|
Weak gravitational lensing creates coherent distortions in
the shapes of background galaxies around massive foreground galaxies
and galaxy clusters, allowing us to map the distribution of matter
around these objects. Standard analyses compress this rich, spatially
structured signal into a circular average around each lens. This
compression is robust and convenient, but throws together noise and
signal in a manner which is less than optimal. In this talk I present
a method that instead models the full two-dimensional lensing pattern.
I forward model the 2D signal and use simulation-based inference (SBI)
to avoid restrictive Gaussian-likelihood assumptions on small to
intermediate scales. I will also discuss additional systematic checks
tailored to two-dimensional lensing measurements, and show where the
method offers the largest gains over standard analyses.
|