Quote:
Originally Posted by Kwalish Kid
So, you are saying that the distant supernova are different enough, in just the right way, that the differences in their their light-curves exactly cancel out the time-delay and the brightness differences when normalized at a redshift of 0.48. In addition we have the unlucky circumstance that Goldhaber et al. happened to pick this magic redshift number.
|
Normalizing at a redshift of z=0.48 provides a balanced fulcrum with the specific set of supernova available at that time. As the population of higher redshift supernova observed increased, and these supernovae were added to the distribution, there was a possible trend towards smaller ‘stretch’ factors (personal private communication) and the data reduction method was abandoned.
To my knowledge, there has only been one study using stretch factors since 2002:
http://arxiv.org/PS_cache/astro-ph/p.../0607363v3.pdf
In this paper on supernova rise times, and they split the sample at a higher redshift (z=0.589) for statistical comparisons. Normalizing about a midpoint is an acceptable analytical technique when the baseline is uncertain, but it is fraught with potential errors. Specifically, off-setting biases will skew the slope of the curve, but provide the analyst with a false sense of security (been there, done that!). I think normalizing about the midpoint is inappropriate for any application where the underlying physics are not completely characterized. It should not be necessary when comparing supernova rise times: Either they are consistent at all redshifts (after time dilation corrections), or they are not.
Quote:
|
Originally Posted by Conley
Using the same analysis technique on a sample of eight nearby SNe Ia (z < 0.1), we derive a value of _r = 19.58+0.22 −0.19 days, where the quoted error incorporates both statistical and systematic errors. These differ at the 1.4 _ level. In other words, using a considerably more precise comparison made possible by a substantially better data set, we find no compelling evidence for any difference between the rise times of nearby and distant SNe Ia. It is important to understand the limitations of this measurement in terms of its constraints on theoretical models. As was the case in R99, AKN00, and G01, the uncertainties presented above are the error in the mean stretch-corrected rise times of the two samples, not the scatter of rise times between individual SNe Ia. However, testing for differences between the two samples is still a very useful check against evolutionary effects that may be affecting cosmological analyses using SNe Ia.
|
I wish they would publish the raw curves - what is the distribution of the rise times before-and-after correction for time dilation, and before-and-after the stretch-factor normalization?