Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
A Debate Over the Physics of Time (quantamagazine.org)
78 points by azuajef on July 23, 2016 | hide | past | favorite | 41 comments


"The laws that underlie these theories are time-symmetric".

I don't get how people can say this. Quantum measurement (collapse of the statevector) is not at all time-symmetric. People seem to always leave this out in such discussions. They either view it is a temporary "problem" with the theory that we will one day fix, or just as something that is plainly not a theory. Or who knows what. The wikipedia article on interpretations of quantum physics has a giant table that illustrates the lack of consensus:

https://en.wikipedia.org/wiki/Interpretations_of_quantum_mec...


But don't quantum measurements appear only to our (classical) perception? If someone (probably not in our Universe i.e. someone who doesn't interact with our Universe in any physical way) hypothetically able to write down the full Hamiltonian of our Universe (assuming that quantum gravity theory exists, so that is possible) then there is no external to this Hamiltonian "measurement devices", so wavefunction of the Universe does not collapse and the time evolution of Universe is reversable as soon as you can just multiply time to -1 in evolution operator.

But I don't understand why do they talk about time reversability of physical systems and don't mention CPT symmetry[1]? Standard model is not actually time symmetrical, but you one should also reverse charge and reflect spatial coordinates to obtain symmetry after time reversal because there are violations of CP symmetry [2].

[1] https://en.wikipedia.org/wiki/CPT_symmetry

[2] https://en.m.wikipedia.org/wiki/CP_violation


Only if you believe in non-unitary collapse, which is probably wrong IMO.

Einselection is my favorite potential solution. Straightforward, clean, unitary, and the math is promising. The idea is that apparent collapse is just emergent behavior of the schroedinger equation, via induced superselection rules between incompatible measurement eigenstates.


> which is probably wrong IMO.

There are so many IMO's floating around when it comes to foundations of QM. I think that is fascinating in and of itself.

> Straightforward, clean, unitary, and the math is promising.

My IMO definitely does not coincide with your IMO :-)


My "IMO" in this case means "by Occam's razor". We don't have enough physical evidence to strongly prefer one theory on the other, so all we have to go by is simplicity.


To me these arguments involving decoherence only show that the classical world is consistent with the quantum world, and indeed place constraints on the two theories in order to achieve this consistency. But in no way does decoherence (exponential suppression of off-diagonal terms in a density matrix) imply collapse of the state vector. In other words, we don't know why anything happens!

It's such a beautiful perplexing quandry, and I just don't understand this attitude of "nothing to see here, move along".


Measuring an unknown datum so that later you know that datum is time asymmetric because your of how your state of knowledge changes. Defining QM in terms of "measurement" automatically imposes the psychological arrow of time into it.

But physics can also be thought of as rules of the from "the joint probabilty distribution of preparations S_j and measurment results R_k is ..."

This in turn will depend on prior-probabilities for the S's and R's, and also the rule which tells us

    P_jk \propto  |<S_j|U|R_k>|^2      (where U is some unitary).
Notice that this rule treats R's and S's symmetrically.

Now the S's and R's are different. The prior for S can have any probability distribution we can engineer, and can even be deterministic. But the prior for the R's must be completely uniform unless we post-select results.

So there is a real asymmetry, but it exists in the past and future boundary conditions that we work with, and not in the mechanics (quantum or classical) of what happens in between.


In sentence preceeding the one you've quoted, the author refers to Eisntein's laws of general relativity and the Standard Model, and with that context the sentence makes sense. These 'classical' models do indeed suggest a time-symmetric universe. The whole problem here is reconciling classical models with quantum mechanical models, which as you've noted do not suggest a time symmetric universe.


The standard model is a quantum theory and so this absolutely includes state vector collapse.


I am not sure what interpretations have to do with symmetry of equations.


Measurement effects are not captured by any QM equation.

The equations are all time symmetric, yes. But the theory is not.


Measurement effects are captured by projection operators (at least in standard QM). This is an equation and it is certainly not time-symmetric.


But who seriously believes the collapse postulate is a fundamentally true law of physics? Unitary time evolution and the collapse postulate are incompatible, so at most one is fundamental and true. And over almost a century we have collected a gigantic amount of experimental data confirming that unitary time evolution is correct.

