Hi Folks,I am currently reading Brian Greene's 'The Fabric of the

[bugatti79]

Hi Folks,

I am currently reading Brian Greene's 'The Fabric of the Cosmos' which is excellent.

I am trying to get my head around the Delayed Choice Quantum Eraser in this book. Is there a website link I could follow up that explains it well? I have already looked at

http://en.wikipedia.org/wiki/Delayed_choice_quantum_eraser

Perhaps there is some applet one can download?

Thanks

bugatti

 

[Naty1]

Getting your "head around" THAT discussion may be impossible...depending on your objective.

These experiments are a magnificent affront to our conventional notions of space and time...by any classical - commonsense- reckoning, that's (the results are) well, crazy...

Greene, Page 199...

There is another description here:

http://en.wikipedia.org/wiki/Quantum_erasure

but don't expect it to make anymore common sense than Greene's discussion.

It's akin to special relativity: Either you accept the experimental results which support the theory ( in SR, that space and time are not fixed and immutable) and use those results and theory in your "new" reality via new ways of thinking or you don't. You will need to accept a new "logic".

I've read that book two or three times now and still find new perspectives upon subsequent reading and discussions in these forums. If this stuff was obvious, everybody would be a quantum theorist!

 

[bugatti79]

Thanks Naty1,

I have printed that link ready for attack later. I haven't the book with me right now but I will indicate which part I don't understand and will reply later.

Cheers

 

[bugatti79]

I have read through some material and I find it mind boggling the weirdness of these experiments.

What amazes me is the very fact of having 'which path' information on the idler photons and then 'erasing' this information will still recover the interference pattern despite the fact that we have not gone near the signal photon.

The nature of the photon appears to depend on what the experimental setting is prior to hitting the screen regardless of whether the photon is a particle or a wave 'after' leaving the down converter.

This interpretation is correct, right?

Thanks

 

[xts]

I am not sure if you got properly what happens in this experiment. Many (most...) people misunderstand it.

If you have quartervawe plates in a lower branch, we never see the pattern on the screen. The photon distribution is uniform (or rather wide blob). The pattern appears only in a correlation between upper (eraser) and lower (double-slit) detectors. You may picture it as the blob you see on a lower screen is a sum of two fringed patterns, displaced by half of the fringe span. Photons creating one pattern are perfectly correlated with photons vertically polarized in the upper branch. Those making second pattern - are correlated with photons hororizontally polarized in upper branch.
Both cases you may explain with simplest high-school wave optics. No mystical erasure is needed, no action on distance, no mysterious particle-wave duality, no 'nature of photons'...

The only remaining mystery is entanglement: however you measure polarisation of photons in both branches, the result is always the same if axes of both polarisators are parallel.

To make it even easier: use normal, untangled light, and put your 'eraser' polariser just by the source. Then analyse what pattern would you see on the screen, if QWP are installed behind slits. Easy. Turn around your polariser.
What entanglement does is the same: correlation between entangled particles ensures that whenever upper detector receives vertically polarized photon, the one in lower branch is also vertical. So restricting the analysis to those events, when upper detector clicks with vertical photon, is equivalent to putting a vertical filter by the source in lower branch.

 

[bugatti79]

I am not sure if you got properly what happens in this experiment. Many (most...) people misunderstand it.

If you have quartervawe plates in a lower branch, we never see the pattern on the screen. The photon distribution is uniform (or rather wide blob). The pattern appears only in a correlation between upper (eraser) and lower (double-slit) detectors. You may picture it as the blob you see on a lower screen is a sum of two fringed patterns, displaced by half of the fringe span. Photons creating one pattern are perfectly correlated with photons vertically polarized in the upper branch. Those making second pattern - are correlated with photons hororizontally polarized in upper branch.
Both cases you may explain with simplest high-school wave optics. No mystical erasure is needed, no action on distance, no mysterious particle-wave duality, no 'nature of photons'...

The only remaining mystery is entanglement: however you measure polarisation of photons in both branches, the result is always the same if axes of both polarisators are parallel.

To make it even easier: use normal, untangled light, and put your 'eraser' polariser just by the source. Then analyse what pattern would you see on the screen, if QWP are installed behind slits. Easy. Turn around your polariser.
What entanglement does is the same: correlation between entangled particles ensures that whenever upper detector receives vertically polarized photon, the one in lower branch is also vertical. So restricting the analysis to those events, when upper detector clicks with vertical photon, is equivalent to putting a vertical filter by the source in lower branch.

