Hole in extraGalactic emission
I. The facts:
When I tried the source
detection methods on the checkout 3 (55 days) data, I noticed that the diffuse model
did not seem to fit well the data. To push the matter further, I looked at
images without the sources (easy to do with the MC_SRC_ID column). It confirmed
the problem. To understand whether the problem came from the exposure map or
from my reconstitution of the Galactic diffuse emission, I split the diffuse
emission into Galactic (MC_SRC_ID=0) and extraGalactic
(MC_SRC_ID=3). I then noted a strange hole in the extraGalactic
photons. In Galactic coordinates it appeared with a sinuous shape, but when I looked
at the same image in Equatorial coordinates (all sky over all energies, 0,0 at center) I saw that:
There exists a clear
deficit of photons near the Equatorial anticenter,
shaped as a disk with sharp edges and radius 65° or so. There is no such feature
in the exposure map, which has a maximum contrast of only 20% between maximum
and minimum, not aligned with that feature. To quantify the deficit, I have
produced a profile in a band of 40° width along the Equatorial plane:
The count rate in the hole
is about constant at one third that elsewhere. I don’t imagine what real effect
could cause that, so I imagine there must be a bug in the simulation.
II. The history
I then wondered whether
the same feature was already present in the checkout 2 (30 days) data. It is
much more difficult to see there because the photons are not tagged with an
MC_SRC_ID column. The full image in the same coordinates (and SQRT distributed
colors carefully chosen to emphasize what I want to show) is:
When one knows where to
look, I think it is clear that the same hole is there as well.
I then looked back at the DC1 data and … it’s not there
! Finally, I looked at the Galactic diffuse emission. No such hole there
either.
III. Proposed fix
Rather than redo the
entire simulation, it would be easy to excise the extraGalactic
diffuse photons (MC_SRC_ID=3) from the simulated data and run a specific
simulation of the extraGalactic diffuse emission (in
the same conditions) after the bug is fixed. Then all one would have to do is
add a fifth (extraGalactic) events list to the four
(excised) already existing ones.
For the checkout 2 data,
it is impossible to remove the already simulated extraGalactic
photons. One would have to simulate only the missing events (in the hole). If this
is possible, then one could add a secondary events list in the same way.
IV. Solution
It was found that the bug
disappeared between ScienceTools v6r0 (used to
simulate the data) and the current v6r0p4.
Because the simulated
data has other problems (incorrect timing, missing 3EG sources), it was decided
to regenerate the full data set from scratch.