You might argue that the same is true for the collapse postulate and the Born rule but I have to disagree. Yes, it gives the correct answer in as many experiments, but those are parts of the experiments that we don't really understand. We have classical system interact with quantum systems and our measurement devices are just such gigantic quantum systems that we can't treat them quantum mechanically.

The collapse postulate is obviously true in a kind of classical limit, but it is really just an emergent law and nothing fundamental. If you could show that a wave function collapse can occurs in quantum system before we measure it, that would change the game. But that is not the case, the wave function collapse only occurs when we throw a gigantic classical system at the quantum system which we can not treat properly due to its complexity.

So I argue people are rightly ignoring wave function collapse when they talk about time symmetry because it is just an emergent phenomenon. And there are several ideas that can explain why it looks like a wave function collapse occurs, so we are not simply throwing a corner stone of our theory out of the window.


How does a non-linear irreversible phenomenon emerges from repeating unitary linear interactions?

Given that a reversible linear mechanics is so much more useful than one that isn't, I think it's way more likely that the universe isn't in fact linear, but the best way people got to mapping it was to shove all the non-linearity at the border of the mechanics, right before they get something from the experiment.


> who seriously believes the collapse postulate is a fundamentally true law of physics?

I would love to find out the answer to this! There needs to be some more surveys of physicists...

> And there are several ideas that can explain why it looks like a wave function collapse occurs

I don't think it is at all as simple as you are implying. This is still very much an open question, even if most physicists don't care to think about it.


Projection operators do not collapse the wave function. Put some mixed state in, and you'll probably get some mixed state out (for some states it may be pure, but not all).


The theory is the equations. Take the Schrödinger equation. It is often misunderstood in discussions of time-symmetry and determinism. The Schrödinger equation describes the evolution of a wavefunction, deterministically, and symmetrically in time. The equations are "indeterministic" in another sense, that of Heisenberg: can't know position and momentum with infinite accuracy. But that's a subtly different point.

The result of a measurement is not dealt with by the equation. Schrödinger only lets you calculate the probabilities of measurement outcomes, but not which one it will be. That's why Einstein proposed we just didn't knew the whole picture. He thought we needed more information.


The theory is whatever tells you what you will observe in the real world.

On the case of QM, that's the set of equations, observable operators, and the instructions of hoe to interpret them. None of those predict anything alone.


The Schrödinger equation alone predicts how a state evolves in time. Just not what the result of a measurement will be.


And that state is a completely made up mathematical artifact. The only thing real on QM are measurements.

Besides, the Schrodinger equation does not predict how the state will evolve after a measurement is done. That's the entire problem.


    And that state is a completely made up mathematical artifact.
The wavefunction is not measurable, but its absolute value squared is. This does not mean it's not "real". I understand what you mean, but you need to provide us with another better answer that supports all observations before saying it's "completely made up".

And yes, the Schrödinger equation predicts exactly how a state evolves, even after a measurement. It does not tell you what the outcome of a measurement is - you give it the outcome, and it gives you the time evolution.

There is a problem, and it lies in what happens during measurement - or what is the exact nature of measurement. But given any state, the SE tells you how it evolves.


Is that an issue for the many-worlds interpretation?


I have no idea what a many-worlds person would say about time asymmetry. Would they postulate a kind of splitting-of-worlds at the moment of measurement? In which case this sounds time-asymmetric to me.


There is no splitting-of-worlds in many-worlds.

The central idea behind many worlds is that the concept of "measurement" is extraneous and can be removed. When to particles interact on a quantum level, they become entangled. The behaviour of this entanglement is well understood and accurately described by pretty simple (but potentially counter-intuitive) math. Our best description of these systems involves them being in a superposition of many states. We have also observed in experiments these superimposed states turn back into pure states, in exactly the way we would expect from the time reversible equations governing quantom mechanics.

The idea behind many-worlds is that "measurement" is just a bunch of these quantom interactions. In theory, we could reverse these interactions in the same way we could reverse them at a two particle level. In practice, this never happens because it is the same idea as the Earth shooting out a giant asteroid. Theoretically possible as the time reverse of an asteriod hitting Earth, but massively improbable.


