A family of maximal subalgebras of the Lie algebra W n ​ ( K ) W_{n}(K) Y.Chapovskyi, A.Petravchuk Abstract . Let K K be an algebraically closed field of characteristic zero and P n = K ⁡ [ x 1 , … , x n ] {P_{n}=K[x_{1},\ldots,x_{n}]} the polynomial ring. Any K K -derivation D D on P n P_{n} is of the form D = ∑ i = 1 n f i ​ ( x 1 , … , x n ) ​ ∂ ∂ x i , {D=\sum_{i=1}^{n}f_{i}(x_{1},\ldots,x_{n})\frac{\partial}{\partial x_{i}}}, where f i ∈ P n . f_{i}\in P_{n}. All such derivations form the Lie algebra W n ​ ( K ) W_{n}(K) over the field K K . We prove that for s = 1 , … , n − 1 s=1,\ldots,n-1 the subalgebra m s ( K ) = { ∑ i = 1 s f i ∂ ∂ x i + ∑ j = s + 1 n g j ∂ ∂ x j ∣ f i ∈ P s , g j ∈ P n } m_{s}(K)=\left\{\sum_{i=1}^{s}f_{i}\frac{\partial}{\partial x_{i}}+\sum_{j=s+1}^{n}g_{j}\frac{\partial}{\partial x_{j}}\mid f_{i}\in P_{s},\ g_{j}\in P_{n}\right\} is a maximal subalgebra of W n ​ ( K ) W_{n}(K) . The ideal I s = { ∑ j = s + 1 n g j ​ ∂ ∂ x j } I_{s}=\left\{\sum_{j=s+1}^{n}g_{j}\frac{\partial}{\partial x_{j}}\right\} of m s ​ ( K ) m_{s}(K) is isomorphic to the Lie algebra P s ⊗ Der ⁡ ( K ⁡ [ x s + 1 , … , x n ] ) P_{s}\otimes\mathrm{Der}(K[x_{s+1},\ldots,x_{n}]) and m s ​ ( K ) / I s ≃ W s ​ ( K ) m_{s}(K)/I_{s}\simeq W_{s}(K) . The Lie algebra W n ​ ( K ) W_{n}(K) is also the free module over the ring P n . P_{n}. Therefore, for any set S ⊆ W n ​ ( K ) S\subseteq W_{n}(K) the rank r ​ k ​ ( S ) rk(S) (over P n P_{n} ) is defined. Some properties of maximal subalgebras of rank n n in W n ​ ( K ) W_{n}(K) are pointed out. 2000 Mathematics Subject Classification 17B65, 17B66, 17B05 1. Introduction Let K be an algebraically closed field of characteristic zero, P_{n}=K[x_{1},\ldots,x_{n}] the polynomial ring in n variables, and R_{n}=K(x_{1},\ldots,x_{n}) the field of rational functions. Denote by W_{n}=W_{n}(K) the Lie algebra of all K -derivations on P_{n} . Any element D\in W_{n} can be uniquely written in the form D=f_{1}(x_{1},\ldots,x_{n})\frac{\partial}{\partial x_{1}}+\cdots+f_{n}(x_{1},\ldots,x_{n})\frac{\partial}{\partial x_{n}},\qquad f_{i}\in P_{n}, so the Lie algebra W_{n} is also the free module of rank n over the ring P_{n} (with the standard basis \frac{\partial}{\partial x_{1}},\ldots,\frac{\partial}{\partial x_{n}} ). From the viewpoint of differential geometry, W_{n} is the Lie algebra of polynomial vector fields on K^{n} . The structure of the Lie algebra W_{n} has been studied by many authors but many open problems remain. For example, the structure of maximal subalgebras was described only in case n=1 , that is, for the Lie algebra W_{1} (see, [ 2 ] ), while for n\geq 2 only limited information is available. Maximal subalgebras of finite codimension from W_{n} were studied in [ 7 ] , [ 8 ] , maximal locally nilpotent and solvable subalgebras were studied in [ 3 ] , see also [ 4 ] and [ 1 ] . In this paper, we point out a family of maximal subalgebras m_{s}(K),s=1,\ldots,n-1 of the Lie algebra W_{n}=W_{n}(K) and prove that they are pairwise non-isomorphic for different values of s (Theorem 1 ). We also study maximal subalgebras M of the Lie algebra W_{n} of rank n over P_{n} . If M is not a submudule of the P_{n} -module W_{n} and contains nonzero ideals of rank 1, in contrast to the case W_{1}, where all such subalgebras are described. The next statement provides a series of maximal subalgebras of W_{n} . Theorem 1 . 