擅长:python、mysql、java
<p>如果要将数据分类为更粗略的分辨率,请尝试分别沿每个维度使用<code>np.reshape</code>和<code>np.mean</code>进行分类:</p>
<pre><code>A = np.array([.....])
s = A.shape # (256, 256, 256)
B = A.copy() # be sure to save original data
# reshape last dimension into new dim and bin-size (int div for index & shape)
B = B.reshape(s[0], s[1], s[2] // 4, 4)
B = np.mean(B, axis=-1) # computes average along last dim
B.shape # (256, 256, 64)
... # repeat for other 2 dimensions.
</code></pre>
<p>这应该是合理的快速,因为您正在使用<code>numpys</code>内部向量化</p>
<p>加:</p>
<p>我不是稀疏矩阵方面的专家,但是如果它真的是SPAR,那么研究一下<a href="https://docs.scipy.org/doc/scipy/reference/sparse.html" rel="nofollow noreferrer">scipy.sparse</a>可能会有用</p>