Hi xts,

Unfortunately, you have gone beyond my capacity as a layperson at this stage. So is it possible that Brian Greene's description is too simplistic such that it invites misinterpretations of this experiment?

cheers

 

[xts]

I don't know Greene's book, so I can't judge it. But virtually all popular text on this issue I ever read were misleading.
It is not a matter of oversimplification. Rather contrary - it is building mysterious structure where it really is not needed.
As Occam's advocate I always prefer simpler explanations...
I also don't like sensational language used in this field: 'teleportation', 'erasure' (working on a distance), etc. - it is awfully misleading especialy for laypeople.

 

[xts]

you have gone beyond my capacity as a layperson at this stage.

OK. So once again:
Imagine you have just a source of light, then linear polarisator, then double-slit with QWP installed behind each slit, then screen - as in QE experiment. If there is no polarisator - there is no pattern on the screen (large blob).If you put the polarisator and turn it axis, fringes appear.
Let's say they appear as the polarisator is set to vertical. As you turn polarisator to horizontal, fringes appear again - but shifted: now you have light in places previously dark and vice versa.
To explain this behaviour you don't need QM - just 19th century wave optics.
So you know, that if incoming light is polarised vertically you have pattern, if it is polarised horizontally - you have antipattern.
Now you remove the polarizator. Pattern vanishes.
You put polarizator into upper branch (eraser branch). Nothing happens - you still see no pattern.
But now you start to count individual photons in both branches. You got a long list of pairs: (was the photon in upper branch vertical? | photon position on the screen in lower branch)
And you take only those cases, where in upper branch the photon was vertical, and plot a dot on the lower screen in the place hit by photon. But only for those cases, when upper one was vertical. Those dots form a pattern - exactly the same, as you saw when you put vertical filter in the lower branch.

Nothing mysterious, weird, spooky, no 'nature of photons' is needed...

The only remaining mystery is that this special crystal, sending two photons at once, always send identically polarised photons in both branches. (actually entanglement involves deeper correlation, but for now that's allowable simplification - let DrChinese forgive me!)

 

[unusualname]

For a start go with Greene's discussion and not xts

xts seems to believe in "elements of reality", ie he believes photons have a definite value of a property such as polarisation before it is measured.

He must have missed the last several decades of experimental physics which shows this is not the case.

People like Wheeler, Aspect and Zeilinger aren't idiots.

The point of modern quantum optics experiments is that they cannot be explained by "19th century wave optics", and anyone who keeps suggesting this here, on physicsforums, ought to be warned methinks.

Once you accept a QM explanation it is simple enough to see that no retrocausality occurs (eg Demystifying the Delayed Choice Experiments )

In fact the experiment is more confusing if you try to understand it using classical concepts (unless you make up your own rules of nature)

But don't be surprised that this stuff is "mystifying", no one really understands QM, for an authoritative overview see a review article like Franck Laloe's

Do we really understand Quantum Mechanics?

 

[xts]

xts seems to believe in "elements of reality", ie he believes photons have a definite value of a property such as polarisation before it is measured.

Oooch? Do I? I think you totally misunderstood me.
What I am trying to advocate in this and several parallel threads is an extreme 'non-realistic' approach. I am trying to convince people that it just makes no sense to even think about such 'entities' as 'photon path', 'photon nature', property existence when not measured, etc...

I have much less realistic approach than Wheeler had! Nothing against Aspect and Zeilinger...

 

[unusualname]

Oooch? Do I? I think you totally misunderstood me.
What I am trying to advocate in this and several parallel threads is an extreme 'non-realistic' approach. I am trying to convince people that it just makes no sense to even think about such 'entities' as 'photon path', 'photon nature', property existence when not measured, etc...

I have much less realistic approach than Wheeler had! Nothing against Aspect and Zeilinger...

Your explanation above sounds pretty realist to me. 19th century wave optics assumes classical realism you know.

 

[xts]

Your explanation above sounds pretty realist to me. 19th century wave optics assumes classical realism you know.

19th century wave optipcs is essentially equivalent to QM. It is absolutely non-realistic! There are no photons, paths, nothing like that. Just a propagation of waves. No particles carrying any properties.

 

[DrChinese]

I have much less realistic approach than Wheeler had! Nothing against Aspect and Zeilinger...

I wouldn't say any of these are realists!

 

[bugatti79]

ok guys,

Thank you for your inputs, I appreciate it. I will look at those links. Cheers

 

[bruce2g]

Actually, the dcqe in Greene's book isn't the one with 1/4 wave plates, it's the other one, where a beam splitter 'erases' the which-path information in the idler photons.

