Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

B Does observation *Determine* the past?

  1. Jul 30, 2017 #1
    Say a photon was emitted from very far away such that it would take billions of years to reach us. According to Feynman's infinite paths, the photon travels every possible path on its way to reach us. The superposition of infinite trajectories collapses when it hits something solid, say an observer on earth for example.

    Then the trajectory of the photon is determined upon observation which determines it's past( relative to us) as well, essentially writing billions of years of history for that photon, which is rather astounding.

    This is just for one photon, but it could work for many as well, in principle.

    Does this translate over to observations of the CMBR? That is, the observations of the photons by us are localizing them to a definite spacetime position in the past? This is very different than saying that we, via CMBR, are observing the past.
  2. jcsd
  3. Jul 30, 2017 #2
    Does observation include interaction with inanimate objects. if so then to reach you, it have to avoid all other items regardless of its super position, if there had been something before you it have happened there etc..
    sorry if this is stupid aha
  4. Jul 30, 2017 #3
    Yes, that is implicit in my example of a particle from far away. I am assuming it avoids everything on it's way here.
  5. Jul 30, 2017 #4
    Ok so by observing this partical you essentially influence the path it has taken, I guess that seems like a problem with applying current human understandings to something that defeys everything we know. Again not trying to be annoying with inferior knowledge, just find all these things rather interesting
    Understanding is probably the wrong word but more our perception of how time works and our linear view of everything
  6. Jul 30, 2017 #5
    Well, quantum is the most successful scientific theory in terms predictive ability. So in a sense, we know more about quantum than what we know about other things.
  7. Jul 30, 2017 #6
    That's almost a bit sad, however knowledge isn't the same as being able to comprehend something. We know that these things happen, even I've heard of them aha. But to truely be able to rap your head around the fact that some thing can be in many places at once is surely not possible, to be able to comprehend infinitey is another example.
  8. Jul 30, 2017 #7
    There are plenty of things that work that challenges us philosophically. Imaginary numbers are difficult to intuit, yet builds a completely self consistent mathematical system that just works logically. Same with negative numbers. When humans found out the earth is round, they had a difficult time wrapping their heads around that, but today we don't. Thats because the notion that the earth is round is ingrained in us since we were toddlers. But Quantum isn't really taught from kindergarden, so thats why many have a hard time accepting it initially.
  9. Jul 30, 2017 #8
    That's all very valid and I accept that. That's not really what I mean though, i can sit and read about it and roughly understand the principle of it. I cannot however in my head think of an object in multiple points in space because like me my thought process can only exist with one object in a linear direction.
    Ultimately I think your probably right in the fact that you do influence the past, it's the only logical outcome given the quantum stuff "aha". But the problem comes in our linear perception and our brains predisposition to believe it cannot be possible to influence the past. It's like newtons laws, pretty much true ha.
  10. Jul 30, 2017 #9


    User Avatar
    Gold Member

    Nor does rapping your head around this "fact" need to be possible, since it isn't true.
  11. Jul 30, 2017 #10


    User Avatar
    Staff Emeritus
    Science Advisor

    That's not quite what it means. The Path Integral Formulation is a way of computing probability amplitudes for quantum systems. It does not mean that a photon (or any other particle) travels every possible path. The idea that a particle has a set path from one point to another does not translate well from the classical to the quantum realm. All we can say is that we will find a photon at a certain place upon observation. We cannot say that it took any one path (or an infinite number of paths) to arrive there.
  12. Jul 30, 2017 #11


    User Avatar
    Staff Emeritus
    Science Advisor

    I'm not that up to speed on retrocausality in the context of quantum theory, but I'd take this conclusion with a grain of salt. As far as I'm aware, there's no way to verify whether retrocausality is correct and the results of retrocausality experiments can be described in other ways that do not invoke or require it.
  13. Jul 30, 2017 #12


    Staff: Mentor

    No, that's not what Feynman's path integral says. The path integral is a way to calculate the amplitude for different possible results of measuring the photon. It is not a literal description of the photon's state as a superposition of trajectories.

    No, that's not correct. See above.

    There are particular measurements you can make that give the photon a definite momentum (or nearly definite in any practical sense, since no measurement has infinite accuracy) after the measurement. On a collapse interpretation, these particular measurements can be thought of as collapsing the photon into a state of definite momentum, which it most likely was not in before the measurement. But that's only on a collapse interpretation; it's not part of the basic math of QM. And these measurements have to be carefully designed; just hitting a solid object isn't enough.

    No, since it's wrong to begin with. See above.
  14. Jul 31, 2017 #13
    Interesting. From my understanding of the double slit experiment, there are two outcome states. |left> and |right>. So the states only refer to the potential outcomes which must be observable? Is that why it is meaningless to ascribe superposition of the trajectories?

