DCQE - how does/can the pattern change?

[Cthugha]

Are you still suggesting your classical phase analysis solves any "mystery" in the DCQE? Because that's what it sound like to me.

I have given you Walborn's opinion on the topic. He uses phases to explain conditional interference patterns. I fully agree with him. By the way it is rather strange to talk about classical phases. When quantifying the em field you also get fields with phases. I already gave you a reference on that which you did not bother to read.

I don't know what you mean

Yes, that was my impression from the beginning of this topic.

Just read up on spatial coherence. It seems to me that you do not even know what spatial coherence means. Do you know how spatial coherence and the visibility of a double slit interference pattern are connected? Then you can easily generalize that to two-photon states. Walborn does all of that in his review paper. All you need to do is read it. So I challenge you again to point out Walborn's error in equation 96 of the paper I linked earlier where the conditional interference patterns are explained in terms of phase relationships.

Please read up on it and/or post some publications in support of your view on the topic or stop trolling.

 

[unusualname]

I have given you Walborn's opinion on the topic. He uses phases to explain conditional interference patterns. I fully agree with him. By the way it is rather strange to talk about classical phases. When quantifying the em field you also get fields with phases. I already gave you a reference on that which you did not bother to read.



Yes, that was my impression from the beginning of this topic.

Just read up on spatial coherence. It seems to me that you do not even know what spatial coherence means. Do you know how spatial coherence and the visibility of a double slit interference pattern are connected? Then you can easily generalize that to two-photon states. Walborn does all of that in his review paper. All you need to do is read it. So I challenge you again to point out Walborn's error in equation 96 of the paper I linked earlier where the conditional interference patterns are explained in terms of phase relationships.

Please read up on it and/or post some publications in support of your view on the topic or stop trolling.

No, you please explain why coincidence counters are used. This is a much more simpler point of the experiment and one that has a simple answer, and one that you have obfuscated by appealing to weird results and papers from all sorts areas.

Once I understand your explanation of the coincidence counters I will be able to continue, otherwise this is going to go the same way as the other discussions.

 

[Cthugha]

No, you please explain why coincidence counters are used. This is a much more simpler point of the experiment and one that has a simple answer, and one that you have obfuscated by appealing to weird results and papers from all sorts areas.

Once I understand your explanation of the coincidence counters I will be able to continue, otherwise this is going to go the same way as the other discussions.

So I should explain it again? Why? It is all in Walborn's overview article. Results are not weird because you dislike them.

However, there are several answers to this question.

You use coincidence counters because you want to identify a two-photon interference pattern. This means that it is present only in the two-photon coincidence count rate, but not in the single photon count rate. Those two are complementary. In detail, you will only be able to see an interference pattern with perfect visibility if you have a momentum eigenstate which is equivalent to having no which-way information and also equivalent to having a high degree of spatial coherence. The most commonly used way of destroying which-way information lies in using a lens and placing the detector in the Fourier plane which means that a detection event at some position of the detector could come from any position of the crystal, but corresponds to some well defined momentum/wave vector. If you move this detector around, you get a different wave vector. So in order to single out a momentum eigenstate (which is the same as having high spatial coherence) you need a detector which does not spread across the whole Fourier plane, but is small enough to pick a momentum eigenstate.

If you do so, it is clear that all detection events at this position correspond to some momentum eigenstate. All detections on the other side (showing up in the coincidence counts) will also belong to some momentum eigenstate (high spatial coherence). As the visibility of a double slit interference pattern is proportional to the spatial coherence of the light beam used, this will give a conditional interference pattern. If you move the first detector out of the Fourier plane, you will notice the visibility of the interference pattern go down as you do not choose a momentum eigenstate subset anymore. Spatial coherence goes down and so does the visibility of the interference pattern. If you move the first detector around, you will notice that the interference pattern moves around as it is now the conditional interference pattern belonging to a different wave vector. To perform DCQE you can now play tricks and insert polarizers wave plates and whatever, but essentially this does not make the experiment more mysterious than entanglement already is.

