Advertisement

Classification of Simple Modules of the Ore Extension \(K[X][Y; f\frac{d}{dX}]\)

  • V. V. BavulaEmail author
Open Access
Article
  • 57 Downloads

Abstract

For the algebras \(\Lambda \) in the title of the paper, a classification of simple modules is given, an explicit description of the prime and completely prime spectra is obtained, the global and the Krull dimensions of \(\Lambda \) are computed.

Keywords

A skew polynomial ring A prime ideal A completely prime ideal A simple module The global dimension The Krull dimension A normal element 

Mathematics Subject Classification

16D60 13N10 16S32 

1 Introduction

Let D be a ring and \(A= D[x; \sigma , \delta ]\) be a skew polynomial ring where \(\sigma \) is an automorphism of D and \(\delta \) is a \(\sigma \)-derivation of D (for all \(a, b\in D\), \(\delta (ab) = \delta (a)b+\sigma (a) \delta (b)\)). The ring A is generated by D and x subject to the defining relations \(xa=\sigma (a) x+ \delta (a)\) for all elements \(a\in D\). When D is a Dedekind domain, a classification of simple A-modules is given in [4]. This is a large class of rings. A machinery is developed in [4] to cover all possible situations (non-commutative valuations, etc).

The algebra
$$\begin{aligned} \Lambda =K[X]\left[ Y; \delta := f\frac{d}{dX} \right] =\bigoplus _{i\ge 0} K[X]Y^i \end{aligned}$$
is a particular example of the ring A where \(\sigma =\mathrm{id}\) is the identity automorphism of the polynomial ring K[X], \(f\in K[X]\) and \(\delta = f\frac{d}{dX}\) is a K-derivation of K[X] (\(\delta (X)=f\)). If \(f=1\) (or, more generally, \(f\in K^\times \backslash \{ 0\}\)) then the algebra \(\Lambda (1)\) is the Weyl algebra
$$\begin{aligned} A_1=K\langle X, \partial \, | \, \partial X-X\partial =1\rangle \simeq K[X]\left[ Y; \frac{d}{dX} \right] . \end{aligned}$$
In 1981, a classification of simple \(A_1\)-modules was obtained by Block (over the field of complex numbers) in [9] (see also [2, 3] for an alternative approach via generalized Weyl algebras in a more general situation).

Recently, classifications of simple weight modules are obtained for some classical algebras (the Euclidean algebra, the Schrödinger algebra, the universal enveloping algebra \(U(\mathrm{sl}_2\ltimes V_2)\)), see [5, 6, 7]. In these classifications, classifications of all simple modules over certain subalgebras of the Weyl algebra \(A_1\) that contain the polynomial algebra K[X] (the, so-called, polyonic algebras) play a crucial role. The polyonic algebras are investigated in [8]. Each polyonic algebra contains the algebra \(\Lambda =\Lambda (f)\) for some non-scalar polynomial \(f\in K[X]\) which play an important role in studying of its properties. This is the main reason why we decided to collect main properties of the algebras \(\Lambda \) in this paper. In particular, a classification of simple \(\Lambda \)-modules is given in Sect. 2 (Lemma 2.1 and Theorem 2.10). This classification can be derived from [4] but we give different and simpler proofs which are based on generalized Weyl algebras rather than skew polynomial rings.

An ideal \(\mathfrak {p}\) of a ring R is called a completely prime ideal if the factor ring \(R/\mathfrak {p}\) is a domain. A completely prime ideal is a prime ideal. The sets of prime and completely prime ideals of the ring R are denoted by \(\mathrm{Spec}(R)\) and \(\mathrm{Spec}_c(R)\), respectively.

In Theorem 1.1, a classification of prime and completely prime ideals of the algebra \(\Lambda \) is given, the Krull and global dimensions of the algebra \(\Lambda \) are found. The algebra \(\Lambda \) is a Noetherian domain of Gelfand-Kirillov dimension 2.

Theorem 1.1

Let K be a field of characteristic zero, \(\Lambda =K[X][Y; \delta := f\frac{d}{dX} ]\) where \(f\in K[X]\backslash K\). Let \(f=p_1^{n_1}\cdots p_s^{n_s}\) be a unique (up to permutation) product of irreducible polynomials of K[X]. Then
  1. 1.

    The Krull dimension of \(\Lambda \) is \(\mathrm{Kdim } (\Lambda )=2\).

     
  2. 2.

    The global dimension of \(\Lambda \) is \(\mathrm{gldim}(\Lambda )=2\).

     
  3. 3.

    The elements \(p_1, \ldots , p_s\) are regular normal elements of the algebra \(\Lambda \) (i.e. \(p_i\) is a non-zero-divisor of \(\Lambda \) and \(p_i\Lambda = \Lambda p_i\)).

     
  4. 4.

