LBL Cosmology Visitor Program 
Weak Lensing workshop 
Day 4 discussions 

* Morning discussion focused on weak lensing as a cluster finding 
tool. 

* Alexandre Refregier presented background on the topic, and Joe 
Hennawi some technical aspects, with David Wittman discussing DLS 
experience.  Martin pointed out that if one wanted to use the cluster 
correlation function, then one would need thousands of square degrees. 
But at this point the observations would be less noisy in the optical 
so WL would be mostly as calibrator.  For leverage on dark energy one 
wants to measure the high redshift end of the function, which drives 
observations to space. 

* Considerable discussion focused on lensing efficiency in cluster 
finding, and completeness vs. contamination, i.e. false negatives 
and false positives, related to the issue of "dark clusters" and 
what fraction of peaks in the maps is real.  Marc Davis expressed 
concern about the high factor of correction needed for selection 
effects - with DEEP2 he would get concerned about a factor of two 
while lensing seemed to require up to a factor of 10.  One idea is 
to use the shape of the counts with redshift rather than the number 
as such, but this will also be sensitive to mass limits. 

* The summary of the opinions of those present seemed to be that 
WL for cluster science would be very tough, and we decided to have 
Josh Frieman do it (he being absent ;-). 

* The lensing role is likely to be predominantly one of calibration, 
and space would be valuable for going deep, to low mass, and giving 
better calibration.  A case of crosscorrelating weak lensing and 
optical surveys needs to be developed in detail. 

- - Afternoon - - 

* One interesting aspect of the STEP program for comparison is that 
the VLT data will enable examination of the dependence on seeing.  
This is expected to be quite steep, as Gary Bernstein's recent 
document demonstrates. 

* Eric pressed the assembled observers on when the variety of 
weak lensing methods would be ready for prime time.  For tomography, 
the feeling was that CFHTLS would be the first step, though having 
only 15 galaxies/arcmin^2 was a little worrying, though less 
important on large scales.  

* For bispectrum, the answer was possibly 
VIRMOS; it was unclear how large an area was required.  The point 
was made that new systematics would enter as one scaled up surveys 
from 1 deg^2 to 100 deg^2 or higher.  Right now the data quality 
is such that one can only aim at a single number, the amplitude of 
the nongaussian signal, whether this is called skewness or M_ap^3 
or whatever.  Martin says this is all one expects as long as you 
stay within the Born approximation.  Concern was also voiced 
regarding holes within the survey area, and the resulting S/N hit. 

* Geometric methods, CCC etc., would be extremely difficult 
without space observations, and possibly with.  Certainly the 
stability in photo-z determination is a critical push to space.