WG 2 and LFIDPC Davide Maino Carlo Baccigalupi
WG 2 and LFI-DPC Davide Maino, Carlo Baccigalupi WG 2 Meeting, Paris, 28 -11/2 -12 2005
Component Separation into the pipeline • Need to compare different algorithms/approaches with respect to well defined figure of merits (e. g. residual, impact on PS estimation) • Interfaces with LFI-DPC almost in place by means of Carlo and Davide • The pipeline end-to-end testing of the LFI-DPC has started, including 1/f noise, CMB only, 70 GHz, Gaussian circular, elliptical and realistic beams, 4 detectors out of 6 WG 2 Meeting, Paris, 28 -11/2 -12 2005
Pipeline testing preliminary results WG 2 Meeting, Paris, 28 -11/2 -12 2005
Pipeline testing preliminary results WG 2 Meeting, Paris, 28 -11/2 -12 2005
Pipeline testing preliminary results WG 2 Meeting, Paris, 28 -11/2 -12 2005
LFI-DPC Principles • Different roles in the LFI Consortium: Ø scientists develop data processing prototypes in a coordinated way (e. g. WG 2) Ø Integration Team work at DPC site to build pipeline, managing code with support from develpoers • Worked well and successfully for different groups (e. g. Santander, IEI, Trieste, Uni. Mi, IPAC) • If you want to feed DPC directly with your own code please contact Carlo and/or Davide WG 2 Meeting, Paris, 28 -11/2 -12 2005
Wish list • It would be useful if WG 2 start comparing their own algorithms/approaches on common simulations with increasing complexity (e. g. including instrument specific signature and residual from systematic effects subtraction) • Feed DPC with comparison results and eventually with code prototypes (look at LFI-DPC policy about intelletual property of codes) • A possible shortcut for these activities would be to test the component separation directly on the simulations produced in the DPC end-to-end tests to begin in April 2006 WG 2 Meeting, Paris, 28 -11/2 -12 2005
Resources • Actual battery of L 3 codes at LFI DPC is made of: Ø Source extraction from TOD and small flat sky patches with wavelets, optimal pseudo-filters (Santander, pipeline integrated) Ø Wiener Filter (SISSA, pipeline integrated) Ø MEM in real space (OAPd – IEI, under construction) Ø Parametric (OAPd – IEI, testing) Ø Independent Component Analysis: Fast. ICA (pipeline integrated) ØLight. ICA (SISSA, Uni. MI, LBL, pipeline integration these days) Ø Master power spectrum estimation (IPAC, pipeline integrated) WG 2 Meeting, Paris, 28 -11/2 -12 2005
Plans • Plans are update the nominal performance on Planck, at least for ICA, and to test those on the pipeline outputs as soon as we get there • Separate figures of merit for CMB, point source and diffuse foreground reconstructions • CMB only pipeline testing will be completed within the first weeks of 2006, preliminary results on L 3 within the next consortium meeting WG 2 Meeting, Paris, 28 -11/2 -12 2005
- Slides: 9