Interference seen in a member of an entangled pair

[msumm21]

Say we have two electrons in a spin entangled state about the z-axis |01> + |10>. One electron travels off to our left, the other to our right. The right electron passes through an inhomogeneous magnetic field with gradient solely in the z direction (Stern-Gerlach type), and subsequently passes through an appropriately placed plate with 2 slits, and then to a screen where its position can be measured. Basically a Stern-Gerlach magnet followed by an appropriately placed double slit experiment.

If we do NOT measure the spin of either electron then isn't it true that, after repeating this experiment several times, the right electron will show an interference pattern on the screen? If we DO measure the spin of the left electron then isn't it true that we will not see the interference pattern (since we know the spin and therefore deflection and which-slit info for the right electron)?

Evidentially there is something wrong with the argument above because, if it were true, then it would be a conceptually trivial way to communicate information faster than c -- someone at a (theoretically) arbitrary distance away can decide to measure the spin of the left electron or not, and another person at the screen after the double slit can see what the other person did (essentially a binary signal).

Question:
What is wrong with the conclusions I'm making about the above experiment?

PS:
I asked this question on this or another form a long time ago, but I'm asking again because the only answers I received were things like "it's impossible to transfer info faster than c." What I am wanting to know is exactly where/why the above experiment breaks down.

 

[colorSpace]

As I remember from reading Anton Zeilingers books a while ago, performing an experiment like this requires an experimental setup in which interference will become visible only by analysis of the data. For example, one needs a relatively large source of light, which causes any interference patterns to overlap such that the data at first appears random, and in fact it is random. The interference is then only discoverable in the correlation to the other particle. Whether such a correlation will exist depends on the measurements being performed, but in either case any particle which is allowed to show interference will only display random behavior.

In order to allow both particles of an entangled pair to show interference, they need to be created in such a way that both particles will show random behavior when examined individually. (At least that's how I understood it).

 

[LaserMind]

If we do NOT measure the spin of either electron then isn't it true that, after repeating this experiment several times, the right electron will show an interference pattern on the screen? If we DO measure the spin of the left electron then isn't it true that we will not see the interference pattern (since we know the spin and therefore deflection and which-slit info for the right electron)?

Evidentially there is something wrong with the argument above because, if it were true, then it would be a conceptually trivial way to communicate information faster than c -- someone at a (theoretically) arbitrary distance away can decide to measure the spin of the left electron or not, and another person at the screen after the double slit can see what the other person did (essentially a binary signal).

Question:
What is wrong with the conclusions I'm making about the above experiment?

PS:
I asked this question on this or another form a long time ago, but I'm asking again because the only answers I received were things like "it's impossible to transfer info faster than c." What I am wanting to know is exactly where/why the above experiment breaks down.

Yes, its a common misconception. I see it in some results of the Quantum Erasure experiment too. Implying FTL as you suggest.
My take is this: When the spin of one of the entangled particles is measured we can say that the other had the opposite spin at that exact time - if we had been clever enough to measure it (at that exact time, remember). But subsequently, its back to a probabilty of either spin because its in superposition still i.e. the wave function has both spins. Its not as if measurement of one 'particle' converts the other particle to one of the opposite spin. It would indeed be useful if that were the case (we could then have FTL).

We would only be certain that the other particle had the opposite spin at one instant only and not know after that time exactly what the spin were - it would be a probability again as given by the amplitude function etc.

 

[Doc Al]

Evidentially there is something wrong with the argument above because, if it were true, then it would be a conceptually trivial way to communicate information faster than c -- someone at a (theoretically) arbitrary distance away can decide to measure the spin of the left electron or not, and another person at the screen after the double slit can see what the other person did (essentially a binary signal).

Question:
What is wrong with the conclusions I'm making about the above experiment?

PS:
I asked this question on this or another form a long time ago, but I'm asking again because the only answers I received were things like "it's impossible to transfer info faster than c." What I am wanting to know is exactly where/why the above experiment breaks down.

Imagine that all of the left members of the entangled pairs go to Lab A and the right ones go to Lab B. Nothing done in Lab A can have any effect on the results of measurements made in Lab B (assuming the Labs do not communicate). Lab B experimenters would have no way of knowing whether Lab A even made any measurements, much less what the results were--for all they know Lab A could have been destroyed before any measurements could be made.

Of course, if you use the results of measurements made in Lab A to filter out a subset of Lab B results, interesting effects can be achieved. But that requires ordinary non-FTL communication between the Labs and cannot be used to send FTL signals.

 

[msumm21]

LaserMind: I believe that, once a particle (or its entangled counterpart) is measured in a particular spin state, it remains in that state until something is done to its spin, like a measurement of spin about x, ... For example, Feynman lectures says that if you pass electrons through a z-oriented Stern-Gerlach (SG) and send the spin up electrons through a second z-oriented SG then all the electrons passing through the second SG will be measured spin up (they retained their spin up property). However, if you pass the spin up electrons through a x-oriented SG before passing them through the second z-oriented SG, then you will see only half of the original spin up electrons measured spin up in the second z-oriented SG.

Doc Al: So you are saying that, whether or not we measure the spin of the particles, it has no effect on the measurements of the right particle hitting the screen. What WILL we see when the right particles hit the screen? I guess it will not be an interference pattern because it seems like we will end up with contradictions if we think that through, for example:
Every 10 seconds we send the entangled pairs out to labs A and B.
Lab A measures and records the Z component of the spin of each (left) electron:
Electron@T=0: up
Electron@T=10: down
...
Electron@T=100: up
Lab B measures and records the location at which each (right) electron hits the screen:
Electron@T=0: position21
Electron@T=10: position15
...
Electron@T=100: position34
Afterwards, Alice in lab A calls Bob in lab B and tells him the spin of each electron. Now Bob saw an interference pattern, but he also knows which slit each electron passed through, a contradiction, right?

Main question: What will Bob see and why? If the answer is that Bob sees no interference pattern, and that the electrons go straight through just as if Bob observed the which-way info, then my next question is how those results remain the same when Alice closes lab A and no longer measures the spin (what causes the interference pattern to disappear when we haven't measured the which-way info)?

Thanks!

 

[DrChinese]

Afterwards, Alice in lab A calls Bob in lab B and tells him the spin of each electron. Now Bob saw an interference pattern, but he also knows which slit each electron passed through, a contradiction, right?

Well, we know there is actually no contradiction and no FTL signaling possible. So let's take a close look.

1. The tacit assumptions present in your example are:

a) Alice and Bob would exhibit "perfectly" symmetric correlations of their path through the slits (i.e. which slit), assuming that you perform exactly the same experiment for both.

b) Entangled particles individually exhibit self-interference when going through a double slit, no differently than an un-entangled particle.

2. So what about these assumptions? a) is true but b) is not. When you check ANY observable on Alice you will ALWAYS see a correlated value for Bob after an identical test. The test on Bob is essentially redundant (since no new information was obtained). This was the starting point for the EPR argument.

But Alice and Bob do NOT exhibit self-interference as do particles that are not entangled! As a reference on this point, see Zeilinger, p.290, figure 2.

3. Entangled particles can be made to self-interfere in some cases. However, in those cases the entanglement will have already been terminated and now assumption a) above is no longer valid.

 

[colorSpace]

3. Entangled particles can be made to self-interfere in some cases. However, in those cases the entanglement will have already been terminated and now assumption a) above is no longer valid.

I'm not sure which cases you are referring to, however I think it is worth mentioning that in the "double-double-slit" experiment (which might be just a thought experiment by A.Zeilinger), both particles can (or are expected to) show an entangled interference pattern, however this pattern is only visible in hindsight, using data exchanged in a classical way, through analysis of the entanglement-based time correlations and a specific experimental setup (which might only analyze one of the patterns at a time, so to speak). Unfortunately quite complicated, this is described in at least one of A.Zeilingers books.

 

[msumm21]

Thanks for the great info. I'm checking out the Zeilinger paper.

 

[debra]

Re Zeilinger Paper, FIG 2. http://www.hep.yorku.ca/menary/courses/phys2040/misc/foundations.pdf

Referring to interference patterns for entangled particles, one of which passes through a double slit.
If one entangled particle travels in the opposite direction then there is no interference pattern given by the
particle that went through the double slit because the superpositon would cancel intensities out exactly - except
at the particle positions, and not primarily because 'there is a possibility of knowing which path'. Using wave packet analysis which path (of a particle!) plays no role - there is no particle its a wave packet as in CI. The 'particle' is a position state observable. Similarly for 'waves', that's another state observable. Cannot we finally get away from the old wave particle duality issue?
I refer to: http://www.hep.yorku.ca/menary/courses/phys2040/misc/foundations.pdf FIG 2.
There were two wave packets (as per CI) that went through each slit, and the resulting interference makes
it look like only particles went through one or the other slit.
If wave packets are analysed instead of trying to think about particles or waves then statements such as 'the possibility of
knowing which path' appear unscientific to me, - the wave packets go down all possible paths allowed by experimental set ups.
The wave packets of entangled particles then have more components to consider which give rise to interference patterns or lack of.

