KDE-Detector

Kernel density estimation for 2D observational data, based on Koposov et al 2008. There are two kinds of statistics:

• 2 Gaussian kernel convolution
• Poisson distribution

Parameters and Database

Before starting density estimation, one shall first set up:

• src/param.py: parameters for density estimation
• src/param_patch_candidate.py: parameters for preprocessing and summary
• wsdb.py: permission for wsdb.

How to use it:

0. Clean the work directory: bash bashtools/clean.sh
1. Preprocess a dwarf list (optional):
   python preprocess.py
2. Get access to the database and enter information in wsdb.py
3. Set up parameters in src/param.py, especially the following:
   • IS_DWARF_LIST = False # use joint list
   • IS_DWARF_SPLIT_LIST = True # use joint-split list
4. Calculate density estimation:
   • python -W ignore main.py
     (if using manual mode)
   • python -W ignore main.py --name_dwarf "Fornax" --gc_size_pc 10
     (if using the joint or joint-split dwarf list: names can be find in the txt files in dwarfs/)
5. Summarize searching result with python -W ignore summary.py
6. If running step 4 and 5 on a cluster, slurm job scripts are provided:
   • bash bashtools/slurm-slurm.sh # make sure the right input txt
   • sbatch bashtools/slurm-summary.sh # make sure all the KDE searches are done and then run this command