Photons Detector not yielding which-path info.

[wampeter]

Photons Detector not yielding "which-path" info.

In Fabric Of The Cosmos, I am just getting through the part where Greene discusses a multi-path setup (7.5?) involving a series of beam splitters and downconverters, whereby photons striking two of four detectors will yield definitive "which-path" information, but, due to the configurations of the optical components, "hits" on the other two detectors will not yield definitive which-path info, and therefore, we preserve the interference pattern that one would expect without any measurement intervention.

It seem that the system "knows" that our experimental setup can not, or does not, provide which-path info and therefore behaves accordingly.

So, what if the detectors were present, but not "ON" or functioning in any way. What would we expect then?

What if they were on, but the wiring was sabotaged such that the result would never be reported reliably?

What if the result were available, but only sent to a computer that was programmed to irretrievably delete the data and never make it available to living beings?

Or what if a long delay line were in the system such that the result were not known for some time after the photon struck the detector. Would the interference pattern break down at the instant the photon struck the detector, or only after the result was "known.?"

Thoughts?

 

[peter0302]

Anton Zelinger discusses this very issue in Experiment and the Foundations of quantum physics (google it). The issue is whether we arrange the experiment in such a way as it is possible - in principle - to determine the which-path information. Even if the detector is off, or if we get the information for a nanosecond but then discard it, the photons behave as though we had the which-path info, and there is no interference pattern.

In other words, nature really doesn't care what the human observes - which should not at all be surprising!

There are "delayed choice" experiments which attempt to show that when we (think we) have the which path information, but then it is destroyed, the interference pattern reemerges. These experiments are very tricky to devise. They only work when it is absolutely impossible - in principle - to recover the information because, say, a beam splitter has sent a photon to one of two random directions which can never be known. But you can't make a quantum eraser by sabotaging the wires or telling the computer to disregard the information. :) The eraser has to ruly be "quantum".

Now a fascinating quantum eraser experiment would be a quantum computer eraser, which randomly - based on a _quantum_ event - discarded the information such that there was no trace of it in the computer's quantum memory. I suspect such an eraser would work.

 

[Usaf Moji]

So, what if the detectors were present, but not "ON" or functioning in any way. What would we expect then?

I'm not familiar with the particular setup described in your post, and I haven't read any of Brian Greene's books. But I know that for the original double-slit experiment, when the detectors are on, but the screen that the observer looks at is off, you get an interference pattern. If you keep everything the same but turn the screen on, you don't get an interference pattern. See this guy's lecture: http://www.youtube.com/watch?v=_OWQildwjKQ&feature=related This shows that only a living creature (and possibly only a human) can collapse the wave function.

I also read somewhere (can't find the source now) that when everything in the setup was identical except that the lighting of the room was just dim enough to prevent the human observers from seeing the screen, the wave function didn't collapse, when otherwise it would have. Again, this shows that it is the human being that collapeses the wave function.

 

[peter0302]

This guy's presentation is incorrect. (BTW, is he in a church??)

When the photon has been detected by one of the detectors, an irreversible measurement is made. That measurement "collapses the wavefunction." If the detector is disconnected from the screen, that does not change the fact that the measurement was made. The state of the system has been altered and merely disregarding the information is not the same as putting the system back into the superpositioned state, whereas with a quantum eraser it is.

I also read somewhere (can't find the source now) that when everything in the setup was identical except that the lighting of the room was just dim enough to prevent the human observers from seeing the screen, the wave function didn't collapse, when otherwise it would have. Again, this shows that it is the human being that collapeses the wave function.

Again, that is also impossible. Whatever source you read that in in bogus.

 

[wampeter]

So, dim lighting and human "awareness" aside...

Where in the "block diagram" of the detector is the wave function caused to collapse?

I don't know the schematic of the detectors that are used, but it is probably safe to assume that there is some sort of light sensitive element which then sends some charge or control signal to something that will indicate that the photon is/was present.

At which moment in time does the wave collapse? Is it upon the launch of this control signal, or only when the information is recorded or preserved? It just seems to me that there is some finite element of chronology that would yield answers.

 

[peter0302]

It collapses the instant the photon is absorbed by an atom in the detector.

The difference between a detector and a mirror or beamsplitter is that, in the latter case, the photons are either reflected or refracted purely, and the state of the device is not permanently altered. (Beautifully, the state actually is altered momentarily, while the electrons "wobble" around a bit, but then goes back to normal - very quantum eraser-like).

A photosensitive detector, however, must alter its state when a photon strikes it in order to be a detector, at which point there's been an unmistakeable collapse of the state of the detector: it was in a superposition of detect/no-detect, and later becomes "detect". I believe most of them use the photoelectric effect, and so a small current is generated when the photon is absorbed. Another photon might well be reemitted and might well continue past the slit, but it will have a well defined position and therefore won't make an interference pattern.

In other words, even if some kind of photon is still passed on throught he slit, it won't be in the same state it was before - no matter whether someone "looks at" the detector or not. That's why the video posted is so off the mark. It will be in a definite state - "Detected by left detector" or "detected by right detector" and will behave accordingly. It most certainly will not remain in left/right superposition.

 

[peter0302]

Since no one else seems to be biting on this (to me) very interesting subject, and because I love to hearmyself talk () I'd like to add some more random observations:

This question really goes to the heart of quantum interpretation. What is the wavefunction? People originally thought it was continuous charge distribution, but quickly realized that couldn't be right, because the charge of an electron or EM field of a photon would then be felt (in small amounts) everywhere. Born figured out that it was probability amplitude, and that the absolute square of the amplitude was the probability density function which, when integrated over a space, gives you the actual odds of finding the particle in that space.

Easy enough right? The big problem then becomes why and how does the wavefunction change when there's a measurement? If it's all just raw statistics with no physical meaning, why *doesn't* whether the detector is connected to the monitor matter? If it were all just probability and Bayes' theorem, then the human observer would be necessary. But he's not. So there's something more at work.