 

[DrChinese]

...however this pattern is only visible in hindsight, using data exchanged in a classical way, through analysis of the entanglement-based time correlations...

Yes, and that is why there are no FTL signaling possibilities (the OP raised that idea). But looking on one side by itself, there is no traditional interference pattern.

 

[DrChinese]

Re Zeilinger Paper, FIG 2. http://www.hep.yorku.ca/menary/courses/phys2040/misc/foundations.pdf

Referring to interference patterns for entangled particles, one of which passes through a double slit.
If one entangled particle travels in the opposite direction then there is no interference pattern given by the
particle that went through the double slit because the superpositon would cancel intensities out exactly - except
at the particle positions, and not primarily because 'there is a possibility of knowing which path'. Using wave packet analysis which path (of a particle!) plays no role - there is no particle its a wave packet as in CI. The 'particle' is a position state observable. Similarly for 'waves', that's another state observable. Cannot we finally get away from the old wave particle duality issue?
I refer to: http://www.hep.yorku.ca/menary/courses/phys2040/misc/foundations.pdf FIG 2.
There were two wave packets (as per CI) that went through each slit, and the resulting interference makes
it look like only particles went through one or the other slit.
If wave packets are analysed instead of trying to think about particles or waves then statements such as 'the possibility of
knowing which path' appear unscientific to me, - the wave packets go down all possible paths allowed by experimental set ups.
The wave packets of entangled particles then have more components to consider which give rise to interference patterns or lack of.

Welcome to PhysicsForums!

Not sure I follow your comment exactly. You can easily tell which slit the entangled photon traverses since there is no interference pattern. Just place a detector behind each slit. Learning that info for Alice will allow you to predict the result for Bob. But that does not violate CI, nor the HUP.

 

[colorSpace]

Yes, and that is why there are no FTL signaling possibilities (the OP raised that idea). But looking on one side by itself, there is no traditional interference pattern.

Yes, no traditional, immediately visible pattern.

The known effects of entanglement are always a symmetric relation of otherwise random events. "Random" means subject to the Heisenberg uncertainty principle. Their uncertainty is coupled. If the experiment doesn't "allow" the particles to be uncertain in some property, then there can't be any entanglement in that property, in so far as is known (or as I know).

------ EDIT: --------

For such experiments where the uncertainty (via interference) itself becomes the observed property, the seemingly unavoidable obstacle is that in order to entangle two particles in way that they can show interference for the entangled property, one needs a different source than one needs to create just one particle with uncertainty (showing interference).

As far as I have understood, this is again due to the HUP, within the source. In order to be entangled in such way that measurement on one particle would influence the interference pattern of the other particle, there needs to be an additional coupled uncertainty regarding, for example in a photon source, the location from where the particles originate. This uncertainty in the source location will cause the interference patterns to overlap such that they appear only as a random patterns (until they are analyzed via time coincidence using classically transmitted data from the other particle's measurements).

 

[msumm21]

I read the Zeilinger paper discussed (and linked to) above, which says that that a member of an entangled pair will only show self-interference if the which-way info for the entangled partner is destroyed. So what I was saying in the original post would NOT work.

Zeilinger (pages 2-3) discusses an experiment by Dopfer in which the measurement of the which-way info of the entangled partner is turned on and off by moving the detector D1 for photon 1. So the interference pattern of the other photon (photon 2) is turned on/off be moving the detector D1. So I have to ask, why can't we use that for FTL communication? Someone can move D1, while someone else (arbitrarily far away) gets photon 2 and runs it through the double slit experiment to look for an interference pattern. What's wrong with this approach? I know you need N copies of this experiment run in parallel to really determine the existence of the interference pattern, but that's simple enough right?

 

[colorSpace]

So I have to ask, why can't we use that for FTL communication?

That is exactly the question I tried to answer above, especially in the "EDIT:" section. I' look forward to see an answer from someone more qualified (than me).

 

[Cthugha]

Zeilinger (pages 2-3) discusses an experiment by Dopfer in which the measurement of the which-way info of the entangled partner is turned on and off by moving the detector D1 for photon 1. So the interference pattern of the other photon (photon 2) is turned on/off be moving the detector D1. So I have to ask, why can't we use that for FTL communication? Someone can move D1, while someone else (arbitrarily far away) gets photon 2 and runs it through the double slit experiment to look for an interference pattern. What's wrong with this approach? I know you need N copies of this experiment run in parallel to really determine the existence of the interference pattern, but that's simple enough right?

See page 290 of the Zeilinger paper:

Therefore, a double-slit interference pattern for photon 2 is registered conditioned on registration of photon 1 in the focal plane of the lens. It is important to note that it is actually necessary to register photon 1 at the focal plane because without registration one could always, at least in principle, reconstruct the state in front of the lens.

The interference pattern in one leg is still conditioned on the registration of the entangled partner in the other leg. So the interference pattern will only show up in coincidence counting, which means that you will still have to get the information from one detector to the other in a traditional communication way. Therefore this setup does not allow FTL communication.

 

[msumm21]

I don't know what is meant by registration, but don't we know that roughly X percent of the photons will be registered when D1 is at the focal plane, and roughly Y percent will be registered when D1 is at the imaging plane? If so, can't we differentiate the patterns behind the double slit when:

Case 1 (D1 at focal plane; trying to destroy the which way info):
100-X percent of the photon's are not registered and effectively follow "patternA" behind the double slit
X percent are registered and they follow "patternB" (interference pattern)
NetPattern1 = (100-X)*patternA/100 + X*patternB/100

Case 2 (D1 at imaging plane; NOT trying to destroy the which way info):
100-Y percent are not registered and follow "patternA"
Y percent are regestered and follow "patternC"
NetPattern2 = (100-Y)*patternA/100 + Y*patternC/100

Can't someone distinguish between NetPattern1 and NetPattern2? With arbitrarily high confidence by increasing N (=the number of experiments performed in parallel)?

 

[DrChinese]

I read the Zeilinger paper discussed (and linked to) above, which says that that a member of an entangled pair will only show self-interference if the which-way info for the entangled partner is destroyed. So what I was saying in the original post would NOT work.

Zeilinger (pages 2-3) discusses an experiment by Dopfer in which the measurement of the which-way info of the entangled partner is turned on and off by moving the detector D1 for photon 1. So the interference pattern of the other photon (photon 2) is turned on/off be moving the detector D1. So I have to ask, why can't we use that for FTL communication? Someone can move D1, while someone else (arbitrarily far away) gets photon 2 and runs it through the double slit experiment to look for an interference pattern. What's wrong with this approach? I know you need N copies of this experiment run in parallel to really determine the existence of the interference pattern, but that's simple enough right?

Following up on what Cthugha said and why you can't get FTL communication:

There ARE some situations in which certain special kinds of interference patterns appear for entangled particles. These are NOT traditional double slit patterns similar to unentangled photons. That is what the Zeilinger reference is saying regarding Figure 2, which is showing the traditional double slit format.

Unfortunately, most of the papers on this type of experiment are a bit complex, and they do not usually explain explicitly that the interference pattern is not the traditional double slit type on one side. In the special cases in which interference patterns can appear (such as Zeilinger's Figure 3), there are two critical factors:

a) Coincidence counting is required involving measurements at both Alice and Bob. Obviously, you cannot get FTL communication if you need to first correlate results at Alice and Bob.

b) The raw pattern that Bob sees is not conditioned on anything Alice does (such as moving Alice's D1 from one focal point to another). The image Bob sees stays the same! Now why is that, since clearly the coincidental counts drop to nearly zero in some of Bob's areas when Alice positions D1 at a precise point?

Registering a photon at the focal point of a lens at Alice occurs randomly, and this occurs only a small fraction of the overall counts. You may find this easier to see if you look at the actual http://www.quantum.univie.ac.at/publications/thesis/bddiss.pdf . Looking at the detail (not given in the Zeilinger reference), I would estimate that only 80 in 300 photons registers this way. The other 220 go towards making up the overall image that looks like a traditional Bell curve. So when you look at the difference between one and the other of the positions Alice chose, there is little difference to see in the raw data Bob sees. By my very rough calculations, it would be right on the threshold of system noise and random variation. So nothing useful for Bob to see by way of a signal.

