python - Speeding up take with numba? -

is possible speed np.take using numba?

here attempt, lot slower. can't use nopython mode not np.empty_like command.

import numba import numpy np timer import timer      def take( x, indices ):     result = np.empty_like( indices, dtype=x.dtype )         in range( len( indices ) ):         result[i] = x[ indices[ ] ]     return result  jtake = numba.jit("f4[:](f4[:],i4[:])" )( take )  if __name__=='__main__':      n = 100000     m = 100     idx = np.random.random_integers( 0, n, m )     x = np.random.randn( n )      num_tests=10000      timer( 'take' ):         in range( num_tests ):                 r0 = take( x, idx )      timer( 'numba take' ):         in range( num_tests ):                 r1 = jtake( x, idx )                      timer( 'numpy.take' ):         in range( num_tests ):             r2 = x.take( idx ) 

which has results:

beginning take take took 2.46 seconds beginning numba take numba take took 1.11 seconds beginning numpy.take numpy.take took 0.04 seconds 

the answer no.

numba not act on compiled functions such np.take() or array methods work fancy indexing, basis algorithm. numba acts on interpreted part of code.

your take() function has more overhead numpy's , numba has improved loops (interpreted).

with numpy > 1.9 code should closer numpy's take since algorithm based on fancy indexing , they improved fancy indexing efficiency level of np.take().