Sunday, December 4, 2016

A measurement can be more than an observer learning the value of a physical observable


Last post created quite a stir and I want to expand on the ideas from it. This will also help me get out of an somewhat embarrassing situation. For months now Lubos Motl tried to get revenge on his bruised ego after a well deserved April Fool's joke and became a pest at this blog. The problem is that although I have yet to see a physics post at his blog that is 100% correct, we share roughly the same intuition about quantum mechanics: I agree more much more with his position than say with the Bohmian, GRW, or MWI approaches. The differences are on the finer points and I found his in depth knowledge rusty and outdated. For his purpose: to discredit the opposite points of view at all costs this is enough, but it does not work if you are a genuine seeker of truth.

So last time he commented here: "A measurement is a process when an observer actually learns the value of a physical observable" which from 10,000 feet is enough. However this is not precise enough, and now I do have a fundamental disagreement with Lubos which hopefully will put enough distance between him and me. 

More important than my little feud with Lubos, I can now propose an experiment which will either validate or reject my proposed solution to the measurement problem. I do have a novel proposal on how to solve the measurement problem and this is distinct from all other approaches. I was searching for months for a case of a novel experimental prediction, but when I applied it to many problems I was getting the same predictions as standard quantum mechanics. Here is however a case where my predictions are distinct. I will not work out the math and instead let me simply present the experiment and make my experimental claim.


Have a box with a single particle inside. The box has a middle separator and also two slits A and B which can be placed next to a two-slit screen. We can then carry two kinds of experiments:

  1. open the two slits A and B without dropping the separator allowing the particle to escape the box and hit a detector screen after the two-slit screen.
  2. drop the separator and then open the two slits A and B allowing the particle to escape the box and hit a detector screen after the two-slit screen.
Next we repeat the experiments 1 or 2 enough times to see the pattern emerge on the final screen. Which pattern would we observe?

For experiment 1 we already know the answer: if we repeat it many times we obtain the interference pattern, but what will we get in the case of experiment number 2?

If dropping the separator constitutes a measurement, the wavefunction would collapse and we get two spots on the detector screen corresponding to two single slit experiments. If however dropping the separator does not constitute a measurement, then we would get the same interference pattern as in experiment 1.

My prediction (distinct from textbook quantum mechanics) is that there will be no interference pattern.


21 comments:

  1. Florin,


    1. Inserting a barrier in a box is not a measurement.

    2. I think you are right that weighing the separated boxes represents a measurement (you find out that the particle is in the heaviest box). You don't need to weigh it, moving it is enough. Accelerating the box and recording the force needed to get that acceleration gives you the mass. Accelerating the box changes the state of the particle inside and the superposition is lost.

    3. Discussion of your proposed experiment:

    a. You need your particle to enter the box from the opposite side so that it has an initial momentum parallel to the separator and perpendicular to the screen. If not, there is no reason for the particle to get in that direction. You will not get two spots and neither interference fringes, but some random detection events anywhere on the screen or the particle might even remain in the box.

    b. If the box is very narrow (in the direction of particle's motion) you will get interference, with or without the separator. It is basically the double slit experiment with a hollow slitted wall.

    b. If the box is large in the direction of particle's motion and the separator is present there is less and less overlap of the waves behind the box so the interference will continuously fade (it only disappears completely for an infinitely large box). But the collapse only takes place when the particle is detected at the screen, not when the barrier is inserted. Even for an infinitely large box with the separator in place the particle is in a superposition of the two possible paths (A and B).

    4. I still do not see the relevance of all this in regards to the locality dilemma. Even if you are right and the proposed experiment (in the previous post) with mobile boxes destroys the superposition there are many ways in which to prepare a particle in a superposition of two or more distinct locations. I am looking forward to see your resolution of the dilemma. Lubos didn't provide one, either.

    Andrei

    ReplyDelete
  2. I wrote a comment on my blog before I saw the second part of your blog post. In that comment, I wrote that it's trivial to make an experiment you want - and it's been done hundreds of times.

    In a double slit experiment, you simply divide the space to 2 parts, with 1 slit on each side, for a picosecond during the time when the particle is moving through the slit(s).

    You say that this insertion of the wall amounts to a measurement of the which-way information and the interference pattern goes away. I am saying that every sane person knows that this claim is totally ludicrous. The particle has 0% probability to be near the plate that separated the space, so obviously the presence of the plate doesn't make any difference at all, and the interference pattern will still be there. There's no disappearance of the pattern or decoherence or anything of the sort.

    ReplyDelete
    Replies
    1. Lubos,

      "The particle has 0% probability to be near the plate that separated the space"

      Not true, the plate is located at the interference maximum so the probability of the particle to be there is large. If the plate is large enough the interference disappears.

