Schur–Horn theorem

(Redirected from Schur-Horn theorem)

In mathematics, particularly linear algebra, the Schur–Horn theorem, named after Issai Schur and Alfred Horn, characterizes the diagonal of a Hermitian matrix with given eigenvalues. It has inspired investigations and substantial generalizations in the setting of symplectic geometry. A few important generalizations are Kostant's convexity theorem, Atiyah–Guillemin–Sternberg convexity theorem and Kirwan convexity theorem.

Statement

edit

Schur–Horn theorem — Theorem. Let   and   be two sequences of real numbers arranged in a non-increasing order. There is a Hermitian matrix with diagonal values   (in this order, starting with   at the top-left) and eigenvalues   if and only if   and  

The inequalities above may alternatively be written:  

The Schur–Horn theorem may thus be restated more succinctly and in plain English:

Schur–Horn theorem: Given any non-increasing real sequences of desired diagonal elements   and desired eigenvalues   there exists a Hermitian matrix with these eigenvalues and diagonal elements if and only if these two sequences have the same sum and for every possible integer   the sum of the first   desired diagonal elements never exceeds the sum of the first   desired eigenvalues.

Reformation allowing unordered diagonals and eigenvalues

edit

Although this theorem requires that   and   be non-increasing, it is possible to reformulate this theorem without these assumptions.

We start with the assumption   The left hand side of the theorem's characterization (that is, "there exists a Hermitian matrix with these eigenvalues and diagonal elements") depends on the order of the desired diagonal elements   (because changing their order would change the Hermitian matrix whose existence is in question) but it does not depend on the order of the desired eigenvalues  

On the right hand right hand side of the characterization, only the values of   depend on the assumption   Notice that this assumption means that the expression   is just notation for the sum of the   largest desired eigenvalues. Replacing the expression   with this written equivalent makes the assumption   completely unnecessary:

Schur–Horn theorem: Given any   desired real eigenvalues and a non-increasing real sequence of desired diagonal elements   there exists a Hermitian matrix with these eigenvalues and diagonal elements if and only if these two sequences have the same sum and for every possible integer   the sum of the first   desired diagonal elements never exceeds the sum of the   largest desired eigenvalues.

Permutation polytope generated by a vector

edit

The permutation polytope generated by   denoted by   is defined as the convex hull of the set   Here   denotes the symmetric group on   In other words, the permutation polytope generated by   is the convex hull of the set of all points in   that can be obtained by rearranging the coordinates of   The permutation polytope of   for instance, is the convex hull of the set   which in this case is the solid (filled) triangle whose vertices are the three points in this set. Notice, in particular, that rearranging the coordinates of   does not change the resulting permutation polytope; in other words, if a point   can be obtained from   by rearranging its coordinates, then  

The following lemma characterizes the permutation polytope of a vector in  

Lemma[1][2] — If   and   have the same sum   then the following statements are equivalent:

  1.  
  2.   and   and   and  
  3. There exist a sequence of points   in   starting with   and ending with   such that   for each   in   some transposition   in   and some   in   depending on  

Reformulation of Schur–Horn theorem

edit

In view of the equivalence of (i) and (ii) in the lemma mentioned above, one may reformulate the theorem in the following manner.

Theorem. Let   and   be real numbers. There is a Hermitian matrix with diagonal entries   and eigenvalues   if and only if the vector   is in the permutation polytope generated by  

Note that in this formulation, one does not need to impose any ordering on the entries of the vectors   and  

Proof of the Schur–Horn theorem

edit

Let   be a   Hermitian matrix with eigenvalues   counted with multiplicity. Denote the diagonal of   by   thought of as a vector in   and the vector   by   Let   be the diagonal matrix having   on its diagonal.

( )   may be written in the form   where   is a unitary matrix. Then  

Let   be the matrix defined by   Since   is a unitary matrix,   is a doubly stochastic matrix and we have   By the Birkhoff–von Neumann theorem,   can be written as a convex combination of permutation matrices. Thus   is in the permutation polytope generated by   This proves Schur's theorem.

( ) If   occurs as the diagonal of a Hermitian matrix with eigenvalues   then   also occurs as the diagonal of some Hermitian matrix with the same set of eigenvalues, for any transposition   in   One may prove that in the following manner.

Let   be a complex number of modulus   such that   and   be a unitary matrix with   in the   and   entries, respectively,   at the   and   entries, respectively,   at all diagonal entries other than   and   and   at all other entries. Then   has   at the   entry,   at the   entry, and   at the   entry where   Let   be the transposition of   that interchanges   and  

Then the diagonal of   is  

  is a Hermitian matrix with eigenvalues   Using the equivalence of (i) and (iii) in the lemma mentioned above, we see that any vector in the permutation polytope generated by   occurs as the diagonal of a Hermitian matrix with the prescribed eigenvalues. This proves Horn's theorem.

Symplectic geometry perspective

edit

The Schur–Horn theorem may be viewed as a corollary of the Atiyah–Guillemin–Sternberg convexity theorem in the following manner. Let   denote the group of   unitary matrices. Its Lie algebra, denoted by   is the set of skew-Hermitian matrices. One may identify the dual space   with the set of Hermitian matrices   via the linear isomorphism   defined by   for   The unitary group   acts on   by conjugation and acts on   by the coadjoint action. Under these actions,   is an  -equivariant map i.e. for every   the following diagram commutes,

 

Let   and   denote the diagonal matrix with entries given by   Let   denote the orbit of   under the  -action i.e. conjugation. Under the  -equivariant isomorphism   the symplectic structure on the corresponding coadjoint orbit may be brought onto   Thus   is a Hamiltonian  -manifold.

Let   denote the Cartan subgroup of   which consists of diagonal complex matrices with diagonal entries of modulus   The Lie algebra   of   consists of diagonal skew-Hermitian matrices and the dual space   consists of diagonal Hermitian matrices, under the isomorphism   In other words,   consists of diagonal matrices with purely imaginary entries and   consists of diagonal matrices with real entries. The inclusion map   induces a map   which projects a matrix   to the diagonal matrix with the same diagonal entries as   The set   is a Hamiltonian  -manifold, and the restriction of   to this set is a moment map for this action.

By the Atiyah–Guillemin–Sternberg theorem,   is a convex polytope. A matrix   is fixed under conjugation by every element of   if and only if   is diagonal. The only diagonal matrices in   are the ones with diagonal entries   in some order. Thus, these matrices generate the convex polytope   This is exactly the statement of the Schur–Horn theorem.

Notes

edit
  1. ^ Kadison, R. V., Lemma 5, The Pythagorean Theorem: I. The finite case, Proc. Natl. Acad. Sci. USA, vol. 99 no. 7 (2002):4178–4184 (electronic)
  2. ^ Kadison, R. V.; Pedersen, G. K., Lemma 13, Means and Convex Combinations of Unitary Operators, Math. Scand. 57 (1985),249–266

References

edit
  • Schur, Issai, Über eine Klasse von Mittelbildungen mit Anwendungen auf die Determinantentheorie, Sitzungsber. Berl. Math. Ges. 22 (1923), 9–20.
  • Horn, Alfred, Doubly stochastic matrices and the diagonal of a rotation matrix, American Journal of Mathematics 76 (1954), 620–630.
  • Kadison, R. V.; Pedersen, G. K., Means and Convex Combinations of Unitary Operators, Math. Scand. 57 (1985),249–266.
  • Kadison, R. V., The Pythagorean Theorem: I. The finite case, Proc. Natl. Acad. Sci. USA, vol. 99 no. 7 (2002):4178–4184 (electronic)
edit