LBL Cosmology Visitor Program 
Weak Lensing workshop 
Day 2 discussions 

* Monday, Aug. 9, morning discussions will focus on using 
weak lensing for cluster finding, and for mapping cluster 
profiles.  The afternoon will either continue this or 
concentrate on space/ground issues. 

* The community benchmark tests of shear extraction by 
various groups on the same catalogs - STEPS? - could take a 
first step at the LBL workshop with the participants next 
week serving to prototype the comparison.  Will space data 
be available? 

* More discussion of the importance of source redshift 
distribution, and the relative slowness of the z^2 e^{-z^1.2} 
drop off.  The z>3 sources make a significant contribution 
in this case.  What are the effects for space/ground?  

This led to a discussion of whether z_median was actually an 
appropriate measure to use in comparing surveys.  Henk pointed 
out that even just z_mean would be superior.  Generally the 
results could be very sensitive to z so we need better 
estimators of the source redshift distribution. 

As a practical concern in simulations, where should the data 
outputs start?  One might likely need to begin dumps at z=5. 

* sigma8 seems to agree within 1-2sigma from the different 
WL experiments, but this is still far from the 1% precision desired. 

* Rule of thumb of the day [Henk]: "If you can measure photo-z's, 
you can measure shapes". 

* Again, how to use multiband information?  At the least, multiple 
bands give effectively increased exposure time, and one would 
presumably weight the shapes from each filter by S/N.  But what 
added value do multiple filters bring for shape determination? 
Are there pros/cons for systematics? 

* Observations - when will it be practical to have sufficiently 
good observations to achieve 1) tomography, 2) CCC/offset-linear 
method.  Henk says VLT gets an effective number density of 
galaxies of 15/min^2.  Once you divide this into bins, you get 
killed as n^2, which you have to make up with a very large area, 
but systematics will limit this at some point. 

* Some discussion of photo-z's and training sets vs. templates.