If Dopfer had seen a change in the pattern at D2 after changing the D1 position, I think she would have been jumping for joy on the way to the patent office. After all, she had the raw data and this experiment is now 10 years old. Nonetheless, John Cramer has proposed that there should be a small differential that might be within the realm of visibility, stating:

"In the Dopfer Experiment, the pattern in the lower detector was selected by requiring events in which detected photon were in good time-coincidence in both detectors. Therefore, it is possible (barely) that with no coincidences, the same raw pattern would appear in the lower detector, independent of the position of the upper detector, with the coincidences from the upper arm selecting from this raw pattern a diffraction pattern in Case 1 and an interference pattern in Case 2.

However, this scenario seems unlikely. Around 85% of the photons in the lower detector should be in coincidence with photons in the upper detector regardless of position. Therefore, while the coincidence requirement should remove the 15% of noise, it should in itself do little else. In particular, it should not be able to thin out the raw pattern enough to produce the minima of the interference pattern.

Unfortunately, the Dopfer thesis does not discuss what was observed in the lower detector with the coincidence requirement removed. For this reason, a crucial test of quantum phenomena would be to re-create the Dopfer Experiment and observe the role of the coincidence requirement on what is observed in the lower arm detector. Several research groups are considering doing this, but there are no results yet."

 

[colorSpace]

I don't know what is meant by registration, but don't we know that roughly X percent of the photons will be registered when D1 is at the focal plane, and roughly Y percent will be registered when D1 is at the imaging plane? If so, can't we differentiate the patterns behind the double slit when:

Case 1 (D1 at focal plane; trying to destroy the which way info):
100-X percent of the photon's are not registered and effectively follow "patternA" behind the double slit
X percent are registered and they follow "patternB" (interference pattern)
NetPattern1 = (100-X)*patternA/100 + X*patternB/100

Case 2 (D1 at imaging plane; NOT trying to destroy the which way info):
100-Y percent are not registered and follow "patternA"
Y percent are regestered and follow "patternC"
NetPattern2 = (100-Y)*patternA/100 + Y*patternC/100

Can't someone distinguish between NetPattern1 and NetPattern2? With arbitrarily high confidence by increasing N (=the number of experiments performed in parallel)?

Preliminary answer from a layman: Netpattern1 (interference) consists (except for some noise) of an superposition of multiple interference patterns. Without patternB, the remaining patternA does not have a completely random distribution anymore.

Netpattern2 does not contain the same patternA, and so you don't get a meaningful result by subtracting patternA from Netpattern2. In both cases, it would look like the result is something like patternB, but in the second case, performing this calculation would be meaningless.

In other words, both Netpattern1 and Netpattern2 are random, but in case 1 it is possible to select an interference pattern (using classically obtained information from measurements of the other particle), where the remaining pattern is not random either, but their sum is random. The superposition of a set of interference patterns which add up to a random pattern. At least that is my understanding of the current theory (in A.Zeilinger's books), and if that understanding is correct, it would take an actual experiment to show if it were otherwise.

 

[DrChinese]

What we are trying to guess is whether the fringe spots with no intensity (when registered) might be detectable in a very tightly controlled setup in which there was a corresponding slight dip in the total at that fringe spot. That is what Cramer wondered, and msumm21 as well.

Obviously, Dopfer noticed nothing. Obviously, there are no other prominent physicists who have published on this in the past ten years with a result supporting FTL communication.

So here is a DrChinese going-out-on-a-limb prediction: The raw data will not show any difference at Bob (when Alice varies her detector positioning) regardless of experimental accuracy. This conclusion is based on a belief that there is no FTL signaling possible. Therefore, for this result to be correct, you merely conclude that the HUP is always at work - if in strange ways.

What observation you choose at Bob must influence what happens at Alice as much as the reverse. So merely changing the Alice setup (between momentum and position observations) is only part of the equation. We still cannot exceed the limits of the HUP. Therefore something else must change. Presumably, there is an offsetting change in the distribution such that:

Alice's detector at position 1:
B1=
Bob's detections at fringe area X, when Alice at fringe area Y +
Bob's detections at fringe area X, when Alice not at fringe area Y

Alice's detector at position 2:
B2=
Bob's detections at fringe area X, when Alice at fringe Y +
Bob's detections at fringe area X, when Alice not at fringe Y

B1 = B2, and this holds for any conditional of Alice, including any fringe area Y, even those where X=Y.

We would expect this to be true, just like we expect polarization results to remain static for Bob regardless of changes in polarization observations at Alice. There is no difference.

But the Dopfer results do show us something. Clearly, the observation at Alice can be chosen after the results are in at Bob - and the coincidence counts will still match predictions. We knew this already, from the earlier Aspect experiments for example. The difference in this case is: the influence of the change at Alice can be discerned (with coincidence counting); while in the Aspect experiment, it was not possible to see the effect of Alice's observation on Bob differently than Bob's observation on Alice (also with coincidence counting). So that is cool in Dopfer.

Presumably, there would be a way to vary the observation at Bob so that either his result occurred either before or after the result at Alice. In that way, it might become clear that the the results are dependent on both entangled Alice and Bob equally, and that the sequence of measurement does not establish a preference in the results. :)

 

[colorSpace]

What we are trying to guess is whether the fringe spots with no intensity (when registered) might be detectable in a very tightly controlled setup in which there was a corresponding slight dip in the total at that fringe spot. That is what Cramer wondered, and msumm21 as well.

Obviously, Dopfer noticed nothing. Obviously, there are no other prominent physicists who have published on this in the past ten years with a result supporting FTL communication.

Right, there are two questions:

1. Is it possible to create an experiment where interference can be detected depending on an entangled particle, but without classically obtained data from that particle. (Apparently not.)

2. If not, what is the explanation a) specifically within any (thought-) experiment, and b) generally, in theory.

So here is a DrChinese going-out-on-a-limb prediction: The raw data will not show any difference at Bob (when Alice varies her detector positioning) regardless of experimental accuracy. This conclusion is based on a belief that there is no FTL signaling possible. Therefore, for this result to be correct, you merely conclude that the HUP is always at work - if in strange ways.

And, you could also base it on the belief that entanglement always only shows a relationship between two otherwise random events (which I understand, is suggested by the theory, their wavefunctions, etc.)

[...]
But the Dopfer results do show us something. Clearly, the observation at Alice can be chosen after the results are in at Bob - and the coincidence counts will still match predictions. We knew this already, from the earlier Aspect experiments for example. The difference in this case is: the influence of the change at Alice can be discerned (with coincidence counting); while in the Aspect experiment, it was not possible to see the effect of Alice's observation on Bob differently than Bob's observation on Alice (also with coincidence counting). So that is cool in Dopfer.

Presumably, there would be a way to vary the observation at Bob so that either his result occurred either before or after the result at Alice. In that way, it might become clear that the the results are dependent on both entangled Alice and Bob equally, and that the sequence of measurement does not establish a preference in the results. :)

As far as I understand, any (experimental, to this date) entanglement-specific results depend on both measurements, and therefore remain in uncertainty until the second one is made. And, they don't appear to depend on the sequence of measurement, each result, observed separately, is apparently random, and therefore (apparently) independent of what happens at the other particle. (This sentence was edited after posting).

This implies, specifically in the case of observing interference, not only that data from the other particle is needed for detection of an effect, but also goes together with certain restrictions on what experimental setups are possible, in this case, that a particle source which is able to produce entangled particles with uncertain properties (which can interfere) will produce only a a set of superimposed interference patterns which in their sum appear random.

Verifying that this is actually the case, in reality, requires verifying that particle sources (in this case a "UV pump", together with the crystal) indeed always have this limitation.

On the other hand, if it were possible to build a particle source that doesn't have this limitation, then that would be a new discovery, also implicating a new or modified theory about entanglement and its basis. In my understanding.

 

[msumm21]

In the previous posts, what is meant by a "random pattern?" I know that, if we have an underlying probability density function F revealing the probability of the photon striking the screen at all places, then a random observation is one in which the photon hits a location on the screen with probability following from F. After repeating the experiment enough times and recording the results we can determine F, which can be a normal distribution, uniform distribution, a type of interference distribution, or anyone of an infinite number of other distributions.

It was stated above that the following two patterns are different:
1) pattern seen by Bob when the detector at Alice is at D1 and the photon is not "registered"
2) pattern seen by Bob when the detector at Alice is at D2 and the photon is not "registered"
Is this true? So even when the detector does not register a photon it influences the pattern at Bob?

Thanks

 

[DrChinese]

It was stated above that the following two patterns are different:
1) pattern seen by Bob when the detector at Alice is at D1 and the photon is not "registered"
2) pattern seen by Bob when the detector at Alice is at D2 and the photon is not "registered"
Is this true? So even when the detector does not register a photon it influences the pattern at Bob?

Thanks

Alice's detector is D1, and it can either be at the focal point or not. So perhaps you intended to ask the following?

