x***u
发帖数: 336 | 1 the matrix with your desired property is called doubly stochastic matrix.
First for an arbitrary matrix you can't always be able to normalize it, say,
the first column is all 1's, and all the other entries are 0.
if you just want to find a doubly stochastic matrix close to the original one,
There is a simple algorithm to do that:
randomly pick up a row and normalize it
ramdomly pick up a column and normalize it
repeat until the matrix is close to what you want.
This algorithm is proved to be con |
|
