The cohomology of Koszul-Vinberg algebra and related topics (Q421861): Difference between revisions

A \textit{Koszul-Vinberg (KV) algebra} over a field \(k\), more widely known as \textit{pre-Lie algebra}, is a pair \(A=(V,\mu)\), where \(V\) is a \(k\)-vector space and \(\mu\) is a \(k\)-bilinear (multiplication) map \(V\times V\to V\), satisfying the condition \[ a\cdot(b\cdot c)- (a\cdot b)\cdot c - b\cdot (a\cdot c)+ (b\cdot a)\cdot c =0 \] for all \(a,b,c\in V\), where \(a\cdot b\) stands for \(\mu(a,b)\). For an arbitrary \(k\)-algebra \((V,\mu)\) the left side of the above condition defines a tri-linear map \(KV_\mu:V\times V\times V\to V\), which measures the failure of the algebra to be Koszul-Vinberg. The above definition is equivalent to requiring that for any \(a,b\in V\) one has \[ \left[ L_a, L_b\right]= L_{[a,b]}. \] Here \(L_a=\mu(a,\quad ):V\to V\) denotes left multiplication by \(a\), \([a,b]:=a\cdot b-b\cdot a\), and the commutator on the left is \(L_a\circ L_b-L_b\circ L_a\). For this reason, the above structure is often referred to as ``left pre-Lie algebra'', and one can similarly define right pre-Lie algebras by passing to right multiplication. If \(V\) is a \textit{graded} vector space, one can define a \textit{graded pre-Lie algebra} by replacing commutators with graded commutators. The defining condition for a pre-Lie algebra is precisely the (graded) Jacobi identity for the (graded) commutator \([\;,\;]\), so to every pre-Lie algebra one can associate a (graded) Lie algebra. Every associative algebra is pre-Lie for trivial reasons. There are two other prototypical examples of pre-Lie algebras coming from geometry. If \(M\) is a smooth manifold with an \textit{affine structure}, i.e, a flat, torsion-free connection \(\nabla\), then the vector space of global vector fields \(V=\Gamma(M,T_M)\) becomes a pre-Lie algebra via \(X\cdot Y:=\nabla_X Y\). The second example comes from symplectic geometry. Suppose \((M,\omega)\) is a symplectic manifold, and \(\mathcal{F}\subset T_M\) a Lagrangian foliation. Then the vector space \(V=\Gamma_{\mathcal{F}}(T_M)\) of vector fields tangent to the foliation acquires a pre-Lie structure by setting \[ \omega(X\cdot Y,Z)=X\omega(Y,Z)-\omega(Y,[X,Z]). \] Although pre-Lie structures have been used by many people in different contexts, they have remained somewhat unpopular. Their origins can be traced back to Arthur Cayley [\textit{A. Cayley}, ``On the theory of analytical forms called trees'', Phil. Mag. 13, 172--176 (1857)]; see also [\textit{D. Manchon},``A short survey on pre-Lie algebras'', in: Noncommutative geometry and physics: Renormalisation, motives, index theory. Based on the workshop ``Number theory and physics'', Vienna, Austria, March 2009. Zürich: European Mathematical Society (EMS). ESI Lectures in Mathematics and Physics, 89--102 (2011; Zbl 1278.17001), Section 3.2] for a modern overview. The name ``pre-Lie'' was coined by \textit{M. Gerstenhaber}. He observed in [``The cohomology structure of an associative ring'', Ann. Math. (2) 78, 267--288 (1963; Zbl 0131.27302)] that if \(A\) is an \textit{associative} algebra, its Hochschild cochain complex \(CH^\bullet(A)\), \(CH^m=\text{Hom}(A^{\otimes m},A)\), carries the structure of a graded pre-Lie algebra. Namely, if \(f\in CH^m\) and \(g\in CH^n\), one defines \(f\cdot g\in CH^{m+n-1}\) by \[ (f\cdot g)(a_1,\dots, a_{m+n-1})=\sum_{i=1}^{m}(-1)^{\sigma_i}f(a_1\otimes \dots\otimes a_{i-1}\otimes g(a_i\otimes\dots\otimes a_{i+n-1})\otimes\dots \otimes a_{m+n-1}), \] \(\sigma_{i}=(n-1)(i-1)\). The passage to the corresponding Lie bracket gives rise to the Gerstenhaber bracket. The cohomology theory of these algebras has been studied by \textit{A. Nijenhuis} [``Sur une classe de propriétés communes à quelques types différents d'algébres'', Enseign. Math., II. Sér. 14, 225--277 (1968; Zbl 0188.08101)] and \textit{J.-L. Koszul} [``Déformations de connexioins localement plates'', Ann. Inst. Fourier 18, No. 1, 103--114 (1968; Zbl 0167.50103)] in the context of deformations of affine manifolds. Pre-Lie algebras also appear in the work of \textit{E. B. Vinberg} [``The theory of convex homogeneous cones'', Trans. Mosc. Math. Soc. 12, 340--403 (1963); translation from Tr. Mosk. Mat. O.