It's important to note that there is a classical connection (a wire) between the idler detectors and signal detectors, because you need to do coincidence counting to reveal the interference pattern when the which-path information is 'erased.' Greene mentions this requirement in the text, but it is missing from his diagram of it, so some people have gotten the (erroneous) impression that you could use this to signal information faster than the speed of light.

Basically, without the final beam splitter, all you get is a big smudge of photons. With the beam splitter, the signal photons show a set of fringes associated with idlers leaving the splitter to the left, and another set of fringes associated with the idlers going to the right.

 

[moving-finger]

Greene's book is indeed excellent.

Best "explanation" of the actual DCQE experiment I have found so far is http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm

One question I have which someone here may be able to help me with, as follows:

The raw position data (without correlation with any of the other detectors) collected at D0 should show an interference pattern, correct?

The various sets of correlated data (obtained via the coincidence counter) are D0-D1, D0-D2, D0-D3 and D0-D4 (depending on whether the idler photon ends up at D1, D2, D3 or D4) - but each set of correlated data is a subset of the raw D0 data - correct?

In other words, if we add up the position data from all of the subsets, we should end up with the raw (uncorrelated) distribution, which should show an interference pattern?

But D0-D3 and D0-D4 correlation data show no interference pattern, and the interference patterns from D0-D1 and D0-D2 are out of phase, so adding them would destroy the pattern.

So how can we add together all the subsets of data and still end up with a raw data interference pattern?

I'm confused. can anyone help me?

 

[xts]

The raw position data (without correlation with any of the other detectors) collected at D0 should show an interference pattern, correct?

Incorrect!
That's very common misunderstanding of the "quantum eraser" and similar experiments.
The full set of data, as collected, never shows any pattern.
The data may be just divided into two subsets, each of them exhibiting a pattern (they differ in phase), using data collected in other branch as a selection trigger.
No timing issues need to be considered. You may just collect independently two series of data in both detectors, store them and compare them offline a week later to produce fringed plots.

 

[moving-finger]

Incorrect!
That's very common misunderstanding of the "quantum eraser" and similar experiments.
The full set of data, as collected, never shows any pattern.

Thanks for the reply, but not sure I follow you.

The raw signal photon data at D0 (prior to any correlation of these data with idler photon detection data) MUST surely show an interference pattern - because this (the raw data at D0, without any other inputs) is precisely the standard 2-slit experiment, with no which-path information.

Indeed, this is exactly what the following website claims:

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

QM predicts that if which-path information is not available at the time of measurement, the pattern will be an interference pattern, as though wave-like photons passed through both slits and "interfered with themselves" to produce the distinctive interference pattern of hits. This is the case at detector D0 at all times.

 

[xts]

The raw signal photon data at D0 (prior to any correlation of these data with idler photon detection data) MUST surely show an interference pattern - because this (the raw data at D0, without any other inputs) is precisely the standard 2-slit experiment, with no which-path information.

That's not true.
Pure D0 (without any coincidence) is a double slit with each slit lit by independent incoherent source. Or - if you prefer - it is a double slit with spatially extended incoherent light source.

 

[moving-finger]

That's not true.
Pure D0 (without any coincidence) is a double slit with each slit lit by independent incoherent source. Or - if you prefer - it is a double slit with spatially extended incoherent light source.

Now you have me really confused. If the source illuminating the double slit is incoherent, how can it ever result in observation of an interference pattern, either with or without inclusion of the coincidence-counter data?

 

[xts]

Take most classic, over 200 years old, Young's experiment.

Wide source - double slit - screen: no pattern, as the source is incoherent (regarding the angle)

Wide source - single slit - double slit - screen: you see the pattern, as the first single slit selects only part of the light (falling at the specific angle) - thus making it coherent (angularily - that's only property important in this experiment). If you shift the single slit, then the pattern remains, but also shifts.

Here you have similar case: the trigger (coincidence) branch is used to select only light incoming at some specific angle.

The only difference is an order. Young first chose a subset of photons (destroying rest of them), and then observed a pattern. As he chose other subset - he observed shifted pattern. As he removed the single slit, the shifted patterns mixed together making bulky spot.
Here you have the same - you just select a subset of the light using coincidence counter, but you may apply this selection trigger at any time (also after collecting the data).