    \(\mathrm{Spec}(\Lambda )=\mathrm{Spec}_c (\Lambda )=\{ 0 , \Lambda p_i, (p_i, q_i)\, | \, i=1, \ldots , s;\; q_i\in \mathrm{Irr}_m(F_i[Y])\}\) where \(F_i:=K[X]/(p_i)\) is a field and \(\mathrm{Irr}_m(F_i[Y])\) is the set of monic irreducible polynomials of the polynomial algebra \(F_i[Y]\) over the field \(F_i\) in the variable Y. If, in addition, the field K is an algebraically closed and \(\lambda _1, \ldots , \lambda _s\) are the roots of the polynomial f then \(\mathrm{Spec}(\Lambda )=\{ 0 , \Lambda (X-\lambda _i), (X-\lambda _i, Y-\mu )\, | \, i=1, \ldots , s;\; \mu \in K\}\).

     

The proof of Theorem 1.1 is given in Sect. 3.

2 Classification of Simple \(\Lambda \)-Modules

In this section, ‘module’ means left module, K is an algebraically closed field of characteristic zero, \(\Lambda =K[X][Y, \delta =f\frac{d}{dX}]\) where \(f\in K[X]\backslash K\). The algebra \(\Lambda \) is a Noetherian domain. The aim of the section is to give a classification of simple \(\Lambda \)-modules (Lemma 2.1 and Theorem 2.10).

The elementfis a regular normal element of\(\Lambda \). It follows from
$$\begin{aligned} fY=Yf-f'f=(Y-f')f,\;\;\mathrm{where}\;\;f'=\frac{df}{dX}, \end{aligned}$$
that the element f is a normal element of \(\Lambda \) (i.e. \(\Lambda f=f\Lambda )\). It determines a K-automorphism \(\omega _f\) of the algebra \(\Lambda \):
$$\begin{aligned} fu=\omega _f(u)f,\;\; u\in \Lambda ,\\ \omega _f:X\mapsto X,\;\;Y\mapsto Y-f'. \end{aligned}$$
The algebra \(\Lambda \) can be identified with a subalgebra of the first Weyl algebra \(A_1\) by the map
$$\begin{aligned} \Lambda \rightarrow A_1,\;\;X\mapsto X,\;\; Y\mapsto f\partial . \end{aligned}$$
(1)
The Weyl algebra\(A_1\)is a generalized Weyl algebra. The Weyl algebra \(A_1\) is a simple Noetherian domain with restricted minimum condition, i.e. any proper left or right factor module of \(A_1\) has finite length, [10].
Definition, [1, 2]. Let D be a ring, \(\sigma \) be an automorphisms of D and a be a central element of D. A generalized Weyl algebra (GWA) \(A=D(\sigma , a)\) of degree 1, is the ring generated by D and by two indeterminates X and Y subject to the relations [1, 2]: For all \(\alpha \in D\),
$$\begin{aligned} X\alpha =\sigma (\alpha )X \;\; \mathrm{and}\;\; Y\alpha =\sigma ^{-1}(\alpha )Y,\;\; \ YX=a \;\; \mathrm{and}\;\; XY=\sigma (a). \end{aligned}$$
The algebra
$$\begin{aligned} A={\bigoplus }_{n\in \mathbb {Z}}\, A_n \end{aligned}$$
is a \(\mathbb {Z}\)-graded algebra, where \(A_n=Dv_n,\)\(v_n=X^n\,\, (n>0), \,\,v_n=Y^{-n}\,\, (n<0), \,\,v_0=1.\)
The Weyl algebra \(A_1\) is a GWA,
$$\begin{aligned} A_1= D(\sigma ,a=H),\; X\leftrightarrow X,\; \partial \leftrightarrow Y,\; \partial X\leftrightarrow H,\;\;D=K[H], \end{aligned}$$
with coefficients from a polynomial ring K[H] where \(\sigma \in \mathrm{Aut}_K\,K[H]\) and \(\sigma :H\rightarrow H-1\).
We denote by \(\Lambda _f\) (resp., \(A_{1,f}\)) the localization of the ring \(\Lambda \) (resp., \(A_1\)) at the powers of the element f, i.e.
$$\begin{aligned} \Lambda _f=S^{-1}_f\Lambda ,\;\;(\mathrm{resp., }\;\;A_{1,f}=S^{-1}_fA_1)\;\;\mathrm{where}\;\; S_f=\{ f^i, i\ge 0\}. \end{aligned}$$
By (1), \(\Lambda _f\) is a subalgebra of \(A_{1,f}\) such that
$$\begin{aligned} \Lambda _f=A_{1,f}. \end{aligned}$$
(2)
The algebras \(\Lambda \) and \(A_1\) can be considered as subalgebras of \(A_{1,f}\),
$$\begin{aligned} \Lambda \subseteq A_1\subseteq \Lambda _f=A_{1,f}. \end{aligned}$$
(3)
The algebra \(A_{1,f}\) is a simple Noetherian domain with restricted minimum condition.
\(\hat{\Lambda }(f-\mathrm{torsion})\). The sets of isoclasses of \(\Lambda \)-modules \(\hat{\Lambda }\) and of \(A_1\)-modules \(\hat{A}_1\) are disjoint unions of f-torsion\((M_f=0)\) and f-torsionfree\((M_f\ne 0)\) simple \(\Lambda \)-modules and \(A_1\)-modules, respectively,
$$\begin{aligned} \hat{\Lambda }= & {} \hat{\Lambda }(f-\mathrm{torsion})\coprod \hat{\Lambda }(f-\mathrm{torsionfree}), \end{aligned}$$
(4)
$$\begin{aligned} \hat{A}_1= & {} \hat{A}_1 (f-\mathrm{torsion})\coprod \hat{A}_1(f-\mathrm{torsionfree}). \end{aligned}$$
(5)
Lemma 2.1 is a classification of simple f-torsion \(\Lambda \)-modules.

