scipy.spatial.distance.squareform的逆函數

Question

是否有scipy.spatial.distance.squareform的逆函數？如果不是，編寫它來處理巨大距離矩陣的最佳方法是什麼？

Answer 1

根據該文檔，squareform是其本身的逆：

http://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.distance.squareform.html

Answer 2

其實，這拉姆達是相當有效率：

In [1]: unsquareform = lambda a: a[numpy.nonzero(numpy.triu(a))] 

例如：

In [2]: scipy.spatial.distance.pdist(numpy.arange(12).reshape((4,3))) 
Out[2]: 
array([ 5.19615242, 10.39230485, 15.58845727, 5.19615242, 
     10.39230485, 5.19615242]) 

和

In [3]: unsquareform(scipy.spatial.distance.squareform(scipy.spatial.distance.pdist(numpy.arange(12).reshape((4,3))))) 
Out[3]: 
array([ 5.19615242, 10.39230485, 15.58845727, 5.19615242, 
     10.39230485, 5.19615242])