Mori 2015

Math. Z.
DOI 10.1007/s00209-015-1407-x Mathematische Zeitschrift
Regular modules over 2-dimensional quantum Beilinson

algebras of Type S
Izuru Mori
Received: 5 November 2014 / Accepted: 13 November 2014

© Springer-Verlag Berlin Heidelberg 2015
Abstract In the study of a finite dimensional hereditary algebra of infinite representation

type, understanding regular modules is essential. Recently, Herschend, Iyama and Oppermann
introduced the notions of d-representation infinite algebra and d-regular module, extending
the above notions to finite dimensional algebras of global dimension d ≥ 1. Since the
Beilinson algebras of AS-regular algebras of dimension d + 1 are typical examples of d-
representation infinite algebras, the purpose of this paper is to study the behavior of d-regular
modules over such algebras. In particular, we will show that the isomorphism classes of
simple 2-regular modules over a 2-representation tame quantum Beilinson algebra of Type
S are parameterized by P2 .
Keywords Regular modules · Beilinson algebras · AS-regular algebras · Representation

infinite algebras · Preprojective algebras
Mathematics Subject Classification 16G60 · 16S38 · 16S36 · 16W50 · 14A22
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1 Notations and terminologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Classes of algebras to study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 The category of d-regular modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1 The conditions (CM) and (LGK) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 The category of d-regular modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 2-Dimensional weighted polynomial algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1 Points of a noncommutative projective scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . .
This work was supported by Grant-in-Aid for Scientific Research (C) 25400037.
I. Mori (B)
Department of Mathematics, Graduate School of Science, Shizuoka University,
Shizuoka 422-8529, Japan
e-mail: [email protected]
123
I. Mori
3.2 Points of a weighted projective line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.3 Regular modules and regular components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4 3-Dimensional quantum polynomial algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1 Geometric algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 d-Representation tame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 2-Regular modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4 2-Regular components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 Introduction
In representation theory of finite dimensional algebras, hereditary algebras and preprojective

algebras are important classes of algebras to study [8,9,28], while in noncommutative alge-
braic geometry, AS-regular algebras are the most important class of algebras to study [4–6].
In fact, the classification of 3-dimensional AS-regular algebras is one of the starting points of
noncommutative algebraic geometry, and they were classified into several types [4,5]. The
Type S in the title of this paper refers to a 3-dimensional AS-regular algebra whose point
scheme is a triangle in P2 . By extending the notions of these classes of algebras, namely,
hereditary algebras, preprojective algebras and AS-regular algebras, they are closely related
as follows [17]: given an AS-regular algebra A of dimension d + 1, we can construct a finite
dimensional algebra ∇ A of global dimension d, called the Beilinson algebra of A, which
is a typical example of a quasi (extremely) Fano algebra of dimension d defined in [16]. A
quasi (extremely) Fano algebra of dimension d was recently renamed as a d-representation
infinite algebra in [11], since a quasi (extremely) Fano algebra of dimension 1 is exactly a
finite dimensional hereditary algebras of infinite representation type [16]. On the other hand,
the notion of preprojective algebra R of a finite dimensional hereditary algebra R was
also extended to an arbitrary finite dimensional algebra R of finite global dimension in [14],
and the preprojective algebra R of a quasi (extremely) Fano algebra R turns out to be an
AS-regular algebra over R defined in [17], which extends the notion of AS-regular algebra to
nonconnected N-graded algebras. By these connections, the above three classes of algebras
became more important to study not only in noncommutative algebraic geometry but also
in representation theory of finite dimensional algebras. In particular, we expect to transfer
many results of AS-regular algebras of dimension d + 1 to those of d-representation infinite
algebras and vice versa.
The purpose of this paper is to study d-regular modules over d-dimensional Beilinson
algebras using the techniques of noncommutative algebraic geometry. First, we will show
that the category of d-regular modules over a Beilinson algebra ∇ A is abelian if A is a
finitely generated graded right coherent AS-regular algebra satisfying the Cohen–Macaulay
property with respect to GKdimension (Theorem 2.9). Then we will give a new proof for
the well-known result that the regular components of a path algebra of a quiver of type

