On 10/22/16 10/22/16 1:30 PM, Prokaryotic Caspase Homolog wrote:
> If you are developing a *test* of SR,
> *** you cannot assume the validity of SR ***
Hmmm. This depends on what you mean.
The essence of an experimental test of a theory is:
a. Think up some physical situation for which the theory makes a prediction
b. implement that physical situation and make measurements
c. calculate the predictions of the theory for the physical situation
d. compare the measurements to the prediction(s) of the theory.
In item c one does indeed assume the theory is valid.
But in modern experimental physics we usually follow a procedure we have found
to be better:
a. Think up some physical situation for which the theory makes a prediction
b. implement that physical situation and make measurements
c. create a test theory of the desired theory, which enhances the theory
with additional terms and factors that contain parameters (the form of
the additional terms and factors must be such that for some specific
set of values the original theory is reproduced).
d. fit the measurements to the test theory, treating the parameters as
unknowns to be determined by the fit; such a fit naturally determines
an errorbar for each parameter.
e. compare the results for the parameters WITH THEIR ERRORBARS to the
values they would have if the original theory is valid
In item d one does indeed assume the test theory is valid.
The challenge of item c is to create a believable and useful test theory. For SR
several are known, especially the one by Mansouri and Sexl. For the standard
model there is the standard model extension. For GR there is the PPN formalism.
There are several advantages to this approach:
1. one obtains a QUANTITATIVE measure of how accurately the theory
describes and agrees with the measurements
2. errorbars are handled in an authoritative and definitive manner
3. some experiments cannot determine the values of all parameters; that's
OK, and they can still set limits on the ones they do determine. One
then needs additional experiments that can determine the values of the
other parameters. So one can naturally and directly combine multiple
experiments to test a theory.
4. The errorbars on the parameters provide a way to compare competing
experiments that determine the same parameters. This is especially
valuable to funding agencies (funding proposals always estimate the
errorbars they expect from their measurement).
Tom Roberts