Notes for the future

[jennykim1016]

• Segments for working null deblender is in deblend.py. I commented out, but it works well. Use it when needed!
• For calculating chi squared and KL distance, I wrote the code in testing.py. No need to write the code again!
• We could try fitting the major/minor axis for the huge gaussian https://stackoverflow.com/questions/9005659/compute-eigenvectors-of-image-in-python/9007249#9007249
• classification of cost function(negative log likelihood) and implementation for CMU deep lens: https://arxiv.org/pdf/1703.02642.pdf