We've got to look for something that physically occurs when the wavefuncion "collapses" to figure out why that happens. And a good starting point is the contrast between a photon striking a mirror and a photon striking a detector. In many experimental setups, one collapses the wavefunction, the other does not. (i.e. you can use mirrors or lenses or beamsplitters or what have you - let's just say "glass" - to bounce photons around as much as you want and still get an interference pattern if you do it right). Yet in both cases (glass vs. detector) electrons are being perturbed by the photon, and new photons are emerging as a result. The same kind of *stuff* is happening. What's the difference?

In the case of glass, the electrons are perturbed, they wobble, and then emit a new photon - conserving energy and momentum. The photon out is (for our purposes) identical to the one that came in. Critically, the photon leaves the glass behind with no trace of it ever having been there. And so maybe it wasn't - because in QM, if there was no evidence that it happened, it might as well not have happened. In any event, we *cannot* know, in principle, whether a photon ever struck that piece of glass. And so bouncing around in a lens or mirror doesn't destory the photon's superpositioned state - because it may not have hit that mirror at all. Put another way, the state of the glass after the photon is gone is identical to the state of the glass before the photon got there - so the state of the glass makes no contribution (and therefore is not entangled with) the state of the photon. (It doesn't matter how many molecules are involved in the photon's "journey" through the lens or bouncing off the mirror - they all wind up in the exact state they were in before.)

Contrast this with a photon hitting a detector. Assuming the detector uses the photoelectric effect, a small current is generated by the strike of the photon. The current takes the form of electric charge - electrons - moving around in a wire, jumping around, getting excited, atoms getting ionized, etc. Billions and billions of molecules are affected even by that tiny little bit of current that was generated as it propagates through the wire.

Electricity is still technically a QM phenominon though, so isn't it possible the wire returns to the state it was in after the current has passed through it? No way - all wires have resistence, which means heat, and even the tiny current from a single photon will generate *some* heat. But even if you ignored that, the current winds up activating electronic circuitry, and once that starts happening there's no going back. The transistors involved contain gazillions of molecules that are interacting and exchanging electrons and generating heat and doing all the things transistors do - all because that photon struck the detector.

So the screen is shut off. All those molecules have still been affected by the photon's strike. They *remember* it. There's evidence of it, and if you really wanted to find it, you could. And therefore the wave function has collapsed. Put another way, the state of the detector - a macroscopic object - has substantially been altered by the hit of the photon. Knowing the state of the detector tells us the state of the photon. The photon has now become entangled with a macroscopic object.

The fancy word for all of that is decoherence. If you let the wave function keep evolving, accounting for every molecule of the wire and each molecule in each transistor, not to mention if you could (somehow) account for the heat as well, you'd find that the wave function would be an absolute mess - but - it would apporximate, very closely, the state of 100% certainty that the photon struck that detector. The state of the detector has become so intertwined with the state of the photon that there's no way the detector could be in that state if the photon hadn't hit there (compared to glass, where the state of the glass means nothing about whether the photon hit). And whether the human looked at the result wouldn't really matter all that much. To be sure, if the human looked, he'd become entangled with the photon too, but the detector's in the state that it's in, and it ain't going back no matter what, so no one really cares whether the human looks.

Decoherence is the subject of a lot of research right now, but IMHO is the leading candidate, if not the outright winner, for explaining the "quantum/classical" transition. (Unfortunately, it's also the leading problem in developing quantum computers). Decoherence doesn't explain everything either - technically speaking, the system is still in a state of superposition, but the other state is so miniscule as to be undetectible. But where is it? That's the subject of debate as well.

 

[wampeter]

Good self talk, Peter0302. Thanks for jumping further with this. You went a long way towards helping my understanding of wave collapse. As you say "We've got to look for something that physically occurs when the wavefunction "collapses" to figure out why that happens.

The issue surrounds the delayed choice quantum eraser experiment (google that and look at the wiki for a great explanation)

If I understand it's fundamental message, they use a network of detectors and down converters such that in 50% of the pathways, the photon strikes the detector, but the interference pattern persists, even though the detector was active and presumably causes, heat, current, collapse, etc, but ALAS, it does not because:

due to the config of the detectors, there are some paths that will not yield to the experimenter any definitive which path info.

So, in this experiment, it appears that the physical impact the detector brings to the photon-detector system is the same as in the more basic experiments, but the interference pattern is still present.

Do I have this right, and if so, why would that be?

 

[Cthugha]

So, in this experiment, it appears that the physical impact the detector brings to the photon-detector system is the same as in the more basic experiments, but the interference pattern is still present.

Do I have this right, and if so, why would that be?

Yes and no. The direct output of the detector will nerver show an interference pattern in delayed choice quantum eraser experiments. You only get the pattern by coincidence counting, so the subset of detected photons, which do not provide which-way information gives you the interference pattern.

The reason for this behaviour lies in the state of the system after the collapse. At first you have a state, where there is a photon present and the detector is not in an excited state. After the detection/collapse the photon has gone and the detector is in an excited state. Due to the fact, that there is no which-way information, the state after the collapse is still a superposition of the two possible paths the photon could have taken and therefore the interference pattern is still present.

 

[peter0302]

The DCQE is a very interresting and complicated experiment. First off, you're dealing with entangled photons, so it's a little more complicated than a simple double slit. And the rule is that the measurements you make on entangled photons have to be consistent. So if you detect which-path on one, you have to get the same result when you measure the other. And therefore you can't see an interference pattern from either if you detect which-path from either. That's basic HUP.

So with that in mind what's going on in that experiment? We've got 5 detectors total - D0 through D4 and two entangled photons, a signal and an idler. D0 gets the signal photon first. That photon lands where it lands, so that's set in stone right? Well, apparently no. Because apparently what happens at D1-D4 actually has a correlation with what happens at D0. I say correlation - not causation - because we don't know what caused what. If we start talking about causation, we are almost trapped into talking about REVERSE causation, so let's not talk about that for now. :) Let's just talk about quantum states.

What does entanglement mean? It means that you cannot talk about the state of "A" without saying something about the state of "B". They are mutually dependent. They're like complimentary angles: if angle A is x degrees, complimentary angle B is 90-x degrees. Period. Causation? Who knows. Not important right now.