> The central idea behind many worlds is that the concept of "measurement" is extraneous and can be removed.

Well obviously I need to read more about MWI because this just looks like standard quantum physics to me. Or perhaps "Church of the larger Hilbert space". Is it just saying superposition of states looks like "many worlds"? This looks to me like a non-interpretation interpretation.


There is no splitting-of-worlds in many-worlds.

TIL the many-worlds interpretation is actually the many-worlds many-interpretations. That's so meta!


What if whatever force makes time asymmetrical also is at the root of volition and consciousness. So by definition we would only be able to experience time in one direction.


Scott Aaronson touches on this idea a bit during a discourse with Roger Penrose[1], commenting on the unification of entropy and information:

"So on the picture that this suggests, to be conscious, a physical entity would have to do more than carry out the right sorts of computations. It would have to, as it were, fully participate in the thermodynamic arrow of time: that is, repeatedly take microscopic degrees of freedom that have been unmeasured and unrecorded since the very early universe, and amplify them to macroscopic scale."

---

[1] http://www.scottaaronson.com/blog/?p=2756


Well, any kind of computation that erases information (for instance, taking two bits and using them to produce one bit in an AND gate) does have to increase entropy, so those two things are already one and the same.


I really like this theory: http://lesswrong.com/lw/fok/causal_universes/

But regardless, what would it even mean for there to not be an arrow of time? Such a universe would have egg shells on the floor, and then waves of energy that just happen to hit the eggshells, at just the right velocity to propel each piece to assemble an unbroken egg. What a remarkable coincidence that would be. It would be very difficult, perhaps impossible, to make a universe like that.

IMO it's just like cellular automata. You start with a "start state" of something simple or random. Perhaps just a single cell. And then it evolves into complex patterns as each time step causes the next time step.

There are reversible cellular automata, that in theory don't have an arrow of time either. You could start at any state, and reverse it, just as easily as advance it. But unless you are remarkably clever, or lucky, at arranging the start state, it will tend to increase in entropy over time. Or rather, as the step counter increases.


>what would it even mean for there to not be an arrow of time?

Thermaldynamic soup? Our universe has an entropy based arrow of time because it started in a low entropy state. If a hypothetical universe started in a high entropy state, than forward time would appear no more probable than backward time. Of course, such a universe would also not have eggshells; it would just be a thermodynamical soup.


Well humans wouldn't exist in such a universe, and couldn't observe it.

But even if you started with a completely random cellular automata, it's often common for order and structure to emerge. You could even say that entropy is going in reverse. But causation and the arrow of time still only flows one direction. Perhaps given enough time, life could evolve in such a universe, and ask why time only goes one direction.


I've been watching quantum for the last 5 years, let's just say the next 15 are going to be a fucking trip. ;)

<3


It's been a trip since at least 1905 when we started looking into the photoelectric effect. But yeah, I can't wait until we get some more understanding of the current questions. That will reveal more questions, of course!



Upgrade the block universe to a blockchain universe and you should see where time's arrow comes from.


Smolin's career seems to basically be entirely comprised of telling the only people who are creating successful theories that their theories can't be right, because he doesn't like the way they make him feel.


The parts of Smolin's career that are easily integrated into a pop science article are entirely comprised of that, but his actual work in physics that doesn't fit into a soundbite does not. He didn't get his position at the Perimeter Institute by talking about how theories make him feel.

Just like pop physics tends to distort people's perception of the field as a whole (e.g. giving the impression that 90% of theoretical physics is string theory) it also makes researchers' body of work look very different than it is.


A hunch can be a powerful thing; sometimes, it's the symptom of a nascent theory that hasn't hit the requisite critical model mass to attain that self-sustaining "eureka" moment in the brain yet.

Of course, this validates truthers as well...


Or it's just nonsense like most hunches. I hope Smolin keeps on trucking but hunch based contrarianism isn't actually a good indicator of truth.


Have you read Smolin's Time Reborn? I've just finished it, and while I don't agree with certain aspects of some of his ideas, I found it quite interesting and valuable.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: