Attendance: SH, Charles, Brandon, Clem, Mathieu, Jacques
Notes
MR reports on his work with Clem's model 90.91. He finds a strong bias. The cause: Clem makes the simulations with nside = 512. MR works with nside=16, so the averaging in the larger pixels strongly biases the result.
What to do?
CL says that Commander2 is used now for Planck. CL will write to the Oslo group. They are back mid-august; can they engage?
MR says that for Comm1 he needs more memory. Suggestion: can you run with nside = 8, then nside=32? Does the bias decrease without changing sigma ( r )? MR is worried that the algorithm will not converge. Will try.
JD points out that the averaging across the sky gives rise to bias, but so does averaging along LOS. He has also discussed engaging the Saclay group with FG separation. Asks what makes sense to do? SH + Clem: models 1,2,3 are progressively more complicated. Proceed in that order.