A are parameterized by P1 using the techniques of noncommutative algebraic geometry
(Theorem 3.19). The main part of this paper focuses on detailed study of quantum Beilinson
algebras of 3-dimensional AS-regular algebras of Type S, which are the path algebras of the
2-dimensional Beilinson quiver with quantum relations. Although 2-dimensional quantum
Beilinson algebras are the simplest examples of 2-representation infinite algebras other than
the usual 2-dimensional Beilinson algebra with commutative relations, the behavior of 2-
regular modules largely depends on the quantum relations. In particular, we will show that if a
2-dimensional quantum Beilinson algebra ∇ A is 2-representation tame, then the isomorphism
classes of simple 2-regular modules are parameterized by P2 , while if it is not 2-representation
123
Regular modules over quantum Beilinson algebras
tame, then they are parameterized by the point scheme of A, which is not P2 in general
(Theorem 4.21). We will also give a characterization for a 2-dimensional quantum Beilison
algebra of Type S to be 2-representation tame using the techniques of noncommutative
algebraic geometry (Theorem 4.17). We hope to study more general Beilinson algebras in
the future.
1.1 Notations and terminologies
First, we fix some notations and terminologies. Throughout this paper, we fix an algebraically
closed field k of characteristic 0. An algebra means an algebra over k. For an algebra R,
we denote by Mod R the category of right R-modules, and by mod R the full subcategory
consisting of finitely presented modules. The category of left R modules is identified with
Mod R o where R o is the opposite algebra of R, and the category of R-R bimodules is identified
with Mod R e where R e := R o ⊗k R is the enveloping algebra of R. For M ∈ Mod R, define
D M := Homk (M, k) ∈ Mod R o . If R is a commutative noetherian algebra, then mod R is
equivalent to the category coh Spec R of coherent sheaves on Spec R. By this reason, for a
noncommutative algebra R, we use the notation Specnc R for an imaginary “locally ringed
space” whose category of “coherent sheaves” is mod R.
In this paper, a graded algebra means a Z-graded algebra A = i∈Z Ai over a field k
although we mainly deal with an N-graded algebra A = ⊕i∈N Ai . We denote by GrMod A
the category of graded right A-modules, and by grmod A the full subcategory consisting of
finitely presented modules. Morphisms in GrMod A are right A-module homomorphisms
preserving degrees. For M ∈ GrMod A, define the Matlis dual of M by D M ∈ GrMod Ao
where (D M)i = D(M−i ). Recall that a graded algebra A is called graded right coherent if
every finitely generated homogeneous right ideal of A is finitely presented. It is known that
A is graded right coherent if and only if grmod A is an abelian category.
Let M = ⊕i∈Z Mi ∈ GrMod A be a graded module. We say that M is locally finite
if dimk Mi < ∞ for all i ∈ Z. Note that if A is a finitely generated N-graded algebra
with dimk A0 < ∞, then A is locally finite, and if A is locally finite, then every finitely
generated graded right A-module is locally finite. The Hilbert series of a locally finite module
M ∈ GrMod A is defined by

HM (t) := (dimk Mi )t i ∈ Z[[t, t −1 ]].
i∈Z

For an integer n ∈ Z, we define the truncation M≥n := i≥n Mi , which is a graded submod-
ule of M if A is N-graded, and the shift M(n) ∈ GrMod A by M(n)i := Mn+i for i ∈ Z. We
say that M is left bounded if M = M≥n for some n ∈ Z, and right bounded if M≥n = 0 for
some n ∈ Z. The rule M → M(n) is a k-linear autoequivalence for GrMod A, called the shift
functor. For M, N ∈ GrMod A, we write the vector space ExtiA (M, N ) := ExtiGrMod A (M, N )
and the graded vector space

ExtiA (M, N ) := ExtiA (M, N (n)).
n∈Z
If A is N-graded, then we define the left exact functor m : GrMod A → GrMod A by

m M := lim Hom A (A/A≥n , M)
n→∞
where m = A≥1 . The cohomologies of its right derived functor are denoted by
Him (M) := Ri m M = lim ExtiA (A/A≥n , M).
n→∞
123
I. Mori
Let M ∈ GrMod A. An element m ∈ M is called torsion if m A ∈ GrMod A is right

