Bachelor thesis on high level analysis for the LST-1

The aict-tools are used to train an apply the random forest algorithm for

• Gamma/Hadron Separation
• Energy Regression
• Reconstruction of origin

An updated and improved version of the analysis performed here can be found in the LST_mono_analysis repository.

Data used

1. The models are trained on Monte Carlo simulations of diffuse protons, diffuse gamma-rays and gamma-rays from a point-like source (prod3b).

2. The trained models are used to analyse the following observational data

   • Crab Nebula: 2.63h ON and 1.33h OFF (taken on 18.01.2020)
   • Markarian 421: 2.22h wobble mode (taken on 20.06.2020)

3. All data (MCs and observations) was processed up to image parameter level (DL1) using lstchain (not part of this thesis).

How to build this thesis

1. Create a new conda environment and install the necessary python packages defined in environment.yaml:

   $ conda env create -n bachelor_thesis_lbeiske -f environment.yaml
   $ conda activate bachelor_thesis_lbeiske
   

2. Download input files and store in data in the root of this repository:

   • Simulations: /net/cta-tank/POOL/projects/cta/LST/Simulations/DL1/20190415 (only %_v0.5.1_% and %_v0.5.2_% files necessary)
   • Observational data: /net/cta-tank/POOL/projects/cta/LST/Data/DL1/v0.5.1

3.  $ make
   

4.  $ mkdir HDD
    $ make OUTDIR=HDD/build_scaling_300 \
    CUTS_CONFIG=config/quality_cuts_300.yaml
   

5.  $ make OUTDIR=HDD/build_noscaling_300 \
    GAMMA_FILE=gamma_south_pointing_20200514_v0.5.1_v01_DL1 \
    GAMMA_DIFFUSE_FILE=gamma-diffuse_south_pointing_20200514_v0.5.1_v01_DL1 \
    PROTON_FILE=proton_south_pointing_20200514_v0.5.1_v01_DL1 \
    CUTS_CONFIG=config/quality_cuts_300.yaml
   

6.  $ make OUTDIR=HDD/build_noscaling \
    GAMMA_FILE=gamma_south_pointing_20200514_v0.5.1_v01_DL1 \
    GAMMA_DIFFUSE_FILE=gamma-diffuse_south_pointing_20200514_v0.5.1_v01_DL1 \
    PROTON_FILE=proton_south_pointing_20200514_v0.5.1_v01_DL1 
   

7.  $ make build/thesis.pdf
   

8. Build the presentation:

   $ cd presentation
   $ make