Mandel & Faster than Light Communication ?

[ShalomShlomo]

In Mandel et Al's most famous experiment (Fig 6 at ‎[URL]http://student.science.nus.edu... one-photon and two-‎photon interference.pdf) the signal beams from two coherent downconverters are ‎observed to interfere only if the two corresponding idler beams are allowed to ‎interfere.‎

Let's say that the a blocking object is inserted or removed in front of idler beam 1 at ‎point A. Let us also say that the signal beams are detected at point B.‎
The direct distance between A and B is x light seconds, and the distance traveled by ‎the idler 1 light beam from the 1 downconverter plus the distance traveled by the ‎signal 1 beam from the 1 downconverter to B is y light seconds.‎

When the object is inserted/removed at A, how long does it take for the interference ‎pattern at B to disappear/appear ?‎

Is it
‎1) Instantaneously
‎2) x seconds later, or‎
‎3) y seconds later ?‎

‎(I know that the detector needs to move back and forth to see interference patterns, ‎but assume it can move back and forth and record very quickly).‎

The above is question 1.‎

Question 2:‎
Since light travels at the speed of light, and from Relativity, something traveling at ‎the speed of light experiences simultaneously different times (unlike all less-than-‎light-speed particles/waves), does it make any sense to say that light communicates ‎and interferes with itself at what an outside inertial observer would call different ‎times, due to those different times being experienced by the light beam ‎simultaneously (ie can General Relativity's understanding of the light beams ‎experience of simultaneous time explain some of the Quantum results seen in ‎Question 1) ?‎
‎(and is there a good [very] basic paper which explains the link ?)‎

 

[werty]

The link didnt seem to work for me, but its sound like it might be the usual delayed choice experiment, which can't be used for faster then light communication.

Furhermore I was under the impression that things moving with the speed of light didnt experience time at all, since they follow delta tau = 0 lines, eg their clocks don't ever change, time is standin still for these things.

 

[ShalomShlomo]

That's not quite right.

Bell showed that communication does travel faster than light, but we just can't use that random-to-random communication to piggyback any information on.

Here's another URL to the same paper - See Fig 6:
http://www.google.com/url?sa=U&star...two-photon%20interference.pdf&e=9707[/PLAIN]
IFABOVEDOESN'TWORKTHENUSEwww.google.com/url?sa=U&start=1&q=http://student.science.nus.edu.sg/~g0203645/Atomic%2520Molecular%2520and%2520Optical%2520Physics/Quantum%2520effects%2520in%2520one-photon%2520and%2520two-photon%2520interference.pdf&e=9707

But in Mandel's now classically quoted experiment, blocking Idler beam 1 does cause a noticeable change elsewhere at the Signal detector where the interference pattern disappears.

What is the time gap between blocking the idler path, and noticing the signal interference pattern disappearing.

Is it

1] Instantaneous (like Bohm's pilot wave) - but this has been proven impossible by Eberhard in similar but not identical cases

2] The direct straight line distance between the cause and effect. Cause is the place where the idler beam is blocked, and effect is the signal beam detector (A & B in my previous post)

3] The actual distance traveled by the light beams between the cause and effect (which is longer than 2] as the light beams take a tortuous route in the experiment - this is the sum of two paths, the idler beam from downconverter to block, and the signal beam from downconverter to detector).

or 4] The distance between where the two idler beams would have first met each other on the one hand, and the effect on the other. This is similar to 3], but with an addition - ie this is the sum of two paths, the idler beam from downconverter to the point it would have first met the other idler beam, and the signal beam from downconverter to detector. After all, according to Mandel, it is the idler beams meeting each other and destroying latent information which allows the interference pattern to occur at the signal detector. (missed this one out the first time).

We can gain (can we not) significant understanding of the 'spooky' quantum effect by determining if the answer is 2), 3) or 4).

Can anyone help me understand which it is.

ShalomShlomo

 

[DrChinese]

2] The direct straight line distance between the cause and effect. Cause is the place where the idler beam is blocked, and effect is the signal beam detector (A & B in my previous post)

The effect you describe travels no faster than c. So case 2 is the correct answer.

It is easy to be confused by the description of the setup. Because it is fundamental that the signal and idler paths are co-mingled. You must also consider that the idler detector plays a role.

In all of the cases Mandel describes in which it appears something happens FTL, you will find that coincidences are an issue - so it is not FTL communication. In all of the cases in which the behavior does not require comparing results from signal and idler, the communication is at c or less.

 

[ShalomShlomo]

The effect you describe travels no faster than c. So case 2 is the correct answer.

It is easy to be confused by the description of the setup. Because it is fundamental that the signal and idler paths are co-mingled. You must also consider that the idler detector plays a role.

Thanks.

But NO on two counts.

1) Mandel has shown that the Idler detector plays no part in what occurs. It is only the theoretical threat of detection, irrespective of whether the detector is there or not, that makes the Signal interference pattern appear or disappear.

2) Bell showed that we live in a non-local universe. Somehow, entanglement is causing FTL communication, it is just a communication that we can never use to send information.

Was your answer 2 based on dogma, or do you have a model to explain why the answer should be 2.

I have a feeling the answer is one of the options other than 2 (2 makes least sense if you think about it).

Looking forward to your reply

ShalomShlomo

 

[DrChinese]

Thanks.

But NO on two counts.

1) Mandel has shown that the Idler detector plays no part in what occurs. It is only the theoretical threat of detection, irrespective of whether the detector is there or not, that makes the Signal interference pattern appear or disappear.

2) Bell showed that we live in a non-local universe. Somehow, entanglement is causing FTL communication, it is just a communication that we can never use to send information.

Was your answer 2 based on dogma, or do you have a model to explain why the answer should be 2.

I have a feeling the answer is one of the options other than 2 (2 makes least sense if you think about it).

Looking forward to your reply

ShalomShlomo

Well, look... This is a system. It is a bit difficult to artifically dissect it as you are attempting. Your "choices" are confusing, so I tried to pick one that came as close as possible. The blocking must occur prior to the detection for the interference pattern to be eliminated. Turning on and off the blocking does not allow you to change the pattern faster than c. So if you are trying to assert something specific, please just go ahead and say it.

The fact is that there is no known or demonstrable FTL signalling occurring in any of the experiments Mandel describes in this paper. It MIGHT be acceptable to say there is an FTL effect, although that is certainly debatable too. The standard dogma, of course, is that there is instantaneous collapse of the wave function relating the two photons. These experiments do not describe any new effects over and above what you could have predicted years ago. They do repackage it nicely so that variations on quantum behavior can be studied in more detail using the entangled photons.

 

[chronon]

I would say that the correct answer is (3). Although the system may be quantum in nature, the interference or lack of it can be explained in terms of wave optics.

 

[ShalomShlomo]

Turning on and off the blocking does not allow you to change the pattern faster than c.

Who says ? If you are you relying on Eberhard's proof, then is it valid to this case ? Aspect's experiment seems to indicate that your understanding doesn't match out what quantum reality has to offer.

The standard dogma, of course, is that there is instantaneous collapse of the wave function relating the two photons.

You have put your finger on the problem. Instantaneous for the two photons, which are at two non-identical positions (ie space divided), means FTL !
That was exactly Bell's 1964 revolution.

I would say that the correct answer is (3). Although the system may be quantum in nature, the interference or lack of it can be explained in terms of wave optics.

Sorry. No way. The world we live in is much stranger than that. Look at Mandel's experiments and results. Changing something at A (blocking one beam) changes another beam at B which has absolutely no connection by wave optics at all to A. This bears no relation to wave optics whatsoever. I wish the world we lived in was that simple.