bounded. We say that M is torsion-free if 0 is the only torsion element of M. The torsion
submodule τ M of M is the smallest submodule of M such that M/τ M is torsion-free. We
say that M is torsion if M = τ M. We denote by Tors A the full subcategory of GrMod A
consisting of torsion modules. Since Tors A is a Serre subcategory of GrMod A, we can
define the quotient category Tails A := GrMod A/ Tors A. The quotient functor is denoted
by π : GrMod A → Tails A, and its right adjoint is denoted by ω : Tails A → GrMod A so
that πω ∼= IdTails A . We often denote by M = π M ∈ Tails A the image of M ∈ GrMod A.
Note that the k-linear autoequivalence M → M(n) preserves torsion modules, so it induces
a k-linear autoequivalence M → M(n) for Tails A, again called the shift functor. For
M, N ∈ Tails A, we write the vector space ExtiA (M, N ) := ExtiTails A (M, N ) and the
graded vector space

ExtiA (M, N ) := ExtiA (M, N (n))
n∈Z
as before.
If A is a graded right coherent algebra, then we define the quotient category tails A :=
grmod A/ tors A where tors A := Tors A ∩ grmod A. If A is a commutative graded algebra
finitely generated in degree 1 over k, then tails A is equivalent to the category coh Proj A of
coherent sheaves on Proj A by Serre. By this reason, for a noncommutative graded algebra
A, we use the notation Projnc A for an imaginary “locally ringed space” whose category of
“coherent sheaves” is tails A.
1.2 Classes of algebras to study
We now define several classes of algebras which

play essential roles in this paper. A connected
graded algebra is an N-graded algebra A = i∈N Ai such that A0 = k. For a connected
graded algebra A, m := A≥1 is the unique maximal homogeneous two-sided ideal of A,
and we view k = A/m ∈ GrMod A as a graded A-module. In noncommutative algebraic
geometry, AS-regular algebras defined below are the most important class of algebras to
study.
Definition 1.1 A connected graded algebra A is called AS-regular (AS-Gorenstein) of
dimension d and of Gorenstein parameter if
(1) gldim A = d < ∞ (id A A = id Ao A = d < ∞), and

(2) Ext A (k, A) =
i i ∼ k() if i = d
∼ Ext o (k, A) =
A
0 if i = d.
A noetherian AS-regular algebra such that H A (t) = 1/(1 − t)d is called a d-dimensional
quantum polynomial algebra.
It is known that a d-dimensional quantum polynomial algebra is an AS-regular algebra of
dimension d and of Gorenstein parameter d.
Example 1.2 Commutative AS-regular algebras are exactly weighted polynomial algebras
and commutative quantum polynomial algebras are exactly polynomial algebras generated in
degree 1. By this reason, we often call Projnc A a quantum projective space of dimension d if A
is a (d+1)-dimensional quantum polynomial algebra. The simplest noncommutative example
of an AS-regular algebra is a weighted skew polynomial algebra A = k x 1 , . . . , xd /(x j xi −
αi j xi x j ) where αi j ∈ k such that αii = αi j α ji = 1 for all 1 ≤ i, j ≤ d.
123
On the other hand, in representation theory of finite dimensional algebras, finite dimen-
sional hereditary algebras and their preprojective algebras are important classes of algebras
to study. Recently, Herschend et al. [11] extended the notion of finite dimensional hereditary
algebra to finite dimensional algebras of global dimension d ≥ 1 and introduced the notion of
d-representation infinite algebra, which is exactly same as a quasi (extremely) Fano algebra
of dimension d introduced in [16].
For an abelian category C , we denote by D(C ) the derived category of C , and by Db (C )
the bounded derived category of C .
Definition 1.3 [11, Definition2.7, Proposition 4.15] Let R be a finite dimensional algebra
of gldim R = d < ∞. We define an autoequivalence νd ∈ Aut k Db (mod R) by νd (M) :=
M ⊗LR D R[−d]. We say that R is d-representation infinite if νd−i (R) ∈ mod R for all i ∈ N.
A d-regular module over a d-representation infinite algebra R is an R-module M ∈ mod R
such that νdi (M) ∈ mod R for all i ∈ Z. The full subcategory of mod R consisting of d-regular
modules is denoted by R(R).
The preprojective algebra of a d-representation infinite algebra R is the graded algebra
defined by R := TR (ExtdR (D R, R)) with the tensor algebra grading.
We will see now that AS-regular algebras, d-representation infinite algebras and prepro-
jective algebras are strongly related. Let A be a graded algebra and r ∈ N+ . Recall that the
r th Veronese algebra of A is a graded algebra defined by