      Florin didn't say the plate remains for a picosecond, but until the particle is detected at the screen. If the plate is inserted and removed before the particle is detected, the interference might remain but this was not the experiment Florin proposed.

      Andrei

      Delete
    2. "You say that this insertion of the wall amounts to a measurement of the which-way information" - not at all

      "and the interference pattern goes away" - that is correct

      I am not taking a realist position or any ontology here: the role of the observer is still paramount. There is a very precise and *unique* mathematical way in which an observer can be introduced.

      My claim is based on my definition of "observer" which is more general than what is usually assumed. A measurement takes place when a certain equivalence relationship is broken. The very presence of the wall breaks the equivalence and this results in a wavefunction collapse. In turn this makes the interference vanish.

      Your comment is reminiscent of Afshar experiment setting. I say cut the box first then open the slits, not when the particle is in flight. By the way, the best criticism of Afshar was done by Ruth Kastner IMHO.

      Delete
    3. You're really a completely clueless crank, Florin. If you place some objects (a wall) at places where a particle is certain not to be located, the effect on the particle's future behavior is obviously non-existent, so the interference pattern cannot disappear. It follows from locality. It's really a completely commonsense example of locality - only unhinged loons may misunderstand it.

      Delete
    4. "If you place some objects (a wall) at places where a particle is certain not to be located, the effect on the particle's future behavior is obviously non-existent"

      Collapse is not a dynamical process, and you objection is bogus.

      Delete
  3. This is interesting, but I doubt that splitting the box amounts to a measurement. Here is an argument.

    I will use a Mach-Zehnder interferometer, with a small modification, please see this picture http://i.imgur.com/K4zVMhj.png

    Suppose you send a particle which is split and goes through both arms (fig 1). The arms are considered to be pipes, so the particle is isolated while inside an arm. Then this amounts to the particle being in two separate boxes. But then, the two "halves" interfere and combine again. Now, you may argue that the pipes are not boxes, that they have open ends.

    However, let me modify the Mach-Zehnder device, by adding separators at the ends of the pipes.
    - Before the particle is fired, let the "in" ends of the pipes be open, and the "out" ends be closed. Then fire the particle (fig. 2)
    - Immediately after the particle enters the pipes, close the "in" ends of the pipes. The particle is now inside two closed pipes, so inside two separated boxes, simultaneously (fig. 3)
    - After a time shorter than it takes the particle to go through the pipes, open the "out" ends of the pipes. Now the particle can interfere with itself and combine back (fig. 4).

    If I still haven't convince you, at least you have an experimental setup that is more practical to implement. In your experiment proposal, the particle is in the box, but to have observable interference, you need to have well-defined frequencies. It is difficult to store the particle in the box while keeping its frequencies and phases, and therefore this makes impossible to predict how the interference looks like. My proposal allows you to do this, and to predict the interference effect, in this case it is simply that the particle hits the target in ~100% of cases.

    ReplyDelete
    Replies
    1. If someone doubts that it is practical to make this experiment, here is another modification. Simply replace the separators at the ends of the pipes with helices, that close the pipes periodically. The length of the pipes can be calculated so that we can be sure that the photons can get out and while inside the pipes there is always a moment when the pipes are closed.

      Delete
    2. Florin, here is an experiment which is almost like yours. Use a one-way mirror to close the photon from the source inside the box at the "in" end, and something like you propose at the out ends, and do the two-slit experiment. Figure here: http://imgur.com/a/9Tvow

      Delete
    3. Hi Cristi,

      Nice to hear from you. You have very nice arguments which are harder to refute, but let me try. On the last image with semi-transparent wall I agree, there will be an interference there. Collapse can happen when the experimental context changes and this is not the case for http://imgur.com/a/9Tvow

      Now on the interferometer example. In the case of helices the interference happens as well but you'll see in a sec why I say so and I (kind of) disagree with your first post.

      On the first post consider this: cut the two arms in half and "bottle up" the photon with two mirrors at the two ends while the photon is in flight. Send one half to London and another to Tokyo. We know that opening them will result in the photon being discovered either in London or in Tokyo. But can we know in which half is the photon without opening them? Yes we do. The moment the photon reaches the end and bounces off the mirror, it will create a recoil on the mirror which will allow its detection. Effectively the end mirror acts as a detector and once this happens, recombining the two arms will no longer result in interference. So for your argument with the particle propagation, as long as we do not know the "which way" information the interference will happen. But when the particle becomes trapped the interference no longer happens.

      Quantum mechanics is contextual: change the experimental context and when a certain equivalence is no longer possible will result in collapse. this is the gist of my argument.

      PS: Lubos objection was trivial to counter, but it took me some time to figure out the answer to your particle in flight challenge.

