Pseudodifferential symbols on Riemann surfaces and Krichever-Novikov algebras (Q2466815): Difference between revisions

Recall the Lie algebra \(\psi DO\) of pseudo-differential operators on the circle \(S^1\): it consists of formal series \[ A(\theta,\partial)=\sum_{k=-\infty}^{n}a_k(\theta)\partial^k \] with \(a_k\in{\mathcal C}^{\infty}(S^1)\). The multiplication of pseudo-differential operators is given by the Leibniz rule \[ \partial^k\circ f\partial^n= f\partial^{n+k}+ \sum_{i=1}^\infty \binom ki f^{(i)} \partial^{n+k-i}, \] where the binomial coefficient \(\binom ki\) is interpreted as usual by \[ \binom ki= \frac{k(k-1)\dots (k-i+1)}{i!}. \] The Lie algebra structure on \(\psi DO\) is given by the commutator \([A,B]=A\circ B - B\circ A\). The Lie algebra \(\psi DO\) has the Lie subalgebra \(DO\) of differential operators on \(S^1\) \[ DO=\left\{\sum_{k=0}^na_k(\theta)\partial^k\mid a_k\in{\mathcal C}^{\infty}(S^1)\right\}. \] These Lie algebras are fundamental in the algebraic theory of integrable systems - for the description of their place in the theory of Lax pairs, let us refer to the book of \textit{L. A. Dickey} [Soliton equations and Hamiltonian systems. 2nd ed. Singapore: World Scientific (2003; Zbl 1140.35012)], and for a thorough description of the Lie algebras, to the forthcoming book of \textit{B. Khesin} and \textit{R. Wendt} [The geometry of infinite-dimensional groups. Berlin: Springer (2009; Zbl 1153.22001)]. The Lie algebra \(\psi DO\) possesses a central extension given by the \textit{logarithmic cocycle} \[ (A,B)\mapsto \text{tr}([\log\partial,A]\circ B), \] where the expression \([\log\partial,-]:\psi DO\to \psi DO\) is an outer derivation of \(\psi DO\), given by the formula \[ [\log\partial,f\partial^n]= \sum_{i=1}^{\infty}\frac{(-1)^{i+1}}{i} f^{(i)}\partial^{n-i}, \] and the trace is given by the integral over \(S^1\) of the residue: \[ \text{tr}\left(\sum_{k=-\infty}^na_k(\theta)\partial^k\right)= \int_{S^1}a_{-1}(\theta)\,d\theta. \] The logarithmic cocycle restricts to (a multiple of) the Kac-Peterson cocycle on \(DO\), and to (a multiple of) the Gelfand-Fuks cocycle on the Lie subalgebra \(\text{Vect}(S^1)\subset DO\) of first order differential operators. The article under review generalizes these and other related structures to a two-dimensional setting. Indeed, let \(\Sigma\) be a (connected) compact Riemann surface and \({\mathcal M}\) the space of meromorphic functions on \(\Sigma\). In order to speak of (symbols of) pseudo-differential operators, one fixes a meromorphic (reference) vector field \(v\not=0\), and denotes by \(D\) the Lie derivation (of meromorphic functions) with respect to \(v\). Then one may introduce the algebra of meromorphic differential operators \[ MDO= \left\{A=\sum_{k=0}^na_kD^k\mid a_k\in{\mathcal M}\right\}, \] and the algebra of meromorphic pseudo-differential symbols \[ M\psi DS= \left\{A=\sum_{k=-\infty}^n a_kD^k\mid a_k\in{\mathcal M}\right\} . \] As before, both algebras become Lie algebras with the commutator bracket. The choice of the meromorphic vector field \(v\not=0\) does not matter up to isomorphy. The authors investigate outer derivations and traces (which are here given by the residue associated to a fixed point on \(\Sigma\)) on the Lie algebra \(M\psi DS\). This leads then to generalizations of the logarithmic cocycle and accordingly to central extensions of \(M\psi DS\). Further, the authors pass to the double extension \(\widetilde{M\psi DS}\) (adding a derivation in order to have a non-degenerate bilinear form), investigate a Manin triple and show that this triple gives rise to a Lie bialgebra structure on some subalgebra of \(\widetilde{M\psi DS}\). In a second part of their paper, the authors fix two distinct points \(P_{\pm}\in\Sigma\) and focus on meromorphic objects which are allowed to have poles only in \(P_{\pm}\) and which are holomorphic elsewhere. This setting is closer to Krichever-Novikov algebras, which are current or vector field Lie algebras on \(\Sigma\setminus \{P_{\pm}\}\), see [\textit{I. M. Krichever} and \textit{S. P. Novikov}, Funct. Anal. Appl. 21, No. 2, 126--142 (1987); translation from Funkts. Anal. Prilozh. 21, No. 2, 46--63 (1987; Zbl 0634.17010)] and [\textit{I. M. Krichever} and \textit{S. P. Novikov}, Funct. Anal. Appl. 21, No. 4, 294--307 (1987); translation from Funkts. Anal. Prilozh. 21, No. 4, 47--61 (1987; Zbl 0659.17012)]. The objects are generalizations of the Virasoro algebra, the central extension of the Lie algebra of meromorphic vector fields on the Riemann sphere with allowed poles in 0 and \(\infty\). One upshot of this theory is that the new algebras are not \({\mathbb Z}\)-graded anymore, but only almost graded (here called generalized graded), i.e. the bracket \([{\mathfrak g}_i,{\mathfrak g}_j]\) of two homogeneous components \({\mathfrak g}_i\) and \({\mathfrak g}_j\) of degrees \(i\) and \(j\) respectively does not fall necessarily into \({\mathfrak g}_{i+j}\), but into a sum \(\bigoplus_{l=i+j-r}^{i+j+s}{\mathfrak g}_l\). The cocycles which permit to lift the almost grading to the central extension are called local. The authors of the present article define an almost grading on their Lie algebra of holomorphic pseudo-differential symbols by showing that the graded components associated to the order filtration are density modules \({\mathcal F}_{\lambda}\) for the Lie algebra of holomorphic vector fields (with possible poles in \(P_{\pm}\)). Then they investigate local cocycles for their Lie algebras. In this part, most proofs reduce the asserted statements to the corresponding theorems of Krichever and Novikov.