def run_test(self, query_func, npoints=1e6, points=None):
"""
test the database cross matching it with a set of fake points on the
sphere.
Parameters:
-----------
query_func: callable
function that is used to query the catalog. It must accept ra and
dec, and the collecion as the arguments:
objs = query_func(ra, dec)
npoints: `int`
number of MC generated points to be used for the macthing.
points: `None` or list/array-like
use this argument to provide points to be used for the testing,
skipping the MC generation. points should be array like and have this
structure: [[ra1, dec1], [ra2, dec2], ...]
Returns:
--------
av_query_time: `float`
average query time for npoints queries.
"""
# method specific import
import tqdm
# generate the random sample of points
if points is None:
from extcats.catquery_utils import random_point_sphere
points = random_point_sphere(int(npoints))
self.logger.info("running test queries using %d random points" %
(npoints))
else:
npoints = len(points)
self.logger.info(
"running test queries using %d user defined points" %
(npoints))
# measure query time
tot_found = 0
start = time.time()
for pp in tqdm.tqdm(points):
buff = query_func(ra=pp[0], dec=pp[1], coll=self.coll)
if not buff is None:
tot_found += 1
end = time.time()
total = end - start
av_query_time = total / float(npoints)
self.logger.info("Total document found for query: %d" % tot_found)
self.logger.info(
"Took %.2e sec for %d random queries. Average query time: %.3e sec"
% (total, int(npoints), av_query_time))
def test_queries(self,
query_type,
method,
rs_arcsec,
npoints=1e4,
points=None,
rnd_seed=42,
**qfunc_args):
"""
run test queries using a set of uniformly distributed points on
a sphere as targets.
Parameters:
-----------
query_type: `str`
identifier for the type of query you want to test. Can either be
'within' (for the findwithin query), 'closest' (for the findclosest),
or 'binary' (for the isanything query).
method: `str`
how to query the database. If
- healpix: use the healpix index of the catalog (findwithin_HEALPix)
- 2dsphere: use mongodb searches in spherical geometry
(findwithin_2Dsphere). In this case a value for the sphere2d_key
has to be passed to the function.
- raw: run a javascript function to compute the angular distances
(use findwithin_RAW)
rs_arcsec: `float`
search radius in arcseconds.
npoints: `int`
number of MC generated points to be used for the macthing.
points: `None` or list/array-like
use this argument to provide points to be used for the testing,
skipping the MC generation. points should be array like and have this
structure: [[ra1, dec1], [ra2, dec2], ...]
rnd_seed: `float` or None
if not None, this seed will be passed to np.random.
Returns:
--------
av_query_time: `float`
average query time for npoints queries measured as (start-stop)/npoints
"""
# method specific imports
import time, tqdm, astropy
# generate the random sample of points if none is given
if points is None:
from extcats.catquery_utils import random_point_sphere
points = random_point_sphere(int(npoints), rnd_seed)
self.logger.info("running test queries using %d random points" %
(npoints))
else:
npoints = len(points)
self.logger.info(
"running test queries using %d user defined points" %
(npoints))
# set up the query function
if 'within' in query_type:
qfunc = self.findwithin
elif 'closest' in query_type:
qfunc = self.findclosest
elif 'binary' in query_type:
qfunc = self.binaryserach
else:
raise ValueError(
"illegal value for parameter 'query_type': %s. Allowed: 'within', 'closest', 'binary'."
% query_type)
self.logger.info(
"running %d test queries using function: %s and method: %s" %
(npoints, qfunc.__name__, method))
# measure query time
tot_found = 0
start = time.time()
for pp in tqdm.tqdm(points):
buff = qfunc(pp[0], pp[1], rs_arcsec, method, **qfunc_args)
if (type(buff) == tuple
and buff == (None, None)) or (buff is None) or (not buff):
continue
tot_found += 1
end = time.time()
total = end - start
av_query_time = total / float(npoints)
self.logger.info("Total document found in queries: %d" % tot_found)
self.logger.info(
"Took %.2e sec for %d random queries. Average query time: %.3e sec\n"
% (total, int(npoints), av_query_time))
