blocks|key|1184342|text|np.vectorize只是一个方便的函数。它实际上并不是make+code+run+any+faster。如果使用np.vectorize不方便，只需编写自己的函数，就可以随心所欲地工作。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|1184343|np.vectorize的目的是将不支持numpy的函数(例如，接受浮点数作为输入，返回浮点数作为输出)转换为可以对numpy数组进行操作(并返回)的函数。|1184344|您的函数f已经支持numpy+--它在其定义中使用了numpy数组，并返回一个numpy数组。所以np.vectorize并不适合你的用例。|1184345|因此，解决方案就是滚动您自己的函数f，它以您所希望的方式工作。|1184346|entityMap|0|LINK|mutability|MUTABLE|url|http://www.mail-archive.com/numpy-discussion@scipy.org/msg00587.html^0|0|C|1M|C|T|O|0|0|0|C|0|4|1|1D|C|0|H|1|0^^$0|@$1|2|3|4|5|6|7|T|8|@$9|U|A|V|B|C]|$9|W|A|X|B|C]]|D|@$9|Y|A|Z|1|10]]|E|$]]|$1|F|3|G|5|6|7|11|8|@$9|12|A|13|B|C]]|D|@]|E|$]]|$1|H|3|I|5|6|7|14|8|@$9|15|A|16|B|C]|$9|17|A|18|B|C]]|D|@]|E|$]]|$1|J|3|K|5|6|7|19|8|@$9|1A|A|1B|B|C]]|D|@]|E|$]]|$1|L|3|-4|5|6|7|1C|8|@]|D|@]|E|$]]]|M|$N|$5|O|P|Q|E|$R|S]]]]

<code>np.vectorize</code> is just a convenience function. It doesn't actually <a href="http://www.mail-archive.com/numpy-discussion@scipy.org/msg00587.html" rel="noreferrer">make code run any faster</a>. If it isn't convenient to use <code>np.vectorize</code>, simply write your own function that works as you wish.

The purpose of <code>np.vectorize</code> is to transform functions which are not numpy-aware (e.g. take floats as input and return floats as output) into functions that can operate on (and return) numpy arrays.

Your function <code>f</code> is already numpy-aware -- it uses a numpy array in its definition and returns a numpy array. So <code>np.vectorize</code> is not a good fit for your use case.

The solution therefore is just to roll your own function <code>f</code> that works the way you desire.

blocks|key|1184386|text|import+numpy+as+np
def+f(x):
++++return+x+*+np.array([1,1,1,1,1],+dtype=np.float32)
g+=+np.vectorize(f,+otypes=[np.ndarray])
a+=+np.arange(4)
b+=+g(a)
b+=+np.array(b.tolist())
print(b)#b.shape+=+(4,5)
c+=+np.ones((2,3,4))
d+=+g(c)
d+=+np.array(d.tolist())
print(d)#d.shape+=+(2,3,4,5)|type|code-block|depth|inlineStyleRanges|entityRanges|data|syntax|javascript|1184387|这应该可以解决这个问题，并且无论您的输入大小是多少，它都可以正常工作。"map“只适用于一维输入。使用".tolist()“并创建一个新的ndarray可以更全面、更好地解决这个问题(我相信)。希望这能有所帮助。|unstyled|1184388|entityMap^0|0|0^^$0|@$1|2|3|4|5|6|7|I|8|@]|9|@]|A|$B|C]]|$1|D|3|E|5|F|7|J|8|@]|9|@]|A|$]]|$1|G|3|-4|5|F|7|K|8|@]|9|@]|A|$]]]|H|$]]

<pre><code>import numpy as np
def f(x):
 return x * np.array([1,1,1,1,1], dtype=np.float32)
g = np.vectorize(f, otypes=[np.ndarray])
a = np.arange(4)
b = g(a)
b = np.array(b.tolist())
print(b)#b.shape = (4,5)
c = np.ones((2,3,4))
d = g(c)
d = np.array(d.tolist())
print(d)#d.shape = (2,3,4,5)
</code></pre>

