Experimantal verification of GR

[paweld]

I would like to understand the reason why GR is commonly believed to be
the best therory of gravity we have. As far as I know there are many
theories of gravity some of which is quite similar to GR. Do we really know
that GR is the best theory (it describes the observed phenomena better
then competing theories) or we decided to use GR only because it's
simpler or more popular (I believed that other theories were rejected
because they don't agree with experiments but I'm not sure).

What are the state of the art experiments indicating how accuretally
GR describes the world. I'm very interested in details concerning
the orders of magnitude. I would be pleased if sb could give me a refernce
to some review describing this issues.

I listed below some experimants which I know (unfortunately I don't have
any information about their accuracy):
(1) The equivalence principle tests (e.g. tests of principle of uniniqueness
of free fall)
(2) Perihelion precession of Mercury
(3) Deflection of light by gravitation
(4) Gravitational redshift of light
(4) Time delay of light in gravitation field
(5) Frame-dragging tests (light traveling in the direction of rotation of the
massive object will move around the object faster than light moving against
the rotation as seen by a distant observer)
(6) Lack of Nordtvedt effect (equality of gravitational and inertial masses,
strong equivalence principle)
(7) Indirect evidences of gravitation wave emmision from binary pulsars

BTW is it true that GR is used in GPS?

 

[Naty1]

BTW is it true that GR is used in GPS?

If you mean are their time corrections that must be made because satellite time passes differently than terrestrial time, then yes. Corrections are also required due to the finite speed of light.

GR is the simplist theory consistent with experimental observations. You can read here and get some ideas about experimental verification accuracy.

http://en.wikipedia.org/wiki/Tests_of_general_relativity

There are other theoretical tests which involve the longer than classically expected life of high speed decay particles.

 

[paweld]

If you mean are their time corrections that must be made because satellite time passes differently than terrestrial time, then yes. Corrections are also required due to the finite speed of light.

I meant if one has to take into account the delay of light in Schwartzschild metric
in order to make GPS work correctly.

 

[nitsuj]

I thought because it was Brand Name stuff lol, jk

 

[Dale]

I would like to understand the reason why GR is commonly believed to be the best therory of gravity we have. As far as I know there are many theories of gravity some of which is quite similar to GR. Do we really know that GR is the best theory (it describes the observed phenomena better then competing theories) or we decided to use GR only because it's simpler or more popular (I believed that other theories were rejected because they don't agree with experiments but I'm not sure).

You are correct, there are many other theories of gravity. Some, like Newton's, are rejected because they don't agree with observations. Others, like Brans-Dicke, also agree with all current observations, but have several tuneable parameters. Most people prefer the theory with a smaller number of tuneable parameters (Occham's razor) if two theories equally explain the data.

AFAIK, GR is the only theory which explains all of the data and has only one tuneable parameter (which is fixed by the Newtonian limit so maybe I shouldn't call it "tuneable").

 

[tom.stoer]

There is one generalization of Einstein-Riemann geometry namely Einstein-Cartan geometry which is obtained by relaxing the constraint that the geometry is torsion-free. There are claims that this framework is more natural as the torsion constraint seems to be artificial, especially when spinors are coupled to gravity. In addition it may be the case that this theory is the classical limit of certain quantum gravity theories.

Einstein-cartan gravity modifies some predictions when a non-vanishing spin current which acts as the source of torsion is present. In that case geodesics are modified (in Einstein-Cartan gravity the two definitions for geodesics, namely i) shortest and ii) straighest lines are no longer identical). As torsion is not dynamical i.e. non-propagating these effects are not visible in vacuum; and due to the smallness of these effects one can not distinguish the two theories experimentally.

So it is a matter of taste, which we view as the simplest theory.

 

[PAllen]

There is one generalization of Einstein-Riemann geometry namely Einstein-Cartan geometry which is obtained by relaxing the constraint that the geometry is torsion-free. There are claims that this framework is more natural as the torsion constraint seems to be artificial, especially when spinors are coupled to gravity. In addition it may be the case that this theory is the classical limit of certain quantum gravity theories.

Einstein-cartan gravity modifies some predictions when a non-vanishing spin current which acts as the source of torsion is present. In that case geodesics are modified (in Einstein-Cartan gravity the two definitions for geodesics, namely i) shortest and ii) straighest lines are no longer identical). As torsion is not dynamical i.e. non-propagating these effects are not visible in vacuum; and due to the smallness of these effects one can not distinguish the two theories experimentally.

