Analysis June 2022 with lstchain v0.9.6 v0.9.7 Jurysek

General information

• Name of the source: LHAASO J2108+5157
• Brief description of the source:

- Object type : Unidentified Galactic PeVatron candidate

- Distance (pc) : Unknown

- RA, Dec in deg (ICRS): 317.22, 51.95

• Analysis by Jakub Jurysek (UNIGE, jakub.jurysek@unige.ch)

Run selection

• Run selection with the use of the notebook from Abelardo Moralejo [1]
• Atmospheric transmission extracted from ELOG
• In total 178 runs selected, obstime 49.3 hours (57% of all observations)
• Interestingly, obstime in the previous analysis was 51.9 hours, even though the number of runs is the same (DL2 for this analysis processed with lstchain v0.9.6, IRFs with lstchain dev (close to v0.9.7))

• Summary of selection cuts:

- Zenith > 55 deg

- Wobble in (0.45, 0.55)

- elapsed_time > 5 min

- transmission_cut > 0.65

- pedestal charge std dev < 1.8 p.e.

- Cosmic rate > 3000 ev/s

- Cosmic rate( > 10 p.e.) > 20 ev/s

- Cosmic rate( > 30 p.e.) > 3 ev/s

- muon ring width std dev < 0.023

• List of selected runs:

- 2021-06-04 : [4913, 4914, 4915, 4916, 4917]

- 2021-06-05 : [4935, 4936]

- 2021-06-12 : [5028, 5029, 5030, 5031]

- 2021-06-30 : [5071, 5072]

- 2021-07-01 : [5080, 5081, 5082, 5083, 5084]

- 2021-07-02 : [5091, 5092, 5093]

- 2021-07-03 : [5101, 5102, 5103, 5104, 5105, 5106, 5107, 5108]

- 2021-07-04 : [5115, 5116, 5117, 5118, 5119, 5120, 5121]

- 2021-07-05 : [5135, 5136, 5137, 5138, 5139, 5140, 5141, 5142]

- 2021-07-15 : [5270, 5272]

- 2021-07-30 : [5411, 5412]

- 2021-08-01 : [5440, 5441, 5442]

- 2021-08-03 : [5461, 5462, 5463, 5464, 5465]

- 2021-08-04 : [5473, 5474, 5475, 5476, 5477, 5478, 5479, 5480]

- 2021-08-05 : [5491, 5492, 5493, 5494, 5497, 5498, 5499, 5500]

- 2021-08-06 : [5505, 5506, 5507, 5508, 5509, 5510, 5511, 5512, 5513, 5514, 5515, 5516, 5517]

- 2021-08-08 : [5576]

- 2021-08-09 : [5590, 5591]

- 2021-08-10 : [5641, 5642, 5643]

- 2021-08-11 : [5681, 5682, 5683, 5684, 5685, 5686, 5687]

- 2021-08-12 : [5707, 5708, 5709, 5710, 5711, 5712, 5713]

- 2021-08-13 : [5727]

- 2021-09-01 : [5947, 5948, 5949, 5950, 5952]

- 2021-09-02 : [5980, 5981, 5982, 5983, 5984, 5985, 5986, 5987, 5988, 5989, 5990, 5991]

- 2021-09-03 : [5999, 6000, 6001, 6002, 6003, 6004, 6005, 6006, 6007, 6008, 6009, 6010]

- 2021-09-04 : [6023, 6024, 6034, 6035, 6036, 6037, 6038]

- 2021-09-05 : [6058, 6059, 6060, 6061, 6062, 6063, 6064, 6065, 6066]

- 2021-09-06 : [6079, 6080, 6082, 6083, 6084, 6085]

- 2021-09-07 : [6130, 6131, 6132, 6133, 6134]

- 2021-09-09 : [6175, 6176, 6177, 6178, 6179, 6180, 6181, 6182, 6183]

- 2021-09-11 : [6230, 6231, 6233]

- 2021-09-12 : [6254, 6255, 6256, 6257]

MC information

- AllSkyMC production

- standard DL1 produced in lstmcpipe v0.7.4 (Training DiffuseGammas and Protons, Testing Ring-like gammas), and v0.8.2 (Testing DiffuseGammas)

- RFs training data reconstruction up to DL2 done manually in lstchain v0.9.6 (lstmcpipe wasn't ready for the full AllSky MC at the time of analysis), IRFs created in dev lstchain (close to v0.9.7)

- For IRFs we used only the closest (in zenith) testing nodes to the LHAASO path

• Nodes used for IRFs:

- node_theta_14.984_az_355.158_

- node_theta_32.059_az_355.158_

- node_theta_43.197_az_262.712_

- node_theta_52.374_az_301.217_

• Standard DL1a MC:

