Can Lorentz Symmetry be Emergent?

[marcus]

http://pirsa.org/11100056/
Can Lorentz Symmetry be Emergent?
Speaker(s): Ted Jacobson
Abstract: I will begin by discussing some of the strongest observational evidence for Lorentz symmetry, and the essential role that Lorentz symmetry appears to play in the consistency of black hole thermodynamics. Next I will discuss some reasons for suspecting that Lorentz symmetry may nevertheless be emergent. And finally I will discuss difficulties with the concept of emergent Lorentz symmetry, and how such difficulties might conceivably be overcome.
Date: 27/10/2011 - 3:30 pm
Collection: Emergence and Effective Field Theories Conference

This might be a good place to start for anyone interested in the question. Ted Jacobson is a central figure in this line of investigation.

Could there be a latent preferred frame in nature, which we just have not detected? Whose effects are too subtle to notice except, say, at very small or very large scale?
Quite probably NOT, but how can we rule this out?
One useful result of Jacobson's work, and others, is that by figuring out all the ways that Lorentz violation might be present but concealed, they stimulate people to test---and perhaps eventually rule the possibility out.

This video is a total 68 minutes, but Jacobson's talk is only the first 27 minutes. Then there are about 20 minutes of questions directed by the conference/workshop audience to Jacobson. At about 47 minutes questions begin to be directed to speakers who presented talks earlier in the day. Occasionally getting back to T.J. but very wide ranging. The focused part of the video is just the first 47 minutes.

 

[qsa]

A silly question. Is the circle emergent?

 

[Fra]

I wanted to listen to the talk but I see there is no mp3 only flash which I can't put on the phone or car. So we'll see when I get time.

From my perspective, all physical observer class symmetries are emergent if you require them to be the result of an inference from information about observer sets. Any imperfect inference (meaning not qualifying as a deduction, but rather something softer) imples imperfect symmetries.

As for lorentz symmetry, I'd say that informally it's expected to be emergent in a small subsystem, in the limit of a infinitely complex and massive observing context. This would for example apply to any scattering experiment, where the scattering information is encoded in the environment that's effectively massive. I practice, all it takes for the manifest symmetry is that the observing context is "large enough" - let's call it "classical", as it's almost the same thing.

OTOH, if one is to consider infering this symmtry from the insider, then this fails just on the grounds that there is an uncertainty in the symmetry. I even think there is a kind of uncertainty relation here between symmetry and the complexity and mass of the observer. At one limit: an infinitely complex observer can infer a sharp, clear symmetry. At the other limit, the obserer can not even distinguish the symmetry at all. I think we understand the first limit, but not hte other limit nor the midrange. The root cause of this is IMHO the same core issue that makes it difficult to define observables.

The only way to mentally maintain perfect symmetries when it's not inferrable, is to consider the symmetries to be structural realist elements - that doesn't NEED to be inferred, or as fixed equilibriums that are distorted in practice. But then I think we are leaving the rationality that is hallmark of science.

/Fredrik

 

[JollyJoker]

I wanted to listen to the talk but I see there is no mp3 only flash which I can't put on the phone or car. So we'll see when I get time.

It may have been added since you looked. There are three links, "Flash Presentation , MP3 , PDF".

 

[Fra]

I listened to the talk in the mp3 format the other day and Ted's ideas on emergence uses a completely different logic that does't quite connect to my thinking. I didn't find it particularly interesting.

He talks about a non Lorentz invariant fundamental theory (whatever that means to him) where the effective or approximate invariance is still emergent at the right observational scale. He also talks about preferred frames in the fundamental theory.

For me the core points is that if the inference, which observer is doing it etc, which is still a different way of thinking. But there are similarirites. In my thinking there relly is not "fundamental theory" where there are preferred frames; instead each observer naturally an unavoidable has it's own frame from which his entiry theory is constructed, but she still infers lorentz symmetry whenever he studies (to her) subsystems.

For me, this isn't a statement of reality, it's a statement of how theories are related and you can only compare two theories, by means of a third theory. One doesn't need a master theory or "fundamental theory" IMO. Thus I think of the concept of possible lorentz violation not as the idea of there in some objective sense exists preferred frames in nature (that I find highly unlikely and unreasonable), I think of it as beeing due to the actual physical limitations on how any given obsever actually makes the actual inference from experience thta a subsystem has lorentz symmetry; and there the idea I entertain is that the infered theory is in fact forged so that it ALWAYS has this symmetry in the limit of subsystem and massive observer.

This point is as I see it not present in Ted's talk, and I'm not sure I understood the plausability in his thinking (well possible yes, but likely no). Maybe someone else did and can explain.

/Fredrik

 

[qsa]

Fred, pls your general explanation has neen going on for too long. Can you be more explicit. Anyway, how on Earth would you model massive observer.

 

[qsa]

Fred, pls your general explanation has neen going on for too long. Can you be more explicit. Anyway, how on Earth would you model massive observer. I don't mean to be disrespectable.

 

[Fra]

Can you be more explicit.

Explicit models with computable predictions that connect to real physics? - that's the ambiton, but atm - No, that takes more development.

Anyway, how on Earth would you model massive observer.

First of all I "model" an observer by an inference system that encodes it's own inferences.
Physically that is like a state that encodes it's own evolution.
In terms of inference the evolution is a random walk relative to the prior state, constrained by the same, like an entropic kind of flow.
The inference rules of system A, is nothing but the interaction laws of that system.
Such evolution has an inerta. A highly massive observer is one with high inertia.

By a massive observer then I mean an inference system that can adapt and encode all the time history of the interation history w/o beeing saturated so that information needs to be discarded. The closest comparasion is information from scattering experiments. But my point is that a real observer only have a finite "computational power" and information capacity, and this limits what is inferrable! This generically applies to any symmetry.

But this only makes sense if you accept that the symmetry is an inferrable position from an interaction history. I take the inferrability conditions as the scientific quality, and thus I demand it.

/Fredrik

 

[qsa]

FRA, PLS check your PM

 

[Fra]

Actually an even simpler presentation of the core of my argument is this:

I reject realist notions of "symmetry". I think it has no place in physics, no more than does "absolute state of matters" in mechanical Newtonian universe.

Instead I required that the only relevant thing; to the observers ACTION, is to the extent a symmetry of a subsystem can be concluded from real experiments by this observer.

When you think about that, you realize that to manifestly demonstrate a symmetry, you need to have information about ALL the instances to which the presumed symmetry relates, and manifestly demonstrate that you can throw away the redundant information and instead compress the information to just equivalence classes that represents the symmetry.

AND all this must be done in a way that respects and sound rules for inference, such as # of samples, confidence levels and confidence intervals. Then I claim that it's not possible for a finite inference machiner to make a 100% confident sharp inference of ANY symmetry. All you get are uncertain expectations of maybe fuzzy symmetries, and it's THESE things that the observers actions are invariant with respect to - not the imaginary limiting symmetries. And the point here is that there is not observer invariant inferrable symmetry, except in limiting cases or as equilibrium cases.

So I'm taking more seriously the idea that SYMMETRY, just like a particle STATE, must not be thought of as absolute matters. That is irrational. The only rational way IMO is to require it to be subject to a generalized measurement process, just like we do for other system states.

This is another level of the issue, thata at least as far as I understood from the mp3, is not expicitly acknowledged in Teds talk.

/Fredrik