Originally Posted by unusualname
The delay is the delay after the s-photons are measured/detected.

Why should a polariser placed in another galaxy affect the s-photon detections, there is a delay of several years before the p-photons will even reach the eraser?

(yes you will have to wait years to do the coincidence match, but there will be an interference pattern if the eraser was in place and there won't be if it wasn't in place, how did the s-photon's "know" that years before)

i am trying to grasp this. maybe cthuga/spectracat can explain how sub-samples

so is the mach zender interference/non-interference also explained by phase?

Hmm, how can I explain this more pedagogically. The polariser placed in a different galaxy does not modify the s-photon detections. There is no retrocausation or such stuff. Have a look at section 6.1. of the arxiv paper from Walborn I linked earlier. That explains the basics of subsampling way better (and with pictures) than I could do by just typing text. Once you understand that it is not a big step to understanding DCQE.

 

[SpectraCat]

i am trying to grasp this. maybe cthuga/spectracat can explain how sub-samples

so is the mach zender interference/non-interference also explained by phase?

It is because the polarizer placed in the p-beam doesn't actually affect the s-photon detection events. It only affects the *coincidence measurements* which are not even generated until after *BOTH* photons have been detected. If you look only at the s-photon detections for the both cases, without considering the coincident statistics, then there is absolutely no difference for sets taken with and without the polarizer in the p-branch. In other words, there is NO interference observed in the detections for the s-photons in any case. The interference fringes are only evident in the coincident measurements.

Notice also, that for the case where the eraser (i.e. the polarizer in the p-beam) is in place, there are two different interference patterns that are observed, depending on whether the polarizer angle is set to match the quarter-wave plate for slit one or for slit two. The two patterns of fringes and anti-fringes (to use the terminology from Walborn's paper) are 180º out of phase ... this is because the of the well-defined phase relationship between the two-photon states, as explained by Cthugha.

 

[unusualname]

So I should explain it again? Why? It is all in Walborn's overview article. Results are not weird because you dislike them.

However, there are several answers to this question.

You use coincidence counters because you want to identify a two-photon interference pattern. This means that it is present only in the two-photon coincidence count rate, but not in the single photon count rate.

So what you mean is the pattern in the coincidence counts shows interference.

Those two are complementary. In detail, you will only be able to see an interference pattern with perfect visibility if you have a momentum eigenstate which is equivalent to having no which-way information and also equivalent to having a high degree of spatial coherence. The most commonly used way of destroying which-way information lies in using a lens and placing the detector in the Fourier plane which means that a detection event at some position of the detector could come from any position of the crystal, but corresponds to some well defined momentum/wave vector. If you move this detector around, you get a different wave vector. So in order to single out a momentum eigenstate (which is the same as having high spatial coherence) you need a detector which does not spread across the whole Fourier plane, but is small enough to pick a momentum eigenstate.

Well I don't think you're correct there, in the Walborn experiment they adjust the p-photon arm by a couple of meters, no worrying about planes there. In recent experiments they have done this stuff across Canary Islands, where I think it would be difficult to accurately find the "Fourier Plane". And I believe fibre optics are/will be used which makes the idea of your planes not really relevant.

If you do so, it is clear that all detection events at this position correspond to some momentum eigenstate. All detections on the other side (showing up in the coincidence counts) will also belong to some momentum eigenstate (high spatial coherence). As the visibility of a double slit interference pattern is proportional to the spatial coherence of the light beam used, this will give a conditional interference pattern. If you move the first detector out of the Fourier plane, you will notice the visibility of the interference pattern go down as you do not choose a momentum eigenstate subset anymore. Spatial coherence goes down and so does the visibility of the interference pattern. If you move the first detector around, you will notice that the interference pattern moves around as it is now the conditional interference pattern belonging to a different wave vector. To perform DCQE you can now play tricks and insert polarizers wave plates and whatever, but essentially this does not make the experiment more mysterious than entanglement already is.

