I Relations resolves singularity -- new paper on arxiv

[windy miller]

A new paper on arXiv today claims that relationsism allows one to evolve the universe through the big bang. Alas I am not familiar with relationsism, is it related to shape dynamics? can anyone explain?
https://arxiv.org/pdf/1607.02460v1.pdf

 

[Some One]

Relationalism is the idea that measurements of quantities have to be made with reference only to entities within the system described. So if you describe the dynamics of the size universe (in this case) you have to also describe the dynamics of the meter stick with which the size is measured. The point of this paper is that if you include the measuring rods and clocks in the evolution itself, there is a well defined and completely deterministic evolution that is entirely in agreement with GR everywhere except when you get to a singularity. There both the size of the universe and the length of any rod with which you measure it go to zero in such a way that the ratio remains well defined throughout.

Shape dynamics is a particular implementation of relationalism, but not all relational systems are shape dynamics. The ideas are very similar in some contexts - in shape dynamics only the conformal elements of the theory count, (e.g. angles in triangles) not sizes (lengths of the triangles' edges), but this paper doesn't require the symmetry trading or preferred frames of simultaneity that shape dynamics normally uses.

 

[windy miller]

Thanks for your reply, this is very interesting. However I am a little confused when you say "both the size of the universe and the length of any rod with which you measure it go to zero in such a way that the ratio remains well defined throughout."
As I understand it the insularity is achieve in Gr when the scale factor goes to zero. So if the singularity is resolved in some framework presumably the scale factor can't go to zero for some reason, perhaps because space is discrete and has a maximum density as in LQC or because you can't exceed the Hagedorn temp as in some string models.
So if this scenario allows the universe to go to zero size how is the singularity resolved?

 

[Some One]

A "singularity" in this sense is not just a term reaching zero (like in the case of the scale factor here) or infinity (the curvature). As an example, think of the opposite case - the recollapse point in a closed universe. There the curvature becomes zero, and Hubble radius infinite, yet we are all happy to continue physics beyond this point - the Friedmann equation tells you what happens next.

The actual problem with a singularity is that the equations of motion in GR cannot be continued beyond this point. What is claimed in this paper is that if you construct dynamics only using relational (dimensionless) variables, there is a complete, predictive evolution that goes beyond this point. It's only the insistence on using certain (not dimensionless) quantities that causes problems. What's found is that the solution has a unique and well defined continuation beyond the zero of scale factor.

As a (poor) analogy, consider system described by df/dx = sec^2 (x) , f(0)=0. The solution is f=tan(x). But as a dynamical system you'd get close to x=pi/2 and think that f(x) and f'(x) are infinite here. A singularity. But if instead you know that what you're actually doing is taking the ratio of the edges of a triangle inscribed on a circle as you move around, you'd be happy to continue beyond this 'singularity'. There's a point where the ratios are undefined (one edge has zero length) but beyond that the ratio is well defined again and so you can continue the solution. Similarly, this relational approach tells you what happens past a=0. There's another, quantitatively different yet qualitatively similar universe on the other side.

 

[windy miller]

Thanks again for this, its really fascinating although hard for me to understand you do a great job of explaining it. It seems they are implying a Janus point on the other side of the bang, as I understand this is contracting universe on the other side from our perspective but its expanding from the perspective of someone on the side of the bang. Is that right? this makes it different from a model say like LQC where its just contracting and then expanding.

 

[spacejunkie]

Thanks again for this, its really fascinating although hard for me to understand you do a great job of explaining it. It seems they are implying a Janus point on the other side of the bang, as I understand this is contracting universe on the other side from our perspective but its expanding from the perspective of someone on the side of the bang. Is that right? this makes it different from a model say like LQC where its just contracting and then expanding.

Yes that is correct. A Janus point indicates a "one past, two futures" model.

 

[Some One]

You have to be a little careful here. The Janus point would be the big bang in this setup, but there are a couple of caveats.

The first is that there isn't a complexity function given like the shape dynamics people talk about. The complexity is a bit like thermal time that Sean Carrol always talks about, a monotonic function going away from the Janus point that the shape people interpret as time. This isn't present in this paper, so it's not sure whether time flows in the other direction on the other side of the big bang. It probably depends on what you interpret as time. The shape dynamics interpretation of time is quite nuanced, being the appearance of "ever better records of events" in some sense. This is only done properly in the n-body stuff so far as I know.

The second is that if you take some function as the equivalent of the complexity function, this should be examined in LQC and other approaches too. LQC normally uses a scalar field clock, and if you do the same with this setup, I think it will come out the same way - just a single direction of time - but if you use the complexity clock in LQC we don't know what will happen. LQC has something like a washout of entropy at the bounce, but this is also somewhat unclear. Likewise Janus points could exist in a lot of other theories.

 

[windy miller]

Thanks again, really helpful.

 

[momo666]

I am correct in saying this paper describes an Eternal Universe?