So, does anyone have an answer from 1, 2, 3 or 4 ? Thanks if you do

ShalomShlomo

 

[werty]

Bell and Aspects exeperiments show (if you trust em, which I do) that "local reality" is nonexistent, they don't show that any signals travel faster then light, eg they don't show that there is a non-local reality just that there arent a combined "local reality", there might be a local nonreality :). Furthermore since these exeperiments (Mandell etc) all follow the laws of QM they must all obey the reduced density matrix idea, that is all that one can locally know about a system is encoded in the reduced density matrix, which in turn is independent of anything you do to the other parts of the system, since that part has been traced out. So if there is indeed FTL between the particles, which hasnt been shown to take place, it can never be used to send any information faster than light. Atleast according to QM, and relativity ofcourse. Since the relativity principle nowadays says that you can't send information faster then light. Signals can go FTL, but you can't use them to anything "usefull" =). Also its interessting to note that since we can't send information faster than light we can never observe or in otherways ascertain that particles communicate FTL, we can't catch em in the act so to speak. For if we could, we could use it to send information FTL.

 

[DrChinese]

Who says ? If you are you relying on Eberhard's proof, then is it valid to this case ? Aspect's experiment seems to indicate that your understanding doesn't match out what quantum reality has to offer.

Your Mandel question is completely equivalent to asking how fast an interference pattern disappears when one of the two slits is blocked in a traditional double-slit setup. There is absolutely no difference in how the interference occurs - it is a superposition of 2 possible paths to the signal detector.

Eberhard's proof has nothing to do with the matter that I can see. The question you are asking can be experimentally tested.

I do not follow the reference to Aspect. Aspect showed that Bell's Inequality is violated. What am I missing?

 

[DrChinese]

I would say that the correct answer is (3). Although the system may be quantum in nature, the interference or lack of it can be explained in terms of wave optics.

Reading the answers again, yeah maybe 3 is better. Anyway, as you say, this is essentially classical at this point and there is nothing weird going on between the blocking and the detection.

 

[ShalomShlomo]

Furthermore since these exeperiments (Mandell etc) all follow the laws of QM they must all obey the reduced density matrix idea

Assuming the reduced density matrix is a full and correct description of quantum mechanics - which it might not be.

Either way, which is the answer ? 1, 2, 3 or 4 ?

ShalomShlomo

p.s.: And I wish I could change the title of this thread. FTL communication was meant to imply, as in Bell's Proof, that the measurement of the wave-function at a point A simultaneously changes the wave-function of an entangled photon at some other space-separated point B, causing observable results. I know we can't send information on this change, but the change happens.

 

[DrChinese]

...the measurement of the wave-function at a point A simultaneously changes the wave-function of an entangled photon at some other space-separated point B, causing observable results. I know we can't send information on this change, but the change happens.

This is standard CI, so who is questioning this? And what does this have to do with the experiment you are referencing? Shouldn't we be talking about an Aspect-like experiment?

 

[vanesch]

But in Mandel's now classically quoted experiment, blocking Idler beam 1 does cause a noticeable change elsewhere at the Signal detector where the interference pattern disappears.

What is the time gap between blocking the idler path, and noticing the signal interference pattern disappearing.

Can you point out exactly which set up you are talking about ? Usually, in order to see "signal beam interference", you need to subselect a signal beam image by an idler detection in one way or another.

cheers,
Patrick.

 

[ShalomShlomo]

Your Mandel question is completely equivalent to asking how fast an interference pattern disappears when one of the two slits is blocked in a traditional double-slit setup.

Interesting analogy. But when blocking one of two slits, we understand that light already through the blocked slit continues to propagate to the detector, and interference-effects only disappear once the time taken for those waves to arrive have been reached. We have a well defined and well understood model for the disappearance of the interference pattern.

Reading the answers again, yeah maybe 3 is better. Anyway, as you say, this is essentially classical at this point and there is nothing weird going on between the blocking and the detection.

Classical !? Whilst we know the blocking of the idler beam propagates to the Signal detector, since we see the interference pattern disappear, we have no model to discover along which route or how soon the cause propagates to the effect ! And you are just guessing by your answers first of 2), and now of 3).

Can you come up with a model justifying answer 1, 2, 3 or 4 ?

ShalomShlomo

 

[ShalomShlomo]

Can you point out exactly which set up you are talking about ? Usually, in order to see "signal beam interference", you need to subselect a signal beam image by an idler detection in one way or another.

Vanesch, the experiment is pretty fully detailed at:

www.google.com/url?sa=U&start=1&q=[PLAIN]http://student.science.nus.edu.sg/~g0203645/Atomic%2520Molecular%2520and%2520Optical%2520Physi cs/Quantum%2520effects%2520in%2520one-photon%2520and%2520two-photon%2520interference.pdf&e=9707[/URL]
IFABOVEDOESN'TWORKTHENUSEwww.google.com/url?sa=U&start=1&q=http://student.science.nus.edu.sg/~g0203645/Atomic%2520Molecular%2520and%2520Optical%2520Physi cs/Quantum%2520effects%2520in%2520one-photon%2520and%2520two-photon%2520interference.pdf&e=9707

See Fig 6 and the text there.

ShalomShlomo

 

[werty]

Assuming the reduced density matrix is a full and correct description of quantum mechanics - which it might not be.

Either way, which is the answer ? 1, 2, 3 or 4 ?

ShalomShlomo

p.s.: And I wish I could change the title of this thread. FTL communication was meant to imply, as in Bell's Proof, that the measurement of the wave-function at a point A simultaneously changes the wave-function of an entangled photon at some other space-separated point B, causing observable results. I know we can't send information on this change, but the change happens.

I don't know what the time gap is, maybe we can say that detection of the path of the idler photon causes immediate collapse and so destroys interference, however I don't like to do that since I feel that it implies that the particles are little balls with FTL intercom. I don't know what they are and I don't think anyone do. We only know how they behave, what will happen in experiments (roughly). There is also the problem of determining when interference have disappeared, you can't do it with only one detection, can you do it with two !? I think that any answer to how long it takes for the interference to disappear has to do with ones idea of what is going on, so it hard to use these answers to figure out what is going on Eg. if you use somekind of wave model than the interference occurs only at the detector and thus the interference is destroyed when one of the waves fails to show up. If you use the particle model, than the interference can be said to be destroyed when the path of the idler photon is detected, collapse is immediate.

 

[werty]

One can also say that interference is destroyed at the moment you decide what experiment to do. This as DrChinese said where similar to the standard two slit experiment. People have tried every imaginable thing to try to fool the photon into giving up what path they took and still have interference. It seems the photon are smarter than us.

 

[vanesch]

See Fig 6 and the text there.

Ah, I didn't know that one. I have to say that I'm surprised about the result, and I have some difficulties with the explanation given in the paper, but you are right, this is one of the few quantum erasure experiments where there is NO subset selection.