So, the signal and idler photons are entangled and D0 always gets the first hit. Now here's the kicker: a hit at D0 tells you _nothing_ about which slit the photon came out of. As you can see from the drawing (figure 2 in the paper, attached), the photon will hit at D0 regardless of which slit it came out of. So while the states of the two PHOTONS are mutually dependent, the state of the detector D0 actually doesn't tell you anything about which-path, and so it doesn't collapse the wavefunction! There's been no measurement of which-path yet. If that was the end of the experiment, it would be impossible, in principle, to determine which path, and so a plot of photons across the x-axis of D0 would show an interference pattern.

But there's the idler photon. The idler photon has a choice: it can either go straight to the detector - D3 or D4 (D4 is not shown in the figure for some reason) depending on the slit - or the eraser - D1 or D2. So if it hits D3, we know it came from slit A, and if it hits D4 we know it came from slit B. If either of those two things happen, BAM the wavefunction is collapsed. There's now a correlation between the irreversible state of the macroscopic detector (D3 or D4) and the idler photon and, because of entanglement, the signal photon as well. So the wavefunction collapses - or the system decoheres - when a hit is registered at D3 or D4 and the plot from D0 shows no interference pattern.

BUT if the idler photon goes to the eraser - D1 or D2 - the wavefunction never collapses. Because of the set up, a hit at D1 or D2 is as ambiguous as a hit at D0. The beamsplitter mucks it up. A hit at D1 is still a 50/50 chance of having come from slit A or B, as is a hit at D2. So the plot from D0 shows the interference pattern because the system stays in the superpositioned state.
Between the fact that the hit at D0 told you nothing about the slit, and the fact that a hit at D1 or D2 told you nothing about the slit, there's now no opportunity - ever - to figure out which slit the photons emerged from.

I'm avoiding any interpretive discussion because inevitably we'll start arguing over causality. How can what happened in the future affect the past? I don't have an answer that I can prove, I can only take some guesses and speculate, no more than anyone else can do right now.

I can tell you that we can't send information into the past using this method, for reasons that are even more copmlicated than the above explanation, but to put it simply, the pattern at D0 never emerges until it's been compared with what happens at D1-D4, which obviously cannot occur until after the experiment has run its course. If you substituted the random beamsplitters BSA and BSB with a switch which caused the erasure to occur, and went back later and looked at the data you'd think you were sending messages into the past. But you can't _read_ the messages until the experiment's over, so they do you no good. :)

 

[ueit]

I say correlation - not causation - because we don't know what caused what. If we start talking about causation, we are almost trapped into talking about REVERSE causation, so let's not talk about that for now. :) Let's just talk about quantum states.

Momentum conservation requires that for any change in the photon's momentum, another particle/group of particles (from the mirrors, slits, etc.) must change their momentum as well. So the immediate cause of the fact that a photon is detected in one place or another is known and it consists in the photon's interaction with whatever we put in front of it. I don't think that a good understanding of DCQE (or any other quantum experiment) can be achieved without a discussion about the microscopic structure of each element that takes part in the experiment. For example it would be interesting to know how the field around one slit changes when the other is closed, or a detector is placed near it.

I think that the requirement for reverse-causality and other such effects comes from the wrong assumption that a change in the experimental setup has no influence whatsoever on the way the photon is scattered during the experiment.

 

[peter0302]

But that's the whole point of designing a delayed choice quantum eraser. The choice whether to erase the which-path information or to destroy it is made after the experiment has been set up. There are no mvoing parts in the experiment. Therefore there's nothing inherent about the physical set up that causes the interference pattern or not - the choice is made after the signal photon has been detected.

So whatever info such a microscopic analysis might yield still cannot change the fact that there is nothing physically different about the set up in the erase case vs. the non-erase case. It's entirely dependent on what the photons do.

 

[ueit]

But that's the whole point of designing a delayed choice quantum eraser. The choice whether to erase the which-path information or to destroy it is made after the experiment has been set up. There are no moving parts in the experiment. Therefore there's nothing inherent about the physical set up that causes the interference pattern or not - the choice is made after the signal photon has been detected.

So whatever info such a microscopic analysis might yield still cannot change the fact that there is nothing physically different about the set up in the erase case vs. the non-erase case. It's entirely dependent on what the photons do.

In my opinion this experiment adds nothing to the classical double-slit experiment, but only complicates things by introducing down-converters, beam-splitters and detectors. Just like in the "simple" experiment, the photons for which you have "which-path" information do not produce interference, the others do.

I think there is an objective difference between the possible paths that a particle can take and this difference comes from the fact that any object produces an EM field. A solid object consists of charged particles and those particles generate an EM field (even if the average value for a neutral object is null). Even before living the source the particle is surrounded by these fields that could provide a sort of map of what the particle will encounter. A wall with one hole produces a different field than one with two holes. A wall with two holes and one detector produces a different field than a wall with no detector. A detector that is switched on, produces a different field than a detector that is off. A detector that can be switched on after one minute is different than a detector that is permanently off (it needs a source of energy for example) etc. My guess is that when all these factors will be taken into consideration a non-puzzling explanation will emerge. Otherwise we'll learn nothing from this or any other quantum experiment just like those trying to device ingenious mechanisms to produce energy from nothing.

It is obvious that a particle for which the "which path" information has been (or even will be) extracted does not produce interference, just like it is obvious that energy must be conserved. What it is laking is not another, more complex experiment, but a microscopic explanation of this observed fact. For energy conservation we have such an explanation (each interaction between two particles must conserve energy). What is the equivalent of it for the "which-path knowledge" law? We don't even have a clear statement of the problem at microscopic level.

 

[peter0302]

What is the equivalent of it for the "which-path knowledge" law? We don't even have a clear statement of the problem at microscopic level.

We do, it's the Schrodinger Equation, and right now it's the best we've got. Unfortunately it defies all our notions of common sense and local realism, which is why all these "interpretations" pop up.