It was stated above that the following two patterns are different:
1) pattern seen by Bob when the detector D1 at Alice IS at the focal point and the photon is not "registered"
2) pattern seen by Bob when the detector D1 at Alice is NOT at the focal point and the photon is not "registered"
Is this true? So even when the detector D1 does not register a photon it influences the pattern at Bob?

If so, I believe the answer must be YES. However, it is tricky here because the sets are not equivalent. I believe that coincidence counting can confirm this, but again no useful information can be encoded.

 

[colorSpace]

In the previous posts, what is meant by a "random pattern?" I know that, if we have an underlying probability density function F revealing the probability of the photon striking the screen at all places, then a random observation is one in which the photon hits a location on the screen with probability following from F. After repeating the experiment enough times and recording the results we can determine F, which can be a normal distribution, uniform distribution, a type of interference distribution, or anyone of an infinite number of other distributions.

It was stated above that the following two patterns are different:
1) pattern seen by Bob when the detector at Alice is at D1 and the photon is not "registered"
2) pattern seen by Bob when the detector at Alice is at D2 and the photon is not "registered"
Is this true? So even when the detector does not register a photon it influences the pattern at Bob?

Thanks

"Random", in the sense of Heisenberg-uncertain, here means that the pattern can be thought of as a superposition (overlay) of multiple interference patterns (at least in the absence of which-way-information) such that the above-average amounts of one interference pattern are canceled out by the below-average amounts of another pattern. The resulting pattern doesn't have any ups and downs as are characteristic for a single visible interference pattern. However an interference pattern can be filtered out (in the absence of which-way information) by counting only those photons at Bob which coincide with certain photons (reaching a specific point) at Alice.

As far as I understand from the books of A.Zeillinger, this is because in the light source, the exact origin location of the photon(s) must be subject to Heisenberg uncertainty, as otherwise they couldn't be entangled regarding their momentum (which must be exactly opposite for this experiment to make sense).

In order to be entangled regarding their momentum, they must also have a matching relationship regarding the origin location. (Momentum and location are complementary). This in turn is only possible if the origin location is subject to Heisenberg uncertainty (it must be a "larger" light source).

Since the origin location is uncertain, this is somewhat similar to having many sources of light, each casting an interference pattern, such that the patterns cancel each other out in terms of highs and lows.

This means that Bob will not be able to see any interference without selecting (filtering out) a specific set of photons by using coincidence counting.

I understand this to be an example of the principle that in entanglement, according to our current knowledge and experiments, any property which is able to show its entangled-ness, must be "uncertain", and the entanglement will be detectable only in hindsight (meaning, by adding classically obtained information) as a relationship of two "uncertain" (random) events.

It seems to me that the Heisenberg uncertainty allows other effects which are not restricted by the speed of light (for example tunneling), but that in each case the randomness of the uncertainty doesn't allow it to be used for achieving targeted effects. It seems this would require some new factor which goes beyond quantum physics as we currently know it. If I had to guess: maybe it would have to do with taking advantage of the tendency in physics for certain things to balance each other out (which one could say entanglement is also an example of).

 

[colorSpace]

Additional to the above comment:

Perhaps more specific to the intent of the question, but still speaking generally in terms of the principle:

In order to detect an interference pattern at Bob, it is necessary to filter out a specific set of photons. The information required to do so must be obtained from Alice. If the photons at Alice are measured in such a way that they reveal which-way information, then they will not also reveal information that allows filtering out an interference pattern.

I hope this helps.

BTW, there are also experiments (which I'm much less familiar with) which demonstrate that not-registering a particle has an effect. I tried to explain this to myself by saying that even if in that case the particle is not influenced by any registration, the wave function (which includes all possible paths) is still affected. Not sure, though, and not sure if this perhaps also applies to some aspect of the above experiment. However, in the double slit experiment, one can see that both slits affect the photon even though it could in the classical sense fly only through one of them.

Thus, the possibility of a path affects the photon, even if it doesn't take that path in the classical sense.

 

[msumm21]

DrChinese, you are right -- I meant to say Alice's detector D1 is at the focal point or moved away from it.

ColorSpace, I think I see what you are saying about the uncertainty in the origin of the original photon and how that could lead to a superposition of several interference patterns, but ultimately there is still some pattern right, even if it's a uniform or near uniform pattern with very small "peaks" and "valleys." That the ultimate pattern is unchanged by moving D1 seems hard to believe at first, but again I don't understand all the details yet. Thanks for the info about the detector location affecting the results even when it fails to register a photon. I think I need to get the Zeillinger book(s) you are mentioning. Could you let me know the title(s)? Thanks.

 

[Dbar_x]

I was looking for information about something else when I discovered this forum. I’m quite sure I know the answer to the initial question you’ve been discussing.

Passage through a Stern-Gerlach magnet is a quantum measurement, resulting in a statistical mixture of spin particles on either one path or the other. No detector screen or filtering slits to separate the two beams are required. Notice that there is an energy and momentum exchange each time the particle absorbs a magnetic field quantum. Information is genuinely physical, as Rolf Landauer emphasized, so this energy exchange is also an information exchange or measurement. Like the sliding slits discussed by Einstein and Bohr in their measurement example, the spin direction of the particle is recorded or registered by the admittedly tiny reciprocal kick given to the magnet when the particle is kicked in the opposite direction.

There are innumerable quanta exchanged as the particle transits the magnetic field, but every one kicks the particle in the same direction, resulting in a macroscopic measurement and macroscopic particle deflection. (Stern and Gerlach found something like 0.1 mm.)

If the particle remained in a superposition of up and down spins after the first field quantum is absorbed, the next quantum would be equally likely to deflect the particle in either direction, and so on. It doesn’t do that because the particle’s spin direction is now definite. This continues for all the other quantum exchanges through the field, resulting in an amplified deflection of the particle and the magnet. (Still extremely small, of course, but most quanta are.) So, whether or not one measures spin direction of the second, entangled particle, the one going through the S-G magnet never emerges in a superposition of spins, and no FTL information transfer can occur.

All the quantum mechanics textbooks I’ve consulted, including Merzbacher and Bohm, say the spinning particle is still in a spin superposition beyond the magnet, until its position is registered by a detector screen. Feynman (in his Lectures) said much the same thing, using a slit screen to “filter” the particles into a statistical mixture. I’m convinced they are mistaken, probably just borrowing the same explanation from a predecessor. No one has done an experiment to determine if spinning particles beyond the magnet are, in fact, still in a superposition. It’s always difficult to observe any difference between a superposition and mixture.

There is, however, a simple, quite inexpensive experiment, that could be done to settle what seems to me a very important issue in quantum measurement. I’ve checked carefully and it has not so far been performed. I’m now a retired physicist, having done experimental particle research and teaching, but am still very interested in quantum measurement, and would like to see such a decisive experiment performed.

DocMike ( Los Alamos)

 

[DrChinese]

All the quantum mechanics textbooks I’ve consulted, including Merzbacher and Bohm, say the spinning particle is still in a spin superposition beyond the magnet, until its position is registered by a detector screen. Feynman (in his Lectures) said much the same thing, using a slit screen to “filter” the particles into a statistical mixture. I’m convinced they are mistaken, probably just borrowing the same explanation from a predecessor. No one has done an experiment to determine if spinning particles beyond the magnet are, in fact, still in a superposition. It’s always difficult to observe any difference between a superposition and mixture.

There is, however, a simple, quite inexpensive experiment, that could be done to settle what seems to me a very important issue in quantum measurement. I’ve checked carefully and it has not so far been performed. I’m now a retired physicist, having done experimental particle research and teaching, but am still very interested in quantum measurement, and would like to see such a decisive experiment performed.

DocMike ( Los Alamos)

Don't think you'll find much there. Every exotic test of QM always yields similar results. There are a lot of experiments which have been performed on photons in the past 10 years that definitely demonstrate that particles are in the superposition until observed. The quantum erasers, for example, allow the way the photon is observed to be varied in many ways. Not sure what you have in mind specifically. But if you can convert your example from electrons to photons (where 98% of the research is performed), most likely there is a similar experiment already documented.

 

[debra]

There is, however, a simple, quite inexpensive experiment, that could be done to settle what seems to me a very important issue in quantum measurement. I’ve checked carefully and it has not so far been performed. I’m now a retired physicist, having done experimental particle research and teaching, but am still very interested in quantum measurement, and would like to see such a decisive experiment performed.

DocMike ( Los Alamos)

What is the experiment, do tell us? (Also, I agree with you that the particles will be disentangled after the first quanta from the field is received).

 

[Dbar_x]

Don't think you'll find much there. Every exotic test of QM always yields similar results. There are a lot of experiments which have been performed on photons in the past 10 years that definitely demonstrate that particles are in the superposition until observed. The quantum erasers, for example, allow the way the photon is observed to be varied in many ways. Not sure what you have in mind specifically. But if you can convert your example from electrons to photons (where 98% of the research is performed), most likely there is a similar experiment already documented.