I think that this is a very interesting experiment, and I don't understand it completely. However, I disagree with the explanation given in the paper as such; indeed, according to the explanation, it wouldn't make any difference if i1 and i2 were combined with i1 going AROUND NL2 or going through NL2. And if that were true, we could obtain paradoxial situations: we could in principle send i1 and i2 on long, long long optical fibers (say, a light minute long), register the interference pattern or not at Ds during a minute, and only decide then whether or not we mix i1 and i2. And according to whether we decide to mix them or not, we would see our results of the past minute CHANGE !
We could even make a paradoxial machine in the following way: if the computer controlling the experiment has statistical evidence for interference during the minute of Ds measuring time, he decides to BLOCK the output of the fiber of i1, in which case we shouldn't see an interference pattern (because we can find out which photons came from NL2, knowing the delay of 1 minute and what comes out of fiber 2). But we SAW an interference pattern. On the other hand, if the computer has statistical evidence for no interference, he can decide to mix the two fiber outputs in a beam splitter, and according to the paper, we now have erasure of the information which is at the origin of the interference pattern, so we should observe interference, but we didn't !

I will try to write down of what I can make up of the setup:

The fact that an interference pattern arises can be understood in a certain way because when a single photon of the argon laser is split at BS(pump), you have that the state of this pump photon is 1/sqrt(2)(|pump1> + |pump2>), and hence the two photon pairs from the NL1 and NL2 are entangled:

1/sqrt(2) Integral ( |pair1> + e^(i theta) |pair2> ) rho(theta) dtheta

Here, theta is the difference in optical path length between the point in NL1 where the conversion |pump1> -> |pair1> took place, and the point in NL2 where the conversion |pump2> -> |pair2> took place. But of course we have to (Feynman path integral) integrate over ALL possible thetas because that conversion point is "floating", within the xtal, and it is this smearing out over different positions which makes that there is no fixed phase relationship between pair1 and pair2 (and hence between a single component of each).

It is somehow my feeling that the fact that the idler photon of pair 1 goes through NL2 changes rho(theta) (gives preferred positions for conversion of |pump2> -> |pair2>. In a non-linear xtal, this is not unexpected. From the moment that rho(theta) is somehow peaked, there is a phase relationship between pair1 and pair2 and you can expect interference.
But I'd be highly surprised that this same experiment works if the combination of i1 and i2 is done without i1 going through NL2 ! A way to find out would also be to change the path length of i1 before it gets to NL2. I'm pretty sure that this SHIFTS the interference pattern at Ds, which means that i1 DID SOMETHING to NL2.

Do you know of more experiments of this kind ? It looks interesting in any case.

cheers,
Patrick.

 

[werty]

we could obtain paradoxial situations: we could in principle send i1 and i2 on long, long long optical fibers (say, a light minute long), register the interference pattern or not at Ds during a minute, and only decide then whether or not we mix i1 and i2. And according to whether we decide to mix them or not, we would see our results of the past minute CHANGE !

No you can't do this because you have already made it possible to in principle decide what path the photon took. This is no different from throwing meassurering results in the trash, its not a quantum erasure.

 

[vanesch]

No you can't do this because you have already made it possible to in principle decide what path the photon took. This is no different from throwing meassurering results in the trash, its not a quantum erasure.

I know. But can you explain me, according to the paper, what is the principal difference between combining i1 and i2, when i1 goes AROUND NL2, or when i1 goes THROUGH i1 ? According to the explanation, it is the fact of combining them that makes it in principle impossible to know if the light (after combination) came from i1 or i2. There is no special status given to the fact that i1 goes through or not, NL2. Do you agree with this ?

cheers,
Patrick.

EDIT: in fact, it is NOT throwing measurements in the trash: after combination of both beams at the output of the two fibers, it is again impossible in principle to find out from which NLx the photon came. I don't see the conceptual difference, in the explanation given, by combining the beams a) when i1 goes through NL2 b) when i1 goes on a short path around NL2 and is combined there c) when i1 goes on a short path around NL2, and i1 and i2 are then sent on long fibers of equal length before being combined.

I don't know enough of the physics of these parametric down converters, but a priori I cannot exclude a "phase locking" (classically speaking) of the conversion amplitude in NL2 by i1, and I would tend to think that this is the explanation ; in which case the setup has a completely classical explanation too. Phase locking is something which happens easily in non-linear systems. This would then also explain the necessity of i1 to have to go THROUGH NL2 of course.

 

[werty]

Im starting to see the complexity of this problem. Its true what you say that it isn't the same thing as throwing the results in the thrash, but to remove the which-path information you need to combine the paths with a 50-50 beamsplitter and we would than have the standard qm-erasure/delayed choice setup. I now think I see the problem you probably noted previously, that in his setup a) (as you specified it) he don't use any such erasure of the which-path information. The information is simply removed in all cases!? But If he used b) instead he would have to use a beamsplitter and there would be no immidiate interference, he would have to use coincidences to pick out interference.

I now see what you hinted at in your previous post, if his reasoning is correct than it should be possible to construct your previously mentioned machine. Which I think is impossible :) Sorry for missunderstanding.

EDIT
Im also clueless how the downconverters work, don't know much about that phaselocking either.

 

[vanesch]

Im also clueless how the downconverters work, don't know much about that phaselocking either.

I have more experience in electronics: a "free running oscillator" which contains the slightest non-linearity very easily locks in on a tiny tiny signal of the same oscillator frequency.

For instance, on a card I had 32 completely independent oscillating circuits (in fact they were amplifiers but due to some error in the amplifier feedback, they were actually oscillators running at 250 MHz). They were in principle totally independent one of the other except for the common power supply, but which was filtered independently for each oscillator. Guess what ? They all were oscillating IN PHASE !
Nevertheless, no signal was seen either on the power supply nor on anything else. So it was just their proximity on the same card, which was quite well done, with shielding ground planes all over the place except for some cm of tracks parallel before they went onto the output connector, which must have been responsible for the phase relationship. The inputs were left open.
So it must have been this small capacitive coupling between the outputs (a few picofarad at most) which was responsible for the lock-in.

A priori I could think of a similar mechanism where the i1 beam makes NL2 down conversion "lock in" on it. But, as I said, I don't know enough of the physics of down converters to know if this is a sensible thing to say.

cheers,
Patrick.

 

[ppnl2]

The blasted links don't work for me. 404 error.

 

[DrChinese]

Interesting analogy. But when blocking one of two slits, we understand that light already through the blocked slit continues to propagate to the detector, and interference-effects only disappear once the time taken for those waves to arrive have been reached. We have a well defined and well understood model for the disappearance of the interference pattern. ...

Classical !? Whilst we know the blocking of the idler beam propagates to the Signal detector, since we see the interference pattern disappear, we have no model to discover along which route or how soon the cause propagates to the effect ! And you are just guessing by your answers first of 2), and now of 3).

Can you come up with a model justifying answer 1, 2, 3 or 4 ?

ShalomShlomo

a. Just as in the double slit experiment, anything that gives us which-path information destroys the interference. I don't think anyone questions this.

b. An oddity of this experimental setup is that the signal and idler paths must match their lengths very closely BEFORE they arrive at NL2 for the interference to be exhibited without blocking of the idler. This may not be clear from the description of the setup, but I believe a careful examination of the setup will confirm this is required. I think this is what Vanesch and werty have been commenting upon in earlier posts.

c. So the question we are now down to is: could the idler be blocked anywhere else other than between NL1 and NL2 and still yield the effect? And I would say the answer must be no. Clearly, there is no difference between blocking the idler photon and placing a detector in the same spot that does the same thing (blocking). (And we already have a detector on the other side of NL2 for the idler anyway, so its path could be lengthened or shortened without changing the result.) So we can only block the idler to turn on/off the interference pattern within a certain limited length range.

d. When we block the idler photon, it instanteneously causes the signal photon to collapse from a superposition to a definite state. That collapse will occur between NL1 and NL2, and the signal photon must have traveled an equal distance from the source (as the idler has) at that point in time - just as in a double slit setup. This happens anytime one of an entangled pair is observed. I do not think anyone will question this.

e. Once the superposition is gone due to blocking of the idler photon between NL1 and NL2, the interference will also be gone when the associated signal photon arrives at the detector. Determine the remaining length for the signal photon to traverse before detection (as of the time when the idler is blocked, since the wavefunction collapse is instantaneous), divide by c, and you know how much longer before the interference pattern disappears. I do not think anyone will question this.