 

[ueit]

We do, it's the Schrodinger Equation, and right now it's the best we've got. Unfortunately it defies all our notions of common sense and local realism, which is why all these "interpretations" pop up.

Yeah, but the microscopic devices (source, detectors, beam-splitters, etc.) do not receive a full quantum description so the interactions I'm speaking about are simply ignored. A bigger problem is that the standard QM does not even allow such a treatment because of the measurement problem. So we have from the start a highly truncated, statistical description of the experiment.

 

[peter0302]

You're absolutely right, but my point is that the way DCQE is designed, such a treatment becomes irrelevant, because neither the items you're talking about - beam splitters, detectors, slits, etc. - nor their configuration change in the erase case vs. the non-erase case. There's no moving parts. The detectors (D0-D4) are all identical. The only difference is the odds, and so the odds are all that matter. DCQE does do something more than the classical double slit - it proves that the experimental environment is irrelevant, and that the schrodinger equation is the only thing at play. Even more importantly, the "choice" that drives everything comes after the first photon is detected. Yet there's an unmistakeable correlation between what happens at D1-D4 (later) and what happened at D0 (earlier). So even if we could scrutinize the set up microscopically, we could never account for this apparent retrocausality in any classical model.

That highly truncated, statistical description turns out to be the only determinative factor, irrespective of space or time!

Now, if you want to disregard the particle model, then sure, the photons behave exactly like waves and so why is this a shock? Well, unfortunately you can't escape the fact that they're particles, because SPDC (the method they use to make entangled photons) is one photon at a time.

 

[peter0302]

Also, ueit, that's not entirely true that the elements are never given microscopic treatment. Using QED and decoherence, you can show exactly what happens when a photon hits a beamsplitter versus a detector (what I was talking about in post #10). With QED you can show that when it hits a beamsplitter, the electrons wobble around and then return to their original state with no evidence of the interaction. With decoherence, you can show that, by contrast, the system enters a chaotic and thermodynamically irreverislbe state when the photon hits the detector and a current starts flowing toward the coincidence counter and starts affecting the transistors in the counter. All this is perfectly consistent with the "statistics-only" approach.

They didn't perform such an analysis in DCQE for reasons I stated in post #16.

 

[ueit]

You're absolutely right, but my point is that the way DCQE is designed, such a treatment becomes irrelevant, because neither the items you're talking about - beam splitters, detectors, slits, etc. - nor their configuration change in the erase case vs. the non-erase case. There's no moving parts. The detectors (D0-D4) are all identical. The only difference is the odds, and so the odds are all that matter.

Not all particles are created identical. Their momenta is slightly different. Not all of them will hit the mirrors in exactly the same place, and even if they do, the microscopic configuration of the mirror is continuously changing. So, by design, the experiment is dividing the incoming particles in two groups (which-path known and which-path unknown). In the end one can separate those groups and find out what it was expected, that the second group does show interference.

DCQE does do something more than the classical double slit - it proves that the experimental environment is irrelevant

I don't think so, see above.

...and that the schrodinger equation is the only thing at play.

Schroedinger's equation, while correct, does not tell us if we have a case of retrocausality or not, it only gives us a statistical prediction of the experimental result. It is not proven that a local-causal interpretation conflicts with it.

Even more importantly, the "choice" that drives everything comes after the first photon is detected.

I don't think we know what "drives" what. It might be that the distribution of matter arround the source "stimulates" it to produce particles with different properties. For example, a matter distribution that corresponds to an experiment that registers which-path info will determine the source to produce particles that do not produce interference.

Yet there's an unmistakeable correlation between what happens at D1-D4 (later) and what happened at D0 (earlier). So even if we could scrutinize the set up microscopically, we could never account for this apparent retrocausality in any classical model.

I think that a microscopic treatment could show how the interactions between all charged particles taking part in the experiment might produce the effect I was speaking above.

That highly truncated, statistical description turns out to be the only determinative factor, irrespective of space or time!

I disagree.

 

[peter0302]

You know what, you're right. It could be the moons of Jupiter at work for all we know and even if we were to disprove your completely non-scientific and unsupported contention that the molecules inside the mirrors, detectors, etc. might be determining whether the photon creates the interference pattern or not, you'd undoubtedly cite something else that could be at work. I can't prove a negative. But at some point you have to pick the most manageable theory that actually produces results, and this is it for now.

By the way, if you watch the Feynman lectures on QED, he discusses this possibility in the context of whether a photon gets reflected or refracted off the surface of water. He said they can direct the photon to individual molecules to test precisely what you're saying - and it still winds up being random. So if you don't believe me believe America's greatest physicist.

 

[nrqed]

Going back to the Kim et al experiment,

I think the key point in all this is the relative phase of pi between the two patterns corresponding to no path information. What introduces the relative phase difference? I guess this comes from the crystal creating the SPDC, right? What's the physical reason for this phase difference?
It's so strange that QM is "saved" by this relative phase. If it was not of this phase, then QM would collapse (no pun intended) because it would be possible to chooses after the reception of the signal photons if there should be an interference pattern or not.

 

[peter0302]

The phase difference is introudced by the beamsplitter. There's four erase scenarios:

slit A -> BS-reflect (+90-deg phase) -> D1
slit B -> BS-transmit (+90-deg phase) -> D1
slit B -> BS-reflect (-90-deg phase) -> D2
slit A -> BS-transmit (-90-deg phase) -> D2

Let's say D0 registered a hit in all four cases (was replaced by a CCD as Cramer suggests). You can easily see that the combined pattern of all four of those scenarios results in a blob; the interference patterns are perfectly 90 degrees out of phase and _constructively_ interfere to make a blob.

And you're absolutely right - I *struggled* hard over this. For the life of me I couldn't understand why this phase issue couldn't be cleverly overcome by rearranging the beamsplitters. Well, I almost drove myself mad trying to do it, so I don't encourage anyone to try. :) It can't be done, or else we'd have a time phone.

