Playing with crl

I'm trying to fix the binning issues I was having with the reconstruction. Alexia suggested playing with the crl file.

In her email on March 10, subject: improving cross-correlation, she says:
"Go to the part of getCoeff in the crl.py file. There are two choices for for the matrix B (for a flat prior and for a linear prior). Comment out the linear and uncomment the flat, and rerun to see if you get the same behavior for a particular binning choice, or if you get the same TYPE of behavior with the new B, but not correlated to the specific bin choice."

I did above and here are some plots (left is with a constant prior and right is with linear prior):

This doesn't seem to change things much. I talked to Alexia on the phone today and she also suggested I change the value of lambda to see how that effects things.

Currently:
if tol > 0
lam*=10
else:
lam/=10

I'm looking at what changing these numbers (lambda) does to the reconstruction...

if lambda is 5:

Flat prior

Linear prior

if lambda is 100:
Flat prior

Linear Prior:

if lambda is 1:
Linear prior

Flat prior

Alexia doesn't think I should worry to much about this right now, but I thought I'd make some plots for her to look and think about.

On the to do list:
Run reconstruction on real data
Check the mesh to see if this is the cause of the correlation function taking so long to compute.