I played with aXeSIM PSFs and tried to compare the PSF in aXeSIM images
with the PSF sent by Benoit.
My goal was to validate that the PSF in images produced with aXeSIM have
the same properties than the zemax PSF.
The attached archive contains the data and the code I used for the tests.
Basicly, the script create a fake spectrum with point like emission
lines at the wavelength of the zemax PSF.
Then it simulate grids of point like sources to get different
observation cases of the same spectrum.
Finally I measured in the image the properties of the PSF and compared
their distribution with the values of the zemax PSF.
I calculated the centroid of the PSF, ellipticity and the orientation
with the moment method.
To compute EE, I oversampled the aXeSIM PSF (with splin, order 3) and
I computed the fraction of the flux in the circle of radius r (I used
the center calculated with the moment method).
There is one figure for each wavelength.
In each figure, I plotted (from left to right and top to bottom):
- histogram of the ellipticity (the red vertical line is the value
measured in the zemax PSF)
- histogram of the orientation
- histogram of the flux (the input flux is 1e-15)
- EE (in red the zemax PSF, in blue the aXeSIM PSF, the black dotted
line is the pixel size)
- histogram of EE50 (radius to encircle 50% of the energy)
- histogram of EE80
For the ellipticity and the orientation, the aXeSIM PSF looked often a
bit biased but the ellipticity is very low
so I not sure this measurements are significants.
For the flux, we systematically lost ~1% of the flux, do you know why?
For the encircled energy, the aXeSIM PSF looked to smooth the structure
in the zemax PSF.
I think it is normal because the scale of those structure are smaller
than the pixel scale
but I'm a bit worry because of the EE50 and EE80 histograms,
EE50 and EE80 are systematically bigger in the aXeSIM PSF than in the
A last point that I don't understand:
my plots EE are very different from the Gregor ones and I don't know
I tried to measured EE directly in the mef file and I got the same
- check flux normalization
- check mef vs benoit's
- check with bilinear interpolation
- check with sampling x10