Julian Barbour on does time exist

Last_Exile

Hi Marcus,

Interesting discussion but I have to admit I don't follow the maths...

I just wanted to say that I remember reading, about 25 years ago or more, in the New Scientist magazine (that was when it was worth reading!) an article relating to the idea of the flow of time as a phase-change.

The general idea was that in a similar way to which a pond freezes over where you can imagine a layer of ice spreading across the surface so does time appear to us. By which I mean that to us the past is fixed (frozen) but the future is mutable and "now" is, of course, where that phase change occurs.

This idea seems to me to be consistent with your I'm sure the original article also had a relevent paper to go with it but it is probably irretrievable now.

Does this idea still sound viable as a layman interpretation of the current discussion?

physicsguy13

If time did not exsist, what would prevent everything from happening at once?

julian

"Time is nature's way to keep everything from happening at once." John Wheeler.

marcus

Thanks Julian, Paulibus, Exile, Detective and others for keeping us wondering about time!

There've been some more interesting papers posted on arxiv that bear on this, but first I want to recall a portion of the passage quoted in post #121, over ten days ago!

==quote page 4 http://arxiv.org/abs/gr-qc/0604045 ==
... Therefore, properly speaking, GR does not admit a description as a system evolving in terms of an observable time variable. This does not mean that GR lacks predictivity. Simply put, what GR predicts are relations between (partial) observables, which in general cannot be represented as the evolution of dependent variables on a preferred independent time variable.

This weakening of the notion of time in classical GR is rarely emphasized: After all, in classical GR we may disregard the full dynamical structure of the theory and consider only individual solutions of its equations of motion. A single solution of the GR equations of motion determines “a spacetime”, where a notion of proper time is associated to each timelike worldline.

But in the quantum context a single solution of the dynamical equation is like a single “trajectory” of a quantum particle: in quantum theory there are no physical individual trajectories: there are only transition probabilities between observable eigenvalues. Therefore in quantum gravity it is likely to be impossible to describe the world in terms of a spacetime, in the same sense in which the motion of a quantum electron cannot be described in terms of a single trajectory.
==endquote==

IOW people have now developed some candidate QG theories, though it's still work-in-progress and as yet there's no consensus as to which nature prefers. But even before we have a fully developed preferred QG theory we can still take the lessons seriously that GR and QM teach us.

Our eventual QG theory will probably NOT be manifold-based. It will not be about the geometry of a space-time continuum---that would be a geometric *trajectory*. Instead it will be about probabilistic correlations between measurements.

That is, the basic math object is probably to be (M,ω) a star algebra and a function from M to the complex numbers that gives the expectation values and correlations, rather than a continuum with fields defined on it. M="measurements" and ω="state" (what we think we know about the both past and future, and our statistical uncertainty therewith.) Our notions of physics THEORY are encompassed in ω, as correlations among possible measurements we might make, as are our ideas about physical constants, past observations, initial conditions etc.

Now this gets into an area of QM foundations where a couple of researchers have recently posted new papers, so I want to quote their abstracts. Here are the links in case you want to check them out:
http://arxiv.org/abs/1211.3062
http://arxiv.org/abs/1212.3606

I have to go out briefly but will try to get back

marcus

I guess the main point to be made in connection with the TIME theme of this thread is that once you have specified (M, ω) i.e. the world of observations/measurements and what we think we statistically know about it, then we automatically get a standard time.

We can compare our own local observer time with that standard time. Sometimes the ratio of rates is physically meaningful.

The standard time is not a pseudo-spatial "fourth dimension", but rather it is a FLOW defined on the star algebra M. That is a one parameter group of automorphisms mapping M → M. "Time" is simply the real number parameter t that parametrizes that flow.

I want to get a quote from one of those QM foundations papers I mentioned. The one by Jeffrey Bub. Here's his introduction paragraph:

This paper is intended to be serious, in spite of the title. The idea is that quantum mechanics is about probabilistic correlations, i.e., about the structure of information, insofar as a theory of information is essentially a theory of probabilistic correlations— not about energy being quantized in discrete lumps or quanta, not about particles being wavelike, not about the universe continually splitting into countless co-existing quasi-classical universes, with many copies of ourselves, or anything like that. To make this clear, it ...

