Why rotation apparently seems to be absolute?

  • Level: Graduate 
  • Thread starter Sushmit Nandy
  • Start date
80 replies · 2K views · last reply
Dale said:
I would say: “Proper acceleration is what an accelerometer measures”. I prefer operational definitions whenever possible.
I prefer logical definitions. This means the following:
1. The notion of acceleration should be defined before the notions of accelerometer and proper acceleration.
2. Proper acceleration should be defined as a special case of acceleration.
3. Accelerometer should be defined as something that measures acceleration, otherwise it should not be called so. If it measures only proper acceleration, then accelerometer should be defined after the definition of proper acceleration, which should be defined after the definition of acceleration.

I can't help it, any other order is illogical to me and creates a conceptual confusion.
 
  • Like
Likes   Reactions: dextercioby and PeroK
Demystifier said:
@PeterDonis and @Dale , how to explain the notion of proper acceleration at a popular physics level, to someone without a physics background?
Proper acceleration is feeling weight. Zero proper acceleration means you're weightless.
 
Demystifier said:
To compute the proper acceleration in relativity theory, one needs to know the metric tensor of spacetime.
If you're taking this approach, proper acceleration is the path curvature of the object's worldline in spacetime.
 
Demystifier said:
I prefer logical definitions. This means the following:
Not at all. Your approach is by no means the only way of reaching proper acceleration logically. Nor do I think it's even a correct way of doing that: see below.

Demystifier said:
1. The notion of acceleration should be defined before the notions of accelerometer and proper acceleration.
2. Proper acceleration should be defined as a special case of acceleration.
But it isn't. If we lump proper acceleration and coordinate acceleration under the general category of "acceleration" (which not all physicists would agree with), neither is a special case of the other. They're two different types of "acceleration".

Demystifier said:
3. Accelerometer should be defined as something that measures acceleration, otherwise it should not be called so.
From a theoretical viewpoint, yes, we would want our theory to tell us "what acceleration is" before we try to construct a model of a device that measures it.

But that doesn't mean your logical steps are necessary to do that. Here's an alternative:

1. Spacetime is a 4-dimensional geometric object that has a metric.
2. Objects have worldlines, which are timelike curves in spacetime.
3. The metric tells you which worldlines are geodesics and which are not, and tells you the path curvature of the non-geodesic worldlines.
4. Proper acceleration is the path curvature of the object's worldline.
5. A correctly designed accelerometer measures this path curvature.
 
  • Like
Likes   Reactions: Dale, PeroK and dextercioby
PeterDonis said:
If you're taking this approach, proper acceleration is the path curvature of the object's worldline in spacetime.
And how do you define a path curvature? Note, in particular, that a "curved" motion of a planet around Sun has a zero proper acceleration.
 
Dale said:
I would say: “Proper acceleration is what an accelerometer measures”. I prefer operational definitions whenever possible.
Suppose that accelerometer is based on a spring. But you can push or pull the spring by your hand, so when you do that would you still say that you are measuring the proper acceleration? I would not, I would say I am measuring the proper force.
 
Demystifier said:
how do you define a path curvature?
Using the metric. Well, the metric and its compatible connection. You can find the math in any GR textbook.

Demystifier said:
Note, in particular, that a "curved" motion of a planet around Sun has a zero proper acceleration.
Wrong. You are talking about curvature of the spatial trajectory in a particular coordinate chart, not path curvature of the worldline through spacetime.

Please note that the things I am saying are elementary GR, and the things you are saying are elementary errors that any GR textbook will correct. This is a Graduate level thread. Please do not clutter it with elementary errors.
 
Demystifier said:
Suppose that accelerometer is based on a spring. But you can push or pull the spring by your hand, so when you do that would you still say that you are measuring the proper acceleration? I would not, I would say I am measuring the proper force.
In order for an object to have proper acceleration, it must have a proper force applied to it. (Note that in GR gravity is not a force.) The two always go together. The object's invariant mass is the proportionality constant between the two.
 
PeterDonis said:
This is a Graduate level thread.
But the thread starter said that he has no background in physics. I understood this thread as a challenge to explain things in an elementary way, at a popular physics level. In fact, I often take such challenges, in spirit of the famous claim often attributed to Einstein, that you don't truly understand things until you can explain it to uneducated people. If, for that reason, I over-trivialized some things, I apologize.
 