Yeah, I have no doubt the pattern moves around, but we're investigating delayed eraser so all we really want is no pattern/some pattern as we remove/put in place the eraser.



The correct answer to the question "Why are coincidence counters used?" is that QM is probabilistic. Even if you could remove all background effects and have an efficient entangled pair source you still have the fact that ~50% of the p-photons will pass through the eraser probabilistically. There is no deterministic way round it, not by phase matching or other weird calculation, otherwise FTL signalling would be possible since you wouldn't need a coincidence match to determine if the eraser was in place or not.

Hmm, how can I explain this more pedagogically. The polariser placed in a different galaxy does not modify the s-photon detections. There is no retrocausation or such stuff. Have a look at section 6.1. of the arxiv paper from Walborn I linked earlier. That explains the basics of subsampling way better (and with pictures) than I could do by just typing text. Once you understand that it is not a big step to understanding DCQE.

No I already understand that QM is either non-local and/or non-separable so I have no problem interpreting the experiment.

You seem to have found a different interpretation that doesn't require non-locality and/or non-separability. You should try to publish this discovery, really.

 

[unusualname]

It is because the polarizer placed in the p-beam doesn't actually affect the s-photon detection events. It only affects the *coincidence measurements* which are not even generated until after *BOTH* photons have been detected. If you look only at the s-photon detections for the both cases, without considering the coincident statistics, then there is absolutely no difference for sets taken with and without the polarizer in the p-branch. In other words, there is NO interference observed in the detections for the s-photons in any case. The interference fringes are only evident in the coincident measurements.

Notice also, that for the case where the eraser (i.e. the polarizer in the p-beam) is in place, there are two different interference patterns that are observed, depending on whether the polarizer angle is set to match the quarter-wave plate for slit one or for slit two. The two patterns of fringes and anti-fringes (to use the terminology from Walborn's paper) are 180º out of phase ... this is because the of the well-defined phase relationship between the two-photon states, as explained by Cthugha.

The problem is that Cthugha's analysis uses classical phases, so it would be unlikely to apply across galaxies, or even the Canary islands with accuracy.

Do you not agree that QM is non-local and/or non-separable?

 

[SpectraCat]

Quo

The problem is that Cthugha's analysis uses classical phases, so it would be unlikely to apply across galaxies, or even the Canary islands with accuracy.

No it doesn't .. I have explained this many times, as has Cthugha, yet you persist to claim that it is true without any support for your position. By the way, what do you mean by "classical phase"? Phase is phase .. it has the same interpretation in both classical and quantum mechanics as far as I can tell.

Do you not agree that QM is non-local and/or non-separable?

Of course ... where did you get the idea that I wouldn't agree with that?

 

[unusualname]

No it doesn't .. I have explained this many times, as has Cthugha, yet you persist to claim that it is true without any support for your position. By the way, what do you mean by "classical phase"? Phase is phase .. it has the same interpretation in both classical and quantum mechanics as far as I can tell.

er, you're being funny right?. No, in QM a phase is assigned to a complex probability amplitude that evolves according to the Schrödinger eqn., in classical EM it is assigned to a wave described my Maxwell's equations. In measurements the intensity predicted by Maxwell's eqns matches the probability predicted by quantum (field) theory, but this is misleading, the Maxwell EM wave is not an ontological wave traveling through space with a well defined phase at all times (so that you might think you can naively interpret phase diagrams for single photons)

Of course ... where did you get the idea that I wouldn't agree with that?

The fact that you think DCQE can be explained by a classical phase argument.

 

[San K]

So I should explain it again? Why? It is all in Walborn's overview article. Results are not weird because you dislike them.


