stevendaryl
Staff Emeritus
I forgot to mention:
if the elevator (rocket) accelerates under the influence of an gravitational field (at the turnaround) then what would we have ????

an inertial frame that accelerates ??

Yes. There are two different notions of "acceleration": coordinate acceleration, and proper acceleration. Coordinate acceleration depends on your coordinate system. For example, the path: $(x=0, y=vt)$ has zero coordinate acceleration in Cartesian coordinates. But if you switch to polar coordinates: $r = \sqrt{x^2 + y^2}$, $\theta = tan^{-1}(\frac{y}{x})$, then the coordinate acceleration in terms of $r$ and $\theta$ is nonzero. Proper acceleration is what is measured by an accelerometer, and it is independent of what coordinate system you use.

even from the point of view of special relativity, this has to be a IRF since the laws of stay unchanged in their simplest form.

Yes. There are two different notions of "acceleration": coordinate acceleration, and proper acceleration. Coordinate acceleration depends on your coordinate system. For example, the path: (x=0,y=vt) (x=0, y=vt) has zero coordinate acceleration in Cartesian coordinates. But if you switch to polar coordinates: r=x 2 +y 2 − − − − − − √ r = \sqrt{x^2 + y^2}, θ=tan −1 (yx ) \theta = tan^{-1}(\frac{y}{x}), then the coordinate acceleration in terms of r r and θ \theta is nonzero. Proper acceleration is what is measured by an accelerometer, and it is independent of what coordinate system you use.
an accelerometer would measure zero in the above case since the rocket would be in free fall.

stevendaryl
Staff Emeritus
an accelerometer would measure zero in the above case since the rocket would be in free fall.

Right, an accelerometer measures proper acceleration, not coordinate acceleration.

Dale
Mentor
2020 Award
an accelerometer would measure zero in the above case since the rocket would be in free fall.
Yes. And therefore the rockets frame is inertial.

PeterDonis
Mentor
2020 Award
an accelerometer doesn't measure acceleration in free fall

It measures zero acceleration in free fall. Zero is a perfectly good measurement result.

You're certainly right, that noninertial frames came up in Newtonian mechanics. So where did the idea that GR was necessary to handle an accelerated reference frame come from?

I think that part of it is the insistence on relativity. [...] Nobody bothered (as far as I know) to try to write Newtonian mechanics in a way that treated all coordinate systems equally. I don't think that the latter was developed until after GR (the Newton-Cartan formulation of Newtonian physics).
I had -and still have- an issue with your "retrospective" because I was not thinking about Newtonian mechanics but about special relativistic mechanics with classical "know-how". As a matter of fact, now that I think of it: Einstein even developed GR based on the understanding that SR can handle accelerated frames! But I think that you are right that many people for some time afterward lacked that understanding, if that is what you meant. It's a mystery to me how this original understanding which was rather well explained in papers could have been lost or mixed up for a while.

I forgot to mention:
if the elevator (rocket) accelerates under the influence of an gravitational field (at the turnaround) then what would we have ???? [...]
even from the point of view of special relativity, this has to be a IRF since the laws of stay unchanged in their simplest form.
It is not a universal IRF. In case you forgot: you can infer my answer (and even infer Einstein's answer), from my earlier reply here:

stevendaryl
Staff Emeritus
I had -and still have- an issue with your "retrospective" because I was not thinking about Newtonian mechanics but about special relativistic mechanics with classical "know-how". As a matter of fact, now that I think of it: Einstein even developed GR based on the understanding that SR can handle accelerated frames! But I think that you are right that many people for some time afterward lacked that understanding, if that is what you meant. It's a mystery to me how this original understanding which was rather well explained in papers could have been lost or mixed up for a while.

I think it was just a matter of figuring out the right pedagogy. Technically, there were no problems in applying SR to noninertial coordinate systems. The issue was how to "frame" what you were doing. Einstein's paper, which does invoke GR to explain the paradox, is an example of bad pedagogy. There is nothing "GR" about it, except for the fact that Einstein maybe was a little unclear about the distinction between the use of noninertial coordinates and gravity. Gravity requires noninertial coordinates, but not the other way around.

[...]
Einstein's paper, which does invoke GR to explain the paradox, is an example of bad pedagogy. There is nothing "GR" about it, except for the fact that Einstein maybe was a little unclear about the distinction between the use of noninertial coordinates and gravity.

I think there was a specific reason Einstein used GR to resolve the twin paradox. He wanted to construct an analogous scenario (via the equivalence principle) in which the "rocket-twin" could say that he was absolutely stationary and unaccelerated during the whole time that the twins were separated. When he fired his rocket engine, he was doing it strictly to counteract the spatially-uniform gravitational field that is somehow momentarily switched on, so that the rocket-twin would remain stationary and unaccelerated. That momentarily switched-on gravitational field causes the "home twin" (the twin who has no rocket) to accelerate, reverse course, and move toward the "traveler". The resulting conclusion using this GR scenario is that the rocket-twin will say that the "home-twin" suddenly gets much older while that gravitational field is switched on.

The exact same result (regarding the rocket-twin's conclusion about the home-twin suddenly getting much older during the turnaround) is obtained without recourse to GR (and without any gravitational fields), purely from SR, using a non-inertial reference frame for the rocket-twin which is formed by piecing together multiple inertial frames that are each momentarily co-moving with the rocket-twin at different instants of his life. The rocket-twin is always at the spatial origin of his non-inertial reference frame, but he never contends that he doesn't accelerate. He knows that he accelerates, and reverses course, when he turns on his rocket. And he knows that it is the home-twin who is unaccelerated for the whole trip.

There is a difference between being "always absolutely at rest" (Einstein's GR scenario for the rocket-twin) versus "being always at the spatial origin of your own personal reference frame, but accelerating at will using your rocket engine" (the SR scenario). But what the rocket-twin says about the home-twin suddenly getting much older during the turnaround is exactly the same for both scenarios (even though it's a different twin doing the turnaround in the two cases).

harrylin
Dale
Mentor
2020 Award
using a non-inertial reference frame for the rocket-twin which is formed by piecing together multiple inertial frames that are each momentarily co-moving with the rocket-twin at different instants of his life.
This is not the only method of forming a non inertial coordinate systems, and as mentioned before it has its own problems.

This is not the only method of forming a non inertial coordinate systems, and as mentioned before it has its own problems.

But the momentary co-moving inertial frames method is the only (SR) method that exactly agrees with the often-cited standard GR method ... both give the result that the rocket-twin says that the home-twin suddenly gets much older during the turnaround. Alternative SR methods that have been proposed don't agree with the standard GR method.

PeterDonis
Mentor
2020 Award
Alternative SR methods that have been proposed don't agree with the standard GR method.

Really? Which ones?

pervect
Staff Emeritus
But the momentary co-moving inertial frames method is the only (SR) method that exactly agrees with the often-cited standard GR method ... both give the result that the rocket-twin says that the home-twin suddenly gets much older during the turnaround. Alternative SR methods that have been proposed don't agree with the standard GR method.

I'm not quite sure what your point is? The momentarily co-moving frames method is popular (deservedly so), but it's not the only method. You'll find some disscusion of Dolby & Gull's "radar time" on Physics Forums and the literature, for instance. See for instance http://arxiv.org/abs/gr-qc/0104077 "On Radar Time and the Twin `Paradox".

In the present paper we recall the definition of ‘radar time’ (and related ‘radar distance’) and emphasise that this definition applies not just to inertial observers, but to any observer in any spacetime. We then use radar time to derive the hypersurfaces of simultaneity for a class of travelling twins, from the ‘Immediate Turn-around’ case, through the ‘Gradual Turn-around’ case, to the ‘Uniformly Accelerating’ case. (The
‘Immediate Turn-around’ and ‘Uniformly Accelerating’ cases are also discussed in Pauri et al.

We show that in all cases this definition assigns a unique time to any event with which Barbara can send and

Editorial note. It isn't obvious, but Barabara can NOT send and receive signals from all space-time events! To give a specific example, if Barbara accelerates at 1g, and Obe stays behind. If Babara leaves in the year 3000 as measured by Obe's calendear, Barbara will never receive a signal sent by Obe in year 3001 or later.

and that this assignment is independent of any choice of coordinates. We then demonstrate that brief periods of acceleration have negligible effect on the radar time assigned to distant events, in contrast with the sensitive dependence of the hypersurfaces implied by Figures 1 and 2. By viewing the situation in different coordinates we further demonstrate the coordinate independence of radar time,
and note that there is no observational difference between the interpretations in which the differential aging is ‘due to Barbara’s acceleration’ or ‘due to the gravitational field that Barbara sees because of this acceleration’.

So to summarize, while the momentarily co-moving frame method is popular (and deservedly so, though I didn't get into it's nice quantities), in some circumstances other methods such as Dolby & Gull's "radar simultaneity" might be better. In the abstract framework of things, the point is that simultaneity is relative, and different simultaneity conventions have different strengths and weaknesses.

Additionally, it's important to note that accelerating observers cannot receive signals from all of space-time, and this in many circumstances effectively prevents an accelerating observer from defining the notion of "at the same time" to certain events behind them, including events that happen at their point of departure after "a long enough time", due to the fact that the accelerating observer can't receive signals from these events as long as they keep accelerating.

Doby and Gull's method isn't an exception to this - while it has some good qualities, it can't handle the situation where Barabara doesn't receive signals from Obe, this is pointed out in the paper but not emphasized.

Last edited:
I think it was just a matter of figuring out the right pedagogy. Technically, there were no problems in applying SR to noninertial coordinate systems. The issue was how to "frame" what you were doing. Einstein's paper, which does invoke GR to explain the paradox, is an example of bad pedagogy. There is nothing "GR" about it, except for the fact that Einstein maybe was a little unclear about the distinction between the use of noninertial coordinates and gravity. [..]
That can't be right. Einstein explains in that very same paper why the twin calculation can hardly be considered paradoxical in SR - at least, it surely wasn't paradoxical for people who correctly understood SR at the time. And Einstein understood rather well how to deal with accelerating frames, as -once more- his "induced gravitational fields" were in fact derived from his calculations with accelerating frames. Working backwards, he did not make any calculation error concerning accelerating frames as far as I can tell, and also according to the Physics FAQ: http://math.ucr.edu/home/baez/physics/Relativity/SR/TwinParadox/twin_gr.html

Perhaps many people who did not correctly understand SR, got the wrong impression that GR had to be used for accelerated objects frames and even accelerated objects, because Einstein argued that GR could be used like that?

Last edited:
stevendaryl
Staff Emeritus
That can't be right. Einstein explains in that very same paper why the twin calculation can hardly be considered paradoxical in SR - at least, it surely wasn't paradoxical for people who correctly understood SR at the time. And Einstein understood rather well how to deal with accelerating frames, as -once more- his "induced gravitational fields" were in fact derived from his calculations with accelerating frames

My point, as I said, is that the calculation has nothing really to do with GR, except in the sense that SR is a limiting case of GR, so any SR calculation is automatically a GR calculation.

Perhaps many people who did not correctly understand SR, got the wrong impression that GR had to be used for accelerated objects frames and even accelerated objects, because Einstein argued that GR could be used like that?

As I said, he's not using GR at all in that calculation. He's using SR in noninertial coordinates.

JesseM
There's a widespread misconception that you need general relativity in situations involving acceleration, but it's just not true; special relativity handles acceleration just fine. You can google for "Rindler coordinates" for one example, and you'll find another example (a clock experiencing uniform circular motion due to the earth's rotation) in Einstein's original 1905 paper to which ghwellsjr gave you a link above.
You can also forego accelerating coordinate systems, and just analyze the time elapsed on an accelerating clock using the coordinates of some inertial frame in which you know the clock's coordinate position and velocity as a function of coordinate time. The trick is to approximate a smoothly-varying path by a polygonal path made up of a bunch of short inertial segments lasting a coordinate time $\Delta t$, that way the time elapsed on the clock on each segment will be $\sqrt{1 - v^2/c^2} \Delta t$, and then you can just add up the clock times on all the segments (using the appropriate v for each segment, which can vary from one to another) to get the total time elapsed on the polygonal path. Then you let the time of each segment become infinitesimal, so the sum becomes an integral and the total time elapsed on a clock with velocity as a function of time v(t) is just $\int \sqrt{1 - v(t)^2/c^2} dt$.

Einstein doesn't go into detail, but he does allude to this method at the end of section 4 of the 1905 paper, when he writes:
It is at once apparent that this result still holds good if the clock moves from A to B in any polygonal line, and also when the points A and B coincide.

If we assume that the result proved for a polygonal line is also valid for a continuously curved line, we arrive at this result: If one of two synchronous clocks at A is moved in a closed curve with constant velocity until it returns to A, the journey lasting t seconds, then by the clock which has remained at rest the travelled clock on its arrival at A will be $\frac{1}{2} tv^2/c^2$ second slow.
(note that the factor he gives is the result of a first-order approximation to the fully accurate formula $t\sqrt{1 - v^2/c^2}$)

My point, as I said, is that the calculation has nothing really to do with GR [..] He's using SR in noninertial coordinates.
I see that you disagree with his argument; and for reasons different from yours, so do I. :)
In fact, I won't discuss disagreements you may have with Einstein about his theory, as you here agree with the point that I tried to make (sorry if that was not clear!): Einstein and contemporaries understood perfectly well how to handle accelerations with SR - even accelerating frames. Furthermore, he did not pretend that GR is required to handle acceleration. Nevertheless, the misconception about acceleration did come about. Thus my suggestion remains that perhaps many people who did not correctly understand SR, got the wrong impression that GR had to be used for accelerated objects frames and even accelerated objects, because Einstein argued that GR could be used like that.

