On 8/14/17 4:12 PM, Dono, wrote:
> On Monday, August 14, 2017 at 11:51:36 AM UTC-7, tjrob137 wrote:
>> On 8/14/17 12:41 PM, rotchm wrote:
>>> In a recent thread about the usual train moving along a track scenario, I claimed that:
>>> " ...there exist an inertial frame such that the train & track have the same speed."
>>> I wonder if this is true. Any inputs?
>>
>> In the context of SR, for any pair of timelike objects moving inertially, it is
>> always possible to find an inertial frame relative to which they have
>> 3-velocities with equal magnitudes and opposite directions. So relative to this
>> frame they have equal speeds.
>>
>> Proof: Sum their 4-velocities; the result will be timelike.
>> The inertial frame in question is the rest frame of the sum.
>> This uses two well-known theorems: a) the sum of two timelike
>> 4-vectors is timelike, and b) every timelike 4-vector has a
>> rest frame in which its spatial components are zero.
>>
>> That frame is unique up to spatial rotations, unless the two objects are both at
>> rest in some inertial frame -- then they have equal speeds relative to any
>> inertial frame.
>
> Here is the counterproof:
It is not a counterproof, it is a mistake on YOUR part (actually two quite
different mistakes plus an omission).
> In a given frame F the two objects have speeds u_1 \ne u_2
> In any other frame F' moving with speed x wrt frame F the speeds are:
> u'_1=\frac{u_1+x}{1+x*u_1/c^2}
> u'_2=\frac{u_2+x}{1+x*u_2/c^2}
> The only value of x that makes u'_1=u'_2 is x=c
MISTAKE #1: You solved the wrong problem by confusing velocity with speed. The
formulas you give are for VELOCITIES along a single axis, not speed.
MISTAKE #2: You made a sign error in x, because we want velocities RELATIVE TO
F', given velocities relative to F. So we must subtract x from u_1 and u_2.
OMISSION: You didn't realize that x=-c is also a solution of your problem.
The statement was about SPEED, which along a single axis is the absolute value
of the velocity along that axis. Equating the SPEEDS RELATIVE TO F':
|(u_1-x)/(1-x*u_1/c^2)| = |(u_2-x)/(1-x*u_2/c^2)| (1)
The absolute values are difficult to handle algebraically, so it is simplest to
enumerate the possible signs in (1):
(u_1-x)/(1-x*u_1/c^2) = (u_2-x)/(1-x*u_2/c^2) (1a)
(u_1-x)/(1-x*u_1/c^2) = -(u_2-x)/(1-x*u_2/c^2) (1b)
-(u_1-x)/(1-x*u_1/c^2) = (u_2-x)/(1-x*u_2/c^2) (1c)
-(u_1-x)/(1-x*u_1/c^2) = -(u_2-x)/(1-x*u_2/c^2) (1d)
Clearly 1a and 1d yield the same values of x as your calculation: x=c and x=-c,
neither of which is valid. But 1b and 1c give a valid value of x:
x = (c^2+u_1*u_2-Sqrt[(c^2-u_1^2)*(c^2-u_2^2)]) / (u_1+u_2)
As it is not obvious that is a valid velocity, set u_1=0.8*c and u_2=0.9*c, then
x=0.857921*c, and the speeds relative to F' are 0.184661*c. This satisfies a
plausibility test, as x is between u_1 and u_2, and the relative speed is
comparatively small. Moreover, it satisfies a numerical test: composing
0.857921*c with +-0.184661*c yields 0.9*c and 0.8*c, as expected.
I used Mathematica to solve (1) for x. There are 4 solutions,
three of which are unphysical with |x| >= c. As I stated,
there is exactly one valid solution as long as u_1 != u_2; if
u_1 = u_2, all values -c < x < c are valid solutions to (1).
My original proof, of course, is valid in (3+1) dimensions, not just along a
single axis. The algebra along a single axis is hairy enough (even using
Mathematica), and I have no interest in doing it for the full (3+1) dimensions.
Later you said to me:
> About half of what you post is , in the words of W. Pauli, "Not even wrong" .
This is not true. It's just that YOU keep making errors and attributing them to
me (and to others around here), as above.
Unlike you, I admit my errors, but this is clearly not one of them.
Will you admit your errors here?
(In related posts in this thread you adamantly do not.)
Tom Roberts