Official: General Relativity Really Accurate

[nescafe]

Two and a half years worth of observations on a double pulsar (http://www.sciencedaily.com/releases/2006/09/060914094623.htm) agree with the predictions made by general relativity to within 5 hundredths of a percent (http://arxiv.org/abs/astro-ph/0609417).

[Yahzi]

5 hundredths of a percent.
You mean... 99.95?

That blows! I expect at least 10 digits of 9's before I give up my cherished woo!

:D

[Cuddles]

You mean... 99.95?

That blows! I expect at least 10 digits of 9's before I give up my cherished woo!

:D

That's astrophysics for you. Now if they'd just stop being so lazy and set up a pulsar near the Moon somewhere we could get lots more 9s and everyone would be happy.:rolleyes:

[Soapy Sam]

-except people with nearshore property. Or just property. Or people who like moons. Or indeed planets. Or anybody in fact.

[Cuddles]

-except people with nearshore property. Or just property. Or people who like moons. Or indeed planets. Or anybody in fact.

Yeah, but there's no pleasing some people.:p

[Nancarrow]

I presume the prediction of any quantity derived from general relativity is gonna involve G, (Newton's) gravitational constant somehow? In which case it's not surprising the accuracy isn't better, since it's the accuracy with which we know G.

[gnome]

So that means there's only a .05% chance of finding it wrong... clearly that is unfair to those attempting to disprove Relativity... demanding odds so strict.

[Ziggurat]

So that means there's only a .05% chance of finding it wrong...

Not exactly. More to the point, it means that if there is a discrepency between the predictions of relativity and our observations, that discrepency will account for less than 0.05% of the observed effect. It says nothing about the chances of finding relativity to be mistaken. An equivalent would be Newtonian mechanics calculations of the orbit of the moon. We KNOW that Newtonian mechanics is wrong, regardless of how accurately it may model the moon's orbit, but if it models the moon's orbit very well then it is a very accurate approximation despite being wrong. These measurements tell us that, at least for the observed phenomenon, general relativity is at least a VERY good approximation. So the "chance" that GR may be wrong remains unknown, which is weaker than your claim in one sense, but on the other hand it is definitely a very good approximation even if it is wrong, which is a stronger statement than your claim.

[jmercer]

Interesting. I would have expected General Relativity to be accurate beyond 2 decimal positions. Isn't that sort of a large variance?

[gnome]

Not exactly. More to the point, it means that if there is a discrepency between the predictions of relativity and our observations, that discrepency will account for less than 0.05% of the observed effect. It says nothing about the chances of finding relativity to be mistaken. An equivalent would be Newtonian mechanics calculations of the orbit of the moon. We KNOW that Newtonian mechanics is wrong, regardless of how accurately it may model the moon's orbit, but if it models the moon's orbit very well then it is a very accurate approximation despite being wrong. These measurements tell us that, at least for the observed phenomenon, general relativity is at least a VERY good approximation. So the "chance" that GR may be wrong remains unknown, which is weaker than your claim in one sense, but on the other hand it is definitely a very good approximation even if it is wrong, which is a stronger statement than your claim.

I was joking anyway :) But I always enjoy a good Relativity lesson...

[Ziggurat]

Interesting. I would have expected General Relativity to be accurate beyond 2 decimal positions. Isn't that sort of a large variance?

The size of the error bars is likely due to imprecision in the measurement, meaning that more precise measurements of the same phenomenon would likely decrease this error bars further.

[hgc]

Not exactly. More to the point, it means that if there is a discrepency between the predictions of relativity and our observations, that discrepency will account for less than 0.05% of the observed effect. It says nothing about the chances of finding relativity to be mistaken. An equivalent would be Newtonian mechanics calculations of the orbit of the moon. We KNOW that Newtonian mechanics is wrong, regardless of how accurately it may model the moon's orbit, but if it models the moon's orbit very well then it is a very accurate approximation despite being wrong. These measurements tell us that, at least for the observed phenomenon, general relativity is at least a VERY good approximation. So the "chance" that GR may be wrong remains unknown, which is weaker than your claim in one sense, but on the other hand it is definitely a very good approximation even if it is wrong, which is a stronger statement than your claim.So it's not really a matter of there being a chance of it being wrong, but a matter of improving the sensitivity of our instruments to the point that the measurable error becomes critically problematic, if it does indeed shake out that way. Then possibly a new theory will give better predictions.

[Cuddles]

Interesting. I would have expected General Relativity to be accurate beyond 2 decimal positions. Isn't that sort of a large variance?

The size of the error bars is likely due to imprecision in the measurement, meaning that more precise measurements of the same phenomenon would likely decrease this error bars further.

As Nancarrow said, the accuracy to which G is known is extremely poor compared to all the other physical constants. I think this would cover a large proportion of the error, so even more accurate measurements won't help much until we work out how to get a better G.

[Ziggurat]

So it's not really a matter of there being a chance of it being wrong, but a matter of improving the sensitivity of our instruments to the point that the measurable error becomes critically problematic, if it does indeed shake out that way. Then possibly a new theory will give better predictions.

Yes.