A(r ) := Ari .
i∈Z
Definition 1.4 [23, Definition 3.7] Let A be a graded algebra and r ∈ N+ . The r th quasi-
Veronese algebra of A is a graded algebra defined by
⎛ ⎞
A(r ) A(1)(r ) · · · A(r − 1)(r )
⎜ A(−1) (r ) A (r ) · · · A(r − 2) ⎟ (r )
⎜ ⎟
A[r ] = ⎜ .. .. .. .. ⎟
⎝ . . . . ⎠
A(−r + 1) (r ) A(−r + 2) (r ) ··· A (r )
r −1
with the multiplication (ai j )(bi j ) := ( k=0 ak j bik ).
Although GrMod A and GrMod A(r ) are hardly ever equivalent, the functor Q :
GrMod A → GrMod A[r ] defined by
⎛ ⎞
M(r − 1)(r )
⎜ .. ⎟
⎜ . ⎟
Q(M) := ⎜ ⎟
⎝ M(1)(r ) ⎠
M (r )
−1
with the action (m i )(bi j ) := ( rk=0 m k bik ) is an equivalence functor for any r ∈ N+ by [23,
Lemma 3.9].
There is another typical graded Morita equivalence. For a graded algebra A and a graded
algebra automorphism θ ∈ Aut k A, we define a graded algebra Aθ by Aθ = A as a graded
k-vector space with the new multiplication a ∗ b := aθ i (b) for a ∈ Ai , b ∈ A j . Similarly,
for M ∈ GrMod A, we define M θ ∈ GrMod Aθ by M θ = M as a graded k-vector space
with the new action m ∗ b := mθ i (b) for m ∈ Mi , b ∈ A j . This gives an equivalence functor
123
I. Mori
(−)θ : GrMod A → GrMod Aθ . Since M ∈ Tors A if and only if M θ ∈ Tors Aθ , it induces

an equivalence functor (−)θ : Tails A → Tails Aθ .
There is another functor induced by a graded algebra automorphism θ ∈ Aut k A. For
M ∈ GrMod A, we define Mθ ∈ GrMod A by Mθ = M as a graded k-vector space with the
new action m ∗ a = mθ (a). This defines an autoequivalence (−)θ : GrMod A → GrMod A,
which induces an autoequivalence (−)θ : Tails A → Tails A. If A is an AS-regular algebra
of dimension d and of Gorenstein parameter , then the canonical module of A is defined
by K A := D Hdm (A). By [17, Theorem 3.12], there exists a graded algebra automorphism
ν ∈ Aut k A, called the generalized Nakayama automorphism, such that K A ∼= Aν −1 (−) as
graded A-A bimodules. We say that A is symmetric if ν = id. We define an autoequivalence
− ⊗L A KA ∈ Aut k D (Tails A) by M ⊗A KA := Mν −1 (−).
L
Definition 1.5 [17, Definition 4.7] The Beilinson algebra of an AS-regular algebra A of
Gorenstein parameter is defined by
⎛ ⎞
A0 A1 · · · A−1
⎜ 0 A0 · · · A−2 ⎟
⎜ ⎟
. ⎟∼
−1
∇ A := (A[] )0 = ⎜ . . . = End A (⊕i=0 A(−i)).
⎝ .. .. . . .. ⎠
0 0 · · · A0
In this paper, if A is a quantum polynomial algebra, then we call ∇ A a quantum Beilinson

algebra by abuse of language. The following theorem is a main source of ideas for this paper.
Theorem 1.6 [17, Theorem 4.12, Theorem 4.14] Let A be an AS-regular algebra of dimen-
sion d + 1 and of Gorenstein parameter . Then the following holds.
(1) The Beilinson algebra ∇ A is a d-representation infinite algebra.

(2) There exists a graded algebra automorphism ν [] ∈ Aut k A[] naturally induced by the
[]
generalized Nakayama automorphism ν ∈ Aut k A such that ∇ A ∼ = (A[] )ν , so that
the composition of functors
[]
Q (−)ν
GrMod A −−−−→ GrMod A[] −−−−→ GrMod ∇ A
gives an equivalence of categories.
(3) The functor
−1
RHomA (⊕i=0 A(−i), −) : D(Tails A) → D(Mod ∇ A)
gives an equivalence of triangulated categories.

