(This is an improved version of a posting that I submitted yesterday
[4-3-22], but which hasn't shown up yet)
In 1907, Einstein published a VERY long paper (in several volumes) on
his "relativity principle". In volume 2, section 18, page 302, titled
"Space and time in a uniformly accelerated reference frame", he
investigated how the tic rates compare for two clocks separated by the
constant distance L, with both clocks undergoing a constant acceleration
"A". He restricted the analysis to very small accelerations (and very
small resulting velocities). His result (on page 305) was that the
leading clock tics at a rate
R = 1 + L A
faster than the rear clock. Note that that result agrees with my
equation, for very small "L" and "A". But he then said:
"From the fact that the choice of the coordinate origin must not
affect the relation, one must conclude that, strictly speaking, equation
(30) should be replaced by the equation R = exp(L A). Nevertheless, we
shall maintain formula (30)."
I've never understood that one sentence argument he gave, for replacing
his linear equation with the exponential equation. But I DID assume he
was right (because he was rarely wrong), until I tried applying his
exponential equation to the case of essentially instantaneous velocity
changes that are useful in twin "paradox" scenarios in special
relativity. Specifically, I worked a series of examples where the
separation of the two clocks is always
L = 7.52 ls (lightseconds)
and where the final speed (with the initial speed being zero) is always
v = 0.866 ls/s.
That speed implies a final "rapidity" of
theta = atanh(0.866) = 1.317 ls/s.
("Rapidity" is a non-linear version of velocity. They have a one-to-one
correspondence. In special relativity, velocity can never exceed 1.0
ls/s in magnitude, but rapidity can have an infinite magnitude. An
acceleration of "A" ls/s/s lasting for "t" seconds changes the rapidity
theta by the product of "A" and "t".)
So in this case, when we are starting from zero velocity and thus zero
rapidity, at the end of the acceleration,
theta = A tau,
where tau is the duration of the acceleration (according to the rear
clock). So, if we know theta and tau, we then know the acceleration:
A = theta / tau.
I do a sequence of calculations, each starting at t = 0, with the two
clocks reading zero, and with zero acceleration for t < 0.
First, I set the duration tau of the acceleration to 1 second. The
acceleration then needs to be
A = theta / tau = 1.317 / 1.0 = 1.317 ls/s/s (that's roughly
40 g's).
So for this first case, the tic rate of the leading clock during the one
second acceleration is
R = exp( L A ) = exp(9.9034), or about 20000.
(I picked that weird value of "L" so that this value of "R" for the fist
case would be a round number, just to make the calculations easier.)
Since R is constant during the acceleration, the CURRENT reading on the
leading clock (which I'll denote as AC, for "Age Change") is
AC = tau R = 2x10sup4 = 20000
(where 10sup4 means "10 raised to the 4th power").
I then start over and work a second case, with ten times the
acceleration (13.17 ls/s/s), but with tau ten times smaller (0.1
second). That keeps the final rapidity the same as in the first case,
and the final speed is also 0.866, as before. For the second case,
AC = 1.02x10sup42.
So when we made the acceleration an order of magnitude larger, and the
duration an order of magnitude smaller, the current reading "AC" on the
leading clock got about 38 orders of magnitude larger.
Next, I start over again and work a third case, again increasing the
acceleration by a factor of 10, and the decreasing the duration by a
factor of 10, so "A" = 131.7 ls/s/s and tau = 0.01 second. Then, AC =
1.27x10sup428. So this time, when we increased "A" by a factor of 10,
and decreased tau by a factor of ten, AC got about 380 orders of
magnitude larger.
AC is not approaching a finite limit as tau goes to zero and "A" goes to
infinity. In each iteration, the change in AC compared to the previous
change gets MUCH larger. Clearly, the clock reading is NOT converging
to a finite limit. It is going to infinity as tau goes to zero.
We can see this, even without doing the above detailed calculations. Since
AC = tau exp(L A),
the tau factor goes to zero LINEARLY as tau goes to zero, but exp(L A)
goes to infinity EXPONENTIALLY as tau goes to zero, so their product is
obviously not going to be finite as tau goes to zero.
So, for the idealization of an essentially instantaneous velocity
change, the change of the reading on the leading clock is INFINITE
during the infinitesimal change of the rear clock. That means that,
when the traveling twin instantaneously changes his speed from zero to
0.866 (toward the home twin), the exponential version of the R equation
says that the home twin's age becomes infinite. But we know that's not
true, because the home twin is entitled to use the time dilation
equation for a perpetually-inertial observer, and that equation tells
her that for a speed of 0.866 ls/s, the traveler's age is always
increasing half as fast as her age is increasing. So when they are
reunited, she is twice as old he is, NOT infinitely older than he is, as
the exponential form of the gravitational time dilation equation claims.
The time dilation equation for a perpetually-inertial observer is the
gold standard in special relativity. Therefore the exponential form of
the gravitational time dilation equation is incorrect.
The correct gravitational time dilation equation turns out to
approximately agree with what Einstein used in his "small acceleration"
analysis, for very small accelerations, but differs substantially for
larger accelerations. And the correct gravitational time dilation
equation agrees with the ages of the twins when they are reunited. It
also exactly agrees with the CMIF simultaneity method for the traveler's
conclusions about the sudden increase in the home twin's age when the
traveler suddenly changes his velocity. The CMIF method provides a
practical way to compute the change in the home twin's age when the
traveler instantaneously changes his velocity. But it is the new
gravitational time dilation equation, and its array of clocks with a
common "NOW" moment, that guarantees that the CMIF result is fully
meaningful to the traveling twin, and that the CMIF method is the ONLY
correct simultaneity method for the traveling twin.