Lemma 2.1

Let \(\lambda _1, \ldots ,\lambda _s\) be the roots of the polynomial f. Then
$$\begin{aligned} \hat{\Lambda }(f-\mathrm{torsion})= \{[ \Lambda /\Lambda (X-\lambda _i,Y-\mu )]\;\;|\;\; i=1,\ldots ,s; \; \mu \in K\}. \end{aligned}$$
All these \(\Lambda \)-modules are 1-dimensional and they are the only simple finite dimensional \(\Lambda \)-modules (by Theorem 2.10).

Proof

Each simple f-torsion \(\Lambda \)-module M is annihilated by the (normal) element f\((fM=0)\). So, in fact, the \(\Lambda \)-module M is a simple module over the factor algebra
$$\begin{aligned} \Lambda /(f)=K[X, Y]/(f) \end{aligned}$$
which is isomorphic to the factor algebra of the polynomial ring K[XY] in two variables at the ideal (f) generated by f, and the equality in the lemma follows.
Let N be a simple \(\Lambda \)-module. Then the map
$$\begin{aligned} f_N:\; N\rightarrow N, \; n\mapsto fn \end{aligned}$$
is either 0 or a bijection (f is normal in \(\Lambda )\). In the second case N is, in fact, a simple \((\Lambda _f\equiv A_{1,f})\)-module, so \(\mathrm{dim }_K\,N=\infty \), since \(A_{1,f}\) is a simple infinite dimensional algebra. If the module N is finite dimensional, then \(fN=0\), i.e. \([N]\in \hat{\Lambda }(f-\mathrm{torsion})\). \(\square \)

\(\hat{\Lambda }(f-\mathrm{torsionfree})\). The sum of all simple submodules of a \(\Lambda \)-module M is called the socle of M which is denoted by \(\mathrm{soc}_\Lambda M \). It is the largest semisimple submodule of M. A \(\Lambda \)-module N is called \(\Lambda \)-socle (or, socle, for short) provided \(\mathrm{soc}_\Lambda N\ne 0\). Denote by \(\hat{A_1}(\Lambda \)-socle) the set of isoclasses of simple \(\Lambda \)-socle \(A_1\)-modules. The proof of the following lemma is evident (see [2, Lemma 3.4] for details).

Lemma 2.2

  1. 1.
    The canonical map
    $$\begin{aligned} (\cdot )_f :\hat{\Lambda } (f-\mathrm{torsionfree})\rightarrow \hat{A}_{1,f} (\Lambda -\mathrm{socle}),\;[M]\mapsto [ M_f:=A_{1,f}\otimes _\Lambda M] \end{aligned}$$
    is a bijection with inverse \([N]\rightarrow [\mathrm{soc}_\Lambda (N)]\).
     
  2. 2.
    Each simple f-torsionfree \(\Lambda \)-module has the form
    $$\begin{aligned} M_\mathfrak {m}:= \Lambda /\Lambda \cap \mathfrak {m}\end{aligned}$$
    (6)
    for some maximal left ideal \(\mathfrak {m}\) of the ring \(A_{1,f} \). Two such modules are isomorphic, \(M_\mathfrak {m}\simeq M_\mathfrak {n}\), iff the \(A_{1,f}\)-modules \(A_{1,f}/\mathfrak {m}\) and \(A_{1,f}/\mathfrak {n}\) are isomorphic. \(\square \)
     

Lemma 2.3

Let \(\lambda _1, \ldots ,\lambda _s\) be the roots of the polynomial f. Then
$$\begin{aligned} \hat{A}_1(f-\mathrm{torsion})=\{ [M_i:=A_1/A_1(X-\lambda _i)]\, | \, i=1,\ldots ,s \}. \end{aligned}$$

Proof

As a vector space the module \(M_i\) can be identified with the polynomial ring K[y] in a variable \(y=\partial +A_1(X-\lambda _i)\) and
$$\begin{aligned} \partial y^j=y^{j+1}, \;\; Xy^j=\lambda _iy^j+\cdots \;\; \mathrm{for}\;\; j\ge 0 \end{aligned}$$
where by three dots we denote the sum of elements of smaller degree in the variable y. Thus the linear operator
$$\begin{aligned} X-\mu : M_i\rightarrow M_i, \;\; m\mapsto (X-\mu )m, \;\;m\in M_i, \end{aligned}$$
is nilpotent if and only if \(\mu =\lambda _i\); otherwise, \(X-\mu \) is an isomorphism of the vector space \(M_i\). From this fact it follows that the \(A_1\)-modules \(\{ M_i\}\) are simple and non-isomorphic.

Now, let \([M]\in \hat{A}_1(f-\mathrm{torsion})\). Then there exists i such that M is an epimorphic image of \(M_i\), hence \(M\simeq M_i\).