Hmm, how can I explain this more pedagogically. The polariser placed in a different galaxy does not modify the s-photon detections. There is no retrocausation or such stuff. Have a look at section 6.1. of the arxiv paper from Walborn I linked earlier. That explains the basics of subsampling way better (and with pictures) than I could do by just typing text. Once you understand that it is not a big step to understanding DCQE.

I have not been able to find the relevant/reference Walborn paper. Can you please paste the link again?


is it this one? http://arxiv.org/abs/quant-ph/0106078

navigating forums is a pain because only 10-15 posts show up per page (click) instead of say 200

edit: it must be this one ----> http://arxiv.org/abs/1010.1236

 

[San K]

It is because the polarizer placed in the p-beam doesn't actually affect the s-photon detection events. It only affects the *coincidence measurements* which are not even generated until after *BOTH* photons have been detected. If you look only at .

what do you mean by "generated" above? did you mean filtered?

because the s-photon position has been generated, the s-quantum has been registered.

the only thing that is to be done is filtering out from the noise (via coincidence counter) to get the proper sub-sample.

 

[San K]

Cthugha's analysis is clearly correct at least about one thing.
Postselection by coincidence counter has a key role in appearance of interference pattern.

That can be easily seen if you replace polarizer in idler beam with polarization beam splitter. Then you will have fringe and antifringe pattern at the same time just by looking at coincidences between signaling detector and one of the two detector at different outputs of PBS.

Ah...this explains it well...good one

 

[unusualname]

Cthugha's analysis is clearly correct at least about one thing.
Postselection by coincidence counter has a key role in appearance of interference pattern.

That can be easily seen if you replace polarizer in idler beam with polarization beam splitter. Then you will have fringe and antifringe pattern at the same time just by looking at coincidences between signaling detector and one of the two detector at different outputs of PBS.

Ah...this explains it well...good one

Yes he's correct in that one thing.

 

[SpectraCat]

what do you mean by "generated" above? did you mean filtered?

because the s-photon position has been generated, the s-quantum has been registered.

the only thing that is to be done is filtering out from the noise (via coincidence counter) to get the proper sub-sample.

No, the coincidence counter is not a simple noise filter ... it is required for proper identification of the photons that were generated as entangled pairs. Since the photon's travel different distances through the apparatus, they arrive at their respective detectors at different times. To keep it simple, consider a case where the A photon travels 2.9979 m to its detector, and the B photon travels twice as far to its detector. In that case there will be a delay of 1.0000 ns between emission and detection of the A photon, and a delay of 2.0000 ns for the B photon. The coincidence counter compensates for these delays so that the proper pair of photon detection events (separated by 1.0000 ns in this case) are paired in the analysis.

So, in order to see the interference pattern in these experiments, you have to wait until both the s-detector and p-detector events for each entangled pair have been registered before they can be properly compared via the coincidence counter. If the polarizer is in place (and set to the appropriate angle), then you will see the interference pattern revealed as an oscillation in the coincidence counts as a function of the s-detector position.

 

[San K]

No, the coincidence counter is not a simple noise filter ... it is required for proper identification of the photons that were generated as entangled pairs. Since the photon's travel different distances through the apparatus, they arrive at their respective detectors at different times. To keep it simple, consider a case where the A photon travels 2.9979 m to its detector, and the B photon travels twice as far to its detector. In that case there will be a delay of 1.0000 ns between emission and detection of the A photon, and a delay of 2.0000 ns for the B photon. The coincidence counter compensates for these delays so that the proper pair of photon detection events (separated by 1.0000 ns in this case) are paired in the analysis.