I’ve no doubt that the wavefunction of a photon passing through a double-slit screen, or through the arms of an interferometer, remains in a superposition of separated paths until detected at the output. From Young’s experiment to the recent quantum eraser (Q.E.) photon experiment done by Scully with his colleagues at the U. of Maryland (Phys. Rev. Lett. 84, 1, 1999), I think that’s undeniable.

And I don’t question the cold neutron interference described by Zeilinger, Rauch, etc., or even the atomic interference reported at MIT. (I do doubt that there is such a thing as real interference between a live and dead cat.) The Stern-Gerlach experiment is a separate phenomenon. Because it remains our pedagogical prototype for quantum measurement, I believe we ought to actually understand what’s going on there.

I think that the source of confusion and controversy in quantum theory is centered on the process of quantum measurement. What precisely is a measurement? What, exactly, is information? How does one describe the process of measurement? Measurement isn’t depicted by Schrodinger evolution because that always maps superpositions into other superpositions, never mapping a superposition into a mixture or into a single eigenfunction. Was von Neumann right about collapse of the wavefunction at measurement? How else can there be a mixture after measurement?

Has anyone else here read with care the original Q.E. paper written by Scully and his colleagues in 1978? (Phys. Rep. 43, 13, p. 485) They analyze a modified S-G experiment, including a bi-level atom in one arm of the magnet as path detector. Under certain assumptions they derive a nearly diagonalized density matrix for the spinning molecule beyond the magnet, without any wavefunction perturbation at measurement. They infer from this that an erasure of information stored in the path detector should restore the interference pattern destroyed when which path was detected.

Because the S-G example is so important for understanding the quantum measurement process, I’d like to promote an experimental demonstration that the spinning particle actually emerges from the magnet in just one of the two paths, not in a superposition of both. It’s a simple, inexpensive experiment. This would contradict what we read in all the quantum mechanics textbooks. It’s different, of course, from the persistent both-way superposition of a single photon’s wavefunction in an interferometer.

What is the experiment, do tell us? (Also, I agree with you that the particles will be disentangled after the first quanta from the field is received).

Doing the experiment I’ve mentioned would require very little funding, almost inconsequential by standards of most such contemporary experiments. I’m sure it could be done in a small undergraduate college lab. If anyone is interested, please contact me via e-mail.

DocMike

 

[debra]

Has anyone else here read with care the original Q.E. paper written by Scully and his colleagues in 1978? (Phys. Rep. 43, 13, p. 485)


DocMike

I cannot find this paper, have you got a web reference?

 

[JesseM]

Has anyone else here read with care the original Q.E. paper written by Scully and his colleagues in 1978? (Phys. Rep. 43, 13, p. 485) They analyze a modified S-G experiment, including a bi-level atom in one arm of the magnet as path detector. Under certain assumptions they derive a nearly diagonalized density matrix for the spinning molecule beyond the magnet, without any wavefunction perturbation at measurement. They infer from this that an erasure of information stored in the path detector should restore the interference pattern destroyed when which path was detected.

Because the S-G example is so important for understanding the quantum measurement process, I’d like to promote an experimental demonstration that the spinning particle actually emerges from the magnet in just one of the two paths, not in a superposition of both. It’s a simple, inexpensive experiment.

I still don't quite understand what this "experimental demonstration" would be--are you just talking about the experiment in the previous paragraph where Scully et al. "infer from this that an erasure of information stored in the path detector should restore the interference pattern", and are you proposing that they would be wrong, that it would be impossible to restore the interference pattern by erasing which-path information? What is the specific setup that they propose would erase this information, is it an interferometer similar to what's used to erase the which-path info of idler photons in the delayed choice quantum eraser?

 

[DrChinese]

Doing the experiment I’ve mentioned would require very little funding, almost inconsequential by standards of most such contemporary experiments. I’m sure it could be done in a small undergraduate college lab. If anyone is interested, please contact me via e-mail.

DocMike

You don't seem to want to address the fact that S-G type experiments are no different than what you can do with a photon and a polarizing beam splitter. These (photons + PBS) are very easy to do and a zillion versions have already been done. So what is so special about S-G? I mean, it's not like electrons display behavior that is fundamentally different than photons with regards to interference or spin. You say you want to understand what is "really" going on, but so do we all. You might be surprised to realize exactly how much new research is done in this area, both theoretically or experimentally. For example, check out any of these 100 articles:

Stern Gerlach

 

[Dbar_x]

I cannot find this paper, have you got a web reference?

Debra, I've searched for the title, "State Reduction in Quantum Mechanics: a Calculational Example", and not found a copy on the web. I got my copy the old way, from the journal, Physics Reports in 1978.

I still don't quite understand what this "experimental demonstration" would be--are you just talking about the experiment in the previous paragraph where Scully et al. "infer from this that an erasure of information stored in the path detector should restore the interference pattern", and are you proposing that they would be wrong, that it would be impossible to restore the interference pattern by erasing which-path information? What is the specific setup that they propose would erase this information, is it an interferometer similar to what's used to erase the which-path info of idler photons in the delayed choice quantum eraser?

There have been some relatively simple quantum eraser experiments published, like the one by Kwiat and Hilmer in Scientific American, "A Do-It-Yourself Quantum Eraser" in 2007, and Schneider and LaPuma in the American Journal of Physics, "A simple experiment for discussion of quantum interference..." in 2002, and Galvez et al. also in the Am. J. Phys. 2005, "Interference with correlated photons...". But the experiment I'm promoting is not a quantum eraser experiment. Rather, as I said, it's a simple experiment to demonstrate that our accepted understanding of the Stern-Gerlach experiment is mistaken. It seems to be taken for granted today, and was also assumed by Scully et al. in 1978, that the spinning silver atoms emerging from the magnet remain in a superposition of the two spin directions. I'm convinced that's incorrect. Each silver atom emerges from the magnet with a single definite spin direction, either up or down, because a measurement has occurred. Magnet deflection (macroscopic momentum transfer) has recorded spin direction, performed a measurement, or observation, and the result is an atom now with a definite eigenvalue of spin.

Essentially all quantum measurement analyses are based on our prototype, the S-G experiment. So it's cruicial, I believe that we actually understand that phenomenon. Thus the simple experiment I have in mind. I believe the results of this experiment must result in rethinking and reanalysis of essentially all our current measurement theories, including quantum erasure and Zurek's decoherence.

The simple experiment I'm proposing is a modification to the original one from the 1920's. So equipment required includes a small vacuum system, a source of polarized particles, like an oven to vaporize silver, and a powerful magnet. The results, I'm convinced, would be convincing. I don't own vacuum equipment and such, but anyone with access to that stuff who would like to discuss the experiment, may contact me via e-mail.

DocMike

 

[Dbar_x]

I still don't quite understand what this "experimental demonstration" would be--are you just talking about the experiment in the previous paragraph where Scully et al. "infer from this that an erasure of information stored in the path detector should restore the interference pattern", and are you proposing that they would be wrong, that it would be impossible to restore the interference pattern by erasing which-path information? What is the specific setup that they propose would erase this information, is it an interferometer similar to what's used to erase the which-path info of idler photons in the delayed choice quantum eraser?

Are you a skeptic of quantum erasure? Perhaps you know that Kwiat wrote about all the original Q.E. experiments, including his own experiments at Berkeley, and those of Mandel and coleagues at Rochester, that “none of these possesses all the attributes of a true quantum eraser.” Senitsky (Phys. Rev. Lett. 73, 3040, 1994) at the laser laboratory at U. of Southern California showed that the experiment Scully finally believed really demonstrated quantum erasure, the somewhat exotic crossed beams done by Zeilnger, et al. (“Frustrated Two-Photon Creation via Interference”, Phys. Rev. Lett. 72 629, 1994) was not an interference experiment at all, but rather an example of parametric amplification that can be completely explained by classical electromagnetic theory. I agree with him.

You don't seem to want to address the fact that S-G type experiments are no different than what you can do with a photon and a polarizing beam splitter. These (photons + PBS) are very easy to do and a zillion versions have already been done. So what is so special about S-G? I mean, it's not like electrons display behavior that is fundamentally different than photons with regards to interference or spin. You say you want to understand what is "really" going on, but so do we all. You might be surprised to realize exactly how much new research is done in this area, both theoretically or experimentally. For example, check out any of these 100 articles:

Happy to addrss it. The cruicial question, I think, is not whether photons and atoms both exhibit interference, but what exactly constitutes a quantum measurement, and what doesn’t. The question has definitely not been settled by physicists, resulting, I believe, in the continued confusion in quantum theory. Many measuement theorests today claim, for example, that what we do or do not know is an essential ingredient in the measurement process. Some, including Kwiat, have written that knowledge and the visability of interference are complementary variables in the Bohr theory (of complementarity). I think that the quantum theory, like all other physical theories, must be observer independent.