\(\square \)

Theorem 2.4

The map
$$\begin{aligned} \hat{A}_1 (f-\mathrm{torsionfree})=\hat{A}_1 \backslash \{ [M_1],\ldots ,[M_s]\}\rightarrow \hat{A}_{1,f}, \;\;[M]\mapsto [M_f] \end{aligned}$$
is bijective.

Proof

The map above is well defined and injective.

Let \([N]\in \hat{A}_{1,f}\). Then \(N\simeq A_{1,f}/J\) for some nonzero maximal left ideal J of \(A_{1,f}\). Then \(I=J\cap A_1\ne 0\) and \(A_1/I\) is a \(A_1\)-submodule of N. The \(A_1\)-module \(A_1/I\) has finite length [10], thus it contains a simple \(A_1\)-submodule, say M. Then \(N\simeq M_f\) which means that the map above is surjective.\(\square \)

Recall that \(D=K[H]\). The localization \(B=S^{-1} A_1\) of the Weyl algebra \(A_1\) at the Ore set \(S=D\backslash \{ 0\}\) is a skew Laurent polynomial ring
$$\begin{aligned} B=K(H)[X, X^{-1}; \sigma ], \;\; \sigma (H)=H-1, \end{aligned}$$
with coefficients from the field K(H) of rational functions. The algebra B is a right and left Euclidean domain with respect to the ‘length’ function
$$\begin{aligned} l: B\backslash \{ 0\}\rightarrow \mathbb {N}:=\{0, 1, 2, \ldots \},\;\; l(\alpha X^m+\beta X^{m+1}+\cdots +\gamma X^n)=n-m,\;\;\alpha \ne 0, \, \gamma \ne 0\in K(H), \end{aligned}$$
hence, it is a right and left principal ideal domain.
We have
$$\begin{aligned} \hat{A}_1=\hat{A}_1 (D-\mathrm{torsion})\coprod \hat{A}_1 (D-\mathrm{torsionfree}) \end{aligned}$$
where a simple \(A_1\)-module M belongs to the first (resp., second) set if \(S^{-1}M=0\) (resp., \(S^{-1}M\ne 0\)).

For \(\lambda \in K\) set \(\mathcal{O}(\lambda ):=\lambda +\mathbb {Z}\). We say that scalars \(\lambda \) and \(\mu \) are equivalent, \(\lambda \sim \mu \), if either \(\mathcal{O}(\lambda )=\mathcal{O}(\mu )\ne \mathbb {Z}\) or both \(\lambda \) and \(\mu \) belong either to \((-\infty ,0]:=\{ i\in \mathbb {Z}\, | \, i\le 0\}\) or to \([1,\infty ):=\{ i\in \mathbb {Z}\, | \, i\ge 1\}\). Then \(\sim \) is an equivalence relation on K. Let \(K/\sim \) be the set of equivalence classes of K under \(\sim \). So, the elements of the set \(K/\sim \) are distinct sets \(\lambda +\mathbb {Z}\) where \(\lambda \not \in \mathbb {Z}\) and the two sets \((-\infty ,0]\) and \([1,\infty )\). Notice that \(\mathbb {Z}= (-\infty ,0]\coprod [1,\infty )\).

Proposition 2.5

([2, Theorem 3.1]) The map
$$\begin{aligned} K/\sim \;\rightarrow \;\hat{A}_1 (D-\mathrm{torsion}), \;\; [\Gamma ]\mapsto [L(\Gamma )], \end{aligned}$$
is a bijection, where
  1. 1.

    If \(\Gamma =\mathcal{O}(\lambda )\ne \mathbb {Z}\), then \(L(\Gamma )= A_1/A_1(H-\lambda )\).

     
  2. 2.

    If \(\Gamma =(-\infty ,0]\), then \(L(\Gamma )= A_1/A_1X\).

     
  3. 3.

    If \(\Gamma =[1,\infty )\), then \(L(\Gamma )= A_1/A_1(H-1, Y)\). \(\Box \)

     

Corollary 2.6

$$\begin{aligned} \hat{A}_1 (D-\mathrm{torsion}, f-\mathrm{torsion})= {\left\{ \begin{array}{ll} \{ [L((-\infty ,0])= A_1/A_1X]\} &{} \text {if 0 is a root of} \; f(X),\\ \emptyset &{} \text {if 0 is not a root of}\; f(X).\\ \end{array}\right. } \end{aligned}$$

Proof

Straightforward. \(\square \)

Corollary 2.7

Let \([M]\in \hat{A}_1 (D-\mathrm{torsion}, f-\mathrm{torsionfree})\).
  1. 1.

    If \(M=A_1/A_1X\) (i.e. 0 is not a root of f, by Corollary 2.6) then M is a simple f-torsionfree \(\Lambda \)-module with \(M=M_f\).

     
  2. 2.

    If \(M\ne A_1/A_1X\) then \(\mathrm{soc}_\Lambda \, M=\mathrm{soc}_\Lambda \, M_f=0\). The set \(\hat{\Lambda }\) (\(D-\)torsion, \(f-\)torsionfree) is equal to \(\{A_1/A_1/X\}\) if 0 is not a root of f and \(\emptyset \), otherwise.