Bub is distinguished prof at U Maryland, same place as Ted Jacobson (top-notch expert on GR and QG, and profoundly original). IMHO with people like Bub and Jacobson you take seriously what they say even if it sounds unusual, or especially if it sounds unusual. He is saying that the Hilbert space doesn't matter and all that paraphernalia, what matters is the structure of correlations. The Hilbert space is just a convenient mathematical device to represent the structure of correlations, and it's not the only possible such framework.

http://en.wikipedia.org/wiki/Jeffrey_Bub born 1942, PhD Uni London 1966, > 100 papers, several books, interpretation of qm and related.
http://carnap.umd.edu/philphysics/bub.html

sshai45

So does that mean there would be an absolute, universal time and simultaneity, and so Einstein was "wrong" in some sense? How does "only 'now' exists" jibe with "'now' depends on the observer"? How do the non-reality of the block universe and the relativity of simultaneity play with each other?

Paulibus

Marcus: your post #141 gave me a belated (by one day) birthday present.

Your reference to Jeffery Bub in his paper "Bananaworld: Quantum Mechanics for Primates", to the effect that: '.....The idea is that quantum mechanics is about probabilistic correlations, i.e., about the structure of information, insofar as a theory of information is essentially a theory of probabilistic correlations— not about energy being quantized in discrete lumps or quanta, not about particles being wavelike, not about the universe continually splitting into countless co-existing quasi-classical universes, with many copies of ourselves, or anything like that...', together with your comment that:
(My emphasis), expresses, more authoritatively than I could, pretty much my sentiments. Bub's paper will compete with my Christmas reading (Bill Bryson: At Home and Paul Theroux's: St Vidia's Shadow).

Thanks for this. I plan to comment later, when I've better absorbed this gem of physics philosophy.

marcus

I'm so glad you were intrigued by Jeffrey Bub's ideas too!
I'm a bit disappointed in myself that I have a hard time following when I get into the middle of the paper. Even though he discusses in very basic terms (bananas, primates) and I'm convinced he has clear important insight, I'm still struggling. Still, it is kind of a fine Christmas present for me too!

My wife likes Bill Bryson's books, and just read "At Home". She passes savory tidbits on to me. He is a good writer and outstanding as a researcher.
==============

I also will have to comment later.

Sshai, I will get back to your comment, time permitting. I think Einstein is still right. We still have observer time. Each observer has a different time (as A.E. said) and it is interesting to compare them.
But also now we have a *state-dependent* time as well. It depends not on a particular observer but on the function omega that summarizes what we think we know (with various degrees of confidence) about the world.

Thanks to certain mathematicians of the second half of the 20th we have a chance at a new way to picture the world, as (M,ω) where M is a star algebra (observables) and omega (state) is a function from M to the complex numbers. Ordinary QFT (quantum field theory) has already been put in star algebra form. And there seems no reason that the dynamic geometry of GR should not also be put into that same form---thus combining the content of QM and GR, combining geometry with matter in a background independent or general covariant way. The (M, ω) is suitable for both.
So this (M, ω) business is quite an interesting development. Of course it is hard to get used to because such a new approach.
However in any case it does not say that "Einstein was wrong". It brings into existence yet ANOTHER version of time, which depends on the state we specify rather than on any particular observer.

And it already seems interesting to COMPARE this time with that of a given observer because it has been shown that the ratio of rates of time-passage can be physically meaningful (corresponding to a geometric temperature discovered by Tolman already in the 1930s and written about in his classic GR treatise). So it can be interesting to compare this version of time with the observer's time. It also seems to be good for other things where you can't use observer-time.

I'll try to get back to this later.

marcus

That's an intriguing question! I don't think introducing the (M,ω) picture and the corresponding state-dependent time flow on the algebra indicates that GR is wrong. But I want to take more time to answer.