 

[moving-finger]

Wide source - double slit - screen: no pattern, as the source is incoherent (regarding the angle)

Wide source - single slit - double slit - screen: you see the pattern, as the first single slit selects only part of the light (falling at the specific angle) - thus making it coherent (angularily - that's only property important in this experiment).

I thought the source illuminating the slits in the DCQE setup was a laser? In which case it should be coherent (regarding the angle)?

Again, I would refer you to the following:
http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm

Cheers

 

[xts]

I thought the source illuminating the slits in the DCQE setup was a laser? In which case it should be coherent (regarding the angle)?

No. That would be true, if our detector D0 would register the light of the laser wavelength, rather than twice bigger.
The source illuminating the slits was not a laser, but one of the pair of photons, created during downconversion in the crystal. The angle of this photon is not well defined (it is bounded by some pretty wide cone). So it is equivalent to illuminating the double slit by angularily large (as seen by slits) source.

We may know the angle measuring the angle of other photon of the entangled pair (as their angles are perfectly correlated).

 

[Cthugha]

I thought the source illuminating the slits in the DCQE setup was a laser? In which case it should be coherent (regarding the angle)?

Again, I would refer you to the following:
http://www.bottomlayer.com/bottom/kim-scully/kim-scully-web.htm

Cheers

A: No, it is down-converted light from a laser which is about as incoherent as it gets.

B: Bottomlayer is a well-known crackpot site. However, the paper cited there is not. Please read the paper instead of that nonsense crackpot site that claims that consciousness has an influence on the experimental results.

edit: Oh, XTS was faster.

 

[moving-finger]

The source illuminating the slits was not a laser, but one of the pair of photons, created during downconversion in the crystal. The angle of this photon is not well defined (it is bounded by some pretty wide cone). So it is equivalent to illuminating the double slit by angularily large (as seen by slits) source.

The setup I am referring to is the Kim et al setup as described here:

http://lanl.arxiv.org/abs/quant-ph/9903047v1

In this setup, the slits are placed immediately after the laser, and before the crystal.

Hence the source impinging on the slits is coherent, hence the photons emerging from the slits should show interference?

Understood the angle at which the photon emits the crystal may vary, but the crystal is placed AFTER the slits, and only photons which emit the crystal at the "right" angle will reach D0?

 

[moving-finger]

B: Bottomlayer is a well-known crackpot site. However, the paper cited there is not. Please read the paper instead of that nonsense crackpot site that claims that consciousness has an influence on the experimental results.

I can accept that Bottomlayer is a crackpot site - is there a non-crackpot site which explains the DCQE better?

Thanks

(ps - I did read the original paper, but its very short on ontological explanation)

 

[xts]

Hence the source impinging on the slits is coherent, hence the photons emerging from the slits should show interference?

Your D0 branch does not utilize interference. As the detector is placed in focal plane of the lens, the image (pattern) detected by D0 reflects just and only the angular direction of the incoming photons. There is no interference at all in this branch.

is there a non-crackpot site which explains the DCQE better? (ps - I did read the original paper, but its very short on ontological explanation)

I doubt. As all the ontology attributed to QE is crap or at most overinterpretation of pretty simple phenomenon...
As I've already shown in the other thread (https://www.physicsforums.com/showthread.php?t=521715), the QE may be fully explained in terms of 19th century optics. It does not utilize even Bell's inequality violations. It uses entanglement just to create pairs of photons correlated by single parameter. Like shooting a series of pairs of balls: green in one direction, red in another. There is no more ontology behind DCQE than behind Young's experiment.BTW. Boys finally built the Quantum Eraser at Home. It works fine!

 

[Cthugha]

Hence the source impinging on the slits is coherent, hence the photons emerging from the slits should show interference?

How does the exact position where coherence is spoiled matter? If you use coherent light in a common double slit experiment, but spoil coherence by placing e.g. some random scattering medium like a rotating teflon disk between the slits and the detectors, the interference pattern will also vanish.

 

[moving-finger]

Your D0 branch does not utilize interference. As the detector is placed in focal plane of the lens, the image (pattern) detected by D0 reflects just and only the angular direction of the incoming photons. There is no interference at all in this branch.

The angular direction of the incoming photons, combined with their phase, is surely what gives rise to the interference pattern in the first place? The image at D0 surely reflects the intensity of the photons in that particular direction - which intensity is determined by diffraction at the slits, hence interference from the slits?