1. Let s be an integer with 1\leq s\leq n-1 . Then the set m_{s}(K)=\left\{\sum_{i=1}^{s}f_{i}\frac{\partial}{\partial x_{i}}+\sum_{j=s+1}^{n}g_{j}\frac{\partial}{\partial x_{j}}\mid f_{i}\in P_{s},\ g_{j}\in P_{n}\right\} is a maximal subalgebra of the Lie algebra W_{n} . 2. The set I_{s}=\left\{\sum_{j=s+1}^{n}g_{j}\frac{\partial}{\partial x_{j}}\mid g_{j}\in P_{n}\right\} is a polynomial ideal of the Lie algebra m_{s}(K) and I_{s}\simeq P_{s}\otimes\mathrm{Der}(K[x_{s+1},\ldots,x_{n}]) . Besides, the quotient algebra m_{s}(K)/I_{s} is isomorphic to the Lie algebra W_{s} . 3. I_{s} is a unique non-trivial ideal of the Lie algebra m_{s}(K) . 4. The subalgebras m_{s}(K) and m_{t}(K) are non-isomorphic for s\neq t , s,t=1,\ldots,n-1 . Proof of (1). One can immediately check that m_{s}(K) is a subalgebra of W_{n}(K) . Let S be a subalgebra of W_{n}(K) strictly containing the subalgebra m_{s}(K) . The subalgebra m_{s}(K) obviously contains the Euler derivation E_{n}=\sum_{i=1}^{n}x_{i}\frac{\partial}{\partial x_{i}} , so by Lemma 2 the subalgebra S is graded, S=\bigoplus_{i\geq-1}S_{i} , where S_{i}=S\cap W_{n}^{[i]} and W_{n}^{[i]} denotes the standard grading component of the Lie algebra W_{n}(K) . Repeating some parts of the proof of Lemma 2 one can show that the set S\setminus m_{s}(K) contains nonzero homogeneous derivations. Fix such a derivation and denote it by D . Write D=\sum_{i=1}^{n}h_{i}\frac{\partial}{\partial x_{i}}, where h_{i}\in P_{n},i=1,\ldots,n. Since D\notin m_{s}(K) , there exists an index j_{0}\in\{1,\dots,s\} such that the coefficient h_{j_{0}} depends on a variable x_{i_{0}},i_{0}\geq s+1, that is, \frac{\partial h_{j_{0}}}{\partial x_{i_{0}}}\not=0 . Write the polynomial h_{j_{0}} as a polynomial in x_{i_{0}} : h_{j_{0}}=u_{0}+u_{1}x_{i_{0}}+\dots+u_{d}x_{i_{0}}^{d} for some d\geq 1 , where u_{t} does not depend on x_{i_{0}} for t=0,\dots d and u_{d}\neq 0. Consider the following homogeneous derivation: D_{1}:=\left[\dots\left[D,\frac{\partial}{\partial x_{i_{0}}}\right],\dots,\frac{\partial}{\partial x_{i_{0}}}\right]=\sum_{l=1}^{n}q_{l}\frac{\partial}{\partial x_{l}}, where \partial/\partial x_{i_{0}} is applied d-1 times). Note that the coefficient q_{j_{0}} can be written in the form q_{j_{0}}=v_{0}+v_{1}x_{i_{0}} with v_{1}\neq 0 , and the polynomials v_{0},v_{1} do not depend on x_{i_{0}} . Clearly, D_{1}\in S\setminus m_{s}(K) as well. Without loss of generality, we may assume that v_{1}\in K^{*} . Indeed, if v_{1}\notin K^{*} , consider a monomial x_{1}^{l_{1}}\dots x_{i_{0}}^{0}\dots x_{n}^{l_{n}} of maximal total degree in the polynomial v_{1} . Acting by the linear operator \ad(\frac{\partial}{\partial x_{1}})^{l_{1}}\dots\ \ad(\frac{\partial}{\partial x_{n}})^{l_{n}},\ l_{i}\geq 0,\ i=1,\ldots,n on D_{1} we can obtain a derivation satisfying the condition v_{1}\in K^{*} . Since D_{1} is homogeneous, it must be a linear homogeneous derivation. Acting by the operator \ad(x_{i_{0}}\frac{\partial}{\partial x_{i_{0}}}) on the derivation D_{1} we obtain the derivation D_{2}:=\left[x_{i_{0}}\frac{\partial}{\partial x_{i_{0}}},D_{1}\right]=\sum_{i=1}^{n}c_{i}x_{i_{0}}\frac{\partial}{\partial