使用array-initialization 来完成交织任务 -
def interweave(a, b):
N = a.shape[1]
M = a.shape[0] + b.shape[0]
out_dtype = np.result_type(a.dtype, b.dtype)
out = np.empty((M,N),dtype=out_dtype)
out[::2] = a
out[1::2] = b
return out
示例运行 -
In [274]: A
Out[274]:
array([[1, 2, 3],
[4, 5, 6],
[7, 8, 9]])
In [275]: B
Out[275]:
array([[ 31, 42, 53],
[ 11, 17, 29],
[100, 59, 32]])
In [276]: interweave(A, B)
Out[276]:
array([[ 1, 2, 3],
[ 31, 42, 53],
[ 4, 5, 6],
[ 11, 17, 29],
[ 7, 8, 9],
[100, 59, 32]])
如果A和B的形状相同,我们也可以stack和reshape-
In [283]: np.hstack((A,B)).reshape(-1,A.shape[1])
Out[283]:
array([[ 1, 2, 3],
[ 31, 42, 53],
[ 4, 5, 6],
[ 11, 17, 29],
[ 7, 8, 9],
[100, 59, 32]])
或np.stack((A,B),axis=1).reshape(-1,A.shape[1])。