    Doesn't Feynman's thought experiment relied on breaking down the wall to hold infinite slits, with many of the same walls behind it, stacked? So if |left> and |right> could be in super position, then why not |first> ,|second>, .....|infinite>?

    Can we not think of every time the photon passes though these infinite slits as somewhat of a measurement occurring, since if it were infinite slits, then the wave function would be continuously collapsing? I thought Feynman was just generalizing the double slit experiment to all of space. Although I could see the issue of needing to collapse infinite states with no appearent cause of collapse at each step of the way.

    Also, so are you saying that quantum works well in the way we think in carefully controlled experiments? Even though we cannot calculate what happens in interstellar space, some quantum effects should still work in principle, since photons are just quantum particles. Well, we can get rid of collapse interpretation issue via MWI but then I wouldn't have posted in the first place since in that case, retrocausality is not even be of consideration to allow for a discussion.
  15. Jul 31, 2017 #14
    It should be inherently unverifiable. But the idea is an interesting curiosity.

    My post was mostly about retrocausality. I probably should have stated that the photon's ontic status was undetermined and only is described via it's wave function until measured.

    I'm confused about the path integral. Feynman's thought experiment appeared to me to be a generalization of the double slit experiment, with infinite walls and infinite slits on each wall, which made me think that at every planar slice of space(i.e imaginary wall), the wave function would somehow collapse into one of those states and that this would happen continuously.
    Last edited by a moderator: Jul 31, 2017
  16. Jul 31, 2017 #15


    Staff: Mentor

    I don't know where you're getting that from, but it's wrong. What textbooks or peer-reviewed papers have you read?

    For a single photon in a double slit experiment, there are an infinite number of possible outcome states, corresponding to all the different possible locations on the detector where the flash of the photon arriving could be detected.

    For a heuristic way to see how the path integral comes about, yes. But the path integral doesn't mean what you think it does, as I've already said.

    I'm saying that in order to apply QM at all, you have to define what the "experiment" is--how you are preparing the quantum system or systems, what interaction or interactions are present, and what measurement or measurements you are making.

    Sure, there could be interactions that the photons could undergo in interstellar space. Astronomers calculate such things all the time. But they make such calculations in order to compare with actual measurements made here on Earth; that's the only actual data we have. We can't make measurements of what photons are doing a billion light years away; we can only measure things that come to Earth possibly carrying information about such happenings.

    If retrocausality is interpretation dependent then there's not much we can discuss about it here, since there's no way to experimentally test things that are interpretation dependent.
  17. Jul 31, 2017 #16
    I started on Susskind's book, but I'm only on the spin part, so trying to apply discrete states to continuous without thinking is a mistake on my part. Yes it makes sense. The wall is continuous. But the probability should be weighted either left or right, making a bimodal distribution? So even though it's not strictly binary, theres still some sense of classification of "leftness" and "rightness" of a particular point on the board due to it's location relative to the clustering pattern.

    Ah ok, got it. I've completely misinterpreted the heuristic. I'll have to look into more on why he set up the infinite walls the way he did.

    Hmm, that makes sense. It would be hard to do experiments on only observational data.
  18. Jul 31, 2017 #17


    User Avatar
    Science Advisor
    Gold Member
    2017 Award

    In addition one should keep in mind that a photon doesn't have a position (at least not in the sense a massive (quantum) particle has one), and the path integral for photons is not taken over paths in configuration space but over all physically distinct field configurations.
  19. Jul 31, 2017 #18


    Staff: Mentor

    No. The probability is whatever QM predicts. You don't apply any "weighting" a priori. You apply QM and figure out what it tells you.

    No. See above.
  20. Jul 31, 2017 #19
    I don't mean that you manually weight the probabilities before hand. There are algorithmic techniques that can analyze clustering of data points without specifying a label on the data a priori. So if you look at a bimodal distribution that is dense around the modes, the algorithm should classify them into two, without even being told that there were two slits, just based on the structure of data alone. I don't know if this is done in practice. I heard that the models for QFT relies heavily on experimental data so I imagine the people at LHC would need some form of clustering algorithms just to do inference on the events surrounding the collisions that produces the particle fragments.
  21. Jul 31, 2017 #20


    Staff: Mentor

    Sure, if the data are a bimodal distribution, the algorithm can tell you that. But the data you get when you run a two-slit experiment with photons is not a bimodal distribution. It's an interference pattern. Have you actually read any descriptions of the experiment and its results in a QM textbook or peer-reviewed paper?
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook