Implements Friends of Friends Sampling function

[atorchylo]

Closes #16

Using fast3tree python package, I implemented the friends of friends sampling function. The algorithm works by precomputing groups for the given table, and then randomly sampling a group according to min_number and max_number requirements. The user has to provide link_distance to the algorithm, which determines the grouping of the galaxies. If the group exceeds the stamp size, the function raises an error. Here is the example code to quickly test:

import btk
import numpy as np
import matplotlib.pyplot as plt

catalog = btk.catalog.CatsimCatalog.from_file('data/input_catalog.fits')

sampling_function = btk.sampling_functions.FriendsOfFriendsSampling(
    min_number = 5, 
    max_number = 10, 
    link_distance=2.5,
    stamp_size=24
)
survey = btk.survey.get_surveys('LSST')
draw_generator = btk.draw_blends.CatsimGenerator(
    catalog,
    sampling_function,
    survey,
    batch_size=8,
    stamp_size=24,
    njobs=1,
    add_noise="background",
    seed=0, # use same seed here
)

blend_batch = next(draw_generator)
ii = np.random.randint(0, 8)

# plot the first blend in the batch, with the r-band
plt.imshow(blend_batch.blend_images[ii, 2, :, :], cmap="gray")

# plot centers
plt.scatter(blend_batch.catalog_list[ii]["x_peak"], 
            blend_batch.catalog_list[ii]["y_peak"], c="r", marker="x")

Sampled galaxies look like this:

[ismael-mendoza]

Looks really good @atorchylo thanks for implementing this. Sorry it took me a while to get back to you. Please find some suggestions below

[ismael-mendoza]

One more thing @atorchylo could you try reducing the link length even further and produce some examples to see if you only get very dramatic blends. Thanks!

[atorchylo]

Hey @ismael-mendoza. I implemented your suggestions and also produced some examples of extreme blends. Code does seem to work as desired! Below are examples of galaxies blends with min number of 3 and link distance of 1.

[codecov]

Codecov Report

Attention: 39 lines in your changes are missing coverage. Please review.

Comparison is base (cc89330) 62.53% compared to head (7b95cf9) 61.15%.

Files	Patch %	Lines
btk/sampling_functions.py	17.02%	39 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #460      +/-   ##
==========================================
- Coverage   62.53%   61.15%   -1.38%     
==========================================
  Files          19       19              
  Lines        1508     1555      +47     
==========================================
+ Hits          943      951       +8     
- Misses        565      604      +39     

Flag	Coverage Δ
unittests	61.15% <17.02%> (-1.38%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.