Static universe - revisited

On Fri, 06 May 11, Steve Willner wrote:
I haven't gone back to check this, but as long as x1 gives s
uniquely, it should be fine. Goldhaber et al. show stretch values
ranging between 0.8 and 1.2, and at first glance your numbers seem to
show lower scatter.

My numbers are in that range, with some outliers to 0.75 and 1.3.
Those SNe with z0.05 average a stretch of 0.923 so if I normalize the
whole data to that, then the following 0.1z bins have:
z=0.0 -- avg stretch of 1.000
z=0.1 -- avg stretch of 1.028
z from 0.2 to 0.9 -- avg stretch of 1.055
z from 1.0 to 1.4 -- avg stretch of 1.081
..... so there is indeed a small dependency of stretch on z.

Eric Flesch wrote: ... we should expect
a Malmquist-caused correlation.

How big do you expect it to be?

Given that rest-frame stretch often reaches 1.25, I'd expect that to
be the average value for the high-z objects.

Neither the redshift nor the stretch depends on cosmology; both are
directly measured from the data.

Well, the stretch has the z component removed. I'm not saying it
shouldn't, but I think there's evidence that too much is removed.
These observations, if normalized to Malmquist expectations (so that
the high-z SNe should have average rest-frame stretch of ~1.25), would
yield a redshift-stretch of about 2/3 of what is assumed. This would
be a significant discovery, and I think it's viable, if only the
researchers would consider it and not treat it as a closed book.

There are several competing groups working on the supernovae, so if
there were any obvious errors, most likely someone would have caught
them. That's no guarantee, of course, but merely asserting
"something must be wrong" is not useful.

Yes, I did not fully analyze the data for my previous posting, because
I had a specific model in mind. So in the finest scientific
tradition, I worked out what stretches my model would yield, then
checked it against the real data. A swing and a miss! So to assert
something is wrong is all I can do at the moment -- I don't agree it's
not useful, surely it's the first step to finding something better.

Eric Flesch

In article ,
Eric Flesch writes:
My numbers are in that range, with some outliers to 0.75 and 1.3.
Those SNe with z0.05 average a stretch of 0.923 so if I normalize the
whole data to that, then the following 0.1z bins have:
z=0.0 -- avg stretch of 1.000
z=0.1 -- avg stretch of 1.028
z from 0.2 to 0.9 -- avg stretch of 1.055
z from 1.0 to 1.4 -- avg stretch of 1.081
.... so there is indeed a small dependency of stretch on z.

OK, that's reassuring.

Given that rest-frame stretch often reaches 1.25, I'd expect that to
be the average value for the high-z objects.

I don't see why. To quantify the expected Malmquist bias, you'd have
to know exactly how all the data were obtained. They are not from
the same or even similar surveys.

SWNeither the redshift nor the stretch depends on cosmology; both are
SWdirectly measured from the data.

Well, the stretch has the z component removed. I'm not saying it
shouldn't, but I think there's evidence that too much is removed.

Dividing by 1+z is just a mathematical operation. You could use any
other measure of the decline timescale. (Other SN groups do in fact
use different measures.) The point is that s, or more generally the
timescale for decline, is measured directly from the data; there is
no cosmological assumption whatsoever in the measurement.

These observations, if normalized to Malmquist expectations (so that
the high-z SNe should have average rest-frame stretch of ~1.25), would
yield a redshift-stretch of about 2/3 of what is assumed.

If you are claiming that unusually bright (large-s) supernovae are
missing from the distant sample, that seems difficult to believe.
More likely would be faint (small-s) supernovae systematically
missing from the nearby samples, but you'd have to look in detail at
the surveys to find out whether that's happening. (Personally I
doubt it, but I could be wrong.)

For derivation of the cosmological parameters, I don't see any reason
the supernova samples have to be complete or even unbiased. All that
is needed is the assumption that nearby and distant supernovae that
have the same timescale and color are the same, and therefore the
absolute magnitudes of the distant supernovae are known. There are
subtle ways that assumption could be false -- among them dependence
on galaxy type or metallicity -- but I don't see Malmquist bias as an
issue.

If you have an unusual cosmological model you want to test, you need
to work out what it predicts for timescale and for distance modulus
and see whether it fits the data.

--
Help keep our newsgroup healthy; please don't feed the trolls.
Steve Willner Phone 617-495-7123
Cambridge, MA 02138 USA

In article , Steve Willner
writes:

There are several competing groups working on the supernovae, so if
there were any obvious errors, most likely someone would have caught
them. That's no guarantee, of course, but merely asserting
"something must be wrong" is not useful.

Indeed. At the beginning, the two big groups were working with
different strategies, different observations, different data reduction
etc. The fact that they got essentially the same results made their
claims (which some pundits were sceptical about at first) more
believable. I think they later reduced each other's data.

Also, none of them set out to find what is now the "standard model" of
cosmology. In fact, many were quite sceptical. This is clearly a case
of the observational data changing what people believe to be the best
model.