Stokes phenomenon, Gelfand-Zeitlin systems and relative Ginzburg-Weinstein linearization (Q1789488)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Stokes phenomenon, Gelfand-Zeitlin systems and relative Ginzburg-Weinstein linearization
scientific article

    Statements

    Stokes phenomenon, Gelfand-Zeitlin systems and relative Ginzburg-Weinstein linearization (English)
    0 references
    0 references
    10 October 2018
    0 references
    The dual Poisson Lie group \(K^\ast\) of a compact Lie group \(K\) with the standard Poisson structure is Poisson isomorphic to the dual of the Lie algebra \(\mathfrak{k}^\ast\) with its canonical linear structure (Ginzburg-Weinstein linearization; see [\textit{V. L. Ginzburg} and \textit{A. Weinstein}, J. Am. Math. Soc. 5, No. 2, 445--453 (1992; Zbl 0766.58018)]). If \(K=U(n)\), \(u(n)^\ast\) carries the Gelfand-Zeitlin integrable system and \(U(n)^ast\) carries the multiplicative Gelfand-Zeitlin system. \textit{A. Alekseev} and \textit{E. Meinrenken} [J. Differ. Geom. 76, No. 1, 1--34 (2007; Zbl 1119.53053)] constructed a Ginzburg-Weinstein linearization which intertwines the Gelfand-Zeitlin systems. In this paper, explicit constructions of such linearizations via Stokes phenomenon of meromorphic connections and Boalch's dual exponential maps are given. Main results are summarized as Theorems 1.1--1.3 (and Proposition 1.4) stated below. The Gelfand-Zeitlin system is explained in \S2. Let \(G\) be a reductive Lie group with Lie algebra \(\mathfrak{g}\), and \(\mathfrak{t}\) the Cartan subalgebra of \(\mathfrak{g}\). Then consider the meromorphic connection \[ \nabla=d-(\frac{\lambda}{z^2}+\frac{1}{2\pi i}\frac{A}{z})dz \] on the trivial holomorphic \(G\)-bundle \(P\) on \(\mathbb{P}^1\). Here, \(A\in\mathfrak{g}\) and \(\lambda\in\mathfrak{t}\). Fixing \(\lambda\), the connection matrix of the monodromy of \(\nabla\) is a function of \(A\) denoted \(C(A)\), called the connection map associated to the irregular type \(\frac{\lambda}{z}\) (\S3). Assume the centralizer of \(\lambda_k\in\mathfrak{t}\) is \(\mathfrak{gl}_{k-1}(\mathbb{C})+\mathfrak{t}\subset \mathfrak{gl}_k(\mathbb{C})\), set \(C=C_1\cdots C_k\), \(C_k:\mathfrak{gl}_k(\mathbb{C})\to \mathrm{GL}_k(\mathbb{C})\), the connection map associated to the irrregular type \(\frac{\lambda_k}{z}\), the following holds (\S4, Theorem 4.1). Theorem. The map \(\Gamma=\mathrm{Ad}_C\circ\exp: \mathrm{Herm}(n)\cong\mathfrak{u}(n)^\ast\to \mathrm{Herm}^+(n)\cong U(n)^\ast\) is a Poisson diffeomorphism compatible with the Gelfand-Zeitlin systems. Here, Herm\((n)\) (Herm\((n)^+\)) denotes the set of (positive definite) Hermitian \(n\times n\) matrices. Technical details are mostly described in \S3. To relate this result to symplectic geometry, as an extended moduli space, the set of equivalence classes of triples \((P,\nabla,\mathbf{g})\) is considered in \S5. Here, \(P\) is a trivial holomorphic \(G\)-bundle, \(\nabla\) is a connection of irregular type \(\frac{\lambda}{z}\) and compatible framing \(\mathbf{g}\). The space of Stokes/monodromy data of \((P,\nabla,\mathbf{g})\) inherits a asymplectic structure [\textit{P. Boalch}, Duke Math. J. 139, No. 2, 369--405 (2007; Zbl 1126.53055)], which is isomorphic to a symplectic ``slice'' \((G\times\mathfrak{t}',\pi_{\mathfrak{g}})\), \(\mathfrak{t}'=\{\Lambda\in\mathfrak{t}\mid \alpha(\Lambda)\notin 2\pi i\mathbb{Z}\}\). \((P,\nabla,\mathbf{g})\) is isomorphic to \(G\times\mathfrak{t}'\), which have a natural symplectic structure \(\pi_{\mathfrak{g}_1}\). Let \(C_{\mathfrak{g}_1}\) be the connection map associated to the irregular type \(\frac{\lambda}{z}\), then the following is proved (\S5, Theorem 5.11) Theorem. The irregular Riemann-Hilbert map \[ \nu_{C_{\mathfrak{g}_1}}: (G\times \mathfrak{t}',\pi_{\mathfrak{g}_1})\to (G\times\mathfrak{t}, \pi_{\mathfrak{g}}); (g,t)=(C_{\mathfrak{g}_1}(Ad^\ast_g t)g,t), \] associating the monodromy data to any triple \((P,\nabla,\mathbf{g})\) is a local symplectic isomorphism. The author remarks this is analogous to the result in [\textit{X. Xu}, ``Irregular Riemann-Hilbert correspondence, Alekseev-Meinrenken dynamical \(r\)-matrices and Drinfeld twists'', Preprint, \url{arXiv:1507.07149}]. But the assumption that \(\lambda\) is regular is dropped. In the rest, relative Ginzburg-Weinstein linearization via dynamical system and the theory of quantization of Lie bialgebra, which generalize various results in [\textit{B. Enriquez} et al., Int. Math. Res. Not. 2005, No. 36, 2183--2198 (2005; Zbl 1092.53057)], are given. In \S6, after explaining relative Ginzburg-Weinstein linearization, the relative gauge transformation equation between \(r\)-matrices \[ r_{\mathrm{AM}_{\mathfrak{g}}}-(\otimes^2\mathrm{Ad}_{\rho^{-1}})(r_{\mathrm{Ad}_{\mathfrak{g}_1}}=(r_{\mathfrak{g}}-r_{\mathfrak{g}_1})^\rho, \] is introduced. Here, \(r_{\mathrm{AM}_{\mathfrak{g}}}:\mathfrak{g}^\ast\to \mathfrak{g}\otimes\mathfrak{g}\) is the Alekseev-Meinrenken dynamical \(r\)-matrix for \(\mathfrak{g}\) (for the precise form of the relative gauge transformation equation, see \S6.4). The connection map \(C_{\mathfrak{g}_1}:\mathfrak{g}^\ast\to G\) satisfies relative gauge transformation equation (Theorem 6.8), which follows from the following. Theorem 1.3. A map \(\rho\) is an \(H\)-equivariant solution of the gauge equation, if and only if \[ \nu_{\rho}: (G\times\mathfrak{t}'.\pi_{\mathfrak{g}_1})\to (G\times\mathfrak{t}', \pi_{\mathfrak{g}}); (g,t)=(\rho(\mathrm{Ad}^\ast_g t)g,t) \] is a relative Ginzburg-Weinstein linearization with respect to \(\mathfrak{g}_1\subset\mathfrak{g}\). In \S7, the last section, a construction of relative linearization via the theory of quantization of Lie bialgebras is given, and the semiclassical limit \(\rho:\mathfrak{g}^\ast\to G\) of the admissible relative Drinfeld twist [\textit{V. Toledano Laredo}, Adv. Math. 210, No. 1, 375--403 (2007; Zbl 1171.17003)] is shown to be an \(H\)-equivariant formal solution of the relative gauge transformation equation (Proposition 1.4, Proposition 7.1). This paper is concluded by asking whether there exists a relative Drinfeld twist whose semiclassical limit is \(C_{\mathfrak{g}_1}\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    meromorphic connection
    0 references
    Stokes phenomenon
    0 references
    connection matrix
    0 references
    Gelfand-Zeitlin systems
    0 references
    Ginzburg-Weinstein linearization
    0 references
    Poisson Lie groups
    0 references
    Alekseev-Meinrenken \(r\)-matrices
    0 references
    relative Drinfeld twist
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references