- Training:

/fefs/aswg/data/mc/DL1/AllSky/20220511_allsky_std/TrainingDataset/dec_4822/GammaDiffuse/dl1_20220511_allsky_std_dec_4822_GammaDiffuse_merged.h5
/fefs/aswg/data/mc/DL1/AllSky/20220511_allsky_std/TrainingDataset/dec_4822/Protons/dl1_20220511_allsky_std_dec_4822_Protons_merged.h5

- Testing (Ring-like MC, offset 0.4 deg):

/fefs/aswg/data/mc/DL1/AllSky/20220511_allsky_std/TestingDataset/dl1_20220511_allsky_std_{NODE}_merged.h5

- Testing (Diffuse MC, only four nodes close to the source path in the sky):

/fefs/aswg/data/mc/DL1/AllSky/20220527_src2_diffgamma/TestingDataset/GammaDiffuse/dl1_20220527_src2_diffgamma_{NODE}_merged.h5

• NSB tuning parameters:

"image_modifier": {
   "increase_nsb": true,
   "extra_noise_in_dim_pixels": 0.979,
   "extra_bias_in_dim_pixels": 0.339,
   "transition_charge": 8,
   "extra_noise_in_bright_pixels": 1.133,
   "increase_psf": false,
   "smeared_light_fraction": 0.0
}

• Tuned DL1b MC:

- Training:

/fefs/aswg/workspace/jakub.jurysek/simulations/DL1/AllSky/20220511_allsky_tune_lhaaso/TrainingDataset/dec_4822/GammaDiffuse/dl1_20220511_allsky_tune_lhaaso_dec_4822_GammaDiffuse_merged.h5
/fefs/aswg/workspace/jakub.jurysek/simulations/DL1/AllSky/20220511_allsky_tune_lhaaso/TrainingDataset/dec_4822/Protons/dl1_20220511_allsky_tune_lhaaso_dec_4822_Protons_merged.h5

- Testing:

/fefs/aswg/workspace/jakub.jurysek/simulations/DL1/AllSky/20220511_allsky_tune_lhaaso/TestingDataset/dec_4822/GammaPoint/dl1_20220511_allsky_tune_lhaaso_{NODE}_merged.h5
/fefs/aswg/workspace/jakub.jurysek/simulations/DL1/AllSky/20220511_allsky_tune_lhaaso/TestingDataset/dec_4822/GammaDiffuse/dl1_20220511_allsky_tune_lhaaso_{NODE}_merged.h5

• Random Forests (source independent):

- Models trained in lstchain v0.9.6, i.e. with "az_tel", "alt_tel" RF features

- cfg file: /fefs/aswg/workspace/jakub.jurysek/simulations/DL1/AllSky/20220511_allsky_tune_lhaaso/lstchain_tune_lhaaso_84.json

/fefs/aswg/workspace/jakub.jurysek/simulations/models/AllSky/20220511_allsky_tune_lhaaso/dec_4822/

• DL2 MC (testing only for IRFs):

/fefs/aswg/workspace/jakub.jurysek/simulations/DL2/AllSky/20220511_allsky_tune_lhaaso/TestingDataset/dec_4822/GammaDiffuse/dl2_20220511_allsky_tune_lhaaso_{NODE}_merged.h5
/fefs/aswg/workspace/jakub.jurysek/simulations/DL2/AllSky/20220511_allsky_tune_lhaaso/TestingDataset/dec_4822/GammaPoint/dl2_20220511_allsky_tune_lhaaso_{NODE}_merged.h5

DL1 data

• DL1a files produced by LSTOSA (lstchain v0.9)
• lstchain v0.9 tailcut8-4 (with cleaning based on pedestal RMS, dynamical cleaning)
• cfg /fefs/aswg/data/real/DL1/{date}/v0.9/tailcut84/log/lstchain_config_tailcut84_v092.json

/fefs/aswg/data/real/DL1/{date}/v0.9/tailcut84/

DL2 data

• data in full range of zenith angles (0, 55) reconstructed with AllSky RFs

/fefs/aswg/workspace/jakub.jurysek/data_analysis/lhaaso_J2108/DL2/allsky_mc/

DL3 data

• Fixed cuts:

"intensity": [50, Infinity],
"width": [0, Infinity],
"length": [0, Infinity],
"r": [0, 1],
"wl": [0.1, 1],
"leakage_intensity_width_2": [0, 1.0],
"event_type": [32, 32]

Cut optimization for source detection (energy dependent)