So it is a matter of taste, which we view as the simplest theory.

Well, it's not really true in GR that geodesics are extremal, except locally and with subsidiary assumptions. They are certainly 'straightest' in parallel transport definition.

For example, in GR, there may be several geodesics connecting two given causally connected events; each is 'straight' from parallel transport definition. Only one is locally extremal. For spacelike (acausal) events it is even more complex. No geodesic is generally even locally extremal - unless restricted to a spacelike hypersurface containing said geodesic.

 

[IsometricPion]

AFAIK, GR is the only theory which explains all of the data and has only one tuneable parameter

Isn't the cosmological constant, $$\Lambda$$, also a tunable parameter of the theory?

 

[PAllen]

Isn't the cosmological constant, $$\Lambda$$, also a tunable parameter of the theory?

Yes. However, that is observationally determined based on cosmology. Thus for GR:

1) G determined by local physics
2) lambda by cosmology.

I anything then doesn't fit, theory must be replaced.

Branse-Dicke:

G, Lambda, and scalar component. Scalar component can be made as small as desired to fit each generation of observations. The upper bound on scalar component has been shrunken continuously to account for more precise observation.

I like a classification posed by Penrose (maybe someone else earlier, but I first saw it by Penrose, arguing that Popper classification is inadequate):

You can have theories that are falsifiable but not verifiable. You can have theories that are verifiable but not falsifiable. Branse-Dicke is an example of verifiable but not falsifiable (verifiable if nonzero scalar component ever found; never falsifiable because scalar component can always be set low enough to account for all observations).

 

[bcrowell]

http://relativity.livingreviews.org/Articles/lrr-2006-3/ , The Confrontation between General Relativity and Experiment, Clifford M. Will

 

[IsometricPion]

You can have theories that are falsifiable but not verifiable. You can have theories that are verifiable but not falsifiable. Branse-Dicke is an example of verifiable but not falsifiable (verifiable if nonzero scalar component ever found; never falsifiable because scalar component can always be set low enough to account for all observations).

Verifiability and falsifiability are dependent on universe of discourse. If one only considers GR and Brans-Dicke (BD) as the possible models of reality (or other sets where BD contains at least one tunable parameter unique to it), I agree with you. However, if one considers a larger set of models, such as those describable via the PPN formalism, both GR and BD are falsifiable and neither are verifiable.

 

[PAllen]

Verifiability and falsifiability are dependent on universe of discourse. If one only considers GR and Brans-Dicke (BD) as the possible models of reality (or other sets where BD contains at least one tunable parameter unique to it), I agree with you. However, if one considers a larger set of models, such as those describable via the PPN formalism, both GR and BD are falsifiable and neither are verifiable.

Point taken.

 

[Matterwave]

Yes. However, that is observationally determined based on cosmology. Thus for GR:

1) G determined by local physics
2) lambda by cosmology.

I anything then doesn't fit, theory must be replaced.

Branse-Dicke:

G, Lambda, and scalar component. Scalar component can be made as small as desired to fit each generation of observations. The upper bound on scalar component has been shrunken continuously to account for more precise observation.

I like a classification posed by Penrose (maybe someone else earlier, but I first saw it by Penrose, arguing that Popper classification is inadequate):

You can have theories that are falsifiable but not verifiable. You can have theories that are verifiable but not falsifiable. Branse-Dicke is an example of verifiable but not falsifiable (verifiable if nonzero scalar component ever found; never falsifiable because scalar component can always be set low enough to account for all observations).

I think the constant in Brans Dicke theory actually should be tending towards infinity for B-D to reduce to GR. The most recent observational bound on it is that it is >40,000.

 

[Dale]

Isn't the cosmological constant, $$\Lambda$$, also a tunable parameter of the theory?

Oops, you are right. I always remember G and forget the cosmological constant.

 

[PAllen]

I think the constant in Brans Dicke theory actually should be tending towards infinity for B-D to reduce to GR. The most recent observational bound on it is that it is >40,000.

Right, I should have said "the effect of the scalar component can be made as small as desired". As the equations are normally given, the scalar effect goes to zero as the constant goes to infinity.