Go to the documentation of this file.
5 retrieved from https://github.com/marcelhaas/python/blob/master/value_locate.py
13 VALUE_LOCATE locates the positions of given values within a
14 reference array. The reference array need not be regularly
15 spaced. This is useful for various searching, sorting and
16 interpolation algorithms.
17 The reference array should be a monotonically increasing or
18 decreasing list of values which partition the real numbers. A
19 reference array of NBINS numbers partitions the real number line
20 into NBINS+1 regions, like so:
21 REF: X[0] X[1] X[2] X[3] X[NBINS-1]
22 <----------|-------------|------|---|----...---|--------------->
23 INDICES: -1 0 1 2 3 NBINS-1
24 VALUE_LOCATE returns which partition each of the VALUES falls
25 into, according to the figure above. For example, a value between
26 X[1] and X[2] would return a value of 1. Values below X[0] return
27 -1, and above X[NBINS-1] return NBINS-1. Thus, besides the value
28 of -1, the returned INDICES refer to the nearest reference value
29 to the left of the requested value.
32 >>> refx = [2, 4, 6, 8, 10]
33 >>> x = [-1, 1, 2, 3, 5, 5, 5, 8, 12, 30]
34 >>> print value_locate(refx, x)
35 array([-1, -1, 0, 0, 1, 1, 1, 3, 4, 4])
38 This implementation is likely no the most efficient one, as there is
39 a loop over all x, which will in practice be long. As long as x is
40 shorter than 1e6 or so elements, it should still be fast (~sec).
44 print (
"TODO: check if refx is monotonically increasing.")
48 loc = np.zeros(len(x), dtype=
'int')
50 for i
in range(len(x)):
52 ind = ((refx - ix) <= 0).nonzero()[0]
55 else: loc[i] = ind[-1]
61 if __name__ ==
"__main__":
63 refx = [2, 4, 6, 8, 10]
64 x = [-1, 1, 2, 3, 5, 5, 5, 8, 12, 30]
67 assert list(res) == [-1, -1, 0, 0, 1, 1, 1, 3, 4, 4]
68 print (
"Test(s) passed!")
70 x= np.random.random(1e6)*20
def value_locate(refx, x)
list(APPEND LIBS ${NETCDF_LIBRARIES}) find_package(GSL REQUIRED) include_directories($