So I would say this answers the question of the timing posed by ShalomShlomo. I am still not sure which of his 4 answers this maps to, but I think it is 3.

f. So the last element is: could the idler photon be any farther away from the signal detector than the signal photon at the time of the collapse? I am assuming no in my answer. And I take it that way because of the oddity of the setup, requiring paths to overlap in such specific ways.

So are c. and f. fair assumptions? My argument would need them to be true, and I believe they are. But to be honest, I am not 100% sure. Of course, this answer can be determined experimentally.

 

[DrChinese]

The blasted links don't work for me. 404 error.

See if this helps:

Quantum Effects in Photon Interference

-DrC

 

[ppnl2]

I have more experience in electronics: a "free running oscillator" which contains the slightest non-linearity very easily locks in on a tiny tiny signal of the same oscillator frequency.

For instance, on a card I had 32 completely independent oscillating circuits (in fact they were amplifiers but due to some error in the amplifier feedback, they were actually oscillators running at 250 MHz). They were in principle totally independent one of the other except for the common power supply, but which was filtered independently for each oscillator. Guess what ? They all were oscillating IN PHASE !
Nevertheless, no signal was seen either on the power supply nor on anything else. So it was just their proximity on the same card, which was quite well done, with shielding ground planes all over the place except for some cm of tracks parallel before they went onto the output connector, which must have been responsible for the phase relationship. The inputs were left open.
So it must have been this small capacitive coupling between the outputs (a few picofarad at most) which was responsible for the lock-in.

A priori I could think of a similar mechanism where the i1 beam makes NL2 down conversion "lock in" on it. But, as I said, I don't know enough of the physics of down converters to know if this is a sensible thing to say.

cheers,
Patrick.

I don't know much about phase locking so forgive what may be a dumb question.

Turn down the intensity on the thing so you only have one photon per year or so. Now when you have a photon in i1 you clearly do not have one in v2 and will not have one for a year or so. How can a photon in i1 cause NL2 to lock phase on a photon that will not arrive for a year?

 

[ppnl2]

See if this helps:

Quantum Effects in Photon Interference

-DrC

Thanks dude!

 

[ShalomShlomo]

Quantum Mechanics depends on the concept of simultaneous collapse of wavefunction of entangled particles at space-separated places.

Relativity depends on the concept of 'no such thing as simultaneous'.

In the experiment described in this thread, the conflict between the two Master Theories is brought to a head. Something has to give.
Our current understanding of them both is contradictory and cannot be complete.

Fortunately, the experiment described can through refinements itself reveal to us where our understanding is wrong.

If 1 turns out to be the right answer, then QM wins, and Relativity is in need of change (and causality is out of the window, and you can build yourself a communications-time-machine to receive messages from the future - terrifying, and therefore unlikely).

If 2 turns out to be the right answer, we have some hard questions to ask QM to understand what on Earth is going on - what is the mechanism for propagation of QM information - something new has to be thought up.

If 3 or 4 turns out to be the right answer, we need to look at exactly how light propagates, and how light experiences time (quick answer - it doesn't, it is at a singularity - but there are different ways of existing inside a singularity and we need to know which one applies to light's propagation through space).

The role of the idler beam entering the NL2 DownConverter is I believe irrelevant, as it seems highly likely that the same results would be obtained if the idler beam was routed around NL2 as Vanesch has also pointed out (I am depending on the credibility of the experimenters - although I have no proof for this). This can be easily checked experimentally.

The described experiment was carried out and published in 1991, 14 years ago.

What has occurred since to further our understanding and resolve the contradictions between QM and Relativity ?

ShalomShlomo

 

[DrChinese]

Quantum Mechanics depends on the concept of simultaneous collapse of wavefunction of entangled particles at space-separated places.

Relativity depends on the concept of 'no such thing as simultaneous'.

In the experiment described in this thread, the conflict between the two Master Theories is brought to a head. Something has to give.
Our current understanding of them both is contradictory and cannot be complete.

...

What has occurred since to further our understanding and resolve the contradictions between QM and Relativity ?

ShalomShlomo

This is an inaccurate portrayal. SR and QM are theories which cover different areas of physics. One does not yield one answer in contradiction to the other. Extension of each past its domain of applicability yields contradictory information, which tells us that we need to properly apply each theory.

There is substantial evidence that c is not the absolute upper speed limit in the sense that some would apply it*. And yet there is no specific equation of SR that is violated. And there is no actual evidence that QM is wrong. So the contradiction is strictly in the eye of the beholder.

This experiment is not now nor has it ever been considered evidence that SR and QM are in conflict. It is just another example of the quantum weirdness with entangled photons.

* Examples: a) Instanteous collapse of wavefunction, which still does not allow FTL signalling; and b) a few celestial objects are receding from us much faster than c, but we are not in the same inertial reference frames.

 

[ShalomShlomo]

This is an inaccurate portrayal. SR and QM are theories which cover different areas of physics. One does not yield one answer in contradiction to the other. Extension of each past its domain of applicability yields contradictory information, which tells us that we need to properly apply each theory.

There is substantial evidence that c is not the absolute upper speed limit in the sense that some would apply it*. And yet there is no specific equation of SR that is violated. And there is no actual evidence that QM is wrong. So the contradiction is strictly in the eye of the beholder.

This experiment is not now nor has it ever been considered evidence that SR and QM are in conflict. It is just another example of the quantum weirdness with entangled photons.

Everything you said above is correct, except the last paragraph.

This experiment does show that our understanding (probably of QM) is greatly lacking. We knew this before this experiment, thanks to Bell and Aspect, but this experiment adds to our lack of understanding greatly.

In the 14 years since the experiment described, has anyone attempted to resolve this hole in our understanding ?

 

[werty]

I think Feynman said to his students during lecture "I don't know why nature behaves in this perculiar way, nobody does!".

And that's that with that, I think we will most probably never understand why nature behaves as it does and that's what I think bells theorem proved, there's no such thing as "local reality". Notice that "local reality" always stand together in these theorems, there's no evidence that there's a "non-local reality" neither that there is a "local-nonreality", (whatever that means). Also I think stringtheory or someother theory tries to explain entanglement as a consequence of tiny wormholes, then there's no need for FTL communication they have an alternative route! And wormholes arent forbidden in GR. So there you go, there are alternative explanations. And until someone shows that you can use Mandells setup to send information FTL, which I believe should have been done by now if the original experiment was done 14 years ago.