Here's part of a short answer I gave to your question in post #144, with a clarifying addition in red:Basically what we are discussing in this thread are theorists' response to what is called the problem of time in GR and also in quantum GR, where the problem is broader and more formidable. Here is the best short statement I know of the problem.
It is from page 4 of Chapter 1 of the 2009 book Approaches to Quantum Gravity, D. Oriti ed. published by Cambridge University Press ( http://arxiv.org/abs/gr-qc/0604045 )

==quote Chapter 1 of Approaches to Quantum Gravity==
... In special relativity, this notion of time is weakened. Clocks do not measure a universal time variable, but only the proper time elapsed along inertial trajectories. If we fix a Lorentz frame, nevertheless, we can still describe all physical phenomena in terms of evolution equations in the independent variable x₀, even though this description hides the covariance of the system.

In general relativity, when we describe the dynamics of the gravitational field (not to be confused with the dynamics of matter in a given gravitational field), there is no external time variable that can play the role of observable independent evolution variable. The field equations are written in terms of an evolution parameter, which is the time coordinate x₀, but this coordinate, does not correspond to anything directly observable. The proper time τ along spacetime trajectories cannot be used as an independent variable either, as τ is a complicated non-local function of the gravitational field itself. Therefore, properly speaking, GR does not admit a description as a system evolving in terms of an observable time variable. This does not mean that GR lacks predictivity. Simply put, what GR predicts are relations between (partial) observables, which in general cannot be represented as the evolution of dependent variables on a preferred independent time variable.

This weakening of the notion of time in classical GR is rarely emphasized: After all, in classical GR we may disregard the full dynamical structure of the theory and consider only individual solutions of its equations of motion. A single solution of the GR equations of motion determines “a spacetime”, where a notion of proper time is associated to each timelike worldline.

But in the quantum context a single solution of the dynamical equation is like a single “trajectory” of a quantum particle: in quantum theory there are no physical individual trajectories: there are only transition probabilities between observable eigenvalues. Therefore in quantum gravity it is likely to be impossible to describe the world in terms of a spacetime, in the same sense in which the motion of a quantum electron cannot be described in terms of a single trajectory.
==endquote==

So the problem is on two levels, classical and quantum. Already at the classical level
there is no observable independent time variable that can be used to describe the evolution of a (general) relativistic system.

And at the quantum level the problem is even more severe, since one cannot realistically assume some fixed metric solution--i.e. a geometric "trajectory".

There's more to say, I'll try to get back to this later. The outstanding thing to notice about the (M,ω) format (for dynamic QG geometry and simultaneously for matter QFT) is that it DOES have an independent time variable that can be used to describe the dynamical evolution of geometry+matter. The point of the quote above is that any observer's time is NOT adequate since the observer's time depends on how the geometry evolves!

When you want to describe the dynamical evolution of a system you need a time variable which is not totally at the mercy of how the system happens to evolve. So observer-time is no good.

This is why the (M,ω) formalism has come up in the context of trying to devise a fully general relativistic treatment of thermodynamics and statistical mechanics. Imagine trying to do statistical mechanics with no possibility of a physically meaningful preferred time variable. That's why it has always been done on a fixed space-time, not in a fully general covariant way. I hope to get back to this. It really interests me.

Paulibus

I've been going through Jeffery Bub's paper that Marcus pointed to in his #141 post (http://arxiv.org/abs/1211.3062). It's about stuff that I'm not at all familiar with: statistics and simplex theory. My comprehension is strictly limited, to put it mildly. Perhaps there is someone here who can straighten my thinking out.

It seems to me that Bub is trying to describing mathematically a world where the speed of information is limited and the observations that guide description are of a statistical nature --- as well as being causal, because they affect this world. He seems to show that observation inevitably generates loss of information, i.e. uncertainty, as is the case in quantum mechanics.

I think he also establishes that entanglement is inevitably associated with such loss of information. He says that all this is an expected consequence of “probabilistic correlations, (and) the structure of information”; all that is needed, I suppose, to formulate a predictive description in a holistically statistical world.

So are the mysteries of quantum mechanics forced on us because causality is a sort of statistical correlation?