So, in order to see the interference pattern in these experiments, you have to wait until both the s-detector and p-detector events for each entangled pair have been registered before they can be properly compared via the coincidence counter. If the polarizer is in place (and set to the appropriate angle), then you will see the interference pattern revealed as an oscillation in the coincidence counts as a function of the s-detector position.

ok... so instead of "generated" we can say "identified/paired" (by the coincidence counter)

because when you said generated, one could assume...as "created"...it can give the sense that position is generated/created...

once s-quantum is registered at Ds, its position is locked, it won't change, right?

thus s-photon/position is not generated but identified (via pairing through coincidence counter)

 

[SpectraCat]

er, you're being funny right?. No, in QM a phase is assigned to a complex probability amplitude that evolves according to the Schrödinger eqn., in classical EM it is assigned to a wave described my Maxwell's equations.

Yeah, that's what I thought .. you don't understand what you are talking about. Phase is just an expression of the relative position of a wave in its cycle ... that's why it is expressed as an angle. So any system with wave-like properties will have a phase, regardless of whether it is quantum or classical. Coherence in any system can be expressed as the persistence of a well-defined phase relationship in time and/or space. So, the spatial coherence that is observed in the double-slit experiment (of which the DCQE is just an extension), occurs because the photon (or massive particle) is interfering with itself, and thus the different paths through the apparatus always have a well-defined phase relationship. The DCQE is more complicated, because it involves phase relationships of the two-photon entangled state as well, but that can still be accounted for, as Cthugha showed in his analysis.

Also, since photons are intrinsically quantum mechanical objects, it seems silly to talk about the phase relationships that Cthugha is presenting as classical. The polarization of photons is ALSO quantum mechanical, however it is analogous to the Jones vector for the classical description of EM radiation.

In measurements the intensity predicted by Maxwell's eqns matches the probability predicted by quantum (field) theory, but this is misleading, the Maxwell EM wave is not an ontological wave traveling through space with a well defined phase at all times (so that you might think you can naively interpret phase diagrams for single photons)

I have no idea what you tried to express above .. what is a "phase diagram" for a single photon?

The fact that you think DCQE can be explained by a classical phase argument.

Nope, I sure don't ... for the umpteenth time, if Cthugha's argument were classical, then a) we would not be talking about photons, and b) there could never be a well-defined phase relationship between the entangled photons, because there is no way of describing that classically.

 

[SpectraCat]

ok... so instead of "generated" we can say "identified/paired" (by the coincidence counter)

because when you said generated, one could assume...as "created"...it can give the sense that position is generated/created...

once s-quantum is registered at Ds, its position is locked, it won't change, right?

Yes, of course.

thus s-photon/position is not generated but identified (via pairing through coincidence counter)

I never said the s-photon position was generated .. I said that the interference pattern is only evident in the coincidence counts, which cannot be generated until detection events for both photons have been registered.

 

[unusualname]

Yeah, that's what I thought .. you don't understand what you are talking about. Phase is just an expression of the relative position of a wave in its cycle ... that's why it is expressed as an angle. So any system with wave-like properties will have a phase, regardless of whether it is quantum or classical. Coherence in any system can be expressed as the persistence of a well-defined phase relationship in time and/or space. So, the spatial coherence that is observed in the double-slit experiment (of which the DCQE is just an extension), occurs because the photon (or massive particle) is interfering with itself, and thus the different paths through the apparatus always have a well-defined phase relationship. The DCQE is more complicated, because it involves phase relationships of the two-photon entangled state as well, but that can still be accounted for, as Cthugha showed in his analysis.

haha, what more can I say, maybe you can use a similar analysis for GHZ states.

Also, since photons are intrinsically quantum mechanical objects, it seems silly to talk about the phase relationships that Cthugha is presenting as classical. The polarization of photons is ALSO quantum mechanical, however it is analogous to the Jones vector for the classical description of EM radiation.

yes of course it is, you guys are right there is no mystery in the DCQE, what was I thinking?

I have no idea what you tried to express above .. what is a "phase diagram" for a single photon?

Nope, I sure don't ... for the umpteenth time, if Cthugha's argument were classical, then a) we would not be talking about photons, and b) there could never be a well-defined phase relationship between the entangled photons, because there is no way of describing that classically.

No there isn't, well done.

