards

 

[gptejms]

The original paper (page 3) clearly says Pi radians phase-shift - that's 180 degrees. A 180 degree phase shift would mean complete loss of fringes (destructive interference) when the data are superimposed - which explains why there is no structure in the D0 data alone. A 90 degree phase shift would not lead to complete loss of interference fringes when you superimpose the data.


OK, thanks for that

Best Regards

Right you are--the phase difference is 180 degrees because of the terms being cos^2 and sin^2(not cos &. sin),but D0 data still has a sinc^2.Anyway,the claim in the paper(s) is about D0,D1 or D0,D2--so what's the problem?

 

[moving finger]

so what's the problem?

you answered it - the BS (ie semi-transparent mirror) is spatially asymmetric, and this results in the fact that idler photons going to D1 are phase-shifted with respect to those going to D2 - which in turn explains why there is no structure in the raw D0 data.

Is there such a thing as a symmetric beamsplitter - one that does NOT result in a 180 degree phase shift? If there were, and we were to use it, this would imply there should be structure in the D0 data? But this would also open the door to paradoxes - so maybe a symmetric beamsplitter (one that does not cause a phase shift) is physically impossible.

Best Regards