Central extensions of current groups (Q1407663)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Central extensions of current groups
scientific article

    Statements

    Central extensions of current groups (English)
    0 references
    0 references
    0 references
    16 September 2003
    0 references
    Let \(G\) be the identity component of the current group \(C^\infty(M,K)\), where \(M\) is a compact manifold and \(K\) a finite- or infinite-dimensional Lie group. The authors consider continuous Lie algebra \(2\)-cocycles \(\omega\) on the Lie algebra \({\mathfrak g}=C^\infty(M,{\mathfrak k})\) of \(G\) which are of ``product type'', meaning that \(\omega(\xi,\eta):=[\kappa(\xi,d\eta)]\in {\mathfrak z} :=\Omega^1(M,Y)/ dC^\infty(M,Y)\) for some continuous, symmetric, invariant bilinear map \(\kappa : {\mathfrak k}\times {\mathfrak k}\to Y\) on the Lie algebra \({\mathfrak k}\) of \(K\), with values in a sequentially complete, locally convex space \(Y\). The article is centered around the question of whether there exists a central Lie group extension \(\widehat{G}\) of \(G\) corresponding to \(\omega\). It is known from the general extension theory developed by the second author [Ann. Inst. Fourier 52, 1365-1442 (2002; Zbl 1019.22012)] that \(\widehat{G}\) exists if and only if the period group \(\Pi_{M,\kappa}\subseteq {\mathfrak z}\) (the image of the period homomorphism \(\text{per}_\omega : \pi_2(G)\to {\mathfrak z}\)) is discrete and the adjoint action of \({\mathfrak g}\) on \({\mathfrak g}\oplus_\omega {\mathfrak z}\) integrates to a smooth representation of \(G\). The second condition is always satisfied in the present context; the Lie algebra action even integrates to a smooth representation of \(C^\infty(M,K)\) (Proposition III.3). As the authors show, for fixed \(\kappa\), the first condition is satisfied for every \(M\) if and only if it is satisfied for \(M={\mathbb S}^1\), i.e., in the case of loop groups (Theorem I.6). Hence \(\widehat{G}\) exists for general \(M\) if and only if it does so for \(M={\mathbb S}^1\). The proof is based on an explicit calculation of \(\Pi_{M,\kappa}\) in terms of \(\Pi_{{\mathbb S}^1,\kappa}\) which is quite remarkable, because \(\pi_2(C^\infty(M,K))\) is not well accessible when \(\dim(M)>2\). The authors then turn to finite-dimensional \(K\); in this case, a central extension \({\mathfrak z}/\Pi_{M,\kappa}\to \widehat{G}\to G\) is shown to exist for \(\kappa : {\mathfrak k}\times {\mathfrak k}\to V({\mathfrak k})\), the universal symmetric invariant bilinear map (Corollary III.7). If \(K\) is semisimple, this implies the existence of a central extension of \(G\) which is universal among all central extensions of \(G\) by abelian regular Fréchet-Lie groups (Theorem IV.5). So far, only a universal central extension of the universal covering group \(\widetilde{G}\) (rather than \(G\)) had been discussed in the literature [\textit{A. Pressley} and \textit{G. Segal}, Loop groups, Oxford University Press (1988; Zbl 0638.22009)], for \(K\) a simply connected, simple, compact Lie group. In a section which can be read independently of the others, the authors also study the problem to associate to a pair of automorphisms of Lie groups \(Z\) and \(G\) an automorphism of a given central extension \(\widehat{G}\) of \(G\) by \(Z\), and they describe sufficient conditions ensuring that actions of a Lie group \(R\) on \(Z\) and \(G\) by automorphisms give rise to a smooth automorphic action of \(R\) on \(\widehat{G}\).
    0 references
    current group
    0 references
    loop group
    0 references
    mapping group
    0 references
    central extension
    0 references
    universal central extension
    0 references
    period group
    0 references
    period homomorphism
    0 references
    obstruction
    0 references
    cocycle
    0 references
    product type
    0 references
    infinite-dimensional Lie group
    0 references
    automorphic action
    0 references

    Identifiers