Mrk 501 Preliminary Analysis with lstchain v0.9.x Priyadarshi w IRF interpolation

From my_wiki
Jump to: navigation, search

Mrk 501 Preliminary Analysis with lstchain v0.9.x Priyadarshi w IRF interpolation[edit]

General information[edit]

  • Name of the source: Mrk 501
  • Brief description of the source:
    • Object type : AGN
    • Redshift (z) : 0.034
    • EHBL
    • RA: 16 53 52.216 (hh mm ss), Dec: +39 45 36.613 (dd mm ss)
    • RA, Dec in deg (ICRS): 253.46756699, 39.76017034
  • Analysis by Chaitanya Priyadarshi (IFAE - cpriyadarshi@ifae.es)

Data-taking information[edit]

Wobble observations as mentioned in AGN_paper_LST1#Mrk_501

Monte Carlo information[edit]

For getting a closer data-MC comparison, we produce MC data along various sky positions (nodes) as a preliminary method to facilitate a better Random Forest model for the reconstruction and to interpolate IRFs from various sky positions to the target position of each Observation Run.

DL1 data[edit]

This includes whether you use LSTOSA, specific versions of lstchain, dllab scripts, cleaning levels, and calibration information.


DL1a files produced by LSTOSA (lstchain v0.9)

  • original real DL1a data
/fefs/aswg/data/real/DL1/{Date}/v0.9/tailcut84/
  • lstchain v0.9 tailcut8-4 (with cleaning based on pedestal RMS, dynamical cleaning)
 /fefs/aswg/data/real/DL1/20220508/v0.9/tailcut84/log/lstchain_config_v0.9.4_from2022onwards.json
  • MC nodes: tailcut8-4 (with cleaning based on pedestal RMS), dynamic cleaning
 /fefs/aswg/data/mc/DL1/AllSky/20220531_dec3476_std/{Testing, Training}Dataset/

Random forest[edit]

Here we use the TrainingDataset nodes, selected along the declination path (or closer to) of the selected source. The "dec3476" part in the production ID refers to the declination angle of +34.76 deg.

In this new RF model production, we include zenith and azimuth pointing parameters as well.

  • lstchain v0.9.6 source-independent, dynamic cleaning
 /fefs/aswg/data/models/AllSky/20220531_dec3476_std/dec_3476/

For IRF interpolation[edit]

Here we create an almost uniform grid of nodes of sky positions for TestingDataset MC, so as to facilitate IRF interpolation in the parameter space of cos zenith and the sin of angle (delta) between the geomagnetic field and the particle shower direction. IRF interpolation is performed using pyirf v0.7.

  • Link to MC DL2 files used:
 /fefs/aswg/data/mc/DL2/AllSky/20220531_dec3476_std/TestingDataset/dec_3476/
  • Particle types: ring-wobble gamma (offset range = [0.3999, 0.4001] deg)

Source pointings Run 2177 8416 std wobble runs.png

DL2 data[edit]

The DL2 files of the real data are stored here - /fefs/aswg/workspace/chaitanya.priyadarshi/real_data/AGN/Mrk501/v09x/tailcut84/DL2/real/src_indep/20220531_dec3476/{Date}

DL3 data selection[edit]

For energy-dependent cuts[edit]

Information about your DL3 data selection.

  • intensity > 50
  • r: [0, 1]
  • wl: [0.01, 1]
  • leakage_intensity_width_2: [0, 1]
  • source-independent
    • gh_efficiency: 0.8
    • theta_containment: 0.8

Off regions Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

  • Energy-dependent Gammaness and Theta cut distribution

Gammaness cuts all Run 2177 8416 std wobble runs.png Rad max all Run 2177 8416 std wobble runs.png

High-level analysis[edit]

  • lstchain Tools to generate source-independent IRF and DL3
  • Science Tool: gammapy 1.0
  • point-like IRF, 1D analysis

For DL3 data reduction to DL4 dataset,

  • OFF regions selected - 1 Wobble
  • Safe Energy mask for systematics - Yes, 39.8 GeV < E < 40 TeV
  • Energy axes -
    • True: 10 GeV to 100 TeV with 5 bins per decade
    • Reco: 10 GeV to 100 TeV with 5 bins per decade