 

[windy miller]

It seems that was, I think Some One should confirm too though, he/she seems a lot more knowledgeable than I am.

 

[Some One]

In a sense, yes - you have to be careful again with what you mean here. The paper describes a continuation of Einstein's equations through a singularity, and connects up to another cosmology on the other side. Now, if these cosmologies are asymptotically flat or open, they'll each expand forever away from the "Janus point". If they're closed, or the cosmological constant is negative etc, they can undergo recollapse, and fall back into another singularity, to continue out of the other side etc etc. This is the same with LQC, ekpyrotic and a lot of other situations.

However, there is a subtlety with regard time in the shape dynamics viewpoint: If you view complexity as time, the recollapsing universes would have Poincare recurrences. That is to say, that if you take the arrow of time as always pointing away from the singularity, then recollapse would mean going backward in time, so from the perspective of complexity as time they would have a finite lifetime - the amount of change between big bang and recollapse. If viewed from some external time perspective, they would seem eternal, of course.

This is true for any system that takes thermal time or complexity time seriously. And of course in the open cases complexity time expands forever.

Sorry to not give you a direct yes/no answer on that, but I hope you can see it's a case of what you mean by 'eternal' - the answer depends on the clock!

 

[Grinkle]

Sorry to not give you a direct yes/no answer on that, but I hope you can see it's a case of what you mean by 'eternal' - the answer depends on the clock!

There have been many PF posts in 2016 that I have read and not been capable of following at all. Among those, this post is definitely my favorite. ;-)

 

[Some One]

Haha, thanks Grinkle! Let me try to be a bit clearer - I might get a bit long-winded in my explanations, so I'm sorry about that, but these are quite subtle issues.

The crux of the difference basically comes down to what you mean by 'eternal' - this means 'for all time' but of course time has to be measured with a clock. Let me give an analogy that I think might help explain the difference:

Suppose you watch a video. You've got an old-fashioned VCR, so at the end of the video, the tape rewinds, and for the purpose of this metaphor, it rewinds at 1-1 speed with its normal playback. If you sit watching this video, it will play forwards, then backwards, then forwards ad infinitum. You check your watch, and the video lasts 3 hours, so you can watch it forwards for 3 hours, back for another 3, forwards for another 3, etc. From your perspective, this process is eternal - it goes on forever.

However, when you deal with a relational system, life is a bit more tricky - you can't measure time from the perspective of some external observer because the universe has to contain everything. You can't use the watch on your wrist to tell time, because there is nothing that exists outside of the video. Fortunately, in the video there's a clock you can see on the screen (what a coincidence! It's almost like I picked this video to make my explanation easier!) so when you're watching it you can use this as your reference time. However if you do this, you see that the clock only goes from 12:00 to 3:00 and back again. So with reference to the on screen clock, this system isn't 'eternal' - it only describes 3 hours of things that happen.

In either case you're perfectly happy to describe the events on the video. The cat goes and gets food at 1:45 on the video clock, or at 1:45, 4:15, 7:45, ... and equivalent times on your watch etc. However, whether or not this describes an 'eternal' set of events depends on whether your refer to the clock on screen (relational time) or the watch on your wrist (absolute, external time).

I hope that makes things somewhat clearer.

 

[Grinkle]

Thanks for the analogy - I appreciate it, it does make the ideas being discussed more clear to me (I hope).

If that analogy can be taken this far, it implies to me that the question of whether our universe is being described as infinite in the self-contained / observable by us model is irrelevant to us, because as far as we can ever observe or measure we only see our clock go once from 12:00 to 3:00. While we are watching the clock, we speculate (as I am doing right now) about how an external-to-the-universe observer might perceive things, drawing an analogy to us as external-to-the-model-under-discussion observers, but this is untestable speculation.

It is also very difficult to visualize speculation - how an observer can be outside the universe and able to observe the universe. But I think you are just saying that to the extent the model describes an eternal universe, this is the manner in which it does so.

 

[Fervent Freyja]

It looks as if the analysis is also intended to support several "cosmological models with inhomogeneity". This may be the first part for more papers to come as it is introducing information that can be used to "open the door for discussions"... I'm not sure if this is an acceptable paper for some of the experts on here?

 

[Some One]

The application to inhomogeneous cosmologies comes through the BKL conjecture - the idea is that (roughly speaking) near to a singularity, spatially separated points stop communicating with one another and each can be well approximated by its own homogeneous (i.e. Bianchi IX/VIII) cosmology. There's a lot of numerical and some analytic support for the conjecture, so what the authors are claiming is that if BKL holds, then you only need to treat the homogeneous case, and apply the resolution of that case everywhere.

This is certainly the first of a series - I'd expect a longer technical paper with all the gory details and probably some other exploring the symmetries in time and parity that is found. There's also likely to be some very interesting contribution to the theory of initial conditions.

As for whether or not it is "acceptable for experts", I don't know. Wait and see the peer reviews, I guess!