 

[nrqed]

The phase difference is introudced by the beamsplitter. There's four erase scenarios:

slit A -> BS-reflect (+90-deg phase) -> D1
slit B -> BS-transmit (+90-deg phase) -> D1
slit B -> BS-reflect (-90-deg phase) -> D2
slit A -> BS-transmit (-90-deg phase) -> D2

Let's say D0 registered a hit in all four cases (was replaced by a CCD as Cramer suggests). You can easily see that the combined pattern of all four of those scenarios results in a blob; the interference patterns are perfectly 90 degrees out of phase and _constructively_ interfere to make a blob.

And you're absolutely right - I *struggled* hard over this. For the life of me I couldn't understand why this phase issue couldn't be cleverly overcome by rearranging the beamsplitters. Well, I almost drove myself mad trying to do it, so I don't encourage anyone to try. :) It can't be done, or else we'd have a time phone.

well...then imagine the following: after the signal photons have struck D0, we replace the beam splitters BSa and BSb by two mirrors sending back the idlers to a single detector. In that case there is no pah information, no phase difference. So we should have a single interference pattern, no? But that can't be!
This is why I was sure the down converter had a key role!

 

[peter0302]

Nah, there's still a phase difference depending on whether it came from slit A or slit B, even if they both go to the same detector. You can't ever escape that no matter what you do downstream. And if you ever tried to sort it out - by, say, getting rid of the results of slit A or slit B, you'd lost the pattern because of hUP. As soon as your experimental setup distinguishes between A and B, you lose the pattern.

You can't just point both photons to the same detector either - that's not enough. You see, even in DCQE, a small minority of the photons are actually getting caught by either branch of the experiment (another reason coincidence counting is needed). You'd need to send ALL the photons to the same detector to guarantee which-path is lost. How would you propose doing that?

My idea was to use two converging lenses, the focal points of which are positioned at the output side of the slits. Theoretically these would emit parallel photons 100% of the time. Then those parallel beams would go to a third, larger converging lens, and theoretically the photons would always arrive at the focal point - where we would place the detector.

Would this work? No way - the HUP will kill you if you try to control the position (slits) and direction (parellel) of the photons so carefully.

PS - I'm not trying to be discouraging - just trying to save you the pain of going mad over this! :)

One thing I agree with you on is that it's all too damned convenient. I have to believe in my gut there is not a cosmic conspiracy at work here but, rather, that in reality the delayed choice is not really affecting what the signal photon does. That's why I understand ueit's skepticism. But no one can find a better explanation than what we have.

 

[nrqed]

Nah, there's still a phase difference depending on whether it came from slit A or slit B, even if they both go to the same detector. You can't ever escape that no matter what you do downstream. And if you ever tried to sort it out - by, say, getting rid of the results of slit A or slit B, you'd lost the pattern because of hUP. As soon as your experimental setup distinguishes between A and B, you lose the pattern.

PS - I'm not trying to be discouraging - just trying to save you the pain of going mad over this! :)

I know you are not trying to be discouraging! I actually appreciate immensely your feedback!

But my point was that I would have eliminated completely all the beam splitters! So where does the phase difference come in now? I have replaced the first beam splitters encountered by the two beams by identical mirrors. Do you see what I mean?

 

[peter0302]

Oh I thought you were keeping the third beamsplitter for D1 and D2.

Ok so you're asking would there be an interference pattern at D0 if we got rid of everything but D3 and D4 and replaced BSA and BSB with mirrors.

In other words you want all the idler photons going to a single detector.

Again you have to arrange your set up so that it's impossible in principle to discern between the two slits. Which means you have to catch _every_ idler photon. This alone is not so hard since SPDC type II is beam-like but you still have to do it in such a way that it's impossible to discern between the two slits. I don't think simply pointing both mirrors at the detector will work, since even if they were a single wavelength out of alignment it'd be possible, again, in principle, to know which side the photon came from.

But I'd encourage you to think about it. Please just don't go mad! :)

 

[nrqed]

Oh I thought you were keeping the third beamsplitter for D1 and D2.

Ok so you're asking would there be an interference pattern at D0 if we got rid of everything but D3 and D4 and replaced BSA and BSB with mirrors.

In other words you want all the idler photons going to a single detector.

Again you have to arrange your set up so that it's impossible in principle to discern between the two slits. Which means you have to catch _every_ idler photon. This alone is not so hard since SPDC type II is beam-like but you still have to do it in such a way that it's impossible to discern between the two slits. I don't think simply pointing both mirrors at the detector will work, since even if they were a single wavelength out of alignment it'd be possible, again, in principle, to know which side the photon came from.

But my setup does not seem much more complicated than the usual setup where a single photon is sent through a beam splitter, may go through either of two paths (which contain mirrors) and then sent to a detector. It seems to be quite feasible.

 

[peter0302]

It's the entangled photons that cause the problem. The fact that there's this entangled photon out there means that the signal photon won't make a pattern unless it's entangled twin's which-path info is destroyed. So any time there's a hit at D0, you have to guarantee that you've caught the idler photon and sent it to a quantum eraser that behaves identically whether the photon came out of slit A or slit B. The DCQE doesn't do that - there's a phase difference between the two slits. So you have to concoct something that uses no beamsplitters, and sends every singler idler photon to the same detector so that there's no chance of discerning where it came from.

 

[ueit]

You know what, you're right. It could be the moons of Jupiter at work for all we know and even if we were to disprove your completely non-scientific and unsupported contention that the molecules inside the mirrors, detectors, etc. might be determining whether the photon creates the interference pattern or not, you'd undoubtedly cite something else that could be at work.

If a strong correlation exists between the motion of Jupiter's moons and DCQE then I would certainly suggest to analyze how this correlation could be explained. Until such an observation is actually made you may rest sure that I will not mention those celestial objects.

If you take a microscopic perspective, the experiment is nothing but a great number of charged particles moving around. It is obviously true that the experimental result must depend somehow on how these particles move. There is nothing "non-scientific and unsupported" in this. Unless you want to argue that it is not true that solid objects consist of charged particles or that charged particles interact on a long range it is you the one who makes the "non-scientific and unsupported contention" that all these interactions do not matter. Can you give me a good reason why are you so sure about this? Also, I think you are wrong in saying that after a particle is reflected by a mirror the mirror's state remains unchanged. The conservation of momentum requires that the mirror will acquire the same momentum it imparts on the particle.