     

Proof

1. As a vector space the module \(M=A_1/A_1X\) has the basis \(\{ y^i=\partial ^i+A_1X, \; i\ge 0\} \), and
$$\begin{aligned} \partial y^i=y^{i+1}, \;\;Xy^i=-iy^{i-1} \;\; \mathrm{and}\;\; Yy^i=f(0)y^{i+1}+\sum _{0\le j\le i}\; \mu _jy^j, \end{aligned}$$
for some scalars \(\mu _j \in K\). Now, it is obvious that the \(\Lambda \)-module M is a simple f-torsionfree \(\Lambda \)-module \((f(0)\ne 0)\). Moreover, the linear map \(f_M:M\rightarrow M\), \(m\mapsto fm\) is a bijection, hence, \(M=M_f\).
2. Since \(\Lambda _f=A_{1,f}\), \(\mathrm{soc}_\Lambda \, M=\mathrm{soc}_\Lambda \, M_f\). Let M belongs to the first (resp., third) class of modules from Proposition 2.5 , i.e.
$$\begin{aligned} M=L(\Gamma ), \;\; \Gamma =\mathcal{O}(\lambda )\ne \mathbb {Z}\;\; (\mathrm{resp}., \;\; \Gamma =[1, \infty )). \end{aligned}$$
The element \(\bar{1}=1+A_1(H-\lambda )\) (resp., \(\bar{1}=1+A_1(H-1, Y )\)) is a canonical generator of the \(A_1\)-module M. In both cases, for \(i\ge 0\), set \(x^i=X^i\bar{1}\). In the first case, for \(i<0\), set \(x^i=\mu _i\partial ^{-i}\bar{1}\), \(\mu _i\in K\). The scalars \(\mu _i\) can be chosen in such a way that (in both cases) \( Xx^i=x^{i+1}\) for all possible i. Degree argument shows that the module M contains a strictly descending chain of \(\Lambda \)-submodules
$$\begin{aligned} M\supset fM \supset \cdots \supset f^nM\supset \cdots \;\;\mathrm{with}\;\; \bigcap _{n\ge 0}\; f^nM=0. \end{aligned}$$
Suppose that \(N:=\mathrm{soc}_\Lambda M\ne 0\), then, in a view of Lemma 2.2 and Theorem 2.4, N is an essential simple \(\Lambda \)-submodule of both \(M_f\) and M, hence \(0\ne N\subseteq \cap _{n\ge 0}\; f^nM=0\), a contradiction. \(\square \)

An element of a ring is called regular if it is not a zero divisor. Given a ring A and a multiplicatively closed subset S of A which consists of regular normal elements. Let \(B=S^{-1} A\) be the localization of A at S.

Theorem 2.8

Let A, B, and S be as above and let \(\mathfrak {m}\) be a maximal left ideal of B. The following are equivalent.
  1. 1.

    The A-module \(M_\mathfrak {m}:=A/A\cap \mathfrak {m}\) is simple.

     
  2. 2.

    The socle \(\mathrm{soc}_A(M_\mathfrak {m})\ne 0\).

     
  3. 3.

    \(A=As+A\cap \mathfrak {m}\) for all \(s\in S\). \(\square \)

     

Remark. If \(S=\{ f^n, n\ge 0\}\) for some regular normal element f of A, then the last condition of this lemma is equivalent to \(A=Af+A\cap \mathfrak {m}\). We shall use this fact in what follows. In general situation, it suffices to check whether the third condition holds only for generators of the monoid S.

Proof

The implications \((1\Rightarrow 2)\) and \((1\Rightarrow 3)\) are obvious.

\((2\Rightarrow 1)\) If \(\mathrm{soc}_A(M_\mathbf{m})\ne 0\) then it is a simple A-module which for some \(s\in S\) is equal to
$$\begin{aligned} (As+A\cap \mathfrak {m})/A\cap \mathfrak {m}\simeq As/As\cap \mathfrak {m}\simeq A/A\cap \mathfrak {m}s^{-1}=\omega _s(A)/\omega _s(A\cap \mathfrak {m})\simeq {}^{\omega _s^{-1} }M_{\mathfrak {m}}, \end{aligned}$$
where \({}^{\omega _s^{-1} }M_{\mathfrak {m}} \) is the twisted A-module \(M_{\mathfrak {m}} \) by the automorphism \(\omega _s^{-1}\) of A (the element s is regular and normal). Since the A-module \({}^{\omega _s^{-1} }M_{\mathfrak {m}} \) is simple, so is \(M_{\mathfrak {m}}\).

\((3\Rightarrow 1)\) If J is a left ideal of A which contains \(A\cap \mathfrak {m}\) but does not coincide with it, then, by the maximality of \(\mathfrak {m}\), \(S^{-1} J=B\). Therefore \(J\cap S\ne \emptyset \). Let \(s\in J\cap S\). Then \(J\supseteq As+A\cap \mathfrak {m}=A\), that is \(M_\mathfrak {m}\) is a simple A-module.

\(\square \)

\(\hat{A}_1(D-\mathrm{torsionfree})\). Let us recall a description of \(\hat{A}_1(D-\mathrm{torsionfree})\) from [2]. In the set \(S=K[H]\backslash \{ 0\}\) consider the relation <: \(\alpha <\beta \) if there are no roots \(\lambda \) and \(\mu \) of the polynomial \(\alpha \) and \(\beta \) respectively and such that \(\lambda -\mu \) is non-negative integer.

