For the first time the cross-correlation between type Ia supernova host galaxies and surrounding field galaxies is measured using the Supernova Legacy Survey sample. Over the z=0.2 to 0.9 redshift range we find that supernova hosts are correlated an
average of 60% more strongly than similarly selected field galaxies over the 3-100 arcsec range and about a factor of 3 more strongly below 10 arcsec. The correlation errors are empirically established with a jackknife analysis of the four SNLS fields. The hosts are more correlated than the field at a significance of 99% in the fitted amplitude and slope, with the point-by-point difference of the two correlation functions having a reduced $chi^2$ for 8 degrees of freedom of 4.3, which has a probability of random occurrence of less than 3x10^{-5}. The correlation angle is 1.5+/-0.5 arcsec, which deprojects to a fixed co-moving correlation length of approximately 6.5+/- 2/h mpc. Weighting the field galaxies with the mass and star formation rate supernova frequencies of the simple A+B model produces good agreement with the observed clustering. We conclude that these supernova clustering differences are primarily the expected outcome of the dependence of supernova rates on galaxy masses and stellar populations with their clustering environment.
We examine recent evidence from the luminosity-redshift relation of Type Ia Supernovae (SNe Ia) for the $sim 3 sigma$ detection of a ``Hubble bubble -- a departure of the local value of the Hubble constant from its globally averaged value citep{Jha:0
7}. By comparing the MLCS2k2 fits used in that study to the results from other light-curve fitters applied to the same data, we demonstrate that this is related to the interpretation of SN color excesses (after correction for a light-curve shape-color relation) and the presence of a color gradient across the local sample. If the slope of the linear relation ($beta$) between SN color excess and luminosity is fit empirically, then the bubble disappears. If, on the other hand, the color excess arises purely from Milky Way-like dust, then SN data clearly favors a Hubble bubble. We demonstrate that SN data give $beta simeq 2$, instead of the $beta simeq 4$ one would expect from purely Milky-Way-like dust. This suggests that either SN intrinsic colors are more complicated than can be described with a single light-curve shape parameter, or that dust around SN is unusual. Disentangling these possibilities is both a challenge and an opportunity for large-survey SN Ia cosmology.