But what he doesn’t clarify is for me the central mystery: the magnitude of what sets the whole shebang up, namely Planck’s constant. Perhaps his interesting approach will eventually lead to our understanding why h is of order 10^-34 J.s.in our real world? I do hope so.

sshai45

By a "preferred" time variable, does this mean that this time forms an "absolute time" in some sense (i.e. a "universal clock" that is not tied to a particular observer, sort of like in "old" pre-Einstein physics), or what?

marcus

Paulibus, bravo for tackling Bub! I'll be interested to know what you make of it. So far what I can get is for the most part merely what delighted me so much in the introduction (and I quoted.) But I'm trying to do too many things at once (family Christmas letters, and our community chorus has given several performances of Moz. mass in c-minor!, not to mention physics-watching) Maybe we need a new word to use instead of "preferred". I think it's different from going back to "old" pre-Einstein time ideas. In the old days there was an absolute time and all observers clocks were supposed to follow it and agree on simultaneity and stuff.

Now we still have all the observer-times and a democracy of disagreeing clocks, but IN ADDITION we have one more clock, which we can COMPARE the various observer clocks with. The ratio of rates can even be physically meaningful, correspond to something measurable.

This one additional clock is distinguished by the fact that it is not observer-dependent it is, instead, state-dependent. It depends on what we think we know about the world--on our degree of (un)certainty about correlations amongst observations---what we posit to be the case, with varying degrees of confidence. If you like, picture the state as a density matrix defining a function on the observables.

Each observer still gets to keep his own individual clock and nobody is presumed to be RIGHT, but there is this one additional clock, which has one particular advantage: the world can be analyzed as a fully relativistic system evolving according to THIS time.

Which is something you CAN'T do with some particular observer time, because the observer's history itself depends on how the system evolves---so there is a kind of logical circularity. The observer's time is not truly an independent variable.

So as I see it, this is not going back to the old picture, but instead is adding one more disagreeing clock to the general temporal madhouse and anarchy---which however has a nifty feature that you can do a fully general relativistic statistical mechanics and thermodynamics using IT as the independent time variable---something you cannot do with any other clock as far as I know. So it is subtly different from going back to the old picture and it does not imply that "Einstein was wrong". Or so I think.

Maybe instead of preferred we could say "distinguished" time variable. Distinguished by the fact that it can be used as independent variable in a fully general relativistic quantum statistical mechanics and suchlike fully relativistic analysis (rather than have to first choose a fixed solution to the Einstein equation and then do the analysis "on curved spacetime".) This gives the variable a definite "distinction" without applying that it is somehow "absolute" and the only right choice :big grin:
The key paper for understanding it in this light is http://arxiv.org/abs/1209.0065
But also maybe re-read the "Chapter 1" quote in post #145. It's short and to the point.

mattmars

Dear Marcus

Hi, I`m new to this thread, and I think it is very well conducted, and please excuse my butting in.

But I would like to respectfully say that I question/disagree with your view...

“I think everybody in the thread would go along with the idea that time is real and vitally important---both in physics and in our everyday experience.”
(post #135)

...at a fundamental level.

If we take the simple working view that ‘time’ is apparently 'real', and thus something that in the most simple terms consists, at least, in some way of components described as ‘the past’ and ‘the future’ then at the most basic level we would need...

A- some initial reason for suspecting or assuming the existence of these things/places ('past', 'future'), and,

B- some proof / reason or experiment, to sensibly show how or why they exist.

It seems to me that if the universe is such that it is just filled with a quantity of matter/energy that ‘just’ (as in ‘only’) exists, moves, changes and interacts, (without leaving a 'past' behind us, and without heading into a 'future'), then this would explain all that we think implies the existence of a past and future – if we misinterpret, or over extrapolate, what we observe.

Specifically – if as that matter moves and changes, it also moves and changes the contents of our minds, we may look at some of the contents of our minds, and ‘call’ those contents ‘memories’, and add to this (possibly wrongly) that those contents (memories) are not just things that exist and prove that matter can exist, but are also proof that another thing called ‘the past’ -also- exists.
(And thus also proof that a thing called 'time' exists).

As such we may (imo wrongly) imply that some existing matter, in a particular formation, gave us good reason to suspect that as things move and change the universe ‘also’ creates and stores some kind of ‘record’ of all events in a place or a thing called ‘the past’.

So – is it correct to say , either,

1- Matter just exists moves and changes, or

2- There is also a (temporal) past, and thus a thing called ‘time’ that also may be considered.