This should fix the problem and it will work regardless of what size your input is. "map" only works for one dimentional inputs. Using ".tolist()" and creating a new ndarray solves the problem more completely and nicely(I believe). Hope this helps.

blocks|key|970016|text|1.12.0中的一个新参数signature完全可以做你想做的事情。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|970017|def+f(x):
++++return+x+*+np.array([1,1,1,1,1],+dtype=np.float32)

g+=+np.vectorize(f,+signature='()->(n)')|code-block|syntax|javascript|970018|然后g(np.arange(4)).shape会给(4L,+5L)。|970019|这里指定了f的签名。(n)是返回值的形状，()是标量参数的形状。参数也可以是数组。有关更复杂的签名，请参阅Generalized+Universal+Function+API。|970020|entityMap|0|LINK|mutability|MUTABLE|url|https://docs.scipy.org/doc/numpy-1.13.0/reference/c-api.generalized-ufuncs.html^0|D|9|0|0|2|L|P|8|0|5|1|A|3|L|2|1H|Y|0|0^^$0|@$1|2|3|4|5|6|7|W|8|@$9|X|A|Y|B|C]]|D|@]|E|$]]|$1|F|3|G|5|H|7|Z|8|@]|D|@]|E|$I|J]]|$1|K|3|L|5|6|7|10|8|@$9|11|A|12|B|C]|$9|13|A|14|B|C]]|D|@]|E|$]]|$1|M|3|N|5|6|7|15|8|@$9|16|A|17|B|C]|$9|18|A|19|B|C]|$9|1A|A|1B|B|C]]|D|@$9|1C|A|1D|1|1E]]|E|$]]|$1|O|3|-4|5|6|7|1F|8|@]|D|@]|E|$]]]|P|$Q|$5|R|S|T|E|$U|V]]]]

A new parameter <code>signature</code> in 1.12.0 does exactly what you what.

<pre><code>def f(x):
 return x * np.array([1,1,1,1,1], dtype=np.float32)

g = np.vectorize(f, signature='()-&gt;(n)')
</code></pre>

Then <code>g(np.arange(4)).shape</code> will give <code>(4L, 5L)</code>.

Here the signature of <code>f</code> is specified. The <code>(n)</code> is the shape of the return value, and the <code>()</code> is the shape of the parameter which is scalar. And the parameters can be arrays too. For more complex signatures, see <a href="https://docs.scipy.org/doc/numpy-1.13.0/reference/c-api.generalized-ufuncs.html" rel="noreferrer">Generalized Universal Function API</a>.

blocks|key|1184470|text|您想要将函数矢量化|type|unstyled|depth|inlineStyleRanges|entityRanges|data|1184471|import+numpy+as+np
def+f(x):
++++return+x+*+np.array([1,1,1,1,1],+dtype=np.float32)|code-block|syntax|javascript|1184472|假设您希望获得单个np.float32数组作为结果，则必须将其指定为otype。在您的问题中，您指定了otypes=[np.ndarray]，这意味着您希望每个元素都是一个np.ndarray。因此，您可以正确地获得dtype=object的结果。|offset|length|style|CODE|1184473|正确的调用应该是|1184474|np.vectorize(f,+signature='()->(n)',+otypes=[np.float32])|1184475|但是，对于这样一个简单的函数，最好利用numpy的ufunctions；np.vectorize只是循环遍历它。|1184476|def+f(x):
++++return+np.multiply.outer(x,+np.array([1,1,1,1,1],+dtype=np.float32))|1184477|这样做速度更快，产生的模糊错误也更少(但请注意，如果您传递一个复数或四精度数，则结果dtype将取决于x，结果也是如此)。|1184478|entityMap^0|0|0|9|A|Y|5|1F|J|2E|A|30|C|0|0|0|J|5|10|C|0|0|16|5|1F|1|0^^$0|@$1|2|3|4|5|6|7|Y|8|@]|9|@]|A|$]]|$1|B|3|C|5|D|7|Z|8|@]|9|@]|A|$E|F]]|$1|G|3|H|5|6|7|10|8|@$I|11|J|12|K|L]|$I|13|J|14|K|L]|$I|15|J|16|K|L]|$I|17|J|18|K|L]|$I|19|J|1A|K|L]]|9|@]|A|$]]|$1|M|3|N|5|6|7|1B|8|@]|9|@]|A|$]]|$1|O|3|P|5|D|7|1C|8|@]|9|@]|A|$E|F]]|$1|Q|3|R|5|6|7|1D|8|@$I|1E|J|1F|K|L]|$I|1G|J|1H|K|L]]|9|@]|A|$]]|$1|S|3|T|5|D|7|1I|8|@]|9|@]|A|$E|F]]|$1|U|3|V|5|6|7|1J|8|@$I|1K|J|1L|K|L]|$I|1M|J|1N|K|L]]|9|@]|A|$]]|$1|W|3|-4|5|6|7|1O|8|@]|9|@]|A|$]]]|X|$]]