I doubt. As all the ontology attributed to QE is crap or at most overinterpretation of pretty simple phenomenon...
As I've already shown in the other thread (https://www.physicsforums.com/showthread.php?t=521715), the QE may be fully explained in terms of 19th century optics. It does not utilize even Bell's inequality violations. It uses entanglement just to create pairs of photons correlated by single parameter. Like shooting a series of pairs of balls: green in one direction, red in another. There is no more ontology behind DCQE than behind Young's experiment.


BTW. Boys finally built the Quantum Eraser at Home. It works fine!

Thanks, I will check these links too.

 

[moving-finger]

How does the exact position where coherence is spoiled matter? If you use coherent light in a common double slit experiment, but spoil coherence by placing e.g. some random scattering medium like a rotating teflon disk between the slits and the detectors, the interference pattern will also vanish.

So... in the DCQE, we are saying that coherence is actually (really, genuinely) "spoiled" by the crystal, such that no interference pattern at all can be observed; and yet somehow this "spoiled" coherence can be magically regenerated, and the interference pattern also regenerated, simply by subsequent detection of idler photons?

Where there was no coherence before, suddenly there is coherence, simply by virtue of retroactive detection of the idler photons? Is that the idea?

 

[xts]

The angular direction of the incoming photons, combined with their phase, is surely what gives rise to the interference pattern in the first place? The image at D0 surely reflects the intensity of the photons in that particular direction - which intensity is determined by diffraction at the slits, hence interference from the slits?

No.
The distribution of photons at D0 plane depends solely on angular direction of the photon leaving the crystal. Their phase (as well as other parameters, like polarisation) are ignored by this experiment. And those photons have an uniform (or rather blobby) angular distribution.

Where there was no coherence before, suddenly there is coherence, simply by virtue of retroactive detection of the idler photons? Is that the idea?

Exactly.
Incoherent light is a mixture of photons having some property (in our case angular direction) of different values. We may make it coherent by selecting only those photons, which have desired value of this parameter. It does not matter if we perform the selection first or retroactively.

Young did that inserting a single slit between source and dual-slit. QE does this retroactively/off-line/on-line by selecting only those hits, which correspond to proper outcome of the counterpart photon, which has always its angle correlated to our one.

 

[moving-finger]

The distribution of photons at D0 plane depends solely on angular direction of the photon leaving the crystal. Their phase (as well as other parameters, like polarisation) are ignored by this experiment. And those photons have an uniform (or rather blobby) angular distribution.

Are you saying there can be no interference pattern whatsoever observed at D0?

For any photon to leave the crystal and be detected at D0 (for any given position of D0), the photon must have a specific angle of emission. The configuration of D0 plus the two "points of emission" on the crystal defines an angle of emission and surely allows for an interference pattern?

Incoherent light is a mixture of photons having some property (in our case angular direction) of different values. We may make it coherent by selecting only those photons, which have desired value of this parameter. It does not matter if we perform the selection first or retroactively.

For any given position of crystal and D0, the angular direction to D0 is well-defined - whether we detect idler photons or not - hence interference should be observed (or not observed), whether we detect idler photons or not.

What difference does it make in this setup whether we detect idler photons or not?

 

[xts]

For any photon to leave the crystal and be detected at D0 (for any given position of D0), the photon must have a specific angle of emission.

True. (Actually not quite true - remember about uncertainity principle - but for our experiment it is allowable assumption)

The configuration of D0 plus the two "points of emission" on the crystal defines an angle of emission and surely allows for an interference pattern?

False. You forgot about lens in the middle, which do not project the image of the slits, but is focused at infinity - projects the angle. Slits are close to each other comparing to detector size, so it may be considered to be a point-like source, emittin angularily distributed light, which is then projected on a screen D0 plane) by out-of-focus lens.

For any given position of crystal and D0, the angular direction to D0 is well-defined - whether we detect idler photons or not - hence interference should be observed (or not observed), whether we detect idler photons or not.

False! You forgot about lens.
True! It is never observed. There is no interference pattern on D0 plane. The pattern appears not on the screen, but on computer printouts, as resulting from combination of the observed blob with selection strategy, making it fringed.

 

[Cthugha]

Just a comment: If you are really interested in this kind of experiments, have a look at Walborn et al., "Spatial correlations in parametric down-conversion", Physics Reports 495, 116 (2010).
Also available freely at arxiv:http://arxiv.org/abs/1010.1236"

This discusses these experiments in depth (see e.g. the part on conditional interference), but it is a rather long review article and you need to examine some of the math involved yourself to get a feeling for the possible regimes of single photon interference and conditional interference, so it takes some endurance to get something out of this paper.