      Delete
    4. Hi Florin,

      Good to hear from you too (well, I hear from you more, because I am watching your posts, but I didn't blog for long time).

      I don't think bouncing on the mirror is a detection, unless you measure the momentum change on the box caused by the bounce. If the bounce constitutes a measurement, then the simple interference experiment with the Mach-Zehnder in Figure 1 will never result in interference. Because there you have a bounce inside the arms, on the mirrors at the corner. It is not frontal, but I think it is still a bounce. And if you wish, we can complicate the Mach-Zehnder device with a frontal bounce at the corner of the arm, and the reflected photon can then be bounced at 90 degrees with a half-way mirror at 45. In fact, frontal bounce takes place in the Michelson interferometer.

      Delete
    5. Also there is bouncing in the two-slit experiment. Knowing the momentum destroys the interference (see Bohr–Einstein debates, discussion surrounding Fig. C.

      Delete
    6. "If the bounce constitutes a measurement, then the simple interference experiment with the Mach-Zehnder in Figure 1 will never result in interference."

      Indeed, I fully agree.

      "I don't think bouncing on the mirror is a detection, unless you measure the momentum change on the box caused by the bounce." - I don't agree.

      The "which way" information is obtain by weighing the arms of the interferometer, not by the momentum transfer. One closed arm is heavier than the other.

      Delete
    7. Hi Florin,

      Of course, I agree that if you weigh one or the arms or the boxes, this constitutes a measurement and destroys the interference. I didn't say that the only way to cause collapse is by measuring the momentum change caused by the bounce :)

      You said "The moment the photon reaches the end and bounces off the mirror, it will create a recoil on the mirror which will allow its detection."

      What I said is that bounce by itself doesn't cause collapse, unless you measure its effect, which you said is the change of momentum of the mirror.

      Weighing is another way to measure in which arms of the interferometer, or in which box, is the particle.

      To show you that the bounce itself is not a measurement, I explained that bounce happens both in the Mach-Zehnder, and in the Michelson interferometers, yet the interference is not destroyed.

      Best regards,
      Cristi

      Delete
    8. Cristi,

      I think we agree in the end. Bouncing by the mirror at the end causes the collapse not by momentum measurement but by Energy-Time uncertainty relationship: bouncing buys time to allow the energy uncertainty to be reduced to the point it will reveal the existence or not of the particle.

      Delete
    9. Hi Florin,

      I don't understand how you apply the energy-time uncertainty so that for my figure 1 bouncing doesn't cause collapse, but for the experiment in figures 2-4 it does. I would be interested if you have more details, in a paper you wrote or perhaps in a future post. You made me curious with this one.

      Best regards,
      Cristi

      Delete
  4. Let me summarize the rebuttal against the 3 kinds of objections above:

    Andrei: "Inserting a barrier in a box is not a measurement." This is just a statement. I have a mathematical formalism which clearly defines what is a measurement. I presented this in prior posts and I will not repeat it here

    Lubos: "If you place some objects (a wall) at places where a particle is certain not to be located, the effect on the particle's future behavior is obviously non-existent" This argument would be valid if collapse would have been a dynamic process. However it is not and the argument is vacuous.

    Cristi: "However, let me modify the Mach-Zehnder device, by adding separators at the ends of the pipes."

    Adding separators acts as detectors when the particle will bounce of them. Once this happens the interference vanishes. While in flight with no interaction there will be interference, but once the transitory period ends we have the which way information killing the interference. The original example assumed a stationary case and my original argument/prediction still stands.

    ReplyDelete
    Replies
    1. The Mach-Zehnder interferometer experiment works, although there is a bounce inside its arms (see Figure 1).

      Delete
    2. Florin,

      It seems to me that you are not interested in having an honest debate anymore. You didn't answer to any of the points I have raised (the fact that the experiment is not doable as presented and the irrelevance of this experiment in regards to locality dilemma).

      I don't feel compelled to participate anymore.

      Andrei

      Delete
  5. Sorry for the delay, I planned to answer yesterday but I was simply too tired and went straight to bed.

    Andrei: I don't get how I could have offended you, but oh well, it's a free village.

    Cristi: so if I understood you correctly, say in the Bohmian interpretation you can "bottle up" the quantum potential in one arm of the interferometer, ship this arm to Tokio and back, recombine, and still get the interference? That would make up for a nice business proposal: "empty" bottles full of quantum potential for sale. ha ha.

    Let me read your posts above and see if I agree...

    ReplyDelete
    Replies
    1. ha ha ha :)))

      I like this, ""empty" bottles full of quantum potential for sale" :)))

      paradoxically, from Bohmian mechanics I would only buy the pilot-wave, not the "particle" :)

      Delete