Noise Model and Sigma Image Without sigma image: s2exp (pixel) = I(pixel)/g + s2bkg ( = Object photon noise + background noise) where s2exp = expected variance in pixel I(pixel) = model flux (DU) in pixel g = effective gain = g1 / N for average of N frames = g1 for sum of N frames s2bkg = Background variance (sky + read-out + others) Note: If wrong gain is used, best-fit parameter values should still be valid, but confidence intervals will not have right sizes With sigma image: s2exp = I2(sigma image) "Initial Condition Filter" (ICF) Þ Needed (or not!) to refine crude initparam estimates Þ Coarsely samples very large volume V0 to determine promising search sub-volume for Metropolis algorithm Þ NICF set on first line of "mdpar" file Þ Choice of NICF is a trade-off between initparam crude, prudence and speed Þ New sampling origin, reduced search volume and cooler Metropolis temperatures are then sent to Metropolis algorithm The cruder the estimates, the higher the number nitparam = yes ("Double-edge sword" i. e. dumb but safe) Þ Recomputed parameters: flux, centroid, sizes, position angles Þ Intensity-weighted centroid is recomputed using minimum area
| Previous slide | Next slide | Back to the index | View Graphic Version |
No notes for slide 5