Many people seem to assume that Relativity tells us something about the nature of a thing called ‘time’.

From what I have read, as far as I can tell, relativity only seems to actually tell us about the way, and ‘rates’ at which things may move and change differently under various conditions.

As far as I can tell no part of relativity ever proves or demonstrates the existence of things (or places etc) such as ‘the past’ or ’the future’. Although it is written in a way that seems to imply or suggest ‘time’ and these places naturally or obviously exist, or make sense.

While relativity seems to correctly show that matter may intrinsically 'change at different/reduced rates' while at velocity, in acceleration, gravity, etc, I don't see any proof that such matter 'sinks into a past' or 'surges into a future', or that relativity indicates the existence of these concepts.

I would suggest 'time' is not real, but only a false idea borne out of us incorrectly interpreting what the contents of our minds prove and do not prove.

If I am wrong could you point me to a link that shows how the existence of these entities ( 'the' past and 'the' future) has been demonstrated (in relativity or otherwise), as opposed to have just been assumed and untested?

Yours M.Marsden, London

marcus

Hi Matt, welcome to PF and thanks for contributing to this discussion. I'll try to respond...
I think your critique is mainly focused on the 4D "block universe" idea. You offer reasons to be skeptical about the existence of time as a DIMENSION, extending back into past and forward into future.

You are skeptical of the idea that pastpresentfuture exists as a kind of 4D crystal with time as a kind of pseudospatial dimension. That's how I read what you say. You seem dubious regarding the presumed real existence of SPACETIME.

You are certainly in good company as far as that goes. There was even a conference in Capetown South Africa this month bringing together people critical of the 4D block spacetime idea. We discussed that earlier in this thread. Your message also doesn't seem basically at odds with what I've been saying.
You may have misunderstood what I've been trying to get across about time as a real process of change, rather than a spacetime dimension.

I think the best argument against the actual existence of spacetime was given in Chapter 1 of that book, also available online at "arxiv.org". I gave the link in post #145, and quoted a passage that comes on page 4 of the essay. A 4D block spacetime seems incompatible with quantum uncertainty. You have to fiddle around with the concepts too much to get them to live with each other.

Look back at post #145 right on this page, where it says:
Therefore in quantum gravity it is likely to be impossible to describe the world in terms of a spacetime...
Which is to say, because we are talking about what is the best most fundamental mathematical description of the world, spacetime does not exist.

This does not mean that TIME AS A PROCESS OF CHANGE isn't real. The math representation of a process of change is as a flow defined on the set of all observations. That is where transition probabilities and correlations between measurements (now and a few seconds later or a few years later) are defined.

When you treat time as a process the math formalities are not as familiar to most people as when you treat it as a dimension, but that just takes getting used to. There is an observable algebra M consisting of all the possible measurements, and a numerical function ω giving correlations amongst the possible measurements, representing our knowledge. It's more abstract than most people are used to, but maybe that will change. (M, ω) is one possible formulation of any quantum theory about the world. John von Neumann is the main person who developed that approach back in 1930s. It does not necessarily assume a space-time.

And then given (M, ω) as the world, TIME is represented as a process that stirs M around.
This kind of picture can be constructed using matrices of numbers. People can build examples of what I'm referring to. Already did this in the 1930s. von Neumann again. and others. Nowadays you can put this kind of model of the world into a computer and run it. Time becomes a process of change defined on M, and is itself represented by matrices of numbers that you multiply other stuff by to change it.

It may be that the "spacetime" or block universe picture simply is not realistic enough and people will have to go with this (M, ω) picture of the world, where time is not a pseudospatial dimension of a block of eternity but is a process of change instead. That block may simply not be realistic enough (with its eternally existing past present future.)

That is what was being argued in that quote I mentioned where it says
Therefore in quantum gravity it is likely to be impossible to describe the world in terms of a spacetime.

Quantum gravity means quantum *geometry*, the quantum theory of the geometry in which matter and everything else lives. So the QG program is aimed at constructing and verifying the most fundamental picture of the world. If you cannot incorporate a block space-time in QG, then that is as much to say a block spacetime does not exist, and time is not a 4th dimension.

