\(\mathrm{Horn}(p, q)\) (Q6132653)

From MaRDI portal
scientific article; zbMATH DE number 7712864
Language Label Description Also known as
English
\(\mathrm{Horn}(p, q)\)
scientific article; zbMATH DE number 7712864

    Statements

    \(\mathrm{Horn}(p, q)\) (English)
    0 references
    0 references
    14 July 2023
    0 references
    The original Horn problem is stated as follows: Assume that the eigenvalues of the Hermitian matrices \(A\) and \(B\) are known. The problem is to determine the set of possible eigenvalues of the sum \(C=A+B\). A representation theory version of this problem is as follows: Given two irreducible representations \(\pi _{\lambda }\) and \(\pi _{\mu }\) of a compact Lie group \(G\) with highest weights \(\lambda \) and \(\mu \). Determine the set of the highest weights occuring in the decomposition of the tensor product \(\pi _{\lambda }\otimes \pi _{\mu }\). According to the correspondance between irreducible representations and coadjoint orbits, Horn problem is transposed as follows: For a Lie group \(G\), determine how the sum of two coadjoint orbits decomposes in a union of coadjoint orbits. Let us consider first the case of the unitary group \(\mathrm{U}(n)\), let \[\mathcal{C}_n=\{(x_1\geq \cdots \geq x_n)\}\subset \mathbb{R}^n,\] and, for \(x\in \mathcal{C}_n\), let \(d_x\) be the diagonal matrix with entries \(ix_1,\ldots ,ix_n\). One defines the Horn cone \[\mathrm{Horn}(n)=\{(A,B,C)\in (\mathcal{C}_n)^3 \mid \mathcal{O}_C\subset \mathcal{O}_A+\mathcal{O}_B\},\] where \(\mathcal{O}_A\), \(\mathcal{O}_B\), \(\mathcal{O}_C\) are the coadjoint orbits associated to \(d_A\), \(d_B\), \(d_C\). It has been proven by \textit{A. A. Klyachko} [Sel. Math., New Ser. 4, No. 3, 419--445 (1998; Zbl 0915.14010)], \textit{A. Knutson} and \textit{T. Tao} [J. Am. Math. Soc. 12, No. 4, 1055--1090 (1999; Zbl 0944.05097)] that the set \(\mathrm{Horn}(n)\) is determined by a set of inequalities which are recursively defined. In the present paper, the author considers the Lie group \(\mathrm{U}(p,q)\). Let \[\mathcal{C}_{p,q}=\{x\in \mathbb{R}^p\times \mathbb{R}^q \mid x_1\geq \cdots \geq x_p>x_{p+1}\geq \cdots \geq x_{p+q} \}.\] One defines the Horn cone \[\mathrm{Horn}(p,q)=\{(A,B,C)\in (\mathcal{C}_{p,q})^3 \mid \mathcal{O}_C\subset \mathcal{O}_A+\mathcal{O}_B\}\] where \(\mathcal{O}_A\), \(\mathcal{B}_B\), \(\mathcal{O}_C\) are orbits of \(\mathrm{U}(p,q)\) associated to \(d_A\), \(d_B\), \(d_C\), contained in an invariant cone \(C(p,q)\) in the Lie algebra of \(\mathrm{U}(p,q)\). The main result of the paper is a recursive description of the Horn cone \(\mathrm{Horn}(p,q)\). In the proof one studies the Hamiltonian action of \(\bigl(\mathrm{U}(p)\times \mathrm{U}(q)\bigr)^3\) on the manifold \(\bigl(\mathrm{GL}(p,\mathbb{C})\times \mathrm{GL}(q,\mathbb{C})\bigr)^2\times \mathbb{C}^p\otimes \mathbb{C}^q\).
    0 references
    Horn's conjecture
    0 references
    pseudo-unitary groups
    0 references
    Horn cone
    0 references
    coadjoint orbits
    0 references
    0 references
    0 references
    0 references

    Identifiers