 

[Cthugha]

edit: it must be this one ----> http://arxiv.org/abs/1010.1236

Yes, I mean this one. Sorry, I thought it was easy to find as I mentioned it only a few posts earlier. Besides section 6, also section 4.2 about ghost interference might help you understand the details better.

Well I don't think you're correct there, in the Walborn experiment they adjust the p-photon arm by a couple of meters, no worrying about planes there. In recent experiments they have done this stuff across Canary Islands, where I think it would be difficult to accurately find the "Fourier Plane". And I believe fibre optics are/will be used which makes the idea of your planes not really relevant.

As I said before you need to create a momentum eigenstate and using the Fourier plane is one of the possibilities. You can also go to far field conditions like in the double slit quantum eraser experiment. The result is the same. You decrease the angular size of the source, get a better defined momentum of your subset and increase spatial coherence. You just need the Fourier plane if you don not wan t to work at far field conditions.

Yeah, I have no doubt the pattern moves around, but we're investigating delayed eraser so all we really want is no pattern/some pattern as we remove/put in place the eraser.

Fine, then you agree with my explanation because if you do not use coincidence counting you will just measure a superposition of all these moced patterns which gives no pattern at all. From this point on the inserting/removing the eraser thing is trivial.

The correct answer to the question "Why are coincidence counters used?" is that QM is probabilistic. Even if you could remove all background effects and have an efficient entangled pair source you still have the fact that ~50% of the p-photons will pass through the eraser probabilistically. There is no deterministic way round it, not by phase matching or other weird calculation, otherwise FTL signalling would be possible since you wouldn't need a coincidence match to determine if the eraser was in place or not.

This is easily refuted because coincidence counting is also needed in experiments without DCQE or polarizers which just rely on basic conditional interference patterns or ghost imaging. The simple fact that relative phases leading to two-photon interference require coincidence counting, is all there is to it. This is not specific to DCQE. Again this is a result explained around equation 96 in Walborn's paper. You still did not tell me where he is wrong. The thing that prevents FTL signaling is the single-photon phase which is not well defined on average and therefore prevents information transfer. It is generally accepted that two-photon interference and single photon-interference are complementary due to the reasons I gave and information contained in coincidence counts therefore cannot be carried by one of these photons alone. I already gave you the references. Do you have any supporting your claim?

No I already understand that QM is either non-local and/or non-separable so I have no problem interpreting the experiment.

You seem to have found a different interpretation that doesn't require non-locality and/or non-separability. You should try to publish this discovery, really.

You do not seem to grasp where non-locality or non-seperability stems from. It comes into play because of the second and the first photon ending up in "compatible" states according to the rules of entanglement even if the second detection is not in the light cone of the first one and that is all there is to it. Most DCQE experiments indeed are not a proof of nonlocality as no Bell tests are performed. The first experiments on DCQE really showing that came in 2004 according to Walborns review paper.

No, in QM a phase is assigned to a complex probability amplitude that evolves according to the Schrödinger eqn., in classical EM it is assigned to a wave described my Maxwell's equations. In measurements the intensity predicted by Maxwell's eqns matches the probability predicted by quantum (field) theory, but this is misleading, the Maxwell EM wave is not an ontological wave traveling through space with a well defined phase at all times (so that you might think you can naively interpret phase diagrams for single photons)

You are wrong here. The phase is (somewhat) well defined in each run of the experiment. It is, however, usually not defined well on average, that means the phase will be different in every run of the experiment. Besides that you can map two-photon Fourier optics to classical Fourier optics using Klyshko's picture as mentioned in the Scarcelli et al. paper I already cited.

 

[unusualname]

OK, Cthuga I'm out, otherwise I may incur further warnings and infractions.

I would suggest you guys put together your explanations of the DCQE (since they are not widely known) in a paper, even just on arXiv so at least it may be argued on a more professional level.

 

[Drakkith]

The correct answer to the question "Why are coincidence counters used?" is that QM is probabilistic. Even if you could remove all background effects and have an efficient entangled pair source you still have the fact that ~50% of the p-photons will pass through the eraser probabilistically. There is no deterministic way round it, not by phase matching or other weird calculation, otherwise FTL signalling would be possible since you wouldn't need a coincidence match to determine if the eraser was in place or not.

I don't see what the 50% has anything to do with a counter. Take the eraser away and you would still need the counter to determine which photons were entangled, correct? It looks like its only job is to count when events are detected at the detectors. Nothing else. Take the counter away, run the experiment, and take the times from each detector for the detection events and compare them to match up the events. You just became a coincidence counter. Correct?

Also, can someone tell me what this purpose of these DCQE's are? Are they to determine if the photons can communicate with each other once one interacts and has to go to a set quantum state after the other one has already been detected?

 

[Cthugha]

Take the counter away, run the experiment, and take the times from each detector for the detection events and compare them to match up the events. You just became a coincidence counter. Correct?

Yes, modern coincidence counters really allow to do so and also offer a mode which gives only timestamps of detections from each photo diode and allows to do the analysis by hand. However, this mode is rarely used as the count rates from each diode are usually on the order of 10^6 per second at least and recording timestamps for every detection piles up huge amounts of data which need to be written to a hard disk quickly.

Also, can someone tell me what this purpose of these DCQE's are? Are they to determine if the photons can communicate with each other once one interacts and has to go to a set quantum state after the other one has already been detected?

Well, from a historical point of view quantum erasers and its delayed choice version were introduced by Scully in 1982. Back then this experiment was aimed at answering the question whether uncertainty or complementarity is more fundamental. So they were aiming at showing that it is not the uncontrolled disturbance introduced by a position measurement in the common double slit experiment that causes the interference pattern to disappear in the double slit experiment and thought of a reversible way to mark the way. The delayed choice version of the quantum eraser was introduced in the same paper - mainly because Aspect's idea of delayed choice experiments in general were intensely debated at that time.

 

[San K]

So they were aiming at showing that it is not the uncontrolled disturbance introduced by a position measurement in the common double slit experiment that causes the interference pattern to disappear in the double slit experiment and thought of a reversible way to mark the way.

it is not the uncontrolled disturbance but rather the way (rules/methods) the sub-samples of photons are chosen/identified ? ...that causes interference pattern to appear/disappear

 

[zonde]

yeah, you are probably right, it would be interesting to see an experiment done this way rather than using a coincidence counter, afaik there is no such published experiment.

There is. Weihs et al experiment http://arxiv.org/abs/quant-ph/9810080"

"Each observer station featured a PC which stored the tables of time tags accumulated in an individual measurement. Long after measurements were finished we analyzed the files for coincidences with a third computer."

 

[Drakkith]

Well, from a historical point of view quantum erasers and its delayed choice version were introduced by Scully in 1982. Back then this experiment was aimed at answering the question whether uncertainty or complementarity is more fundamental. So they were aiming at showing that it is not the uncontrolled disturbance introduced by a position measurement in the common double slit experiment that causes the interference pattern to disappear in the double slit experiment and thought of a reversible way to mark the way. The delayed choice version of the quantum eraser was introduced in the same paper - mainly because Aspect's idea of delayed choice experiments in general were intensely debated at that time.

What did they find out?

 

[unusualname]

yeah, you [SpectraCat] are probably right, it would be interesting to see an experiment done this way rather than using a coincidence counter, afaik there is no such published experiment.

There is. Weihs et al experiment http://arxiv.org/abs/quant-ph/9810080"

"Each observer station featured a PC which stored the tables of time tags accumulated in an individual measurement. Long after measurements were finished we analyzed the files for coincidences with a third computer."

Thanks zonde, I assume this should be easier now than in 1998, but the lengths the experimenters go to close "loopholes" is impressive.