So we have to see how the QG research program goes, and how it turns out.

sshai45

So if I'm getting this right, then the difference between this and "pre-Einstein" universal time is that in the latter, everyone's clocks must agree with the universal time, but that is not the case for this kind of "universal" time. Does this time also have a rest frame associated with it? Also, I notice you mention that "omega" represents "our knowledge". Does this mean that as we get "more knowledge", then it further "refines" this universal time?

However, I'm curious about that bit about "spacetime does not exist", the "block universe doesn't exist": I thought that the block universe was essentially necessitated by the fact that observers could disagree on what constituted the past, present, and future. So that all three would have to exist "eternally". How is this handled in this "spacetime-less" theory? If you replace "time as a dimension" with "time as 'change'", then that would mean there would have to exist a universal "now", no? And that "now" would be the only thing that exists, everchanging (as we have no spacetime, so the past and future don't eternally exist). And if that "now" is the only thing that exists, then how can some observer in it include events in the non-existent universal past as part of their "now"?

marcus

Now you are getting down to the basic similarities and differences with the earlier picture. These are good questions, I think.M consists of all possible measurements, ever. I think you are RIGHT that we will want to use a different state function ω when we have improved physics theories and more precise estimates of the constants. ω is a probabilistic idea of the state---correlations between measurements also embodying uncertainties about the fundamental constants, earlier conditions and even what the applicable equations are.

The idea is, we have to go with the best ideas and knowledge we have, and predict the measurements that we consider to be in our future, based on our knowledge and the odds we ascribe to it.

So exactly as you say, as future humans refine ω so would this idea of time (the Tomita flow on M) be refined.

I do not think that the Tomita flow has any idea of simultaneity belonging to it. There is no distinguished time-slice associated with it, that you could somehow "date from".

this is jumping way ahead, but if LQC (with its bounce) were ever implemented in a (M,ω) model then it would acquire a reference time-slice, the bounce. But we already know that when standard Friedmann cosmology is implemented one recovers standard time used in cosmology. Cosmologists use a universe time or "Friedmann time" in their standard expansion model. And , no surprise, (M,ω) reproduces it. Tomita = Friedmann. but that's jumping ahead.

The basic answer is NO there is no reference timeslice in the (M,ω) picture. there is the Tomita flow but no universal starting place for it.

In a sense M takes the place of the 4D spacetime of GR. but it has no geometry. the measurements all embody uncertainty and can assume various values. We can only imagine making a FINITE NUMBER of measurements. Like knowing where a particle went but only at a finite number of points along the way---not knowing the entire continuous trajectory.

M is very different from a space-time with a metric describing its geometry, in the sense that we make only a finite number of measurements (of areas, of angles, of distances, of matter density, of charge, etc)---and make a finite number of predictions based on that---beyond that we don't presume. The geometry is obviously quantum and uncertain because the geometric measurements themselves are quantum observables. But more than that, we do not presume that an overall classical geometry even exists.

I'm trying to interpret from the Connes Rovelli paper http://arxiv.org/abs/gr-qc/9406019 as best I can, and also from the recent one
http://arxiv.org/abs/1209.0065

Yes we all want observers to be able to disagree! But does this actually necessitate a "block universe". I think that is only ONE POSSIBLE data structure that permits them to systematically disagree. I think you are asking an extremely good question and one which, since the (M,ω) way of representing the world is new to me, as is the Tomita flow idea of time, I cannot competently answer. I would like to see more examples.
the Connes Rovelli paper shows a bunch examples but I would like to be clearer. How do different observers disagree harmoniously within the (M,ω) context? A researcher at Perimeter Institute named Laurent Freidel has been working on something he calls "Relative Locality" in which no global spacetime exists but there is Lorentz symmetry locally. Could this be encompassed in the (M,ω) picture?

The model itself does not force any division into past present future. But how for example is Lorentz symmetry implemented? Wish I could do a better job answering.

sshai45

Thanks for the response.

I'm curious about that "bounce". Does it imply that the future of the universe is to recollapse ("Big Crunch") and bounce again? If so, how does that jibe with dark energy? Does dark energy disappear at some point, or does it "reverse" itself somehow (so as to become attractive instead of repulsive in effect)?