Demystifier said:
Do you mean geodesic deviation?
You don't need geodesic deviation to know a worldline is "curved" (non-geodesic). You only need to know the world line does not parallel transport its own tangent vector. ##u^a \nabla_a u^b \neq 0## assuming affine parameterization, otherwise, you can technically have a ##u^b## on the rhs.
 
  • Like
Likes   Reactions: PeterDonis, Dale and Demystifier
Sushmit Nandy said:
First, I would like to say that I have no formal training in physics, so pardon me please, if I am mistaken.

Maybe this should be asked directly: if you have no formal training in physics, do you still want this thread marked "graduate" level? That tag means you are expecting answers here at the level explained to a graduate student in physics.

Many answers here have been aimed at a graduate level audience, with terms of art not explained in detail.

The topic is graduate level. But do you want the answers to be at a graduate level?
 
  • Like
Likes   Reactions: Dale and PeterDonis
Here's an example leveler setter, and a bit of exposition expanding on things mentioned earlier in this thread:

The kinematic decomposition of a geodesic congruence is as follows. I use abstract index notation.

Take ##\nabla_b u_a##, where ##u^a## is a vector field whose congruence are time like geodesic curves. This expression can be broken down into:

$$
B_{ab} := \nabla_b u_a = \frac{1}{3}\theta h_{ab} + \sigma_{ab}+\omega_{ab}
$$

Where ##h_{ab}=g_{ab}+u_a u_b## is the spatial metric orthogonal to ##u^a## -- try contracting it with ##u^a## to see that it gives you 0. ##\theta=\nabla_a u^a##, called expansion, is the trace of ##B_{ab}##, ##\sigma_{ab}=B_{(ab)}-\frac{1}{3}\theta h_{ab}##, called shear, is the symmetric trace free part, and ##\omega_{ab}=B_{[ab]}##, called vorticity, is the antisymmetric part.

You don't need to have known before hand this decomposition (many graduate students wouldn't) but can you at least follow some of the steps of that? For example, do you understand why I could freely move indices from being upstairs to downstairs? What do symmetric vs antisymmetric mean? What is a trace?

The kinematic decomposition was mentioned earlier and specifically the vorticity was highlighted. The vorticity has to do with, loosely speaking, rotation of nearby geodesics around each other. And you can see, it is a geometric quantity--being a proper covariant derivative of a geodesic vector field.

Rotation in GR is a difficult subject. I would have a very hard time explaining it at a non graduate level and that could very well be my own limitation and the fact that I don't understand the subject well enough.

I guess one thing worthy of note is that nowhere in here did I talk about non-geodesic motion. Vorticity is already well defined on geodesic congruences.

It's really easy to get spun around in circles here (pun intended).
 
Last edited:
Demystifier said:
Do you mean geodesic deviation?
Sigh. Again, this is elementary GR, and a Graduate level thread shouldn't need to be cluttered with an explanation of it. If you absolutely insist, see Misner, Thorne & Wheeler, Section 6.2, for a simple definition and formula for proper acceleration (they call it "4-acceleration" there, but it's the same thing).
 
Demystifier said:
the thread starter said that he has no background in physics. I understood this thread as a challenge to explain things in an elementary way, at a popular physics level.
If the OP wants the thread level adjusted, he can say so. And then I suppose we can split off the parts of this discussion that are indeed at Graduate level into a separate thread. If "popular physics level" really means "High School" in terms of PF thread levels, we might not be able to say much of anything, since practically nothing we've said so far is at that level (including your questions about proper acceleration, which I would say are Undergraduate level--but as @Matterwave has pointed out, even at that level the subject of rotation can only be discussed to a limited extent).
 
Matterwave said:
The kinematic decomposition of a geodesic congruence
Note that the kinematic decomposition is not limited to geodesic congruences. And since in this thread we have talked about objects that are not in free fall, we need to include non-geodesic congruences in our analysis. For a non-geodesic congruence, you would need to add a proper acceleration term to the decomposition.
 