I can't prove a negative.

There is no negative to be proven.

But at some point you have to pick the most manageable theory that actually produces results, and this is it for now.

We have such a theory. The question is if the theory requires us to accept backwards causality or not.

By the way, if you watch the Feynman lectures on QED, he discusses this possibility in the context of whether a photon gets reflected or refracted off the surface of water. He said they can direct the photon to individual molecules to test precisely what you're saying - and it still winds up being random. So if you don't believe me believe America's greatest physicist.

We are limited by uncertainty to repeat the experiment with the same starting conditions, therefore no conclusion can be extracted from it. If you repeat a classical experiment with different initial parameters you get different results. This doesn't prove that classical physics requires randomness at fundamental level.

By the way, Feynman used to make a classical analogy of the double slit experiment involving bullets and a concrete wall. This was supposed to show that classical particles could not reproduce the quantum observations. I find this analogy to be deeply flawed. The classical analogous wold be charged spheres sent towards an array of dipoles. Maxwell's equations should be used to calculate particle's motion. I do not claim that in this way an interference pattern would appear, I have no idea, but Feynman's analogy remains wrong.

 

[peter0302]

If you take a microscopic perspective, the experiment is nothing but a great number of charged particles moving around.

Ok, photons aren't charged. The only relevant charged objects are the electrons in the detectors and mirrors.

It is obviously true that the experimental result must depend somehow on how these particles move. There is nothing "non-scientific and unsupported" in this.

Yes there is. You're guessing because of what you believe in your gut should be right and you're ignoring 75 years of science that says you're wrong.

Can you give me a good reason why are you so sure about this? Also, I think you are wrong in saying that after a particle is reflected by a mirror the mirror's state remains unchanged.

The reason I am so sure of this is that 75 years of experiments have not proven otherwise.

You *think* I am wrong that the mirror's state is unchanged??

The conservation of momentum requires that the mirror will acquire the same momentum it imparts on the particle.

Yes indeed, which becomes embodied in a new photon reflected off the surface.

There is no negative to be proven.

The negative you wish me to prove is that the microscopic workings of the experimental set up work together in a grand conspiracy to yield the results we see and that they're not truly random. I can't prove that.

We have such a theory. The question is if the theory requires us to accept backwards causality or not.

Indeed. But no mainstream theory believes that the microscopic movements of the molecules in the experimental components are the answer.

By the way, Feynman used to make a classical analogy of the double slit experiment involving bullets and a concrete wall. This was supposed to show that classical particles could not reproduce the quantum observations. I find this analogy to be deeply flawed. The classical analogous wold be charged spheres sent towards an array of dipoles. Maxwell's equations should be used to calculate particle's motion. I do not claim that in this way an interference pattern would appear, I have no idea, but Feynman's analogy remains wrong.

I wonder why the double slit works with uncharged neutrons then...

 

[nrqed]

It's the entangled photons that cause the problem. The fact that there's this entangled photon out there means that the signal photon won't make a pattern unless it's entangled twin's which-path info is destroyed. So any time there's a hit at D0, you have to guarantee that you've caught the idler photon and sent it to a quantum eraser that behaves identically whether the photon came out of slit A or slit B. The DCQE doesn't do that - there's a phase difference between the two slits. So you have to concoct something that uses no beamsplitters, and sends every singler idler photon to the same detector so that there's no chance of discerning where it came from.

Thanks Peter.

I am not trying to be apain in the neck but I still don't see the problem with my proposal:
After the signal photons have hit the screen, you take away the beam splitters BSa and BSb and you insert mirrors that send ll idlre photons to a single detector. Of course in real life there will be efficiency issues but in theory, all which path information has been destroyed. And since there is no beam splitter involved at all, I don't see any phase shift anywhere. Everything is completely symmetrical. So what gives?

 

[peter0302]

Oh it's not a pain, it's just bringing back painful memories of going mad! :)

Right I know what you're saying. I'll make it even more simple. Take away the prism and instead use a converging lens and put a single detector at the focal point. What gives?

What gives is that it's still only a small minority of the photons that are being detected. The rest float off into space and help destroy the interference pattern. The coincidence circuitry, among other things, isolates the photons that actually _are_ detected so that the analysis can performed on just those. Without that, if you just looked at D0, there'd be far too much noise.

You've got to get the overwhelming majority of idler photons from the BBO crystal to hit the detector. I don't know how you can do that without disturbing the interference pattern because if you place anything too close to the crystal you learn which-path (in principle) but if it's too far away there's too much uncertainty in the location of the photons to capture enough of them.

That's why I thought of using lenses behind the slits to focus the beams but I don't think the HUP will let us control both the position and direction of the photons with such precision.

 

[nrqed]

Oh it's not a pain, it's just bringing back painful memories of going mad! :)

Right I know what you're saying. I'll make it even more simple. Take away the prism and instead use a converging lens and put a single detector at the focal point. What gives?

Right

What gives is that it's still only a small minority of the photons that are being detected.
.

Why? We are taking about a gedanken experiment so I am not sure if your point is about a practical limitation or a fundamental one. Why can't we assume that they are all detected?
If th elimitation is only a question of experimental precision, then it makes the rest of the argument shaky

The rest float off into space and help destroy the interference pattern. .

how?

.

The coincidence circuitry, among other things, isolates the photons that actually _are_ detected so that the analysis can performed on just those. Without that, if you just looked at D0, there'd be far too much noise.
.

Why? It seems to me that every single pair now has gone through a set up in which the which-way information is not available. Hence an interference pattern. I don't see how losing some of the photons could change the result. The ones that are not counted, it seems to me, would still be part of the interefrence pattern since no which way information si available for those as well. So we simply take an interference pattern and removes some of the photons, we simply get a weaker (in the sense of less intense) interference pattern, no?


.

