report of problems - the Fermi main page

Transcription

report of problems - the Fermi main page
The Galprop model
●
●
●
Galprop model 6002029RE: optimized 2D/4 kpc halo with gammas to 10^8 MeV (galprop 49)
Galprop model 6002029lt: as 6002029RE with updated ISRF, and isotopic abundances (galprop 50) Gamma­ray skymaps for pi0 decay, bremsstrahlung and ICS from 10 Mev to 656 Gev (17 logarithmic slices) in the Chameleon cloud region (l= 279.5 ­> 320.5 deg, b= ­42.5 ­> +0.5 deg, 0.5 deg steps)
pion decay
bremsstrahlung
ICS
LAT observation simulations
●
●
●
●
Simulation for 1 year observation with LAT using gtobssim (Science Tools v7r6p1) for different components
Circular region 12 deg centered on Chameleon cloud (l= 300 deg, b= ­16 deg)
Model for galactic diffuse components: from galprop output sky maps converted into MapCubes with cp_strip http://sirad.pd.infn.it/glast/ground_sw/cloud/code/cp_strip_c.txt fluxes calculated with integrate http://sirad.pd.infn.it/glast/ground_sw/cloud/code/integrate_c.txt DC2 response functions gtobssim bug (offset)
I found a bug in gtobbsim reading MapCube files: I modified the original MapCubes to match simulation results
original map
simulation histogram
modified map
an example with a test source
Components fits
●
●
●
First step: fit LAT simulated data with the same model used to generate them (corrected to match spatial bug described in the previous slide) separately for each component
Binned analysis from 30 MeV to 200 GeV (21 energy bounds) in 64x64 bins (0.25 deg x 0.25 deg) centered on l= 300 b= ­16 deg Summary table:
Pion decay
Bremsstrahlung
ICS
Expected value
Fit value
Error
1
1
1
0.951
0.950
0.937
0.006
0.007
0.006
Fit of pion data
An example: the pion decay gamma rays fit
Flux from MapCube:
0.237 m^­2 s^­1
(entered as flux in obssim)
Value from fit: ratio between observed flux and model flux 0.951±0.006
Binned analysis (results)
●
I carried out different binned analysis tasks over the summed galactic diffuse background trying to separate the cloud related components (pion decay and bremsstrahlung) and the smoother ICS ●
I used as fit model the same MapeCube used to generate data ●
Results:
Angular bin
0.5
0.5
0.25
0.25
0.125
Energy bounds CR components
ICS
10
0.22±0.02
1.64±0.02
21
0.72±0.02
1.25±0.02
21
0.71±0.02
1.26±0.02
41
0.79±0.02
1.20±0.02
0.82±0.02
1.19±0.02
99
Decreasing the bin width results are more similar to simulation input
model
fit
Binned analysis (plots)
model
Model versus simulation counts before fit
(“value” =1)
fit
FIT
Angular bin 0.25 deg ●
41 energy bounds
●
Unbinned analysis
Unbinned analysis results:
●
●
cloud related components (pion decay and bremsstrahlung) 0.73±0.02
ICS 1.30±0.02
Questions
●
●
●
●
●
Gtobssim makes an offset error reading MapCubes
Fit results do not match with simulation input; I modified MapCubes to avoid the gtobssim bug so spatial discrepancies are not so likely
Residuals always have the same feature, I have too many counts at low energies and a few counts at higher energies (both respect to the original model and the fit results) DC2 energy dispersion has biases; do they affect results?
Are fit errors reliable or are we ignoring some systematics? Note that I checked Likelihood behavior near the minimum and it is consistent with errors given by Minuit.