Demystifier said:
I prefer logical definitions.
Yes, I guessed as much 😁

That is definitely not my preference whenever possible, but certainly it is a valid preference. A good theory-first definition follows Dr. Schuller's approach in his lecture series here: https://www.youtube.com/@thewe-heraeusinternational2060

Basically, you introduce all of the relevant mathematics of manifolds, and then acceleration is ##a=\nabla_v v## where ##v## is the velocity (tangent vector) and ##\nabla_v## is the directional derivative in the ##v## direction. @Matterwave wrote the component expression ##a^\nu = v^\mu \nabla_\mu v^\nu## (Edit: see below, I accidentally misrepresented @Matterwave’s post)

In Schuller’s approach there is only the proper acceleration which is simply called acceleration, and coordinate acceleration is physically irrelevant and doesn't even deserve a name and especially doesn't deserve to be called acceleration. Then an accelerometer is a device that measures acceleration.
 
Last edited:
Demystifier said:
But you can push or pull the spring by your hand, so when you do that would you still say that you are measuring the proper acceleration?
That is a silly question. What would make you believe that I would still say that?

The definition I gave was "Proper acceleration is what an accelerometer measures”. This means that proper acceleration is the measurand of an accelerometer. The measurand is the thing that is being measured, not the measurement. If a device is not operating within specifications then its readout doesn't reflect the measurand.
 
Last edited:
Dale said:
Basically, you introduce all of the relevant mathematics of manifolds, and then acceleration is ##a=\nabla_v v## where ##v## is the velocity (tangent vector) and ##\nabla_v## is the directional derivative in the ##v## direction. @Matterwave wrote the component expression ##a^\nu = v^\mu \nabla_\mu v^\nu##
Note that I was using abstract index notation. It is the notation used by Penrose and Wald. It looks a lot like component expressions in many situations but it's not. 😂

Specifically, something like ##u^a## is read as an actual vector and not the components of the vector. Breaking it down into components looks like:

$$
u^a=\sum_\mu u^\mu (e_\mu)^a
$$
 
  • Informative
  • Like
Likes   Reactions: PeterDonis and Dale
Matterwave said:
Note that I was using abstract index notation. It is the notation used by Penrose and Wald. It looks a lot like component expressions in many situations but it's not. 😂

Specifically, something like ##u^a## is read as an actual vector and not the components of the vector. Breaking it down into components looks like:

$$
u^a=\sum_\mu u^\mu (e_\mu)^a
$$
Oops, sorry, I did not intend to misrepresent your post. Then I introduced the component expression myself
 
Dale said:
Oops, sorry, I did not intend to misrepresent your post. Then I introduced the component expression myself
All good!

I prefer abstract index notation because it keeps the distinction between vectors/forms/tensors and their components clear. You use latin ##a,b,c..## indices to denote actual geometric objects. You use Greek ##\mu,\nu,...## indices to denote components (or raw labels, like when you label a basis vector).

And you can write something like ##{{T^{a}}_{bc}}^{d}## without twisting yourself into knots lol.
 
  • Like
Likes   Reactions: PeterDonis and Dale
The original question was:
Sushmit Nandy said:
Why does this asymmetry exist in nature?
The first response in the thread, that rotation is an acceleration, was a good start, but then the question becomes why the asymmetry between velocity and acceleration?
I would not call that an asymmetry, but still one can ask, why the difference?
In GR, it does seem to come down to frame dragging (I found https://iopscience.ios.org/article/10.1088/1742-6596/82/1/012003/pdf very readable, which I need), but that as an answer is still just "that's the way it is".

Dale said:
The universe also does not appear to conform to Mach’s principle
Would you expand on that, please? I understand that GR does not require it to, but is it known that it doesn't?

Btw, it should really be called Berkeley's principle, predating Mach by a couple of centuries (https://en.wikipedia.org/wiki/George_Berkeley).

A philosopher, one Bishop Berkeley*
Remarked, metaphysically darkly,
That what we don't see cannot possibly be
And the rest's altogether unlarkly.

*Pronounced BARKlee.
 
PeterDonis said:
Sigh. Again, this is elementary GR, and a Graduate level thread shouldn't need to be cluttered with an explanation of it. If you absolutely insist, see Misner, Thorne & Wheeler, Section 6.2, for a simple definition and formula for proper acceleration (they call it "4-acceleration" there, but it's the same thing).
Well, you confused me by calling it "path curvature". I am not aware that somebody calls it that way, which I asked you to define what you mean. If you meant 4-acceleration, then I'm fine.
 