 

[ShalomShlomo]

How can we experimentally measure the sought time delay between the blocking of ‎the Idler 1 beam, and the corresponding disappearance of the Signal interference ‎pattern (the cause/effect time delay) ?‎

‎Method 1:‎
Send the Signal beams along fibre-cables (maybe once or half way around the world) ‎before combining them to observe their interference (or non-interference) pattern, ‎depending on the blocking of the idler beam. This will enable confirmation or ‎elimination of answer 1.‎
Then send the Idler beams instead of the Signal beams along the fibre-cables before ‎combining/blocking them. The time delay from the journey along the cables help will ‎ascertain which answer from 2,3 or 4 is correct.‎

‎Method 2 (if Method 1 is experimentally unworkable - for example, if the ‎coherence of the light beams is not maintained during their journey through the ‎cables):‎
Create a feedback system between the Idler 1 blocker and the Signal Interference ‎pattern, by changing the Signal interference pattern from one detector, into 2 slits (one ‎fed by each of the Signal beams) - since the Signal beams are coherent, the two ‎separately fed slits will produce an interference pattern. Position a fast detector at one ‎of the peaks or troughs of the Signal Interference pattern, such that once a threshold ‎value close to the peak/trough value is detected, an electric (or light) signal is sent ‎through a flexible fixed length wire to the Idler-Blocker to block the Idler Beam. The ‎Idler Blocker blocks the beam (it also needs to be fast – along the timescales of ‎Aspect’s experiment). Once the Idler beam is blocked, the Signal interference pattern ‎will automatically disappear, and when the Signal detector finds that the threshold ‎value has been breached it will send a signal along the wire to the Idler-blocker to ‎stop blocking. This will result in the Signal interference pattern reappearing, which ‎will cause the Signal detector to again reach its interference threshold value, and an ‎endless loop will commence, with the Idler-Blocker switching on and off.‎
From the frequency of the switching on and off of the Idler Blocker ‎it may be possible to determine the sought cause/effect delay (or non-delay) of the ‎time between blocking and interference disappearance (by changing the experimental ‎lengths of the beams and observing the differences in resonant frequency).‎
It may well be found, however, that the time delays in the experimental signal ‎processing drown out the desired cause/effect time delay which may well turn out to ‎be much smaller. If so, a further refinement can be made to discover the sought ‎cause/effect time delay regardless of the larger experimental delays.‎
Using identical equipment, set up an identical second experimental system, but the ‎blocker blocks the signal beam itself (there need be no idler beams, and the detector ‎will just react to the presence of the signal beam itself). As it is almost identical, the ‎resonant frequency of the second set of apparatus will be almost identical to the first ‎set of apparatus. By combining the on/off beat from the two resonant blockers in the ‎two sets of apparatus, in effect combining two almost identical signals with a very ‎slight difference in frequency, the sought cause/effect delay can be established by studying the harmonics of the resultant pattern, and ‎altering the light path lengths in the first apparatus and noting the change in ‎harmonics caused.‎

A picture is worth a thousand words, and if anyone would like a diagram of the above ‎experimental set-up clarifying the notes above, I will gladly send it (sorry, I don’t know if it ‎is possible to post the diagram to the forum or not).‎

And so, using one of the two above methods, a startling experimental result can be ‎found that I think will shape QM for the next decade or so.‎

Who will be the first to carry out this or a similar experiment, which is crying-out to be completed ‎‎? Has it been carried out already ?‎

ShalomShlomo

 

[DrChinese]

I think Feynman said to his students during lecture "I don't know why nature behaves in this perculiar way, nobody does!".

And that's that with that, I think we will most probably never understand why nature behaves as it does and that's what I think bells theorem proved, there's no such thing as "local reality". Notice that "local reality" always stand together in these theorems, there's no evidence that there's a "non-local reality" neither that there is a "local-nonreality", (whatever that means).

Everything you said above is correct, except the last paragraph.

This experiment does show that our understanding (probably of QM) is greatly lacking. We knew this before this experiment, thanks to Bell and Aspect, but this experiment adds to our lack of understanding greatly.

What werty says is accurate. Local non-reality is a possibility. Most interpretations of QM say that there is something special going on with the act of observation.

ShalomShlomo: QM provides a description that nicely matches experiment. At the current time there is no known hole, although the current description may be unsatisfying.

 

[vanesch]

Turn down the intensity on the thing so you only have one photon per year or so. Now when you have a photon in i1 you clearly do not have one in v2 and will not have one for a year or so. How can a photon in i1 cause NL2 to lock phase on a photon that will not arrive for a year?

I don't know, but I'd like to add 2 comments.
First of all, downconversion is a classically non-linear process, so you cannot turn down indefinitely the pump beam (and hence the production of pairs).
Second, the two conversion events are in fact in superposition: let us not forget that the pump photon that is to be converted (given a classical intensity of the pump beam!) is in a superposed state of going to NL1 and NL2. Of course, the exact location of that conversion is somehow a priori arbitrary within NL1 and NL2 (which makes that we don't see raw interference of the outgoing signal photons because they have random phase relations due to different optical paths before conversion). But nevertheless the two conversions are entangled (because originating from two terms in the single pump photon state). The paper seems to forget that: it says that these are "spontaneous conversions" ; if that were true, there wouldn't be any difference between using coherent pump beams (a split beam) or using two independent pump lasers. The very fact that coherence can be restored in certain circumstances between the two signal beams means that they were in an entangled state. This entanglement finds its origin in the superposition of the photon states pumping NL1 and NL2.
So you cannot say that "the photon came from NL1 so clearly it didn't come from NL2" in the same way as you cannot say, in a Young's experiment that the photon went through 1 slit so clearly it didn't come through slit 2.
The presence of one term in the superposition of the converted photon in NL1 could (I don't know how but the possibility is not excluded) induce some phase locking with the other term (the term, in the superposition, of the conversion in NL2), in the same way as stimulated emission does (although people checked that it is not exactly stimulated emission in that no MORE photons come out) ; but why cannot they "lock in" the existing emission, which, I argued, cannot really be "spontaneous" (otherwise there's no reason to use coherent pump beams for NL1 and NL2).


cheers,
Patrick.

 

[ShalomShlomo]

I don't understand.

We're spending billions of dollars on super powerful particle accelerators, to try and understand the fundamental make up of our quantum mechanical particles.

And yet, the problem described here is an even more fundamental gap in our understanding, with the prize of a significantly deeper understanding of quantum mechanics, far beyond what we know today, and no-one is interested in discovering what is in front of our noses ?

Why ?


ShalomShlomo

 

[DrChinese]

How can we experimentally measure the sought time delay between the blocking of ‎the Idler 1 beam, and the corresponding disappearance of the Signal interference ‎pattern (the cause/effect time delay) ?‎

...

Who will be the first to carry out this or a similar experiment, which is crying-out to be completed ‎‎? Has it been carried out already ?‎

ShalomShlomo

When there is a serious suggestion that there is a superluminal signal device involved, you can bet it will be tested. This particular experimental setup precludes such result, since the idler travels (more or less) "alongside" the signal photon during much of the journey - at least that is my interpretation. QM itself does not predict any FTL signalling possibilities, at least at this time.

There is in fact a lot of experimentation going on in the area of parametric down conversion and entangled photons. Desktop versions of EPR tests are appearing in undergraduate labs as the price is dropping (currently under $50,000). See this slide presentation by Mark Beck (Whitman College) showing about 12 variations on these tests. I think it is safe to say that a lot more permutations will be tested in the near future. In fact, I suspect there is a lot of stuff already in the preprint archives that is up this alley. If only I weren't so lazy...

 

[vanesch]

And yet, the problem described here is an even more fundamental gap in our understanding, with the prize of a significantly deeper understanding of quantum mechanics, far beyond what we know today, and no-one is interested in discovering what is in front of our noses ?

Except for the specific experiment of which I'm convinced the explanation in the paper is not correct, and which must have something to do with the properties of non-linear optical materials, these experiments don't add much to our understanding of quantum mechanics per se: they just ILLUSTRATE where quantum mechanics is at work. Most of these experiments have very clear predictions in quantum theory and are also confirmed, so they only indicate that quantum mechanics is ALSO valid in those particular cases.
What they learn us is where we already DON'T have to look for deviations from quantum mechanics (although the strange predictions of quantum mechanics would make us think that this "cannot be"). They only add to our understanding by EXCLUDING possible semiclassical explanations.