Definition, [2]. An element \(b=Y^m\beta _{-m}+\cdots +\beta _0\in A_1\), \(m>0\), all \(\beta _i\in D\), is called l-normal if \(\beta _0<\beta _{-m}\) and \(\beta _0<H\), (i.e. the polynomial \(\beta _0\) has no root from \(\{ 0, 1, 2, \ldots \}\) and there are no roots \(\lambda \) and \(\mu \) of the polynomials \(\beta _0\) and \(\beta _m \) respectively with \(\lambda -\mu \in \{ 0, 1, 2, \ldots \}\)).

Theorem 2.9

([2, Theorem 3.8]) Let \(b=Y^m\beta _{-m}+\cdots +\beta _0\in A_1\), \(m>0\), all \(\beta _i\in D\), be an l-normal and irreducible element in B. Then
$$\begin{aligned} \mathcal{M}_b:=A_1/A_1\cap Bb \end{aligned}$$
is a simple D-torsionfree \(A_1\)-module. Two such \(A_1\)-modules are isomorphic, \( \mathcal{M}_b\simeq \mathcal{M}_c\), iff \(B/Bb\simeq B/Bc\) as B-modules. Each simple D-torsionfree \(A_1\)-module is isomorphic to some \( \mathcal{M}_b\). \(\square \)
Set
$$\begin{aligned} B_f:=S_f^{-1}B=A_{1,f}\otimes _\Lambda \,B=\Lambda _f\otimes _\Lambda \,B \end{aligned}$$
for the localization of the (left)\(\Lambda \)-moduleB at \(S_f\). Then the algebra \(A_{1,f}= \Lambda _f\) can be considered as a (\(A_{1,f}= \Lambda _f\))-submodules of \(B_f\). For any nonzero \(b\in B\), \((Bb)_f=B_fb\).

Theorem 2.10 is a classification of simple f-torsionfree \(\Lambda \)-modules.

Theorem 2.10

Let \(b=Y^m\beta _{-m}+\cdots +\beta _0\in A_1\), \(m>0\), all \(\beta _i\in D\), be an l-normal and irreducible element in B such that
  1. 1.

    \(\Lambda =\Lambda f+\Lambda \cap B_fb \; (= \Lambda f+\Lambda \cap Bb )\), and

     
  2. 2.

    the simple B-module B / Bb is not isomorphic to any of modules \(B/B(X-\lambda )\) where \(\lambda \) runs through the nonzero roots of f.

     
Then
$$\begin{aligned} \mathcal {M}_b:=\Lambda /\Lambda \cap Bb\; \, (=\Lambda /\Lambda \cap B_fb) \end{aligned}$$
is a simple f-torsionfree \(\Lambda \)-module. Two such \(\Lambda \)-modules are isomorphic, \( \mathcal {M}_b\simeq \mathcal {M}_c\), iff \(B/Bb\simeq B/Bc\) as B-modules.

Each simple f-torsionfree \(\Lambda \)-module is isomorphic either to some \( \mathcal {M}_b\) or to the module \(M=A_1/A_1X\) from Corollary 2.7, if 0 is not a root of f (the \(\Lambda \)-module M is not isomorphic to any \(\mathcal {M}_b\)). The condition 1 above is equivalent to the condition that \(\Lambda = \Lambda (X-\lambda _i)+\Lambda \cap B_fb \; (= \Lambda (X-\lambda _i)+\Lambda \cap Bb )\) for all roots \(\lambda _i\) of the polynomial f.

Each simple f-torsionfree \(\Lambda \)-module is infinite dimensional.

