From: Greg Aldering (aldering@panisse.lbl.gov)
Date: Tue Jul 27 2004 - 01:25:52 PDT
Greetings,
At the SCP collaboration meeting Viltaliy drew my attention to a
preprint by Williams & Song (astro-ph/0403680), which claims to detect
a weak lensing signal in the Tonry 2003 (HZSST & SCP) dataset. Today I
quickly read the paper and started to wonder whether the HZSST and SCP
datasets were both consistent with this measurement. The paper plots SN
brightness deviation from the best-fit cosmlogy as a function of the
relative surface density of foreground galaxies as given by the APM
catalog. Their linear fit finds a significant correlation. The
size of the effect is about 0.25 magnitudes peak-to-peak (i.e. from
the lowest to highest foreground galaxy densities).
Two things motivated me to wonder about possible photometry errors as
the source of the detected effect. First, the effect is detected on an
angular scale similar to the size of our images - for larger angular
scales it disappears. Second, when doing photometry on older SCP SN -
when clusters were preferentially targeted - I found problems because
many extended galaxies where included in forming the fiducials ratios; for
such objects the fractional flux in a 2*FWHM radius aperture does not
scale with the seeing as our photometry procedure assumes and so a
significant zeropoint error can (and in at least one case, did) result.
As most of the HZSST SN have photometry uncertainties < 0.28 mag (as
reported by Tonry 2003), while those SCP SNe considered have photometry
uncertainties > 0.28 mag (again, as reported by Tonry), I was able to
distinguish between HZSST and SCP SNe using the size of their plotted
error bars in Figure 1 of Williams & Song. This showed that, for the
most part, the HZSST and SCP SNe residuals are intermixed. There is no
trend of the SCP residuals, however they also consistent with the
measured trend. However, it seems fairly clear that if the 3 brightest
HZSST SNe - all having high foreground densities - were to be
eliminated from the sample, the evidence for weak lensing would largely
disappear.
So, if the detection is spurious, it is likely due to the HZSST SNe,
not SCP photometry errors. Put another way, the significance of the
detection comes from the HZSST due to the presence in their sample of
some brighter SNe and high foreground densities *and* their smaller
quoted uncertainties.
This type of analysis is something we can easily improve upon ourselves
with data now being analyzed.
Cheers,
Greg
This archive was generated by hypermail 2.1.4 : Tue Jul 27 2004 - 01:27:12 PDT