Most of these experiments are confirming extremely simple quantum mechanical models which can be worked out on 2 or 3 pages, as a simple exercise in quantum theory. (the i1 through NL2 xtal experiment is an exception to it, that's also why I think that some more complicated physics must be going on there). There IS a fundamental issue to solve in QM, but these experiments only show us that they are NOT exploring it, and that QM is still "alive and kicking" at the level where they explore nature.

So although it is always good to do experiments, and to check them against QM predictions, especially when they are "spectacular", they don't learn us much in fact. On the other hand, big accerator experiments try to explore NEW stuff, of which we are NOT sure that we know what is going on (unfortunately, for about 20 years, they all agree with the predictions of the standard model in as far we can make predictions). Hopefully LHC will show unexpected things ; in any case it WILL learn us something (supersymmetry, or not, Higgs, or not).

cheers,
Patrick.

 

[ShalomShlomo]

I guess the dispute between us is - Is there anything in this experiment beyond our current understanding of Quantum Mechanics ?

I say yes, a major hole in our knowledge is staring us in the face, and we are fools to ignore it and not grasp the opportunity.

You both say no, the experiment is indeed described by our current knowledge of Quantum Mechanics.

History will show who was right (If there is only one history :!) ).

ShalomShlomo

 

[werty]

Theres not much else to do is there. For us to answer your question of what time it takes for interference to disappear we must consult QM and let it give us the answer and than you want to use that answer to figure out if QM is wrong or not ? Seems to be a kind of logical circle don't you think ? It would be better to search for experiments that show that there is FTL communication.

 

[James Jackson]

May I refer people so David Deutsch's paper 'Information Flow In Entangled Quantum Systems.', available here: http://www.qubit.org/people/david/index.php?path=Documents/Research Papers

 

[DrChinese]

History will show who was right (If there is only one history :!) ).

ShalomShlomo

Anyway, entangled photons via PDC give us a great opportunity to probe reality. I am sure there are exciting experiments yet to come, perhaps som with unexpected results!

 

[ppnl2]

Except for the specific experiment of which I'm convinced the explanation in the paper is not correct, and which must have something to do with the properties of non-linear optical materials...
Patrick.

Why? Is it violating some principle of QM?

Ok what if we change the experiment a little. What if we made the thing symmetric across V1 and V2 so that we are generating one interference pattern with S1 and S2 and another interference pattern with I1 and I2. This should be possible as long as the path lengths are equal. Now if we block I1 it will stop interference at Di. But it should also stop interference at Ds. But I1 never goes through NL2 here.

I would really like to get an understanding of this experiment nailed. I have been trying since I first saw it in scientific american years ago.

 

[vanesch]

Why? Is it violating some principle of QM?

Not if something is changed in NL1. But yes if not, for the reason I will line out after your next comment.

Ok what if we change the experiment a little. What if we made the thing symmetric across V1 and V2 so that we are generating one interference pattern with S1 and S2 and another interference pattern with I1 and I2. This should be possible as long as the path lengths are equal. Now if we block I1 it will stop interference at Di. But it should also stop interference at Ds. But I1 never goes through NL2 here.

By blocking i1 somewhere, you should, according to QM, not be able to change the result of s1 and s2 (except if i1 DID something, like interacting in NL2 or so of course).
The reason in QM is the following: if rho is the global density matrix, and rho1 is Tr_2(rho) (the density matrix, traced over the space H2 which corresponds to the i1 and i2 degrees of freedom), then ALL observables A1 which are local to H1 (the s1 and s2 degrees of freedom) have the following expectation values: <A1> = Tr(A1 rho1).
In the same way, observables which are local to H2 (only measure things about i1 and i2) are given by <A2> = Tr(A2 rho2).
That means that, no matter what measurement we do on i1 and i2, this should not influence any outcome of measurement (expectation value of operator) at s1 and s2. Only CORRELATIONS can see influences (because these are expectation values of observables that are NOT local to H1 and H2 respectively). Now, Interference, or not, of s1 and s2 is clearly a local observable of H1, and should hence be described by Tr(A1 rho1), IRRESPECTIVE of what is measured at i1 and i2, because (in Copenhagen view) applying a weighted projection in just any basis (any measurement) in H2 does not change the trace over H2, and hence leaves rho1 invariant.
This fundamental property of QM makes that superluminal information transfer is not possible in principle (though Bell violating *correlations* are possible).

Now, if your setup (or the explanation in the paper) were correct, then you could use this setup to do some FTL signalling of course, by deciding, or not, to block i1 (very late after emission).
That's why I think that it is not possible (and I'm SURE it is not possible within quantum theory). It might of course, in which case we found a deviation from QM. But Occam's rasor points us first to consider that i1, traversing NL2, simply DOES something in NL2, like locking in s2.

cheers,
Patrick.

 

[ppnl2]

By blocking i1 somewhere, you should, according to QM, not be able to change the result of s1 and s2 (except if i1 DID something, like interacting in NL2 or so of course).
The reason in QM is the following: if rho is the global density matrix, and rho1 is Tr_2(rho) (the density matrix, traced over the space H2 which corresponds to the i1 and i2 degrees of freedom), then ALL observables A1 which are local to H1 (the s1 and s2 degrees of freedom) have the following expectation values: <A1> = Tr(A1 rho1).
In the same way, observables which are local to H2 (only measure things about i1 and i2) are given by <A2> = Tr(A2 rho2).
That means that, no matter what measurement we do on i1 and i2, this should not influence any outcome of measurement (expectation value of operator) at s1 and s2. Only CORRELATIONS can see influences (because these are expectation values of observables that are NOT local to H1 and H2 respectively). Now, Interference, or not, of s1 and s2 is clearly a local observable of H1, and should hence be described by Tr(A1 rho1), IRRESPECTIVE of what is measured at i1 and i2, because (in Copenhagen view) applying a weighted projection in just any basis (any measurement) in H2 does not change the trace over H2, and hence leaves rho1 invariant.
This fundamental property of QM makes that superluminal information transfer is not possible in principle (though Bell violating *correlations* are possible).

Ok this is mostly over my head. I don't really understand what a density matrix is. But from what very little grasp I have of it I'm suprised its applicable here. I'm stuck trying to visualize the unitary evolution of a single photon as it goes through the experiment. From this perspective if you have "which way" information then you have no interference pattern. If you combine I1 and I2 then you don't have that information and should have interference. As you have pointed out this may not be takeing into accout the physics of the down converters.

As you can see my understanding of QM is less than adequate.

Now, if your setup (or the explanation in the paper) were correct, then you could use this setup to do some FTL signalling of course, by deciding, or not, to block i1 (very late after emission).
That's why I think that it is not possible (and I'm SURE it is not possible within quantum theory). It might of course, in which case we found a deviation from QM. But Occam's rasor points us first to consider that i1, traversing NL2, simply DOES something in NL2, like locking in s2.

cheers,
Patrick.

As you say my setup would allow ftl signaling. A real bad thing that I will have to think about.

(Hmm, my setup is a two photon state. But in the real experiment when you combine I1 and I2 you are reduceing back down to a one photon state. Aren't you? Wouldn't this make a difference?)

