1. Introduction
Let denote the set of all real column vectors with n coordinates. A real matrix A is called non-negative, denoted by , if all its entries are non-negative. Let denote the set of all doubly stochastic matrices, i.e., real non-negative matrices, with each row sum and column sum equal to 1. For a vector , let denote the vector obtained from by rearranging the coordinates in nonincreasing order. For vectors is said to be majorized by y, denoted by , if for all positive integers k such that
It is well known that for two vectors
iff
for some
In [
1], the polytope of doubly stochastic matrices
D for which
was investigated. The notion of vector majorization is naturally extended to that of matrix majorization as follows. For two matrices
X and
Y in
,
X is said to be majorized by
Y, denoted by
, if
for some
In [
2], a new notion of matrix majorization was introduced, called weak matrix majorization, where
, if there exists a row stochastic matrix
R such that
Additionally, relations between this concept and strong majorization (usual majorization) are considered. In [
3], the polytope of row-stochastic matrices
R for which
was investigated and generalizations of the results for vector majorization were obtained. The notion of matrix majorization was referred to as multivariate majorization in [
4]. In [
4], it was proved that, if
for
, then
for any real matrix
C with
m rows.
For , if then certainly for all However, the converse does not hold in general, as is easily seen with the matrices and
For
, let
denote the set of all
satisfying
The set
is known to contain various special types of doubly stochastic matrices [
5,
6]. While quite a lot of progress has been made in the theory of vector majorization, very little is known about multivariate majorization. In [
7], it is proved that
X is majorized by
Y if
is majorized by
for every real
n vector
v, under the assumption that
is nonnegative, where
e denotes the
m-vector of ones and
denotes the Moore–Penrose generalized inverse of
In [
8], a new matrix majorization order for classes (sets) of matrices was introduced, which generalizes several existing notions of matrix majorization. Matrix majorization order has several applications in mathematical statistics. Some applications of the matrix majorization were discussed by Marshall, Olkinbarry and Arnold [
4] and Tong [
9]. Majorization of (0,1) matrices have important applications in classification theory and principal/dominant component analysis. Eduard Jorswieck and Holger Boche [
10] reviewed the basic definitions of majorization theory and matrix monotone functions, describing their concepts clearly with many illustrative examples and then proceeded to show their applications in wireless communications. In [
11], an algorithm was developed for the problem of finding a low-rank correlation matrix nearest to a given correlation matrix. The algorithm was based on majorization. The problem of rank reduction of correlation matrices occurs when pricing a derivative dependent on a large number of assets, where the asset prices are modeled as correlated log-normal processes. Mainly, such an application concerns interest rates. Matrix majorization also has applications in the comparison of eigenvalues [
12].
Hwang and Park [
13] obtained some necessary and sufficient conditions for
to satisfy
for the case that the rank of
Y is any one of
or
n and for the case that
. In
Section 2, we obtain some necessary and sufficient conditions for
to satisfy
for the case that the rank of
Y is
and for the general case that the rank of
Y is
. We also obtain some necessary and sufficient conditions for
X to be majorized by
Y with some conditions on
X and
Additionally, we obtain some necessary and sufficient conditions for
X to be doubly stochastic majorized by
Y with some conditions on
X and
Dahl, Guterman and Shteyner [
14] obtained several results concerning matrix majorizations of
matrices and characterizations for certain matrix majorization orders. We extend these results for
matrices in
Section 3. We introduce a new concept of column stochastic majorization. A matrix
X is said to be column stochastic majorized by
denoted as
if there exists a column stochastic matrix
S such that
We obtain some characterizations for column stochastic majorization and doubly stochastic majorization of
matrices.
2. Matrix Majorization
Let denote the identity matrix of order n. For two real matrices of the same size, let (resp. ) denote that each entry of A is bigger (resp. less) than or equal to the corresponding entry of B. Let denote all l vectors in . For matrix A, let and denote the sum of all of the entries of A, the row sum vector of A and the column sum vector of A, respectively. A vector is called a stochastic vector if and .
Theorem 1. Let Then if the following hold.
Proof. Let us assume . Then there exists satisfying It is easy to see that . Let with Then Since , we obtain that there exist stochastic vectors and such that Since , we obtain that there exist stochastic vectors and such that . Conversely, suppose (1) and (2) hold. For the vectors in (2), let , where . Then . It remains to show that . We see from that From (1) and (2) we have Additionally, since and This implies that □
The above theorem can be extended to any of rank .
Theorem 2. Let Then if the following hold.
There exist stochastic vectors such that and
Proof. Proof is similar to the proof of Theorem 1 extending to k vectors. □
In the next theorem, we give a necessary condition for , for any two matrices .