You've got to get the overwhelming majority of idler photons from the BBO crystal to hit the detector. I don't know how you can do that without disturbing the interference pattern because if you place anything too close to the crystal you learn which-path (in principle) but if it's too far away there's too much uncertainty in the location of the photons to capture enough of them.

Again it seems to me as if you are talking about practical limitations, not fundamental ones. And again, if there is no which path information for any of the photons, it seems to me that it does not matter if we don't catch all of them anyway. I don't see how to take away photons from an interference pattern and leaves something that will look like the combination of two interference pattern shifted by pi! (and *even* if that was possible, it would be incredible to think that the photons missed are just the right ones to accomplish this!). So there is something I am clearly missing!

That's why I thought of using lenses behind the slits to focus the beams but I don't think the HUP will let us control both the position and direction of the photons with such precision.

But we could instead simply use some mirrors to redirect the idler photons to a single detector.


Thanks for your input.

 

[ueit]

Ok, photons aren't charged. The only relevant charged objects are the electrons in the detectors and mirrors.

The electron shells from which the photons originate are charged. The photon's properties are therefore correlated to whatever fields reach the source.

Yes there is. You're guessing because of what you believe in your gut should be right and you're ignoring 75 years of science that says you're wrong.

Can you be more precise what experiment/theoretical result has shown that?

The reason I am so sure of this is that 75 years of experiments have not proven otherwise.

Same comment as above.

You *think* I am wrong that the mirror's state is unchanged??

That's a fair assessment.

Yes indeed, which becomes embodied in a new photon reflected off the surface.

The incoming photon has a momentum m. The reflected one has -m (on the axis that makes a 90 degree angle with the mirror). Now, -m does not equal m, does it?

The negative you wish me to prove is that the microscopic workings of the experimental set up work together in a grand conspiracy to yield the results we see and that they're not truly random. I can't prove that.

So, if you can't prove that, why do you still claim that "75 years of science" have proven me wrong? Please decide which of the statements:
1. I cannot be proven wrong.
2. I have already been proven wrong,

better reflects your position because both of them cannot be true at the same time.

A "conspiracy" is usually imposed top -to- down. My suggestion is nothing of this sort, on the contrary.

Indeed. But no mainstream theory believes that the microscopic movements of the molecules in the experimental components are the answer.

A theory has no "beliefs". It is your belief that by applying the theories we have to the whole experimental setup (making, of course, simplifications, but being careful to base them on the microscopic behavior and not on a macroscopic general behavior) nothing interesting could emerge. May be you could add some substance to this claim.

I wonder why the double slit works with uncharged neutrons then...

The neutrons are made of charged quarks and they also have a non-zero magnetic moment.

 

[peter0302]

The electron shells from which the photons originate are charged. The photon's properties are therefore correlated to whatever fields reach the source.


Can you be more precise what experiment/theoretical result has shown that?

Feynman's online QED lectures talk about this in the context of reflecting light off of water. He addresses your argument: "Isn't it just a matter of precision? Couldn't we aim a photon at a precise spot on the water - say directly at an electron - and guarantee reflection or refraction?" and he goes on to state that "no, we've tried with very precise instruments and we cannot do that."

Now, I don't know if he's being literal here but I understand his point. The uncertainty principle won't allow you to aim a photon directly at a particular electron and guarantee a hit. That uncertainty carries over into the reflection probability and therefore you cannot in principle accomplish what you're talking about doing. In other words you can build a classical model of photons and electrons as billiard balls and try to analyze a beam-splitter by saying "what are the odds of the photon hitting an electron and being reflected versus the odds of it being passed through?" And you can't do it because no matter how close you look the odds will be what the odds will be, because of the HUP.

I can't point you to specific experiments. Maybe someone else here can?

The incoming photon has a momentum m. The reflected one has -m (on the axis that makes a 90 degree angle with the mirror). Now, -m does not equal m, does it?

You just answered your question. m + -m = 0, which is the change in momentum of the mirror after the reflection. Hence, the state of the mirror is unchanged.

So, if you can't prove that, why do you still claim that "75 years of science" have proven me wrong? Please decide which of the statements:
1. I cannot be proven wrong.
2. I have already been proven wrong,

If the HUP is right you are wrong, and since in 75 years the HUP has not been proven wrong, so far the odds aren't looking good for you. That doesn't mean you are wrong necessarily but I think the burden is on you at this point to prove otherwise.

A "conspiracy" is usually imposed top -to- down. My suggestion is nothing of this sort, on the contrary.

Well, the conspiracy could have begun at the big bang with the breath of god. Science can never disprove that which is why we don't like to talk about it.

A theory has no "beliefs". It is your belief that by applying the theories we have to the whole experimental setup (making, of course, simplifications, but being careful to base them on the microscopic behavior and not on a macroscopic general behavior) nothing interesting could emerge. May be you could add some substance to this claim.

I think any delayed choice experiment is the substance to my claim.

 

[peter0302]

Oh, nrqed, I am going to respond to your post when I have more time. Got super busy at work today. :) But by the way, I just want you to know that I hate you. You've awoken the beast in me like I knew you would!

 

[ueit]

Feynman's online QED lectures talk about this in the context of reflecting light off of water. He addresses your argument: "Isn't it just a matter of precision? Couldn't we aim a photon at a precise spot on the water - say directly at an electron - and guarantee reflection or refraction?" and he goes on to state that "no, we've tried with very precise instruments and we cannot do that."

Now, I don't know if he's being literal here but I understand his point. The uncertainty principle won't allow you to aim a photon directly at a particular electron and guarantee a hit. That uncertainty carries over into the reflection probability and therefore you cannot in principle accomplish what you're talking about doing. In other words you can build a classical model of photons and electrons as billiard balls and try to analyze a beam-splitter by saying "what are the odds of the photon hitting an electron and being reflected versus the odds of it being passed through?" And you can't do it because no matter how close you look the odds will be what the odds will be, because of the HUP.