Last edited:
stevendaryl
Staff Emeritus
I see that you disagree with his argument.

No, I don't disagree with his argument. I disagree that his argument involves General Relativity. Going from SR to GR consists of two steps:
1. Allowing the metric tensor $g_{\mu \nu}$ to be dynamic, instead of constant everywhere.
2. Describing how the metric tensor is affected by mass/energy/momentum (the field equations--these describe how spacetime curvature is affected by stress-energy, but in the most basic version of GR, the curvature tensor is a function of the metric only).
Those are the only two additions to SR to produce GR, as far as I know. (There is perhaps a little ambiguity in #1, because it's not always straight-forward how to generalize the physics of flat spacetime to curved spacetime.) Neither is relevant in the twin paradox.

No, I don't disagree with his argument. I disagree that his argument involves General Relativity. [..].
Once more (and with this last repetition I end my discussion about this side topic): that is your main disagreement with Einstein. He admitted that " it is certainly correct that from the point of view of the general theory of relativity we can just as well use coordinate system K' as coordinate system K " , and the rest of that paper is his argument in defence of that position against critics.

stevendaryl
Staff Emeritus
He admitted that " it is certainly correct that from the point of view of the general theory of relativity we can just as well use coordinate system K' as coordinate system K " , and the rest of that paper is his argument in defence of that position against critics.

I agree that it is correct from the point of view of GR, but it's ALSO correct from the point of view of SR.

I agree that it is correct from the point of view of GR, but it's ALSO correct from the point of view of SR.
Einstein clarified that if one would make the mistake to consider the accelerating system K' as a valid "rest" system from the point of view of SR, in which only Galilean reference systems such as K are equivalent, one would create a twin paradox in SR. That has now been explained on this forum many times, even in this thread. The point from which you distracted is the fact that Einstein and his contemporaries knew very well how to handle acceleration in SR.

stevendaryl
Staff Emeritus
Einstein clarified that if one would make the mistake to consider K' as a valid "rest" system from the point of view of SR, in which only Galilean reference systems such as K are equivalent, one would create a twin paradox in SR.

Yes, the equations do not have the same form in a non-inertial coordinate system. But that's a fact about SR. The derivation that Einstein gave is an SR derivation. Of course, it's valid in GR, as well, but there is nothing specifically GR about it.

His derivation is an SR derivation, because he did not make use of any of the differences between SR and GR.

stevendaryl
Staff Emeritus
Yes, the equations do not have the same form in a non-inertial coordinate system. But that's a fact about SR. The derivation that Einstein gave is an SR derivation. Of course, it's valid in GR, as well, but there is nothing specifically GR about it.

His derivation is an SR derivation, because he did not make use of any of the differences between SR and GR.

The equations of SR have the same form in any inertial frame. But to figure out the form in a noninertial frame, all it takes it calculus. I consider that still SR to treat the accelerated frame as "at rest"--that's just a coordinate transformation. The fact that when using noninertial coordinates, the metric tensor becomes position-dependent just falls right of the coordinate transformation. So position-dependent time dilation in noninertial coordinates is inherent in SR. There is nothing "General Relativistic" about it. To call the position-dependence of the metric a "gravitational field" is just picturesque language. There is still no additional assumptions involved, as far as I can tell, beyond those of SR. So it's a mistake to call it a "GR" resolution, because there is nothing in it that isn't already implicit in SR.

I suppose that GR is involved when you say that the fake gravitational field that results from noninertial coordinates is no less real than the gravitational field due to planets, but since the equations don't depend on how "real" the gravitational field is, I just don't see how the derivation could be considered a GR derivation.

Yes, the equations do not have the same form in a non-inertial coordinate system. But that's a fact about SR. The derivation that Einstein gave is an SR derivation. Of course, it's valid in GR, as well, but there is nothing specifically GR about it.

His derivation is an SR derivation, because he did not make use of any of the differences between SR and GR.
K' is invalid as "rest frame" in SR. Despite our differing disagreements with Einstein, we agreed a long time ago on the point that I made, which is that Einstein and others understood acceleration in SR. It was not in "retrospect" that General Relativity was not needed to calculate the twin problem. Einstein never suggested that GR would be needed for the calculation. However, from my discussion with you I now slightly change my hypothesis about how that misunderstanding may have come about. For it now seems plausible to me that many people may have misunderstood Einstein's arguments in his papers from 1916-1918 that GR could be used for accelerated frames and even accelerated objects, so that they misconstrued that according to Einstein GR had to be used. And that's all that I will hypothesize about that. :)

stevendaryl
Staff Emeritus