Astronomical Software

In article , Charles Francis
writes:

There seems
to be some issue with SN1997ff anyway, because correction for
gravitational lensing may not be accurate (see abstract below - any
comment, Philip? this is your field isn't it).

astro-ph/0105355, Gravitational Lensing of the Farthest Known Supernova
SN 1997ff, Mörtsell, E.; Gunnarsson, C.; Goobar, A. ApJ 561, Issue 1,
pp. 106-110.

Abstract
We investigate the effects of gravitational lensing due to intervening
galaxies on the recently discovered Type Ia supernova at z~1.7, SN
1997ff, in the Hubble Deep Field North. We find that it is possible to
obtain a wide range of magnifications by varying the mass and/or the
velocity dispersion normalization of the lensing galaxies. In order to
be able to use SN 1997ff to constrain the redshift-distance relation,
very detailed modeling of the galaxies to control the systematic effects
from lensing is necessary. Thus, we argue that, based on our current
limited knowledge of the lensing galaxies, it is difficult to use SN
1997ff to constrain the values of OM and OLAMBDA, or even to place severe
limits on gray dust obscuration or luminosity evolution of Type Ia
supernovae.

This is of course a real concern, more important for higher-redshift
supernovae (because it is more likely that it is lensed, and, if so,
less likely that the lensing galaxy or galaxies would be detected).
However, the "interesting" result with the m-z relation for supernovae
is the fact that the best fit has a positive cosmological constant, the
universe is accelerating etc. This is due to the fact that supernovae
are FAINTER than they would be in either the Einstein-de Sitter universe
or a low-density universe without a cosmological constant. Thus,
"hidden" gravitational lens effects leading to amplification of a
supernovae would tend to bias the results AGAINST the current best-fit
model.

If the matter along the line of sight to the supernovae is less dense
than average, then this will also make the objects fainter than
expected, so in some sense this could make a lambda-dominated universe
look more plausible than it should be. However, it's not just a
question of fainter/less faint but rather the detailed shape of the m-z
curve. In particular, if the current best-fit model is correct, then at
higher redshift supernovae will become BRIGHTER than in competing
models. This is a testable prediction and difficult to mimic with other
effects (less-than-average density, absorption by grey dust etc) which
make things fainter at low redshift, since for these the effect will
tend to increase with redshift.

In principle, one could introduce the inhomogeneity parameter as a
separate free parameter and simultaneously fit for it. In practice, one
would require many more data points. Also, it's not really a fixed
parameter, or even a parameter with a fixed dependence on redshift, but
more like an additional source of error for each data point.

It is important to realise, though, that the current best-fit model can
be obtained from a combination of other cosmological tests. In other
words, take away ALL the supernova results and one would still be left
with a lambda-dominated low-density universe. Thus, any theory which
indicates that the supernova results should result in different
cosmological parameters would have to explain why combinations of other
data sets lead to the result they do.

I think the situation is quite similar to that of 100 years ago when the
question was "what is Avogadro's number?", coupled with the question
"are atoms real?". The issue was decided not by any one decisive test,
but rather the convergence of many different methods on the same result.
Even though objections could be raised to any one method, any
explanation as to how several different methods could conspire to give
the same wrong result would have to be very ad-hoc.