Demystifier said:
you confused me by calling it "path curvature".
Meaning, the degree to which the path (worldline) is not straight (geodesic). That's what 4-acceleration/proper acceleration is, geometrically speaking.

It's true that "path curvature" is not as common a term; the main usage of it that I've seen is in references that use the Frenet-Serret formalism. The key distinction to keep in mind is that it's the curvature of the path in spacetime, not "space".
 
  • Informative
  • Like
Likes   Reactions: Matterwave and Demystifier
haruspex said:
In GR, it does seem to come down to frame dragging
Frame dragging is one particular manifestation of the rotation of a gravitating body, or more precisely the effects of that rotation on the spacetime geometry surrounding the body.

But that's not the kind of rotation the OP is talking about. The OP is talking about rotation of a test object, one whose effect on the spacetime geometry is negligible. In the simplest such scenarios, spacetime is treated as flat; that's what the OP scenario was doing. There is no frame dragging in flat spacetime, but the manifestation of rotation of an object as vorticity is still there.
 
haruspex said:
I understand that GR does not require it to, but is it known that it doesn't?
I think whether or not our actual universe (as opposed to particular theoretical models in GR) conforms to Mach's principle is still a subject of debate among physicists.
 
Dale said:
That is a silly question. What would make you believe that I would still say that?

The definition I gave was "Proper acceleration is what an accelerometer measures”. This means that proper acceleration is the measurand of an accelerometer. The measurand is the thing that is being measured, not the measurement. If a device is not operating within specifications then its readout doesn't reflect the measurand.
We understand each other about physics, but I want to nitpick a bit about logic. How can I know that a device is not operating within specifications, if you didn't tell me any specifications of the accelerometer? You only told me that "Proper acceleration is what an accelerometer measures”, but that is a definition of "proper acceleration", not a specification of the accelerometer.
 
haruspex said:
Would you expand on that, please? I understand that GR does not require it to, but is it known that it doesn't?
Brans Dicke gravity has a parameter that represents how Machian the universe is (with lower values being more Machian and GR being the limit as it goes to infinity). The current lower bound is 40000, with no upper bound.
 
Demystifier said:
I want to nitpick a bit about logic
Yeah, that doesn’t sound particularly enjoyable.

So instead of getting involved in nitpicking, I will just point out that many, perhaps most, types of measuring devices were developed before the corresponding theory. So, if you want to allow science to actually happen in the real world instead of just in theory, then you must allow operational definitions in the absence of a theory.
 
In Newtonian theory, rotation, of any kind really, is absolute. The only inertial reference frames are straight line constant velocity motions.

In GR, not everything one might naively think of as "rotation" comes from proper acceleration. This is because curvature of spacetime already gives rise to what one might think of as rotation. It's why we have to be so careful of terminology when discussing GR.

For the simplest example, a free fall orbit in Newtonian theory is non-inertial because gravity is a force. In GR, such motion is geodesic. "A circular orbit" (in free fall) may not count as rotation in the sense of there being something measurable with a (pure) accelerometer. But I think in every day use, it's so easy to get it confused.

The geodetic precession which comes from parallel transporting a spin vector around an orbit is another purely geometrical effect, with no proper forces anywhere. Frame dragging must be sourced by a rotating mass, but the analysis on the test mass still doesn't need proper acceleration. You can still do the analysis via parallel transport of a spin vector along a geodesic.

And indeed, even vorticity (Wald calls this twist btw) can be defined on purely geodesic congruences where there is no proper acceleration anywhere.

But, as Peter correctly already pointed out, you can define vorticity on a non geodesic congruence as well. There you need to consider that the congruence does not have the nice property ##u^a\nabla_a u^b=0##. When you stand in the box (feet planted), from the OP and the box rotates, you feel an actual proper acceleration. That one an accelerometer will measure. The Thomas precession, as was originally pointed out, is a local effect of the non inertial reference frame. That one arises if we analyzed the non-inertial reference frame as a series of Lorentz boosts in different directions and note that pure boosts are not closed under composition of boosts in different directions.

I'm not sure if that was helpful or confusing (apologies if confusing), but at least my head was spinning a bit trying to figure out which parts of the discussion required an accelerometer vs not.