(4) If A is graded right coherent, then the above functor restricts to an equivalence of
triangulated categories
−1
RHomA (⊕i=0 A(−i), −) : Db (tails A) → Db (mod ∇ A).
The above theorem provides strong interactions between noncommutative algebraic geom-
etry and representation theory of finite dimensional algebras. We expect to transfer many
results of AS-regular algebras of dimension d + 1 to those of d-representation infinite alge-
bras. The purpose of this paper is to study d-regular modules over d-dimensional Beilinson
algebras using the techniques of noncommutative algebraic geometry.
123
2 The category of d-regular modules
It is not clear from the definition that the category of d-regular modules is abelian. In this
section, we will show that the category of d-regular modules over a Beilinson algebra ∇ A of
a finitely generated graded right coherent AS-regular algebra A is abelian if A satisfies the
condition (CM).
2.1 The conditions (CM) and (LGK)
Definition 2.1 Let A be a graded algebra. For M ∈ GrMod A, we define the following
numbers:
• the Gelfand–Kirillov dimension (GKdimension for short) of M by
GKdim M := inf{ρ ∈ R+ | dimk Mi ≤ i ρ−1 for all i 0}.
• the grade of M by
j (M) := inf RHom A (M, A) = inf{i ∈ Z | ExtiA (M, A) = 0}.
If A is N-graded, then we define the following numbers:
• the depth of M by
depth M := inf R m (M) = inf{i ∈ Z | Him (M) = 0}.
• the local dimension of M by
ldim M := sup R m (M) = sup{i ∈ Z | Him (M) = 0}.
We say that M is Cohen–Macaulay if depth M = ldim M < ∞, that is, for some m ∈ N,
Him (M) = 0 for all i = m and Hm
m (M) = 0.
Let A be a graded algebra and M ∈ GrMod A. By the definition of GKdimension, if M

is right bounded, then GKdim M = 0, and if M is not right bounded, then GKdim M ≥ 1,
so GKdim M will not take a value strictly between 0 and 1. If A is a locally finite connected
graded algebra and M ∈ GrMod A, then it is known that
depth M = inf RHom A (k, M) = inf{i ∈ Z | ExtiA (k, M) = 0}.
Definition 2.2 We say that a graded right coherent algebra A satisfies the Cohen–Macaulay
property with respect to GKdimension (CM) if
j (M) + GKdim M = GKdim A
for all 0 = M ∈ grmod A.
Example 2.3 We expect that every noetherian AS-regular algebra satisfies (CM). In fact, an
iterated graded Ore extension of k is a typical example of a noetherian AS-regular algebra
satisfying (CM) by [15, Lemma p184]. In particular, every weighted skew polynomial algebra
satisfies (CM). Every noetherian AS-regular algebra of dimension 3 satisfies (CM) by [6,
Theorem 4.1], and many examples of 4-dimensional quantum polynomial algebras are known
to satisfy (CM) (cf. [32, Corollary 1.6], [27, Corollary 1.8]).
Definition 2.4 We say that an N-graded right coherent algebra A satisfies (LGK) if ldim M =
GKdim M for all 0 = M ∈ grmod A.
123
I. Mori
We expect that every noetherian AS-regular algebra satisfies (LGK) by the following
theorem. Note that a connected graded right coherent algebra A is finitely generated as a
graded algebra if and only if k ∈ grmod A is finitely presented as a graded right A-module
if and only if A is Ext-finite, that is, dimk ExtiA (k, k) < ∞ for all i ∈ N (cf. [24]).
Theorem 2.5 Let A be either

(1) a noetherian AS-Gorenstein algebra, or
(2) a finitely generated graded right coherent AS-regular algebra.
Then A satisfies (CM) if and only if A satisfies (LGK).
Proof If A is of dimension d and of Gorenstein parameter , then D R m (A) ∼

= A(−)[d] in
D(GrMod A) by [12, Theorem 1.2] in the first case and by [17, Theorem 3.12] in the second
case. Since A is Ext-finite in either case,
D R m (M) ∼
= RHom A (M, D R m (A)) ∼
= RHom A (M, A)(−)[d]
by local duality [33, Theorem 5.1].
Since
ldim A := sup R m (A) = sup D(A(−)[d]) = − inf A[d] = d,
we have
j (M) := inf RHom A (M, A) = inf(D R m (M))[−d]()
= inf D(R m (M)[d]) = − sup R m (M)[d]
= d − sup R m (M) = ldim A − ldim M,
so if A satisfies (LGK), then A satisfies (CM).
Conversely, suppose that A satisfies (CM). Since k ∈ grmod A in either case,
GKdim A = j (k) + GKdim k = j (k) = ldim A − ldim k = ldim A.
For M ∈ grmod A,
GKdim M = GKdim A − j (M) = ldim A − (ldim A − ldim M) = ldim M,
hence A satisfies (LGK).