You want to vectorize the function

<pre><code>import numpy as np
def f(x):
 return x * np.array([1,1,1,1,1], dtype=np.float32)
</code></pre>

Assuming that you want to get single <code>np.float32</code> arrays as result, you have to specify this as <code>otype</code>. In your question you specified however <code>otypes=[np.ndarray]</code> which means you want every element to be an <code>np.ndarray</code>. Thus, you correctly get a result of <code>dtype=object</code>.

The correct call would be

<pre><code>np.vectorize(f, signature='()-&gt;(n)', otypes=[np.float32])
</code></pre>

For such a simple function it is however better to leverage <code>numpy</code>'s ufunctions; <code>np.vectorize</code> just loops over it. So in your case just rewrite your function as

<pre><code>def f(x):
 return np.multiply.outer(x, np.array([1,1,1,1,1], dtype=np.float32))
</code></pre>

This is faster and produces less obscure errors (note however, that the results <code>dtype</code> will depend on <code>x</code> if you pass a complex or quad precision number, so will be the result).

<code>numpy.vectorize</code> takes a function f:a->b and turns it into g:a[]->b[].

This works fine when <code>a</code> and <code>b</code> are scalars, but I can't think of a reason why it wouldn't work with b as an <code>ndarray</code> or list, i.e. f:a->b[] and g:a[]->b[][]

For example:

<pre><code>import numpy as np
def f(x):
 return x * np.array([1,1,1,1,1], dtype=np.float32)
g = np.vectorize(f, otypes=[np.ndarray])
a = np.arange(4)
print(g(a))
</code></pre>

This yields:

<pre><code>array([[ 0. 0. 0. 0. 0.],
 [ 1. 1. 1. 1. 1.],
 [ 2. 2. 2. 2. 2.],
 [ 3. 3. 3. 3. 3.]], dtype=object)
</code></pre>

Ok, so that gives the right values, but the wrong dtype. And even worse:

<pre><code>g(a).shape
</code></pre>

yields:

<pre><code>(4,)
</code></pre>

So this array is pretty much useless. I know I can convert it doing:

<pre><code>np.array(map(list, a), dtype=np.float32)
</code></pre>

to give me what I want:

<pre><code>array([[ 0., 0., 0., 0., 0.],
 [ 1., 1., 1., 1., 1.],
 [ 2., 2., 2., 2., 2.],
 [ 3., 3., 3., 3., 3.]], dtype=float32)
</code></pre>

but that is neither efficient nor pythonic. Can any of you guys find a cleaner way to do this? 

Thanks in advance!

Using Numpy Vectorize on Functions that Return Vectors

numpy.vectorize接受函数f:a->b并将其转换为g:a[]->b[]。当a和b是标量时，这很好用，但我想不出为什么它不能将b用作ndarray或list，即f:a->b[]和g:a[]->b例如：import numpy as npdef f(x):    return x * np.array([1,1...

问在返回向量的函数上使用Numpy Vectorize
EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问在返回向量的函数上使用Numpy VectorizeEN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问在返回向量的函数上使用Numpy Vectorize
EN