-va 12, 303--358 (1963; Zbl 0138.43301)], under the name ``left-symmetric algebras''. It is well-known that various algebraic structures, for example, associative or Lie algebras, \(A_\infty\) or \(L_\infty\) algebras, etc. arise as algebras over an operad. Then their deformation-obstruction theory, Maurer-Cartan equations, various (co)homology theories, Koszul duality, etc., carry many common features due to the common operadic origin. For an overview, see [\textit{J.-L. Loday}, ``La renaissance des opérades'', Astérisque. 237, 47--74, Exp. No. 792 (1996; Zbl 0866.18007)]. Pre-Lie structures are not an exception: they arise as algebras over the ``rooted tree operad'', as shown by \textit{F. Chapoton} and \textit{M. Livernet} [``Pre-Lie algebras and the rooted trees operad'', Int. Math. Res. Not. 2001, No. 8, 395--408 (2001; Zbl 1053.17001)]. They study (following Ginzburg and Kapranov) how the operadic setup can be used to define pre-Lie homology, as well as its relation to Chevalley-Eilenberg homology. Earlier, homology of pre-Lie algebras has been introduced by Nijenhuis and Dzhumadil'daev [\textit{A. Dzhumadil'daev}, ``Cohomologies and deformations of right-symmetric algebras'', J. Math. Sci., New York 93, No. 6, 836--876 (1999; Zbl 0938.17002)]. The present article is an overview of various known results on pre-Lie algebras, both by the author and by other people, without any recourse to operads. The article also indicates some appearances and possible applications of pre-Lie structures to different parts of mathematics. In Section 2 the author defines (two-sided) modules over a pre-Lie algebra and the KV-complex of a KV algebra. Namely, for a 2-sided module \(W\) over a pre-Lie algebra \(A\), the author defines \(C^\bullet(A,W)\) by \(C^i(A,W)=0\) for \(i<0\), \(C^i(A,W)= \text{Hom}(A^{\otimes i},W)\) for \(i>0\) and \(C^0(A,W)= \{w\in W: a\cdot(b\cdot w)-(a\cdot b)\cdot w=0\}\). The differential \(\delta\) sends \(f\in C^i, i>0\) to \(\delta f\in C^{i+1}\), defined by \[ \begin{multlined} (\delta f)(a_1\otimes\dots \otimes a_{i+1})\\ = \sum_{j\leq i}(-1)^j\left[(a_j f)(a_1\otimes \dots \otimes\widehat{a_j}\otimes\dots\otimes a_{i+1}) + f(a_1\otimes \dots\otimes \widehat{a_j}\otimes \dots\otimes a_i\otimes a_j)a_{i+1} \right]. \end{multlined} \] In degree zero, the coboundary operator is defined by \((\delta w)(a)=-aw+wa\). The author calls the cohomology of this complex \textit{Koszul-Vinberg cohomology} and denotes it by \(H^\bullet_{KV}(A,W)\). The author shows (\textbf{Theorem 2.6}) that \(H_{KV}^2(A,W)\) is in bijection with the set of (equivalence classes of) KV-algebra extensions of \(A\) by \(W\), where \(W\) is thought of as a trivial KV-algebra. However, the expected analogous result on extensions of KV-modules does not hold, as shown in Section 2.3. In low degree there is a relation between Koszul-Vinberg cohomology and Hochschild cohomology \(HH^\bullet(A,W)\): for \(q=0,1\), \(H^q_{KV}=HH^q\), and \(HH^2\subset H^2_{KV}\), as shown in Section 5.1. Section 3 is a review of some general facts about manifolds with affine structure: hyperbolicity, completeness and Hessian metrics. Section 4 is a review of work of Koszul and Nijenhuis, and Koszul's complex of ``high order differential forms'' on a (smooth) vector bundle with a flat connection over a smooth manifold. Section 5, titled ``Other relevant relations'', is a review of various facts from differential and symplectic geometry, which, according to the author, can be obtained using KV cohomology techniques. These include the relationship between KV and de Rham cohomology on an affine manifold (subsection 5.1), Poisson structures and Dirac reductions (subsection 5.2) and Kähler metrics on solvmanifolds, etc. (subsection 5.3). Despite the occasional typographical errors and stylistic peculiarities, the article gives a fresh overview of certain aspects of this under-appreciated topic. Perhaps it would have been beneficial if it also contained a more detailed comparison between the results obtained by the author and the results of Dzhumadil'daev, Chapoton and Livernet, cited above.