# Doubly stochastic matrix

In mathematics, especially in probability and combinatorics, a doubly stochastic matrix (also called bistochastic), is a square matrix $A=(a_{ij})$ of nonnegative real numbers, each of whose rows and columns sum to 1, i.e.,

$\sum_i a_{ij}=\sum_j a_{ij}=1$,

Thus, a doubly stochastic matrix is both left stochastic and right stochastic.[1]

Such a transition matrix is necessarily a square matrix: if every row sums to one then the sum of all entries in the matrix must be equal to the number of rows, and since the same holds for columns, the number of rows and columns must be equal.

## Birkhoff polytope and Birkhoff–von Neumann theorem

The class of $n\times n$ doubly stochastic matrices is a convex polytope known as the Birkhoff polytope $B_n$. Using the matrix entries as Cartesian coordinates, it lies in an $(n-1)^2$-dimensional affine subspace of $n^2$-dimensional Euclidean space. defined by $2n-1$ independent linear constraints specifying that the row and column sums all equal one. (There are $2n-1$ constraints rather than $2n$ because one of these constraints is dependent, as the sum of the row sums must equal the sum of the column sums.) Moreover, the entries are all constrained to be non-negative and less than or equal to one.

The Birkhoff–von Neumann theorem states that this polytope $B_n$ is the convex hull of the set of $n\times n$ permutation matrices, and furthermore that the vertices of $B_n$ are precisely the permutation matrices.

## Other properties

The inverse of a nonsingular doubly stochastic matrix need not be doubly stochastic.

Sinkhorn's theorem states that any matrix with strictly positive entries can be made doubly stochastic by pre- and post-multiplication by diagonal matrices.

For $n=2$, all bistochastic matrices are unistochastic and orthostochastic, but for larger $n$ it is not the case.

Van der Waerden conjectured that the minimum permanent among all n × n doubly stochastic matrices is $n!/n^n$, achieved by the matrix for which all entries are equal to $1/n$.[2] Proofs of this conjecture were published in 1980 by B. Gyires[3] and in 1981 by G. P. Egorychev[4] and D. I. Falikman;[5] for this work, Egorychev and Falikman won the Fulkerson Prize in 1982.[6]