Proof

By Lemma 2.2,
$$\begin{aligned}{}[M]\in \hat{\Lambda }(f-\mathrm{torsionfree}) \Leftrightarrow [M_f]\in \hat{A}_{1,f}(\Lambda -\mathrm{socle}) \end{aligned}$$
and \(M=\mathrm{soc}_\Lambda (M_f)\simeq \Lambda /\Lambda \cap \mathfrak {m}\) for some maximal left ideal \(\mathfrak {m}\) of \(A_{1,f}\). By Corollary 2.7, either \(M_f\simeq A_1/A_1X\) (0 is not a root of f) or \(M_f\in \hat{A}_{1,f}(D-\mathrm{torsionfree}, \Lambda -\mathrm{socle})\). In the first case, \(M=\mathrm{soc}_\Lambda (M_f)=M_f=A_1/A_1X\) (Corollary 2.7).
In the second case, by Theorems 2.4 and 2.9,
$$\begin{aligned} M_f\simeq (\mathcal{M}_b)_f=A_{1,f}/A_{1,f}\cap B_fb \end{aligned}$$
for some l-normal irreducible element b from Theorem 2.9. Note that the left ideal \(\mathfrak {m}=A_{1,f}\cap B_fb\) of \(A_{1,f}\) is maximal. By Lemma 2.3 and Theorem 2.9, \([\mathcal{M}_b]\in \hat{A}_1(D-\mathrm{torsionfree}, f-\mathrm{torsionfree})\) iff the second condition of the theorem holds. Now,
$$\begin{aligned} \mathrm{soc}_\Lambda (M_f)=\mathrm{soc}_\Lambda (\mathcal{M}_b)_f=\mathrm{soc}_\Lambda (\Lambda /\Lambda \cap A_{1,f}\cap B_fb)=\mathrm{soc}_\Lambda (\Lambda /\Lambda \cap B_fb). \end{aligned}$$
(7)
By Theorem 2.8 and by the Remark after it,
$$\begin{aligned} \mathrm{soc}_\Lambda (M_f)\ne 0\;\; \mathrm{iff}\;\; \Lambda =\Lambda f+\Lambda \cap (A_{1,f}\cap B_fb)= \Lambda f+\Lambda \cap B_fb. \end{aligned}$$
In this case,
$$\begin{aligned} \mathrm{soc}_\Lambda (M_f)=\Lambda /\Lambda \cap (A_{1,f}\cap B_fb)=\Lambda /\Lambda \cap B_fb. \end{aligned}$$
Let us show that (in this case) the natural \(\Lambda \)-module epimorphism
$$\begin{aligned} \varphi :M_b=\Lambda /\Lambda \cap Bb\rightarrow \Lambda /\Lambda \cap B_fb, \;\; \lambda +\Lambda \cap Bb\rightarrow \lambda +\Lambda \cap B_fb, \end{aligned}$$
is an isomorphism. Note that
$$\begin{aligned} \mathrm{ker}\, \varphi =\Lambda \cap B_fb/\Lambda \cap Bb. \end{aligned}$$
The \(A_1\)-module \(\mathcal{M}_b\) is a submodule of \((\mathcal{M}_b)_f\simeq M_f\). So,
$$\begin{aligned} \mathrm{soc}_\Lambda (M_f)=\mathrm{soc}_\Lambda (\mathcal{M}_b)=\mathrm{soc}_\Lambda (\Lambda /\Lambda \cap Bb). \end{aligned}$$
By assumption \(\mathrm{soc}_\Lambda (M_f)\ne 0\), then it is a simple essential f-torsionfree \(\Lambda \)-submodule of \(M_f\). If \(\mathrm{ker}\, \varphi \ne 0\), then \(\mathrm{soc}_\Lambda (M_f)\subseteq \mathrm{ker}\, \varphi \), but \(\mathrm{ker}\, \varphi \) is an f-torsion \(\Lambda \)-module, a contradiction.
Let \(\mathcal{M}_b \) and \(\mathcal{M}_c\) be as in the theorem. By Lemma 2.2, \(\mathcal{M}_b \simeq \mathcal{M}_c\) as \(\Lambda \)-modules \(\Leftrightarrow \)\(A_{1,f}\otimes _\Lambda \,\mathcal{M}_b \simeq A_{1,f}\otimes _\Lambda \, \mathcal{M}_c\) as \(A_{1,f} \)-modules. Since
$$\begin{aligned} A_{1,f}\otimes _\Lambda \,\mathcal{M}_b \simeq A_{1,f}/A_{1,f}\cap B_fb\simeq (\mathcal{M}_b)_f, \end{aligned}$$
by Theorem 2.4, the above \(A_{1,f}\)-modules are isomorphic iff \(\mathcal{M}_b \simeq \mathcal{M}_c\) as \(A_1\)-modules, so, by Theorem 2.9, \(B/Bb\simeq B/Bc\) as B-modules.

The condition 1 of the theorem is equivalent to the condition that \(\Lambda = \Lambda (X-\lambda _i)+\Lambda \cap B_fb \; (= \Lambda (X-\lambda _i)+\Lambda \cap Bb )\) for all roots \(\lambda _i\) of the polynomial f (since the elements \(X-\lambda _i\) are regular normal elements of \(\Lambda \) and \(\lambda _i\) are the roots of f).

By Lemma 2.1, each simple f-torsionfree \(\Lambda \)-module is infinite dimensional. If 0 is not a root of f, then the modules \(M=A_1/A_1X\) and \(\mathcal{M}_b\) (from the theorem) are not isomorphic, since the linear map \(X_M: M\rightarrow M\), \(m\mapsto Xm\) is locally nilpotent but \(\mathrm{ker}\, X_{\mathcal{M}_b}=0\). \(\square \)

3 The Prime Ideals, the Krull and Global Dimensions of the Algebra \(\Lambda \)

In this section, K is a field of characteristic zero (not necessarily algebraically closed) and \(f=p_1^{n_1}\cdots p_s^{n_s}\) is a nonscalar polynomial of K[X] where \(p_1, \ldots , p_s\) are irreducible, co-prime divisors of f (i.e. \(K[X]p_i+K[X]p_j=K[X]\) for all \(i\ne j\)). The aim of this section is to give a proof of Theorem 1.1.

Proof of Theorem 1.1

3. The elements \(p_1, \ldots , p_s\) are regular normal elements of the algebra \(\Lambda \) since
$$\begin{aligned} Yp_i=p_i(Y-p_i^{-1}f)\; \; \mathrm{and}\;\; Xp_i=p_iX. \end{aligned}$$
4. The algebra \(\Lambda \) is a domain, hence \(0\in \mathrm{Spec}_c (\Lambda )\).
Since
$$\begin{aligned} \Lambda / \Lambda p_i\simeq F_i[Y] \end{aligned}$$
(8)
is a polynomial algebra with coefficients in the field \(F_i\) (since \(YX-XY=f\in \Lambda p_i\)), the ideal \(\Lambda p_i\) is a completely prime ideal of \(\Lambda \).

