| #cython: language_level=3 | |
| from libc cimport math | |
| cimport cython | |
| cimport numpy as np | |
| ctypedef fused dtype: | |
| np.uint8_t | |
| np.uint16_t | |
| np.uint32_t | |
| np.uint64_t | |
| np.int8_t | |
| np.int16_t | |
| np.int32_t | |
| np.int64_t | |
| np.float32_t | |
| np.float64_t | |
| np.longdouble_t | |
| @cython.boundscheck(False) | |
| @cython.nonecheck(False) | |
| @cython.wraparound(False) | |
| @cython.cdivision(True) | |
| cpdef double ks_2samp(dtype[:] data1, dtype[:] data2): | |
| cdef: | |
| size_t i = 0, j = 0, n1 = data1.shape[0], n2 = data2.shape[0] | |
| dtype d1i, d2j | |
| double d = 0, mind = 0, maxd = 0, inv_n1 = 1. / n1, inv_n2 = 1. / n2 | |
| while i < n1 and j < n2: | |
| d1i = data1[i] | |
| d2j = data2[j] | |
| if d1i <= d2j: | |
| while i < n1 and data1[i] == d1i: | |
| d += inv_n1 | |
| i += 1 | |
| if d1i >= d2j: | |
| while j < n2 and data2[j] == d2j: | |
| d -= inv_n2 | |
| j += 1 | |
| mind = min(mind, d) | |
| maxd = max(maxd, d) | |
| return maxd - mind | |