xpipe.tools.selector.selector

xpipe.tools.selector.selector(pps, limits)

Applies selection to array based on the passed parameter limits

Selection is defined like

Examples

First make some mock data:

np.random.seed(seed = 5)
nrows = 100000
ids = np.arange(nrows, dtype=int)
ra = np.random.uniform(low=0., high=360., size=nrows)
dec = np.random.uniform(low=-60., high=10., size=nrows)
z = np.random.uniform(low=0.1, high=1.0, size=nrows)
lamb = np.random.uniform(low=20, high=40, size=nrows)

data = np.vstack((ra, dec, z, lamb)).T

then define binning edges:

ra_edges = (0., 100., 300.)
dec_edges = (-60., -30., 10.)
z_edges = (0.2, 0.4, 0.8)
lamb_edges = (20., 30., 45., 60.)

edges = (ra_edges, dec_edges, z_edges, lamb_edges)

This selection divides the data into 24 bins in 4-D space, which we can obtain perform as:

sinds, bounds, plpairs = sl.selector(data, edges)

Let’s inspect the output, sinds contains the boolean indexing arrays to select rows by, while bounds shows the corresponding parameter boundaries, for example:

>>> bounds[13]
((0, (100.0, 300.0), 1), (1, (-60.0, -30.0), 0), (2, (0.2, 0.4), 0), (3, (30.0, 45.0), 1))

where the first number indicates the column of the data file it corresponds to, and the tuple contains the parameter boundaries for that column. The last element is an index for the original parameter edge. Indeed we find:

>>> data[sinds[13], :].min(axis=0)
array([ 100.03881951,  -59.96803764,    0.20010631,   30.02361767])
>>> data[sinds[13], :].max(axis=0)
array([ 299.68649402,  -30.00117535,    0.39994843,   44.99060231])

which are the boundaries we were expecting

Parameters

• pps (np.array N-D) – numpy array with N rows, containing the parameters to split the sample by

• limits (list of lists) – list of parameter limits, each element contains limits for a columna of pps

Returns

list of boolean indices (one for each selection), list of param limits corresponding to each selection parameter limits expanded from sequence to list of tuples

Return type

list, list, list