What you are saying is perfectly true. HUP does not allow us to use experiment to decide on the probabilistic nature of QM. This is what I was saying in one of my previous posts. Anyway, I don't see the relevance of this to my point that the emission of a photon is influenced by an electric field. This is true, regardless of any interpretation. On the other hand a realistic interpretation of QM is probably required to understand the causal chain, but again, HUP is not a problem as the existence of such an interpretation (Bohm's) has shown us.

I can't point you to specific experiments. Maybe someone else here can?

I can wait.

You just answered your question. m + -m = 0, which is the change in momentum of the mirror after the reflection. Hence, the state of the mirror is unchanged.

No, in this example momentum is not conserved. "Conserved" means it remains the same. If you apply the above calculation to a free, non-interacting particle at two time points you get m(initial momentum) + m (final momentum, unchanged because nothing happened) = 2m. So your particle will accelerate without any force acting on it which is clearly wrong.

If the HUP is right you are wrong, and since in 75 years the HUP has not been proven wrong, so far the odds aren't looking good for you. That doesn't mean you are wrong necessarily but I think the burden is on you at this point to prove otherwise.

The HUP is right, we know that but I don't see why I should be wrong. A charged particle still interacts with other charged particles.

Well, the conspiracy could have begun at the big bang with the breath of god. Science can never disprove that which is why we don't like to talk about it.

I don't see any analogy between your poetic words and the fact that charged particles interact with other charged particles.

I think any delayed choice experiment is the substance to my claim.

Well, you didn't show me any argument, just a restatement of your original position

 

[peter0302]

No, in this example momentum is not conserved. "Conserved" means it remains the same. If you apply the above calculation to a free, non-interacting particle at two time points you get m(initial momentum) + m (final momentum, unchanged because nothing happened) = 2m. So your particle will accelerate without any force acting on it which is clearly wrong.

That's just nonsense. Go back and study mechanics.

As for the rest, I don't know what else to say to convince you. Sorry.

 

[ueit]

That's just nonsense. Go back and study mechanics.

I think you should go back to introductory arithmetics.

From wiki:

"The law of conservation of linear momentum is a fundamental law of nature, and it states that the total momentum of a closed system of objects (which has no interactions with external agents) is constant."

Do you agree with the above statement?

If yes, let's calculate the total momentum before reflection:

Mi (initial total momentum) = m (initial photon's momentum) + 0 (assume the mirror doesn't move in that direction) = m
Mf (final total momentum) = -m (photon's momentum after reflection) + 0 (you claim that the mirror remains unchanged) = -m

So, does m equal -m?

 

[peter0302]

So, does m equal -m?

For purposes of what we've been talking about, yes.

Do you understand vectors?

 

[ueit]

Do you understand vectors?

I think I do.

 

[dst]

The magnitude of momentum is conserved in all cases.

Let's apply vectors to it in momentum space, mirror is unchanged -

Before - m(p) = sqrt(x^2 + y^2)
After - m(p) = sqrt((-x)^2 + (-y)^2) = sqrt(x^2 + y^2).

So it is conserved.

 

[hyperon]

I am not exactly convinced that finding which path information necessarily collapses the photon's wavefunction. If a single detector is placed in front of only one slit, this does collapse the wavefunction if the photon passes through that slit, since the photon would have interacted with the detector.

However, if the photon passed through the other slit, there would be nothing for it to interact with before reaching the screen. This situation is then equivalent to one with no detector in front of the first slit. Which path information is still preserved because the nondetection of the photon at the detector implies that it can ideally only have passed through the second slit. In this case, we would know which path information without collapsing the photon's wavefunction.

Perhaps I am treating this too classically. Please correct me if I am wrong.

On further thought I tried considering the possibility that the detector somehow disturbs the entire wavefunction of the photon from one end of the setup where it is placed. But this would not follow from peter0302's interpretation of collapse of wavefunction by an irreversible interaction with macroscopic objects in the experiment.

 

[ueit]

The magnitude of momentum is conserved in all cases.

Let's apply vectors to it in momentum space, mirror is unchanged -

Before - m(p) = sqrt(x^2 + y^2)
After - m(p) = sqrt((-x)^2 + (-y)^2) = sqrt(x^2 + y^2).

So it is conserved.

Do you claim that only the magnitude and not also the direction of the momentum vector is conserved?

 

[Count Iblis]

You'll never get a paradox if you try to use a mirror to extract the "which path information" by measuring the mirror's recoil. To get an interference pattern the wavefunction of the mirror must be such that the mirror is confined within some fraction of the photon's wavelength. This means that the mirror's wavefunction in momentum space will have a width that is larger than the photon's momentum (this is the uncertainty principle).

So, by measuring the momentum of the mirror, you cannot be sure whter or not the mirror has recoiled. Now, you could still make statistical predictons, because when the mirror recoils its's wavefunction does change. The sharper peaked the mirrors wavefunction in momentum space is the better you see if the mirror has recoiled. But then the wavefunction in ordinary space will become wider and the interference pattern will become less visible anyway.

According to quantum mechanics, the interference term is proportional to the overlap of the wavefunctions of the mirror when it recoils and when it doesn't. The interference pattern can only vanish completely if this overlap is zero, i.e. if the two states of the mirror are orthogonal. In that case they can be considered to be eigenstates of an operator which upon measurement would yield the which path information. But if the two states are not orthogonal, then you cannot have an operator such that the two states are the eigenstates because two eigenstates corresponding to different eigenvalues are always orthogonal.

 

[peter0302]

Do you claim that only the magnitude and not also the direction of the momentum vector is conserved?

Ueit, you're overthinking this. When you bounce a billiard ball off the side of the table, the ball bounces off with almost all of the momentum it had before. If this were in a vacuum and there was no such thing as sound or thermodynamics, the momentum would be exactly what it was before. Do you not think momentum is conserved in that case?

Thinking classically, the momentum from the photon is SO SMALL compared to the force of the other atoms in the mirror that 99.99999999999% of the momentum is imparted back to the new photon. So there might be a miniscule change in the mirror but it's still indistinguishable from random molecular movement.

Thinking quantum mechanically, what Count Iblis said.