x_{i}}=x_{i_{0}}\sum_{i=1}^{n}c_{i}\frac{\partial}{\partial x_{i}}, where c_{i}\in K , c_{j_{0}}\neq 0 . Again, D_{2}\in S\setminus m_{s}(K) . We may assume that c_{i}=0 for all i>s , since \sum_{i=s+1}^{n}c_{i}x_{i_{0}}\frac{\partial}{\partial x_{i}}\in m_{s}(K) . Note that \left[x_{j_{0}}\frac{\partial}{\partial x_{i_{0}}},D_{2}\right]=\left[x_{j_{0}}\frac{\partial}{\partial x_{i_{0}}},x_{i_{0}}\sum_{i=1}^{s}c_{i}\frac{\partial}{\partial x_{i}}\right]=x_{j_{0}}\sum_{i=1}^{s}c_{i}\frac{\partial}{\partial x_{i}}-x_{i_{0}}c_{j_{0}}\frac{\partial}{\partial x_{j_{0}}}. Since the first term in the latter sum belongs to the subalgebra m_{s}(K) and c_{j_{0}}\neq 0, we conclude that x_{i_{0}}\frac{\partial}{\partial x_{j_{0}}}\in S . Furthermore, for any p,1\leq p\leq s , we have the equality \left[x_{j_{0}}\frac{\partial}{\partial x_{p}},x_{i_{0}}\frac{\partial}{\partial x_{j_{0}}}\right]=-x_{i_{0}}\frac{\partial}{\partial x_{p}}. So, we can find in the subalgebra S all derivations of the form x_{i_{0}}\frac{\partial}{\partial x_{p}} , 1\leq p\leq s . Take an arbitrary element of the form f(x_{1},\dots,x_{n})\frac{\partial}{\partial x_{i_{0}}} and let j\leq s . Then \left[f(x_{1},\dots,x_{n})\frac{\partial}{\partial x_{i_{0}}},x_{i_{0}}\frac{\partial}{\partial x_{j}}\right]=f(x_{1},\dots,x_{n})\frac{\partial}{\partial x_{j}}-x_{i_{0}}\frac{\partial f}{\partial x_{j}}\frac{\partial}{\partial x_{i_{0}}}. Since the second term lies in m_{s}(K) , we conclude that f\frac{\partial}{\partial x_{j}}\in S . Therefore, S=W_{n}(K) , which shows that m_{s}(K) is a maximal subalgebra of W_{n}(K) . ∎ Proof of (2). It is easy to check that I_{s} is an ideal of the Lie algebra m_{s}(K). Let D\in{I}_{s} . Then D can be written in the form D=f_{s+1}\frac{\partial}{\partial x_{s+1}}+\dots+f_{n}\frac{\partial}{\partial x_{n}}, where f_{i}\in P_{n},i=s+1,\ldots,n. Note that any polynomial in P_{n}=K[x_{1},\dots,x_{n}] can be regarded as a polynomial in variables x_{s+1},\dots,x_{n} with coefficients in P_{s}=K[x_{1},\dots,x_{s}] . In particular, P_{n}\cong P_{s}\otimes K[x_{s+1},\dots,x_{n}]. Therefore: {I}_{s}\cong P_{s}\otimes\left(K[x_{s+1},\dots,x_{n}]\frac{\partial}{\partial x_{s+1}}+\dots+K[x_{s+1},\dots,x_{n}]\frac{\partial}{\partial x_{n}}\right)\cong P_{s}\otimes W_{n-s}. The quotient algebra m_{s}(K)/{I}_{s} is clearly isomorphic to the Lie algebra generated by the first s partial derivatives with coefficients in P_{s}, since the terms involving \partial/\partial x_{j} for j>s lie in the ideal {I}_{s} . Hence, m_{s}(K)/{I}_{s}\simeq W_{s} . ∎ Proof of (3). Proceed step by step. 1. I_{s} is a maximal ideal of the Lie algebra m_{s}(K) . Indeed, m_{s}(K)/I_{s}\simeq W_{s} , which is a simple Lie algebra. 2. I_{s} is a minimal ideal of m_{s}(K). To show this take an arbitrary nonzero element D\in I_{s} . Consider the ideal J in m_{s}(K) generated by D . Let us show that {J}=I_{s} . Indeed, applying the adjoint operators \ad\frac{\partial}{\partial x_{i}},i=1,\ldots,n on D sufficiently many times we can obtain the element \overline{D}=c_{s+1}\frac{\partial}{\partial x_{s+1}}+\dots+c_{n}\frac{\partial}{\partial x_{n}},\quad c_{i}\in K. The elements \frac{\partial}{\partial x_{1}},\dots,\frac{\partial}{\partial x_{n}} lie in m_{s}(K) , so \overline{D}\in{J} . Note that \overline{D} can be chosen nonzero since D\not=0, so for some k,s+1\leq k\leq n the coefficient c_{k} is nonzero. It is obvious that c_{k}^{-1}x_{k}\frac{\partial}{\partial x_{k}}\in m_{s}(K). Therefore [D,\ c_{k}^{-1}x_{k}\frac{\partial}{\partial x_{k}}]=\frac{\partial}{\partial x_{k}}\in J. Then for an arbitrary derivation of the form f(x_{1},\ldots,x_{n})\frac{\partial}{\partial x_{i}},s+1\leq i\leq n one can write the derivation -\int f(x)dx_{k}\frac{\partial}{\partial x_{i}} and this derivation lies in the ideal I_{s}. Then we obviously have f(x)\frac{\partial}{\partial x_{i}}=\left[-\int f(x)dx_{k}\frac{\partial}{\partial x_{i}},\frac{\partial}{\partial x_{k}}\right]\in{J}. This demonstrates that {J}=I_{s} . 3. Uniqueness. If we assume there exists a maximal ideal {J}\neq I_{s} , then the sum {J}+I_{s} must be m_{s}(K) . Therefore, we have W_{s}(K)\simeq m_{s}(K)/I_{s}=({J}+I_{s})/I_{s}\simeq{J}/({J}\cap I_{s}) Since I_{s} is a minimal ideal of the Lie algebra m_{s}(K) it holds {J}\cap I_{s}=0. The latter implies m_{s}(K)\simeq W_{s}(K)\oplus I_{s} . In [ 4 ] , it was proved that the derived length of solvable subalgebras in W_{n}(K) is at most 2n . The known example of solvable subalgebras that achieves this bound was pointed out in [ 5 ] , this is the solvable subalgebra s_{n}(K)=(P_{0}+x_{1}P_{0})\frac{\partial}{\partial x_{1}}+\dots+(P_{n-1}+x_{n}P_{n-1})\frac{\partial}{\partial x_{n}}, see also [ 3 ] . Denote by d^{\star}(L) the maximum of derived lengths of solvable subalgebras from a Lie algebra L . Note that the Lie algebra m_{s}(K) contains the subalgebra s_{n}(K), so d^{\star}(m_{s}(K))=d^{\star}(W_{n}(K))=2n. But we also have d^{\star}(m_{s}(K))=d^{\star}(W_{s}(K)\oplus I_{s})=\max(d^{\star}(W_{s}(K)),d^{\star}(I_{s}))=\max(2s,2n-2s)\neq 2n The obtained contradiction proves the uniqueness of the ideal I_{s} in the Lie algebra m_{s}(K). ∎ Proof of (4). Suppose, on the contrary, that m_{s}(K)\simeq m_{t}(K) for s\neq t and let \varphi:m_{s}(K)\to m_{t}(K) be an isomorphism. By the above proven I_{s} is a maximal ideal of the Lie algebra m_{s}(K) . Therefore, \varphi(I_{s})=I_{t} and hence I_{s}\simeq I_{t} . By the above mentioned I_{s}\simeq W_{n-s}\otimes P_{s}\quad\text{and}\quad I_{t}\simeq W_{n-t}\otimes P_{t}. It is easy to see that d^{\star}(W_{n-s}\otimes P_{s})=d^{\star}(W_{n-s})=2(n-s) and d^{\star}(W_{n-t}\otimes P_{t})=d^{\star}(W_{n-t})=2(n-t). Therefore I_{s}\not\simeq I_{t} , which implies m_{s}(K)\not\simeq m_{t}(K) for s\neq t . The proof is complete. ∎ Remark 1 . One may also consider a partition of the set \{1,\ldots,n\} into a disjoint union \{1,\ldots,n\}=I\cup J,I=\{i_{1},\ldots,i_{s}\},J=\{i_{s+1},\ldots,i_{n}\} and define the subalgebra m_{I}(K)=K[x_{i_{1}},\ldots,x_{i_{s}}]\frac{\partial}{\partial x_{i_{1}}}+\cdots+K[x_{i_{1}},\ldots,x_{i_{s}}]\frac{\partial}{\partial x_{i_{s}}}+P_{n}\frac{\partial}{\partial x_{i_{s+1}}}+\cdots+P_{n}\frac{\partial}{\partial x_{i_{n}}}. This subalgebra is also maximal in W_{n} since it is conjugate to m_{s}(K) via the automorphism \theta\in\mathrm{Aut}\,P_{n} defined by \theta(x_{j})=x_{i_{j}} , j=1,\ldots,n . Remark 2 . One could attempt to generalize the previous theorem: for example, for integers s_{1},s_{2},1\leq s_{1}