 

[zonde]

There is something puzzling for me about Walborn Quantum eraser experiment.

Without quarter wave plates coincidence postselection increases spatial coherence. Otherwise interference is just a single beam interference.
At least that's how it seems from Walborn explanation (http://arxiv.org/abs/1010.1236" 4.1)

But in that case placing polarizer in idler beam when there are no quarter wave plates before slits should change nothing. Interference pattern should still be there.

Does it seems right?

 

[SpectraCat]

There is something puzzling for me about Walborn Quantum eraser experiment.

Without quarter wave plates coincidence postselection increases spatial coherence. Otherwise interference is just a single beam interference.
At least that's how it seems from Walborn explanation (http://arxiv.org/abs/1010.1236" 4.1)

But in that case placing polarizer in idler beam when there are no quarter wave plates before slits should change nothing. Interference pattern should still be there.

Does it seems right?

Yes, that is correct. Note also that with the QWP's in place, the angles of the QWP's seem to define a preferred basis for the polarizer to recover the interference patterns. The authors only report results for the polarizer in the QWP basis (45 and 135 degrees), but it would be interesting to see how changing the angle of the polarizer changes the experimental results. From an (admittedly casual) analysis of the mathematical treatment earlier in the paper, it seems like moving the polarizer angle away from 45 towards 90 would cause the interference fringes to gradually lose intensity, and eventually disappear at 90 (or 0) degrees. The reason I find this somewhat striking (assuming it is correct) is that it seems to work in the opposite sense of other experiments where a particular polarization basis is associated with "which path" information, in that interference is not observed in those cases until the polarizer angle is chosen to mix the basis states.

 

[zonde]

Yes, that is correct. Note also that with the QWP's in place, the angles of the QWP's seem to define a preferred basis for the polarizer to recover the interference patterns. The authors only report results for the polarizer in the QWP basis (45 and 135 degrees), but it would be interesting to see how changing the angle of the polarizer changes the experimental results. From an (admittedly casual) analysis of the mathematical treatment earlier in the paper, it seems like moving the polarizer angle away from 45 towards 90 would cause the interference fringes to gradually lose intensity, and eventually disappear at 90 (or 0) degrees. The reason I find this somewhat striking (assuming it is correct) is that it seems to work in the opposite sense of other experiments where a particular polarization basis is associated with "which path" information, in that interference is not observed in those cases until the polarizer angle is chosen to mix the basis states.

Well, I am quite sure that source determines preferred basis. Source produces H and V modes in certain basis. QWP mixes H and V modes so that interference (at detector) happens between them. And polarizer at +45° or -45° mixes H and V modes in similar way as two QWPs.
When you rotate polarizer you get more of one mode and less of the other. Finally at 0° and 90° you get pure H or V mode (so no interference between modes after polarizer).

That's how I see it.

 

[SpectraCat]

Well, I am quite sure that source determines preferred basis. Source produces H and V modes in certain basis. QWP mixes H and V modes so that interference (at detector) happens between them. And polarizer at +45° or -45° mixes H and V modes in similar way as two QWPs.
When you rotate polarizer you get more of one mode and less of the other. Finally at 0° and 90° you get pure H or V mode (so no interference between modes after polarizer).

That's how I see it.

No .. the source is entangled, so there is no preferred basis for detection until the QWP's are added. This is the fundamental results showed by the Aspect experiments, and since confirmed many times.

 

[San K]

There is something puzzling for me about Walborn Quantum eraser experiment.

Without quarter wave plates coincidence postselection increases spatial coherence. Otherwise interference is just a single beam interference.
At least that's how it seems from Walborn explanation (http://arxiv.org/abs/1010.1236" 4.1)

But in that case placing polarizer in idler beam when there are no quarter wave plates before slits should change nothing. Interference pattern should still be there.

Does it seems right?

Well then in this case you are getting which-way (via polarizer in idler) and also getting interference pattern...