用numpy最快的方式对内存最少的上三角元素求和 [英] Fastest way in numpy to sum over upper triangular elements with the least memory

问题描述

我需要对对称矩阵执行类型i<j的求和.这等于对矩阵的上三角元素求和，不包括对角线.

I need to perform a summation of the kind i<j on symmetric matrices. This is equivalent to sum over the upper triangular elements of a matrix, diagonal excluded.

给出A一个对称的N x N数组，最简单的解决方案是np.triu(A,1).sum()，但是我想知道是否存在更快的方法，它们需要更少的内存. 似乎在大型数组上(A.sum() - np.diag(A).sum())/2更快，但是如何避免从np.diag创建甚至N x 1数组呢? 双重嵌套的for循环不需要额外的内存，但是显然这不是在Python中使用的方式.

Given A a symmetric N x N array, the simplest solution is np.triu(A,1).sum() however I was wondering if faster methods exist that require less memory. It seems that (A.sum() - np.diag(A).sum())/2 is faster on large array, but how to avoid creating even the N x 1 array from np.diag? A doubly nested for loop would require no additional memory, but it is clearly not the way to go in Python.

推荐答案

您可以将np.diag(A).sum()替换为np.trace(A)；这不会创建临时的Nx1数组

You can replace np.diag(A).sum() with np.trace(A); this will not create the temporary Nx1 array

这篇关于用numpy最快的方式对内存最少的上三角元素求和的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！