Obviously, I believe that the spin direction of silver atoms exiting a S-G magnet has been measured, so resulting in each atom going either one way or the other, not both. There is no experimental evidence of S-G interference, or anything else, yet, to refute or confirm my contention. That includes all the photon interference experiments.

If the atoms from a S-G magnet had not been measured, and actually remained in a spin superposition, there would not be the two separated traces left on the detector. If the atom continued in its spin supeposition it would be kicked back and forth, randomly, by each quantum in the macroscopic field, and would emerge still centered on the magnet, not at one of two separated spots. Instead, the measurement is amplified by each successive quantum exchange, certainly not something done in a photon beam splitter. A beam splitter cannot record (or register, or measure) which way the photon went. If it did measure which way, the photon would be reduced to just one path, with no interference.

 

[JesseM]

I still don't quite understand what this "experimental demonstration" would be--are you just talking about the experiment in the previous paragraph where Scully et al. "infer from this that an erasure of information stored in the path detector should restore the interference pattern", and are you proposing that they would be wrong, that it would be impossible to restore the interference pattern by erasing which-path information? What is the specific setup that they propose would erase this information, is it an interferometer similar to what's used to erase the which-path info of idler photons in the delayed choice quantum eraser?

Are you a skeptic of quantum erasure?

I see no reason to be, why would you imagine that? I was just asking whether you thought the way to test whether an electron (or atom) passing through a Stern-Gerlach device is in a superposition of states before its direction is measured would be to do some sort of erasure of the which-path information which according to orthodox QM should allow us to observe an interference pattern (which would presumably show that the wavefunction could not have been collapsed due to the SG device alone). If that's not what you were getting at with your comment "They infer from this that an erasure of information stored in the path detector should restore the interference pattern destroyed when which path was detected", then what would you consider a valid experimental demonstration that the electron is indeed in a superposition of states after leaving the SG device (but before its deflection angle is measured by us)? The only way I can think of to demonstrate superposition experimentally is to show that some type of interference effects can be measured...

Perhaps you know that Kwiat wrote about all the original Q.E. experiments, including his own experiments at Berkeley, and those of Mandel and coleagues at Rochester, that “none of these possesses all the attributes of a true quantum eraser.” Senitsky (Phys. Rev. Lett. 73, 3040, 1994) at the laser laboratory at U. of Southern California showed that the experiment Scully finally believed really demonstrated quantum erasure, the somewhat exotic crossed beams done by Zeilnger, et al. (“Frustrated Two-Photon Creation via Interference”, Phys. Rev. Lett. 72 629, 1994) was not an interference experiment at all, but rather an example of parametric amplification that can be completely explained by classical electromagnetic theory. I agree with him.

I don't know anything about the specifics of that particular experiment, but are you claiming that all quantum eraser experiments can be explained in a classical way? What about the delayed choice quantum eraser, for example?

 

[debra]

It seems to me that the crux of the matter is to decide if entangled electrons passing through a magnetic field removes the entanglement effect between the electrons.

 

[QuantumBend]

Particle no entangling after magnet field. We know this. Many wrong thing in theory - no good.

 

[DrChinese]

... The cruicial question, I think, is not whether photons and atoms both exhibit interference, but what exactly constitutes a quantum measurement, and what doesn’t. The question has definitely not been settled by physicists, resulting, I believe, in the continued confusion in quantum theory. Many measuement theorests today claim, for example, that what we do or do not know is an essential ingredient in the measurement process. Some, including Kwiat, have written that knowledge and the visability of interference are complementary variables in the Bohr theory (of complementarity). I think that the quantum theory, like all other physical theories, must be observer independent.

Obviously, I believe that the spin direction of silver atoms exiting a S-G magnet has been measured, so resulting in each atom going either one way or the other, not both. There is no experimental evidence of S-G interference, or anything else, yet, to refute or confirm my contention. That includes all the photon interference experiments.

If the atoms from a S-G magnet had not been measured, and actually remained in a spin superposition, there would not be the two separated traces left on the detector. If the atom continued in its spin supeposition it would be kicked back and forth, randomly, by each quantum in the macroscopic field, and would emerge still centered on the magnet, not at one of two separated spots. Instead, the measurement is amplified by each successive quantum exchange, certainly not something done in a photon beam splitter. A beam splitter cannot record (or register, or measure) which way the photon went. If it did measure which way, the photon would be reduced to just one path, with no interference.

OK, I guess this makes sense... I guess I didn't follow (and still don't) where you are going with this. I think you are asking WHEN is the observation to have occurred, and I can see your point that this is not specified precisely in current theory.

S-G is usually used to orient particle spin in a particular direction and/or show that the value of the spin is quantized. A PBS does the same basic thing with light, so I guess I am not seeing your point here either... I don't see how the physical impact of particle on the S-G apparatus is fundamentally any different than that of the PBS.

You have indicated your preference for the view that reality is observer independent, so I am guessing your proposed experiment is intended to show this one way or the other. After reading your posts again, it seems like you are well aware of key work in the field. So that leaves me just a bit confused, because there have been so many experiments that have demonstrated observer dependence - or at a bare minimum, failed to demonstrate observer independence where it might have been expected. I don't fault anyone for believing in observer independence, mind you, I just question how another experiment is likely to provide a result any different than all the others. But I acknowledge that there is always the better idea out there, hopefully yours is the one. :)

 

[Dbar_x]

It seems to me that the crux of the matter is to decide if entangled electrons passing through a magnetic field removes the entanglement effect between the electrons.

Particle no entangling after magnet field. We know this. Many wrong thing in theory - no good.

If I understand entanglement (I hope I do), Debra, the single electron, or silver atom, moving through a S-G magnet is not entangled, as QuantumBend said. It’s just a single particle. Entanglement always entails two or more particles, components of what von Neumann called a composite object, described by a single wavefunction. Think of the two photons in Aspect’s experiment, moving away in opposite directions. Because of the property of the composite system, total spin in that case, each photon must exhibit an opposed spin direction when measured.

I believe the crux of the matter, Debra, is whether the S-G magnet is the measuring apparatus which reduces the particle’s spin superposition to a single eigenvalue. I’m convinced that no detector screen is necessary.

DocMike

 

[JesseM]

believe the crux of the matter, Debra, is whether the S-G magnet is the measuring apparatus which reduces the particle’s spin superposition to a single eigenvalue. I’m convinced that no detector screen is necessary.

And again, what sort of experimental result would it take to convince you that you are wrong about this?

 

[Dbar_x]

I am skeptical of quantum erasure. I’m a committed believer in rational argument and directed, repeatable experimentation as the fundaments of science. Perhaps everyone else is also. I think we ought not believe anything presented as scientific fact unless it’s consistent, comprehensive, comprehensible, and has been subjected to experimental tests by refutation. (Not just confirmation, I would suggest, since a confirmatory test usually can be explained by another theory, as Senitzky did with the Zeilinger-Kwiat experiment.) I think it’s a profound mistake for us to accept something as scientific fact just because nearly everyone else is saying the same thing.

We must know how wrong the majority can be, even in science. Remember the difficulties Galilleo endured, or Boltzmann, whose rash belief in atoms was treated with disdain by nearly every scientist of his day. Plate techtonics is a more recent example.

So, I want to see the scientific evidence for quantum erasure, not just scores of similar publications. As I’ve said, the years of Q.E. experiments done at Berkeley and Rochester were dismissed afterwards by the experimenters themselves. If they don’t believe it...

some sort of erasure of the which-path information which according to orthodox QM should allow us to observe an interference pattern

I don’t think, JesseM, that information erasure at the detector, causing interference restoration, is orthodox quantum mechanics. What’s the basis in theory? There can be no reversible measurement via Schrodinger development; mixtures do not continuously evolve from superpositions. And Q.E. advocates do not accept von Neumann’s collapse theory. The closest I’ve seen to a plausible theory of Q.E. is from Fedorh Herbut this year. He suggests that the Everett-like relative state interpretation may be necessary. Or what he calls relative reality of unitarily evolving states.

I don't know anything about the specifics of that particular experiment, but are you claiming that all quantum eraser experiments can be explained in a classical way? What about the delayed choice quantum eraser, for example?

The recent delayed choice quantum eraser experiment, in which Scully participated, (Phys. Rev. Lett. 84, 1; “Delayed ‘Choice’ Quantum Eraser” ) is supposed to be the implementation of the original micromaser quantum eraser proposed by Scully and Druhl in 1982. If one reads the article critically (not simply accepting its conclusions) and carefully, it’s easy to see that not even their mathematical development is credible.