But in the real setup there is no place that you could block I1 that would make the signal FTL. You may be able to change it in some way to do it but it isn't clear how. You could use mirrors to route I1 around NL2 but then you would need to make I1 and I2 colinear again and in phase.

I understand that it has been proved that you cannot use QM to signal FTL but has it been proved that you cannot use QM to signal at light speed in some similar way?

 

[vanesch]

Ok this is mostly over my head. I don't really understand what a density matrix is. But from what very little grasp I have of it I'm suprised its applicable here.

The reason is simply that a density matrix describes a statistical mixture, which can also be a pure state of course. Now, when you do a measurement, then you transform a density matrix of the situation before measurement (written in the eigenbasis of the observable of which we are performing the measurement) into a density matrix with the same elements on the diagonal, and 0 elsewhere. That's the Born rule in "density matrix" language.

For a pure state, you can understand this. Imagine that we have an initial pure state given by |psi> = a |e1> + b|e2> and we are going to measure the E quantity with eigenvectors |e1> and |e2>.
The density matrix corresponding to psi is simply
|psi><psi| = (a |e1> + b |e2>) (<e1| a* + <e2|b*) which gives the 4 terms corresponding to the 4 elements of the density matrix:
{ a a* ; a b*}
{ b a* ; b b*}

After measurement, we know that we have a MIXTURE, given by:
probability a a* to be in state |e1> and probability b b* to be in state |e2>, so the expectation value of A is aa* e1 + bb* e2.

This state is represented by a density matrix
{ a a* ; 0 }
{ 0 ; b b*}

So you see, keeping the diagonal elements and putting 0 everywhere else is the effect of a measurement (the Born rule).
In this basis, of course, it is simple to see that A is a diagonal matrix with e1 and e2 on the diagonal, so that rho A equals:

{ a a* e1 ; a b* e2}
{ a* b e1 ; b b* e2}

and the trace of it, Tr(rho A) = aa* e1 + bb* e2, the expectation value of A.

This seems trivial of course, because we were working in the right basis. But a property of the Trace is that it is invariant wrt a change of basis, so this expression is valid in EVERY basis !

Now, in the case of a composite system H1 x H2, the states can be written in a kronecker product basis (local basis). If an observable A only observes on system 1 then A can in fact be written as A x 1 (1 = unit operator in H2). In the same way, if B only observes a property of H2, it can be written as 1 x B (here, 1 is unit operator in H1).
Now, let us consider that the overall state |psi> = a |e1> |f1> + b |e2> |f2> where e1 and e2 are eigenvectors of A (but f1 and f2 are not basis vectors of H2, they just happen to be vectors in H2). Clearly |e1>|f1> is an eigenvector of A x 1 and so is |e2> |f2> (and both are orthogonal, even if f1 and f2 aren't). The probability of finding e1 is a a* and the probability of finding e2 is b b*.
As a density matrix, it can be written as:

{ a a* |f1><f1| ; a b* |f1><f2|}
{ b a* |f2><f1| ; b b* |f2><f2|}

where the |f><f| stand for submatrices written out in a basis of your choice in H2.

The subtrace of this matrix with respect to H2 is obtained by tracing out each of the 4 submatrices:

rho1 =
{a a* Tr(|f1><f1|) ; a b* Tr(|f1><f2|) }
{b a* Tr(|f2><f1|) ; b b* Tr(|f2><f2|) }

Now, it is a property of Tr(|a><b|) that it is equal to <b|a>.
Remember that f1 and f2 are not basis vectors of H2, they are just 2 arbitrary vectors in H2, so <f1|f2> is not necessarily 0. However, <f1|f1> = 1 because they are supposed to be normalized.

So rho1 =

{ a a* ; a b* x*}
{a* b x; b b*}

with x = <f1|f2>

and we see that our expectation value for A is given by:

<A x 1> = Tr( A rho1)

in the same way as before.

Now, the important part is that the basis in H2 in which we did this (worked out the trace) DIDN'T MATTER. Also, it wouldn't matter if we applied a measurement in H2 or not, which would have (as we saw before) transformed the submatrices |f1><f1| and so on into their DIAGONAL matrices in the basis corresponding to the measurement. Indeed, the only quantities needed to calculate a trace are the diagonal elements, and they don't change, with or without measurement.
So we can conclude 2 things:
1) the expectation value of A doesn't change, whether we apply, or not, a measurement in H2 (full submatrices, or only diagonal)
2) if we apply a measurement in H2 it doesn't matter which one (the eigenbasis in which we write out f1 and f2).

I'm stuck trying to visualize the unitary evolution of a single photon as it goes through the experiment. From this perspective if you have "which way" information then you have no interference pattern. If you combine I1 and I2 then you don't have that information and should have interference.

It is not because you don't have the information that you necessarily have interference ! It is if you USE that information to subselect a sample and then you are working out expectation values of a CORRELATION, which is not a local observable anymore acting only on H1 or H2.

(Hmm, my setup is a two photon state. But in the real experiment when you combine I1 and I2 you are reduceing back down to a one photon state. Aren't you? Wouldn't this make a difference?)

No, the 2-photon state is |signal> |idler>
and we have a SUPERPOSITION of 2-photon states:
|s1>|i1> + exp(i phi) |s2>|i2>
(this superposition is due to the pump photon being in a superposed state: one state going to NL1 and the other going to NL2)
So s1 and s2 are part of one and the same "photon" (in 2 different states).
The trick of combining the i1 and i2 beams into one single beam is to evolve unitarily |i1> and |i2> into the same state (in order to keep unitarity then this must in fact result into 2 different states but in such a way that they are indistinguishable for one detector: |i1> goes to |u> + |v> and |i2> goes to |u> - |v> and we detect |u> and |v> ; this is typically done with a beamsplitter).
By doing so, we rewrite the wave function as

(|s1> + exp(i phi) |s2> ) |u> + (|s1> - exp(i phi) |s2>) |v>

If we now look at |u> detections of the idler we find an interference pattern of the two different states of the "signal" photon, and if we look at the |v> detections, we find a complementary interference pattern.

But normally, WITHOUT subsampling (tagging on u or on v) an ENTANGLED state can never give rise to interference in the whole sample (that's the entire content of decoherence!).
However, if some physics in the NL2 converter makes that the |u> state (it is there if i1 gets into the converter) can LOWER the |s2> phase (that's possible, a unitary phase rotation), and the |v> state can INCREASE the |s2> phase, then both interference patterns are NOT perfectly complementary anymore, and the overall sample will show a partial interference pattern. But that is due to an interaction of the idler photon states (u and v) with the signal photon state s2. This is possible because the 3 states are PRESENT, LOCALLY, IN NL2. Mind you that it cannot do anything to s1 (which is far away, in NL1).

cheers,
Patrick.

 

[Hans de Vries]

I don't understand.

We're spending billions of dollars on super powerful particle accelerators, to try and understand the fundamental make up of our quantum mechanical particles.

And yet, the problem described here is an even more fundamental gap in our understanding, with the prize of a significantly deeper understanding of quantum mechanics, far beyond what we know today, and no-one is interested in discovering what is in front of our noses ?

Why ?

But it IS technically easy enough (and cheap!) to make timing measurements
to distinguish between mere Correlation and actual "Action on a Distance"

Analyzing detection times and actuator times afterwards should reveal
the propagation speeds you're after in the "Action on a Distance" case.
(Apart from the question if one can transmit information FTL) If one can
detect correlation afterwards then it is also possible to analyze the timing
of the correlated events.