By (3), \(\Lambda _f=A_{1,f}\) is a simple algebra (as a localization of a simple Noetherian algebra). If \(\mathfrak {p}\) is a nonzero prime ideal of the algebra \(\Lambda \) then \(f^n\in \mathfrak {p}\) for some natural number \(n\ge 1\). Hence, \(p_i\in \mathfrak {p}\) for some i, by statement 3. By (8), \(\mathfrak {p}= (p_i, g_i)\) for some monic irreducible polynomial \(g_i\) of the polynomial algebra \(F_i[Y]\).

1. By [11, Theorem 6.5.4.(i)], \(\mathrm{Kdim } (\Lambda ) \le \mathrm{Kdim } (K[X])+1=1+1=2\).

Since \(p_i\) is a regular normal element of the algebra \(\Lambda \),
$$\begin{aligned} \mathrm{Kdim } (\Lambda ) \ge \mathrm{Kdim } (\Lambda / \Lambda p_i)+1{\mathop {=}\limits ^{(8)}} \mathrm{Kdim } ( F_i[Y])+1=1+1=2, \end{aligned}$$
by [11, Theorem 6. 5.9]. Therefore, \(\mathrm{Kdim } (\Lambda )=2\).

2. By [11, Theorem 7.5.3.(i)], \(\mathrm{gldim}(\Lambda ) \le \mathrm{gldim}(K[X])+1=1+1=2\).

By (8), \(\mathrm{gldim}(\Lambda / \Lambda p_i)= \mathrm{gldim}( F_i[Y])=1<\infty \). Now, by [11, Theorem 7.3.5.(i)],
$$\begin{aligned} \mathrm{gldim}(\Lambda ) \ge \mathrm{gldim}(\Lambda / \Lambda p_i)+1{\mathop {=}\limits ^{(8)}} \mathrm{gldim}( F_i[Y])+1=1+1=2. \end{aligned}$$
Therefore, \(\mathrm{gldim}(\Lambda )=2\). \(\square \)

Notes

References

  1. 1.
    Bavula, V.V.: The finite-dimensionality of Ext\({}^n\)’s and Tor\({}_n\)’s of simple modules over a class of algebras. Funktion. Anal. i Prilozhen. 25(3), 80–82 (1991)MathSciNetGoogle Scholar
  2. 2.
    Bavula, V.V.: Generalized Weyl algebras and their representations. Algebra i Analiz 4(1), 75–97 (1992). English transl. in St.Petersburg Math. J. 4(1) (1993), 71–92Google Scholar
  3. 3.
    Bavula, V.V.: Generalized Weyl algebras, kernel and tensor-simple algebras, their simple modules, representations of algebras. In: Dlab, V., Lenzing, H. (eds.) Sixth International Conference, August 19–22, 1992. CMS Conference proceedings, vol. 14, pp. 83–106 (1993)Google Scholar
  4. 4.
    Bavula, V.V.: The simple modules of the ore extensions with coefficients from Dedekind ring. Commun. Algebra 27(6), 2665–2699 (1999)MathSciNetCrossRefGoogle Scholar
  5. 5.
    Bavula, V.V., Lu, T.: Prime ideals of the enveloping algebra of the Euclidean algebra and a classification of its simple weight modules. J. Math. Phys. 58(1), 011701 (2017).  https://doi.org/10.1063/1.4973378 MathSciNetCrossRefzbMATHGoogle Scholar
  6. 6.
    Bavula, V.V., Lu, T.: Classification of simple weight modules over the Schrödinger algebra. Canad. Math. Bull. 61(1), 16–39 (2018)MathSciNetCrossRefGoogle Scholar
  7. 7.
    Bavula, V.V., Lu, T.: The universal enveloping algebra \(U(_2\ltimes {V}_2)\), its prime spectrum and a classification of its simple weight modules. J. Lie Theory 28(2), 525–560 (2018)MathSciNetzbMATHGoogle Scholar
  8. 8.
    Bavula, V.V., Cai, Y., Lü, R., Tan, H., Zhao, K.: The Polyonic Algebras and their simple modules (work in progress) Google Scholar
  9. 9.
    Block, R.E.: The irreducible representations of the Lie algebra \(sl(2)\) and of the Weyl algebra. Adv. Math. 39, 69–110 (1981)MathSciNetCrossRefGoogle Scholar
  10. 10.
    Dixmier, J.: Sur les algébres de Weyl, II. Bull. Sci. Math. 94(2), 289–301 (1970)MathSciNetzbMATHGoogle Scholar
  11. 11.
    McConnell, J.C., Robson, J.C.: Noncommutative Noetherian rings. With the cooperation of L. W. Small. Revised edition. In; Graduate Studies in Mathematics, 30. American Mathematical Society, Providence, RI (2001)Google Scholar

Copyright information

© The Author(s) 2019

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors and Affiliations

  1. 1.School of Mathematics and StatisticsUniversity of SheffieldSheffieldUK

Personalised recommendations