Bug #468
closed
Convolution in spectral direction for per spectral pixel calculation
0%
Description
The way the different spectra are convolved is taking in consideration only the spectral binning size, by convolving with a constant averaging kernel with this size. This calculation is not enough for the source spectrum. In this case the kernel should not be constant but it should follow the spatial distribution (the PSF for point sources or the convolution of the PSF with the surface brightness profile for extended sources).
Note that this problem exist only for per spectral pixel calculations and not for per spectral resolution element calculations, because for at second case there is only one calculation done for each spectral binning, which makes the averaging kernel calculation valid.
Updated by Anonymous about 11 years ago
- Tracker changed from Action to Bug
Changed the simulated signal calculator to not use an averaging kernel convolution, but a convolution of a variable function kernel. This function is the PSF (or the PSF convolved with the surface brightness profile for extended sources), bounded in the range of the spectral binning, and normalized to 1. This should fix the bug (it will be available on 0.5.1). I am not updating yet the status to resolved, to wait for a confirmation from Zulien that it works for Euclid. Also changed the type from Action to Bug.
Updated by Christian Surace almost 7 years ago
- Status changed from In Progress to Closed