For spectral analysis,

  • Model used -
    • Log Parabola + EBL (Dominguez) (6.92 sigmas preference over Power Law + EBL)
  • Reference energy used by calculating the decorrelation energy with a Power Law fit, 563 GeV (Full dataset)
  • Spectral Fit energy range - 10 GeV to 100 TeV with 5 bins per decade
  • Spectral fit with gammapy default optimization - minuit


For LC,

  • E_min = 100 Gev

Analysis Results[edit]

Cross Check with Crab Nebula[edit]

Analysis Crab Nebula Performance Paper Dataset w IRF interpolation Priyadarshi

Theta2 plot[edit]

  • Single energy bin

Theta2 counts 0 Mrk501 153 std wob obs Run 2177 to 8416 1 en bins cut 04 std.png Theta2 excess 0 Mrk501 153 std wob obs Run 2177 to 8416 1 en bins cut 04 std.png

  • 5 bins per decade in energy

Theta2 counts 0 Mrk501 153 std wob obs Run 2177 to 8416 20 en bins cut 04 std.png Theta2 excess 0 Mrk501 153 std wob obs Run 2177 to 8416 20 en bins cut 04 std.png

Significance map[edit]

Excess map[edit]

Spectral results[edit]

For energy-dependent cuts[edit]

Preliminary LC and SED

  • Full Dataset

Model parameters correlation Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

Model parameters Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

LC min 100 max 100000 Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

SED Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png


LC Flux Variability check[edit]

To check the LC flux variability, fluxes for low energy range (100 GeV < E < 500 GeV) and high energy range (500 GeV < E < 100 TeV) are compared. Ratio of these fluxes is called as "hardness ratio". Comparisons and ratios involving UL values are excluded here.

  • Low energy LC flux vs high energy LC flux

LC 2 compare 100 to 500 GeV v 500 GeV no ul to 100.00 TeV Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

LC 2 compare night 100 to 500 GeV v 500 GeV no ul to 100.00 TeV Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

  • Hardness ratio for each data

LC hardness ratio 100 to 500 GeV v 500 GeV to 100.00 TeV Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

LC hardness ratio night 100 to 500 GeV v 500 GeV to 100.00 TeV Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png


  • Hardness ratio vs low energy LC flux

LC hardness ratio 100 to 500 GeV v 500 GeV to 100.00 TeV vs lc flux 100 to 500 GeV wo quiver Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

LC hardness ratio night 100 to 500 GeV v 500 GeV to 100.00 TeV vs lc flux 100 to 500 GeV wo quiver Run 2177 8416 std wobble runs w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png


LC with Bayesian Block time intervals[edit]

Using astropy's bayesian blocks to re-evaluate the time intervals to estimate the LC flux, by using the "measures" fitness function (measured sequence with Gaussian errors), and with argument, p0 = 0.05, false alarm probability to compute the prior.

  • LC flux for the Bayesian Block periods overlaid with the night-wise LC flux

LC bayesian block trial p0 0.05 min 100 max 100000 Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

  • LC flux for the Bayesian Block periods only

LC bayesian block only p0 0.05 min 100 max 100000 Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

  • Photon index evaluated at the reference energy vs LC of the Bayesian Block periods

Photon indices vs bb lc flux for 13 bayesian blocks Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

  • LC flux, photon index and statistical significance of the flux estimation for each time bin

LC bb flux spectral index sqrt ts per bayesian block.png


SED for different Bayesian Block time intervals[edit]

Using the same reference energy as for the total dataset, the datasets are stacked together as per the Bayesian Block time intervals from the previous section, and re-run the Fit and Flux Point Estimation process for each sub-datasets. No other spectral model was found to be preferred over PL+EBL in a significant way (LP+EBL has 1-4.6 sigmas throughout the sub-datasets, but for uniformity, I keep using PL+EBL)

  • SEDs for some close Bayesian Block periods, LC for the same period are shown in the right plots

SED w const reference energy for 13 bayesian blocks Run 2177 8416 std wob only lp ebl dominguez w cust safe en mask min 39.8 GeV 1 off 5 energy bin p dec.png

Retrieved from "https://www.lst1.iac.es/wiki/index.php?title=Mrk_501_Preliminary_Analysis_with_lstchain_v0.9.x_Priyadarshi_w_IRF_interpolation&oldid=7025"