This folder contains the Jupyter Notebooks for tests and experiements showed in the paper.
Tests of locator can be found in test_locator.ipynb
Tests of estimator can be found in test_cdemdn.ipynb (the estimator without optimization), test_opt.ipynb (the optimization step) and test_lcplot.ipynb (plots of light curve examples).
Tests of the joint pipeline are in loc+cdemdn.ipynb.
An example of applying MAGIC to a real event (KMT-2019-BLG-0371) is given in test_KMT.ipynb.
The ./KMT/ folder contains some attempts of applying MAGIC to more real events.
Experiments in the extended abstract can be found in analysis.ipynb and plot.ipynb.
rescale.ipynb explores the differences between the impact of different microlensing parameters.
opt.ipynb and downhill_optimization.ipynb explores how to automate the optimization step for batches of light curves.
locate_and_scale.py is a automated script for transforming (i.e. shifting and rescaling) light curves given
plot_triangle.py is a package for drawing contours of a Gaussian mixture.
test_embedding.ipynb utilizes the Embedding Projector in Tensorboard to visualize the latent space of neural CDE. This enables further exploration like clustering.
Note that the python scripts (ending with .py) are normally the massive production version of the coressponding Jupyter notebooks.