Look at Fig. 2 and their mathematical notation. They call L₀ the optical path length from the location on the BBO crystal where the bi-photon pair is created to the position on the interference pattern. (D₀, that is, where the signal photon is detected.) Crucially, the authors treat L₀ as a constant. It surely is not, but depends on which slit produces the photon pair and on location x. On the other hand, L_j (just below equation 3) is treated as a variable, dependent on x, but is, instead a constant optical distance. Contrary to what they say, t is the time when the photon pair is produced at the crystal.

Moreover, there’s no information erasure, at all, that occurs in this experiment, is there? The which way information is stored at either detector D₃ or D₄ (presumably on a computer disk). It is never erased, but is used, at a future time, to correlate with those coincident photons at D₀ which produce no interference.

I’ve done as much calculation as I can based on their limited published physical parameters, and I’m convinced that all their results, figures 3, 4, and 5, are accounted for by two pump photons at the crystal simultaneously creating four, not two, down-converted photons. If the detector thresholds are set for two simultaneous photons, (interpreting a single photon signal as jitter) one sees all the results they interpret as quantum erasure. If it were just one signal and one idler photon, as claimed, the coincidence rate in figure 4 ought to be about 30,000 per second, not about 150. That’s based on the singles rate of 300,000 they show (figure 4) and an overall detection efficiency of about 10 per cent. Because two simultaneous pump photons at the crystal are much less likely than just one, they actually get the low coincidence rate of about 150.

I think you are asking WHEN is the observation to have occurred, and I can see your point that this is not specified precisely in current theory.

I believe the key, DocChin, is not when the measurement occurs, but which object is the measuring apparatus. In this case, does the S-G magnet, itself, record spin direction due to the macroscopic energy and momentum transfer that occurs? Or is it necessary, as our textbooks say, to insert a detector screen to complete the measurement?

I don't fault anyone for believing in observer independence, mind you,

Let’s recall the history of physics. Galillean relativity is fundamental to all of classical physics. Einstein’s special and general theories are founded on the basic notion that all the laws of science are covariant. Form invariant, that is, in the reference frame of every observer. Does it make any sense to discard observer independence because some of the many, contradictory, proposed theories of quantum measurement require observer dependence?

what would you consider a valid experimental demonstration that the electron is indeed in a superposition of states after leaving the SG device (but before its deflection angle is measured by us)? The only way I can think of to demonstrate superposition experimentally is to show that some type of interference effects can be measured...

No interference pattern is required, JesseM. Contact me via e-mail if you’d like to discuss the experiment.

DocMike

 

[JesseM]

I don’t think, JesseM, that information erasure at the detector, causing interference restoration, is orthodox quantum mechanics. What’s the basis in theory? There can be no reversible measurement via Schrodinger development; mixtures do not continuously evolve from superpositions.

Can you elaborate on why you think quantum erasure would require "reversible measurement" or mixtures evolving from superpositions? The delayed choice quantum eraser involves the "signal" photon that goes through the slits being entangled with an "idler" which can tell you which slit the signal photon went through if it goes to one set of detectors (D3 or D4 in Scully's paper) but this information will be lost if it's detected at a different set of detectors (D1 or D2). If the signal photons are detected first, then you could use the 2-particle wavefunction for the signal/idler pair to calculate the probability distribution for just the signal photon to end up at different points on the screen; if the idlers are detected first, then according to the standard procedure for calculating probabilities this would cause a discontinuous collapse in the 2-particle wavefunction which would alter the probabilities of the signal photon landing at different points on the screen. However, I presume that if you calculate P(signal photon detected at position X|idler detected at detector D1)*P(idler detected at detector D1) + P(signal photon detected at position X|idler detected at detector D2)*P(idler detected at detector D2) + P(signal photon detected at position X|idler detected at detector D3)*P(idler detected at detector D3) + P(signal photon detected at position X|idler detected at detector D4)*P(idler detected at detector D4), then you should get almost exactly the same answer as you'd get for P(signal photon detected at position X) in the case where the signal photon is detected first so there was no prior state reduction in the 2-particle wavefunction caused by the detection of the idler (I say almost exactly because there may be cases in which the idler just misses all 4 detectors, but you should be able to make it exact by adding additional detectors so that there was no possible direction the idler could go without hitting a detector).

No interference pattern is required, JesseM. Contact me via e-mail if you’d like to discuss the experiment.

Required for what, exactly? I'm specifically asking for an experiment that could falsify your claim that the atoms are not in a superposition after traveling through the SG device, not one that you think would "prove" that claim (if you're thinking of the latter type of experiment, it may be that your idea of proof is based on a misconception, and that in fact orthodox QM would predict exactly the same result that you predict even if we assume the particle does remain in a superposition after passing through the device, with the collapse not occurring until we actually detect its position). So, are you in fact thinking of an experiment that would falsify your claim, and that would provide evidence that the atom was in a superposition after it passed through the SG device?

 

[QuantumBend]

I believe the crux of the matter, Debra, is whether the S-G magnet is the measuring apparatus which reduces the particle’s spin superposition to a single eigenvalue. I’m convinced that no detector screen is necessary.

DocMike

OK, so particle not interfere with itself after go out from magnet field. Screen show no interference. You want proove this - yes / no?

 

[msumm21]

Back to an earlier discussion within this thread regarding the Dopfer experiment. I wanted to clarify why I would be surprised if the pattern seen by Bob after performing the experiment a million times were the same, independent of the location of Alice's detector D1. Here are some definitions similar to the ones given before:

D1 at focal plane:
A1 - Probability distribution function A1 is revealed behind the double slit when D1 fails to register (failure to register occurs with probability P1).
B - Probability distribution function B is revealed behind the double slit when D1 registers a photon (as colorspace pointed out this may be a superposition of numerous interference patterns).

D1 at imaging plane:
A2 - Probability distribution function A2 is revealed behind the double slit when D1 fails to register (failure to register occurs with probability P2).
C - Probability distribution function C is revealed behind the double slit when D1 registers a photon.

Note: initially I said A=A1=A2, but it was pointed out by colorpsace that A1 and A2 are indeed different.

Now, given that A1 and A2 are different, and evidentially B and C are different, why must it be true that P1*A1+(1-P1)*B (the distribution when D1 is at the focal plane) is the same as P2*A2+(1-P2)*C (the distribution when D1 is at the imaging plane)? For B, C, P1, and P2 to take on values in just the right way to make the equality hold seems too remarkable.

Also, I believe someone said these results (overall net probability distribution functions) were not given in Dopfer's thesis. Is that true (I can't read it unless there's an English version)? I would think that this would be THE most interesting result of the experiment, unless there is already some theory explaining why P1*A1+(1-P1)*B must be exactly equal to P2*A2+(1-P2)*C.

 

[Dbar_x]

Can you elaborate on why you think quantum erasure would require "reversible measurement" or mixtures evolving from superpositions?

Marlan Scully is the father of quantum erasure. His mathematical analysis from 1978 with Shea and McCullen is the genesis of all the QE experiments that have followed. So, let me quote from an article by Scully and Walter, “An Operational Analysis of Quantum Eraser and Delayed Choice,” Found. of Phys. 28, 3, 1998. On p. 400 they say, “...if we put a Welcher Weg detector in place (so we lose interference even if we don’t look at the detector) and then erase the which way information after the particles have passed...such a ‘quantum eraser’ process (would) restore the interference fringes.” (Notice that in this instance QE is described as observer independent, since the detector is responsible for destroying interference whether or not any person is looking.)

The detector is a measurement apparatus and the measurement destroys the interference. Measurement implies a single real eigenvalue for the property observed; which path in this case. So, the density matrix is diagonalized. Von Neumann showed that diagonalization of the density matrix is a thermodynamically irreversible process. But, by erasing the information now stored in the apparatus, QE is supposed to restore interference. That’s a reversal of the measurement.

this information will be lost if it’s detected at a different set of detectors (D1 and D2).

As we know, the wavefunction predicts the probability that the idler photon might be absorbed at either D3 or D4 (or D1 or D2). If the idler is not absorbed at D3 or D4 (but at D1 or D2 instead) no information about which way the signal photon went is recorded at a detector. The which way information isn’t erased or lost, it never was.

If the signal photons are detected first

In the Kim, ... Scully experiment the signal photons are always detected first.

if the idlers are detected first

The idler photons are never detected first in this experiment. The authors say, on the third page, that “...photon 2 must be at least 7.7 ns later than the registration of photon 1.”

I'm specifically asking for an experiment that could falsify your claim that the atoms are not in a superposition after traveling through the SG device

You’re asking the wrong person for such an experiment. I know of no experiment that will do it. There is, however, a simple experiment that will refute or falsify the widely-held belief (hypothesis) that the particle remains in its spin-direction superposition through the magnetic field.

So, are you in fact thinking of an experiment that would falsify your claim, and that would provide evidence that the atom was in a superposition after it passed through the SG device?