• lstchain_create_irf_files tool in lstchain doesn't support energy dependent cuts optimized on sensitivity (only gamma efficiency)
• Energy dependent gammaness cut optimized on Crab detection significance in each energy bin, point-like source assumption
• For energy bins (0.1-1.0 TeV): Crab data from 2021, zenith between 20-40 deg, 13.6h, (2/3 of LHAASO data in this range), good runs selected by Abelardo [2]
• For energy bins (1.0-100.0 TeV): All 121 good runs selected by Abelardo, 35.6h [3]
• DL1->DL2 (Crab) with AllSky RFs tuned on the Crab field

models: /fefs/aswg/workspace/jakub.jurysek/simulations/models/AllSky/20220511_allsky_tune_crab/dec_2276/
DL2 data: /fefs/aswg/workspace/jakub.jurysek/data_analysis/crab/DL2/allsky_mc/

Cut optimization for 1D spectral analysis in gammapy (global cuts)

• Crab data from 2021, zenith < 35 deg, good runs selected by Abelardo [4]
• Global gammaness and theta2 cuts optimized on Crab detection significance, point-like source assumption
• As gamma efficiency optimized cuts on diffuse gammas are not optimal for point source analysis, we also need global gammaness and theta2 cuts for DL3 analysis (see also IRF description).
• DL1->DL2 (Crab) with AllSky RFs tuned on the Crab field (see above)

global_gh_cut: 0.84
global_theta_cut: 0.2 (0.04 deg^2)

IRFs

• Full enclosure
• Four different IRFs merged with DL2 data depending on the run zenith angle
• Global cuts optimised on Crab data used
• Energy dependent cuts not yet possible. For these we would need point-like IRFs produced from ring-like MC. The problem is that the LHAASO source was observed with variable offsets from 0.45 - 0.55 deg, while point-like IRFs in v0.9.6 have a hard-coded offset axis [0.3, 0.5] deg, which means that IRF cannot be used for many runs in data in gammapy analysis.
• Produced in lstchain dev (close to v0.9.7)

• IRF path: /fefs/aswg/workspace/jakub.jurysek/data_analysis/IRFs/v0.9.7/AllSky/

irf_allsky_{NODE}_int50_leak10_gh084_th004_diffuse.fits.gz

DL3 files

/fefs/aswg/workspace/jakub.jurysek/data_analysis/lhaaso_J2108/DL3/v0.9.7/AllSky/int50_leak10_gh084_th004_diffuse/

High-level analysis

Theta2 distribution

- Energy dependent gammaness cuts optimized on Crab detection significance + global theta2 cut (also optimized on Crab significance, but not bin-wise)

- 3 OFF regions

- Three energy bins blind, for E>3TeV we have 3.7 sigma

1D Spectral analysis

• Performed with gammapy-v0.19

bkg_maker = ReflectedRegionsBackgroundMaker(exclusion_mask=exclusion_mask,
                                           min_distance=1 * u.rad,
                                           max_region_number=2
                                          )
safe_mask_masker = SafeMaskMaker(methods=["aeff-max"], aeff_percent=10)

e_min, e_max = 0.1, 100
nbin = 6
energy_axis = MapAxis.from_energy_bounds(
   e_min, e_max, nbin=nbin, per_decade=False, unit="TeV", name="energy"
)
energy_axis_true = MapAxis.from_energy_bounds(
   0.05, 100, nbin=20, per_decade=False, unit="TeV", name="energy_true"
)

• Spectral fitting of stacked LST dataset
• Performed for point-like source assumption
• LST-1 data alone: Power-law spectral model
• Joint likelihood forward folding with LHAASO SED data points: Exponential cutoff power-law spectral model
• Results are compared with Analysis April 2022 with lstchain v0.9.X Jurysek

Power-law model of LST-1 data

Joint likelihood forward folding with LHAASO flux points

Analysis of systematic

• True energy in testing diffuse MC gammas scaled by a factor in a range [0.8, 1.2] for all four nodes close to the LHAASO path

/fefs/aswg/workspace/jakub.jurysek/simulations/DL2/AllSky/20220511_allsky_tune_lhaaso/TestingDataset/dec_4822/GammaDiffuse/Energy_scaled/

• DL3 created with the closest IRFs along the LHAASO path for all energy scaling factors

/fefs/aswg/workspace/jakub.jurysek/data_analysis/IRFs/v0.9.7/AllSky/Escale/
/fefs/aswg/workspace/jakub.jurysek/data_analysis/lhaaso_J2108/DL3/v0.9.7/AllSky/Escale{0o8..1o2}/int50_leak10_gh084_th004_diffuse

• Full enclosure IRFs, point-like source assumption (ON region=0.02 deg), 1D spectral analysis, stacked datasets, PL model

Results

- 20% change in energy => 2% change in PL spectral index, 5% change in flux normalization