[英]Vectorized way to arrange vector into matrix (numpy)
I have 4 vectors of the same dimentions (say 3) 我有4个相同维数的向量(例如3)
a= [1, 5, 9]
b= [2, 6, 10]
c= [3, 7, 11]
d= [4, 8, 12]
What i want to do with numpy is to create a matrix of dimensions 3x2x2 that has this structure 我想用numpy做的是创建一个具有这种结构的尺寸为3x2x2的矩阵
so the resultan matrices will be like this 所以结果矩阵将是这样的
[
[[1, 2],[3,4]],
[[5, 6],[7,8]],
[[9, 10],[11,12]],
]
I know that it is pretty easy using a for loop but I'm looking for a vectorized approach. 我知道使用for循环非常容易,但是我正在寻找一种向量化方法。
Thanks in advance 提前致谢
np.stack
is handy tool for combining arrays (or in this case lists) in various orders: np.stack
是用于以各种顺序组合数组(或本例中的列表)的便捷工具:
In [74]: a= [1, 5, 9]
...: b= [2, 6, 10]
...: c= [3, 7, 11]
...: d= [4, 8, 12]
...:
...:
Default without axis parameter is like np.array
, adding a new initial dimension: 没有轴参数的默认值类似于
np.array
,添加一个新的初始尺寸:
In [75]: np.stack((a,b,c,d))
Out[75]:
array([[ 1, 5, 9],
[ 2, 6, 10],
[ 3, 7, 11],
[ 4, 8, 12]])
But the order isn't what you want. 但是订单不是您想要的。 Lets try
axis=1
: 让我们尝试
axis=1
:
In [76]: np.stack((a,b,c,d),1)
Out[76]:
array([[ 1, 2, 3, 4],
[ 5, 6, 7, 8],
[ 9, 10, 11, 12]])
Order looks right. 订单看起来正确。 Now add a reshape:
现在添加一个重塑:
In [77]: np.stack((a,b,c,d),1).reshape(3,2,2)
Out[77]:
array([[[ 1, 2],
[ 3, 4]],
[[ 5, 6],
[ 7, 8]],
[[ 9, 10],
[11, 12]]])
Another approach is to join the lists, reshape and transpose: 另一种方法是加入列表,重塑和转置:
In [78]: np.array([a,b,c,d])
Out[78]:
array([[ 1, 5, 9],
[ 2, 6, 10],
[ 3, 7, 11],
[ 4, 8, 12]])
In [79]: _.reshape(2,2,3)
Out[79]:
array([[[ 1, 5, 9],
[ 2, 6, 10]],
[[ 3, 7, 11],
[ 4, 8, 12]]])
In [80]: _.transpose(2,1,0)
Out[80]:
array([[[ 1, 3],
[ 2, 4]],
[[ 5, 7],
[ 6, 8]],
[[ 9, 11],
[10, 12]]])
In [81]: __.transpose(2,0,1)
Out[81]:
array([[[ 1, 2],
[ 3, 4]],
[[ 5, 6],
[ 7, 8]],
[[ 9, 10],
[11, 12]]])
We can try to be systematic about this, but I find it instructive to experiment, trying various alternatives. 我们可以尝试对此进行系统化,但是我发现尝试各种替代方法很有帮助。
np.reshape()
will do it: np.reshape()
可以做到:
np.reshape(np.array([a,b,c,d]).T,[3,2,2])
will produce the desired result. 将产生期望的结果。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.