numpy.array 和 array以及list的效率对比

#coding:utf-8
import timeit

common_for = """
for d in data:
    s += d
"""

common_sum = """
sum(data)
"""

common_numpy_sum = """
numpy.sum(data)
"""

def timeit_list(n, loops):
    list_setup = """
import numpy
data = [1] * {0}#生成n个数的列表
s = 0
""".format(n)
    print 'list:'
    print timeit.timeit(common_for, list_setup, number = loops)#'common_for'是要执行语句 list_setup导入执行语句需要的数据，loops表示执行语句最后要执行的次数
    print timeit.timeit(common_sum, list_setup, number = loops)
    print timeit.timeit(common_numpy_sum, list_setup, number = loops)

def timeit_array(n, loops):
    array_setup = """
import numpy
import array
data = array.array('L', [1] * {0})#unsigned long，这里的初始化必须先给类型
s = 0
""".format(n)
    print 'array:'
    print timeit.timeit(common_for, array_setup, number = loops)
    print timeit.timeit(common_sum, array_setup, number = loops)
    print timeit.timeit(common_numpy_sum, array_setup, number = loops)

def timeit_numpy(n, loops):
    numpy_setup = """
import numpy
data = numpy.array([1] * {0})
s = 0
""".format(n)
    print 'numpy:'
    print timeit.timeit(common_for, numpy_setup, number = loops)
    print timeit.timeit(common_sum, numpy_setup, number = loops)
    print timeit.timeit(common_numpy_sum, numpy_setup, number = loops)

if __name__ == '__main__':
    timeit_list(50000, 500)
    timeit_array(50000, 500)
    timeit_numpy(50000, 500)

我这里的结果是

list:
0.743401847752
0.244681626678
1.00076942461
array:
1.83873936113
1.36051825384
4.02721759923
numpy:
3.86211620789
3.34397739435
0.0103776763518

所以对于numpy，我们尽量使用它的内置函数吧！