Tutorials

Our workshop contains several hands-on tutorial sessions. On Tuesday afternoon, we will showcase a pipeline for beam simulation.

On Wednesday, we are offering two consecutive tutorials, one on methods for maps with rectangular pixels, and a tutorial to learn how to calibrate a CMB experiment based on timestreams of planet observations. 

Using a jupyter notebook, you should be able to follow those along on your own computer using one of two options:

1. If you have a NERSC account, we have made a conda environment with all the required dependencies.

It can be set up as a kernel for your notebooks with the following lines: 

module load conda
conda activate /global/cfs/cdirs/sobs/users/aeadler/cmb_cal
python -m ipykernel install –user –name cmb_cal –display-name cmb_cal

2. You can use your own server or your laptop. In that case, you will require the following python dependencies:
3. Standard libraries: os, pickle, math, time

2. Pip-installable packages:

·         numpy for array and matrix operations

·         matplotlib for plotting

·         scipy for a number of mathematical methods and statistics

·         healpy for map projection

·         astropy for units and ephemerids

·         so3g for quaternion math

·         pysm3 to simulate foregrounds

·         h5py to read in h5 files

·         pandas for that extra kick of data analysis

·         pytables for pandas to be happy

2. Git repositories to clone into:

·         map_multi_tool 

·         beams_pipeline

·         sotodlib for more coordinate maths

3. Where the data and notebooks are
   For the TOD to beam fitting we have grouped the data and the notebook here: https://drive.google.com/drive/folders/1Lu5oR7shxIjb0gGangEnxFk_lyzI6lad?usp=sharing
   However, the data is too large to be fourier transformed in google colab. You could either try to use a subset of the data, or use any place with more than ~16Gb of RAM. If you have a professional account for google colab you probably have the necessary ressources for memory.