Theorem 3. Let If then there exists such that for and
Proof. Suppose so that for some Let be a matrix. Then the j-th element in the row vector is . Let , …
This implies
Since □
Example 1. Let and Then and are such that they satisfy the required conditions of the theorem and so this example validates Theorem 3.
The converse of Theorem 3 is not true.
Example 2. Let and .
Take and Then and satisfy the conditions of the above theorem but there exists no matrix S such that Assume that there exists a matrix S such that Since when we multiply the first row of S with the first column of Y, we obtain This implies that Similarly for we have This implies that Similarly for we have This implies that and However, this is a contradiction to Therefore, there exists no such matrix S such that In the next theorem, we show sufficient conditions for , for any two matrices .
Theorem 4. Let . A sufficient condition for X to be majorized by Y is that and for and there exists such that for and for
Proof. Assume that and such that the conditions given in the theorem hold. Let Then S is a doubly stochastic matrix. When we multiply the first row of S and the first column of Y, we will obtain Hence, This implies that X is majorized by □
Example 3. Let and
Then X and Y satisfy the conditions of Theorem 4 and so X is majorized by Take and Then and So X is majorized by
A matrix X is said to be doubly stochastic majorized by Y, denoted by , when there is such that In the next theorem, we prove a necessary condition for X to be doubly stochastic majorized by
Theorem 5. Let If then there exists such that for and
Proof. Suppose so that for some The j-th element of the vector is Let
This implies, Since □
Example 4. Let and Then and are such that they satisfy the required conditions of the theorem and so this example validates Theorem 5.
The converse of Theorem 5 is not true.
Example 5. and Take and Then and satisfy the conditions of the above theorem, but there exists no matrix such that Assume that there exists a matrix S such that Since when we multiply the first row of Y with the first column of we obtain This implies that Similarly for we have This implies that Similarly, for we have This implies that which is a contradiction since and Therefore, there exists no such matrix S such that
Theorem 6. Let A sufficient condition for X to be doubly stochastic majorized by Y is that and for and there exist such that for and
Proof. The proof is similar to the proof of Theorem 4. □
Example 6. Let and Then X and Y satisfy the conditions of Theorem 6 and so X is doubly stochastic majorized by
Take and Then and So X is doubly stochastic majorized by
3. Majorization for (0,1) Matrices
There are two main motivations for the study of matrix majorization for
matrices. First, it is of interest to see if this restriction to the subclass of
matrices leads to simpler characterizations of the majorization order in question. Secondly,
matrices are essential to represent combinatorial objects, and therefore one may look at the meaning of such a matrix majorization order (each associated with a
matrix). In [
14] weak, directional and strong majorizations of
matrices are characterized. Matrix majorization on
matrices is investigated.
Definition 1. A matrix X is said to be column stochastic majorized by denoted as if there exists a column stochastic matrix S such that
Theorems 7 and 8 give characterizations of and for matrices.
Theorem 7. Let and Then for every the number of in the i-th column of X is equal to the number of in the i-th column of
Proof. By assumption, for some Then where Since both matrices are the i-th entry of is the number of in the i-th column of The same holds is for Hence the result follows. □
Theorem 8. Let Assume but Then
- (i)
There exists satisfying such that S contains a zero row, and for each row sum of S it holds that either or
- (ii)
If Y does not contain a zero row, then for any satisfying it holds that S contains a zero row, and for each row sum of R it holds that or .
- (iii)
X contains a zero row.
Proof. (i) Suppose where S is column stochastic. Since , the sum of all elements in S is Since by assumption there is i such that the ith row sum of and there is k such that the kth row sum of If then is a zero row. Indeed, each element of is less than or equal to Hence, it is zero. We are going to modify S in order to construct such that and is zero. Fix some k such that Suppose Since we obtain Consider arbitrary We consider the matrix which is obtained from S by changing to and to We do the same for the rest of the nonzero elements in Finally, we obtain such that is a zero row, and for We repeat this procedure for every q with After several such substitutions, we obtain such that for every q either or contains a zero row, as required. (ii) Suppose Y does not contain a zero row. Let i be such that From is zero. and all summands are non-negative, and it follows that for all Finally if then and S contains a zero row. (iii) By (i) we obtain that contains a zero row, and, as a consequence, contains a zero row. □
In [
14], it was proved that if
then
, but matrix majorization cannot be described in terms of row/column inclusion. The following examples show that strong majorization also cannot be described in terms of row/column inclusion. The first example shows that column inclusion does not follow from
, and the second one shows that the converse implication does not hold as well.
Example 7. Let
Then so Also,
Example 8. Let and Then and But it is easy to verify that Suppose that there exists such that Since Since it follows that Then and For similar reasons, and then R is not a row-stochastic matrix.