No. For what I think are obvious reasons, I’m not willing to discuss details of the experiment in this public forum.

OK, so particle not interfere with itself after go out from magnet field. Screen show no interference. You want proove this - yes / no?

Essentially, yes.

 

[JesseM]

Marlan Scully is the father of quantum erasure. His mathematical analysis from 1978 with Shea and McCullen is the genesis of all the QE experiments that have followed. So, let me quote from an article by Scully and Walter, “An Operational Analysis of Quantum Eraser and Delayed Choice,” Found. of Phys. 28, 3, 1998. On p. 400 they say, “...if we put a Welcher Weg detector in place (so we lose interference even if we don’t look at the detector) and then erase the which way information after the particles have passed...such a ‘quantum eraser’ process (would) restore the interference fringes.” (Notice that in this instance QE is described as observer independent, since the detector is responsible for destroying interference whether or not any person is looking.)

But presumably any calculation that shows that erasing the information from the detector would "restore the interference fringes" depends on modeling the detector's interaction with the particle as creating an entangled detector/particle system rather than the detector collapsing the wavefunction.

The detector is a measurement apparatus and the measurement destroys the interference. Measurement implies a single real eigenvalue for the property observed; which path in this case. So, the density matrix is diagonalized. Von Neumann showed that diagonalization of the density matrix is a thermodynamically irreversible process. But, by erasing the information now stored in the apparatus, QE is supposed to restore interference. That’s a reversal of the measurement.

Have you studied at the actual mathematical analysis of how erasing the information in the apparatus is supposed to restore interference? Again, I would assume that such an analysis would not involve the assumption that the detector caused a collapse of the wavefunction when it made the original measurement, but would instead model the detector as becoming entangled with the particle it measured, would then show that if you don't assume the detector's state is measured (and its wavefunction collapsed) until after the which-path information has been erased, in this case you can show there will be interference effects in the system. Think about how the situation would be analyzed if you treat the idler as a miniature "detector" which carries information about which path the signal photon went through, but where the idler's information can later be erased if it is measured in a certain way--I'm pretty sure you don't assume that when the signal photon passes through the slits, the which-path information potentially available in the idler causes an immediate collapse of the 2-particle wavefunction!

As we know, the wavefunction predicts the probability that the idler photon might be absorbed at either D3 or D4 (or D1 or D2). If the idler is not absorbed at D3 or D4 (but at D1 or D2 instead) no information about which way the signal photon went is recorded at a detector. The which way information isn’t erased or lost, it never was.

But that's just the sort of situation the words "quantum eraser" are used for! What's "erased" if the idler ends up and D1 or D2 is the potential that existed at earlier times to gain the which-way information by measuring the idler in the right way. After all, if you really want to know the which-way information immediately after the signal-idler pair are created, you can always guarantee that the idler will end up at a detector that gives you the which-way information by altering the experimental setup at that moment--in the diagram of the experiment in Scully's paper, just imagine removing the two beam-splitters BSA and BSB, in which case no photons will be deflected towards D1 or D2, they'll all go to D3 or D4. By making the path length of the idler long enough in comparison to the path length of the signal photon, you could even set it up so you didn't have to decide whether you wanted to have a guarantee of getting which-path information (by putting nothing in the spot of BSA and BSB so all the idlers go to D3 or D4) or if you wanted to have a guarantee of no which-path information (by putting mirrors in the spots of BSA and BSB so all the idlers go to D1 or D2) until after you'd already detected the signal photon at D0. If you make the latter choice, I'd assume Scully would deem this a choice to "erase" the which-path information.

Similarly, if you could have a macroscopic detector which was completely isolated from interactions with the external environment (decoherence) that might carry away information about its state (a practical impossibility similar to Schroedinger's cat), and the detector worked in such a way that if you waited long enough the information would be unavailable (imagine Schroedinger's cat makes a measurement and writes it down, then you have the option to either open the box immediately or else wait a few trillion years until the cat and everything else in the box have gone to a state of maximum entropy and become some sort of featurless liquid/gas combination), then if you do in fact wait, one could say as you did above that "the which way information isn't erased or lost, it never was". It's really just a question of what you mean by the words "erased", but whatever you say about this situation, you should say the same about the idler which also contains the potential to give you the which-way information at earlier times but where it may not give you any which-way information if you wait until later times.

I'm specifically asking for an experiment that could falsify your claim that the atoms are not in a superposition after traveling through the SG device

You’re asking the wrong person for such an experiment. I know of no experiment that will do it. There is, however, a simple experiment that will refute or falsify the widely-held belief (hypothesis) that the particle remains in its spin-direction superposition through the magnetic field.

That doesn't really make sense to me. If there are two alternative possibilities (in this case, either the atoms are in superposition or aren't after passing through the SG device), then the only way to "falsify" either possibility is to show that they yield different predictions about a given experiment, and then demonstrate one prediction is confirmed while the other isn't. If you don't even know what an advocate of the "atoms do remain in superposition after passing through the SG device" view would predict about your own experiment, how can you possibly say that if the experiment has the results you predict it'd falsify this view? Maybe you're just thinking about it wrong and an advocate of the superposition view would actually predict the same thing you're predicting. And if you do know what they'd predict and it's different from what you predict, then if the actual results matched their prediction but not yours, wouldn't that be a falsification of your own view that they don't remain in superposition?

No. For what I think are obvious reasons, I’m not willing to discuss details of the experiment in this public forum.

The reasons are not obvious to me, you could probably get a lot more useful feedback than you will with your current cryptic approach. If you're afraid you'd run afoul of the rules against challenging mainstream theories, you can just present the experiment as a way of distinguishing the two views on superpositions without making any confident predictions about what the results would be, or asking for feedback about what orthodox QM would predict about your experiment. If you're afraid someone will steal your idea and take credit, then I don't really understand why it would be any better to share it with anonymous users via PM as you have offered to do, and in any case a public post outlining your idea would show that you were the first to publicly propose this experiment.

 

[debra]

No. For what I think are obvious reasons, I’m not willing to discuss details of the experiment in this public forum.

Essentially, yes.

But presumably any calculation that shows that erasing the information from the detector would "restore the interference fringes" depends on modeling the detector's interaction with the particle as creating an entangled detector/particle system rather than the detector collapsing the wavefunction.

Well I kind of agree with DBar_x but some of the details get lost on me.
I am re-reading them.

In this well explained but sensationally written offering:
http://grad.physics.sunysb.edu/~amarch/

I am question the conclusions. See the well known diagram:

http://grad.physics.sunysb.edu/~amarch/PHY5657.gif

Isn't it true that a proportion of the photons passing through
both quarter wave plates will have the same polarization - by
probability? - and will interfere normally? Then the filter
labelled Polarizer examines those only. Not exotic at all? I
may be wrong here and its all wrapped up and proved.


Has this been done with electrons to check the erasure as polarizers can
throw confusing results? Can we believe what they claim in this
reference above? (photons sending secret messages about filters etc).
I am to be convinced - but maybe its all accepted and wrapped up?

The SG discussion seems similar to me, but maybe I am off topic.

 

[zonde]

Can we believe what they claim in this
reference above? (photons sending secret messages about filters etc).

If you consider that optically active devices can send electromagnetic "signatures" along the same path that photons go (only in opposite direction) down to BBO then you will have more sensible picture than these secret messages.

Of course it is not what QM says.

 

[debra]

If you consider that optically active devices can send electromagnetic "signatures" along the same path that photons go (only in opposite direction) down to BBO then you will have more sensible picture than these secret messages.

Of course it is not what QM says.

Quantum erasure experiments often speak about backwards in time messages and I was being sarcastic about 'secret messages' even though they actually claim this in the above reference. I am questioning the experimental set up and analysis of results and am hoping the SG discussion in this thread will throw some clarity on these entanglement and beam splitting questions without resorting to FTL and similar exotics.

 

[JesseM]

Isn't it true that a proportion of the photons passing through
both quarter wave plates will have the same polarization - by
probability? - and will interfere normally? Then the filter
labelled Polarizer examines those only. Not exotic at all? I
may be wrong here and its all wrapped up and proved.

According to the chart on the page underneath the diagram you posted, it's true that some photons going through slit 1 have polarization R, and some photons going through slit 2 also have polarization R...but the ones that have polarization R after going through slit 1 are all matched to entangled p photons that had polarization x, while the ones that have polarization R after going through slit 2 are all matched to entangled p photons that had polarization y. So by matching up each s photon at detector Ds with its entangled p photon at detector Dp, you should always be able to tell which of the slits the p photon went through, so there should be no interference. And you can see on the graph underneath the chart that when they actually performed this experiment and looked at the data, they didn't see an interference pattern:

http://grad.physics.sunysb.edu/~amarch/PHY5658.gif