The total absence of any such experimental timing results after all these
years points strongly towards Correlation and against "Action on a Distance"



Like Patrick, I don't agree with the author's conclusion on the experiment
depicted by figure 6. Partial reflection of i1 to the s2 path would simply
explain the results.

Worse, The author in fact assumes One photon interference with a single
path! First he excludes simultaneous down conversion by both PDC's as
"very improbable" He then excludes an i1 to s2 path and goes on with:

"one might not expect to see ONE-photon interference at Ds , but, ...,
interference fringes were actually observed"

Interference with a single path? Later then he talks about interference
again between s1 and s2 which he first ruled out. This is TWO photon
interference... from two different PDC's which are unlikely to be always
coherent with the same phase since the down conversion can take place
anywhere in the PDC resulting in phase differences.


Regards, Hans

 

[vanesch]

Like Patrick, I don't agree with the author's conclusion on the experiment
depicted by figure 6. Partial reflection of i1 to the s2 path would simply
explain the results.

This was also my first idea, but it cannot simply be partial reflection, given that the frequencies of idlers and signals are different, and that there is a filter before the s1+s2 detector, filtering out any idler contribution. That's why I suspect a frequency change i2 -> s1 which is not excluded, because for that you need a non-linear interaction, and NL2 IS a nonlinear material. Moreover, the interference only appears when i2 is EXACTLY aligned in such a way, that the momentum and energy relations in the NL2 xtal are satisfied to allow for such a i2 -> s1 transition. But, as I said, I don't know enough of the detailled physics of these xtals to work out exactly what is going on.

BTW, I also agree fully with you that the author makes quite some errors by considering the s1 and s2 photon states as "independently generated" by a "spontaneous process" : in that case, in no way ever there could be any interference. They are produced in superposition (by the pump photon), but, as you also point out, with random phase relation due to different points of conversion.

cheers,
Patrick.

 

[ppnl2]

The reason is simply that a density matrix describes a statistical mixture, which can also be a pure state of course. Now, when you do a measurement, then you transform a density matrix of the situation before measurement (written in the eigenbasis of the observable of which we are performing the measurement) into a density matrix with the same elements on the diagonal, and 0 elsewhere. That's the Born rule in "density matrix" language.

...

Patrick.

Sorry, mostly over my head. I think I get the gest of it. But this stuff is just a hobby for me and I have always had trouble with the notation used by the people who know what they are talking about. Still, I'm going to disagree with you.

First, I have real problems believeing in your phase locking. I know less about the physics of down converters than you but still I just don't buy it.

Second, I think I have worked out a way for it to work as advertised and still protect us from FTL communication. Follow this and tell me where I'm going wrong.

We start with a single photon from the argon laser. This is split into V1 And V2. This is a one photon state, a superposition of (V1 OR V2). If we were to check now we could get a visible interference pattern with V1 and V2. Except that at this point we only have one photon.

Now we send V1 and V2 through NL1 and NL2 creating a two photon state that is in superposition ((S1 AND I1) OR (S2 AND I2)).

(Sorry, I understand boolean speak much better than that ket and bra matrix crap you guys throw around.)

Now at this point if we were to check for interference between S1 and S2 we would not see it because we have a two photon state. I assume you gave me a good explanation of why and I think I get the gest of it. Simply put it is as you say decoherence.

But that isn't the end of the story. We now combine I1 and I2 in phase creating the state ((S1 AND I2) OR (S2 AND I2)). The I2 photon is the same in both so it no longer plays a role in differentiating between the state. So we drop it leaveing us with the state (S1 OR S2). This is a SINGLE PHOTON STATE. We have interference again!

But we are protected from FTL signaling. To see why you have to see why the way we (or I anyway) have been thinking about it is wrong. Usually we think of the choice to measure one of the Idler beams causes the interference pattern to colapse. But this is exactly wrong and counter to the whole idea of decoherence. It is the choice to combine I1 and I2 thus reduceing us to a one photon state that creates the interference pattern.

And here is the key that prevents FTL signaling. The idler beams must be combined before the entangled photon in the signal beams hits the signal detector or we still have a two photon state. We must destroy entanglement before detection in order to see an interference pattern!

So what do you think?

 

[vanesch]

We start with a single photon from the argon laser. This is split into V1 And V2. This is a one photon state, a superposition of (V1 OR V2). If we were to check now we could get a visible interference pattern with V1 and V2. Except that at this point we only have one photon.

Ok, that's the same as what I said.

Now we send V1 and V2 through NL1 and NL2 creating a two photon state that is in superposition ((S1 AND I1) OR (S2 AND I2)).

(Sorry, I understand boolean speak much better than that ket and bra matrix crap you guys throw around.)

Ok, apart from the strange boolean notation, we're on the same line...
except one point: there is a random phase shift between the first and the second pair, due to different optical lengths where the conversion v1->s1 and i1 on one hand, and v2 -> s2 and i2 on the other hand take place. These conversions can be anywhere in each of the NL xtals. This is important for what follows.

Now at this point if we were to check for interference between S1 and S2 we would not see it because we have a two photon state. I assume you gave me a good explanation of why and I think I get the gest of it. Simply put it is as you say decoherence.

Ok.

But that isn't the end of the story. We now combine I1 and I2 in phase creating the state ((S1 AND I2) OR (S2 AND I2)). The I2 photon is the same in both so it no longer plays a role in differentiating between the state. So we drop it leaveing us with the state (S1 OR S2). This is a SINGLE PHOTON STATE. We have interference again!

This is fundamentally impossible. You see, whatever you do to i1 and i2 must be a unitary transformation (any optical action, like a mirror, a lens, a beam splitter, a double slit etc... gives rise to a time evolution operator, derived from a hermitean hamiltonian).
Now, a unitary transformation CONSERVES IN PRODUCT. Clearly, <i1|i2> = 0 (they are completely different modes). So no matter what you do to i1 and i2, using a unitary transformation, you cannot have an U that transforms i1 and i2 into the same mode, because then <U i1 | U i2> would be 1.
The thing that comes closest is to transform (as I showed) i1 into something like u + v and i2 into something like u - v (in that way, <u+v|u-v> = 0) and only detect u, for instance. But that comes down to selecting a subset of the sample of s1 + s2.

But we are protected from FTL signaling. To see why you have to see why the way we (or I anyway) have been thinking about it is wrong. Usually we think of the choice to measure one of the Idler beams causes the interference pattern to colapse. But this is exactly wrong and counter to the whole idea of decoherence. It is the choice to combine I1 and I2 thus reduceing us to a one photon state that creates the interference pattern.

I agree with you that this collapse business is not correct and that one should consider unitary evolution all the way. So indeed, if you could "unify" i1 and i2 you would indeed undo the entanglement and in principle, allow for interference (if there wasn't also a random optical path length difference)...
Except that you can't do that, with a unitary transformation.
Another way to see it is that a unitary operator is reversible in principle. Now if i1 -> i2 and i2 -> i2, you cannot reverse that.

And here is the key that prevents FTL signaling. The idler beams must be combined before the entangled photon in the signal beams hits the signal detector or we still have a two photon state. We must destroy entanglement before detection in order to see an interference pattern!

Even if it were true, it would still allow for FTL signalling. The source could be 1 light year from the signal detector, and in the opposite direction, we could let the idlers travel, for say, 3/4 of a light year. You could combine them or not, according to your choice, and 3 months later, the detection would take place at the signal detector. But the decision was made at a distance of 1.75 light years, and the effect was seen only 3 months later ! That's FTL.
But as it isn't possible with unitary transformation, there is no such problem.

cheers,
Patrick.