2.2 The category of d-regular modules
Let A be an N-graded right coherent algebra. In [11], the full subcategory grmod≤i A =
{M ∈ grmod A | ldim M ≤ i} of grmod A and the quotient category tailsi A :=
grmod≤i+1 A/ tors A were introduced for each i ∈ N to prove the following theorem.
Theorem 2.6 [11, Theorem 6.5] If R is a d-representation infinite algebra such that its
preprojective algebra T := R is graded right coherent, then the equivalence functors
πT (− ⊗LR T ) : Db (mod R) ←→ Db (tails T ) : RHomT (T , −)
quasi-inverse to each other restrict to equivalence functors
πT (− ⊗ R T ) : R(R) ←→ tails0 T : HomT (T , −)
quasi-inverse to each other.
123
Let A be a graded right coherent algebra. We define another full subcategory grmod≤i A =
{M ∈ grmod A | GKdim M ≤ i} of grmod A for each i ∈ N. For a short exact sequence
0→L→M →N →0
in grmod A, GKdim M = max{GKdim L , GKdim N }, so M ∈ grmod≤i A if and only if

L , N ≤ grmod≤i A. It follows that grmod≤i A is a Serre subcategory of grmod A, so we can
define the quotient categories tailsi A := grmod≤i+1 A/ grmod≤i A.
Lemma 2.7 If A is a graded right coherent algebra, then tailsi A is an abelian category for
every i ∈ N.
Proof Since grmod≤i+1 A is a Serre subcategory of an abelian category grmod A, it

is an abelian category. Since grmod≤i A is a Serre subcategory of an abelian category
grmod≤i+1 A, tailsi A := grmod≤i+1 A/ grmod≤i A is an abelian category.

Lemma 2.8 If A is a finitely generated connected graded right coherent algebra satisfying
(LGK), then grmod≤i A = grmod≤i A for every i ∈ N. In particular, tails0 A = tails0 A.
Proof By the condition (LGK), grmod≤i A = grmod≤i A for every i ∈ N. Since A is finitely
generated connected graded algebra, every M ∈ grmod A is locally finite and left bounded.
It follows that grmod≤0 A = tors A is the full subcategory of grmod A consisting of finite
dimensional modules, so
tails0 A := grmod≤1 A/ tors A = grmod≤1 A/ grmod≤0 A =: tails0 A.

Theorem 2.9 Let A be a finitely generated graded right coherent AS-regular algebra.
(1) For every i ∈ N, tailsi A ∼
= tailsi ∇ A and tailsi A ∼
= tailsi ∇ A.
(2) The equivalence functor
−1
RHomA (⊕i=0 A(−i), −) : Db (tails A) → Db (mod ∇ A)
restricts to an equivalence functor

−1
HomA (⊕i=0 A(−i), −) : tails0 A → R(∇ A).
(3) If A satisfies (CM), then R(∇ A) ∼

= tails0 A is an abelian category.
Proof Let be the Gorenstein parameter of A, ν ∈ Aut k A the generalized Nakayama

[]
automorphism of A, T := ∇ A, and n := T≥1 so that T ∼
= (A[] )ν by Theorem 1.6 (2).
(1) The composition of equivalence functors
[]
Q (−)ν
F : GrMod A −−−−→ GrMod A[] −−−−→ GrMod T
restricts to an equivalence functor F : grmod A → grmod T . For any M ∈ grmod A,
−1
dimk F(M) j = i=0 Mj+i , so GKdim F(M) = GKdim M. It follows that F restricts to
an equivalence functor F : grmod≤i A → grmod≤i T , so F induces an equivalence functor
F̄ : tailsi A → tailsi T for every i ∈ N.
We now show that ldim F(M) = ldim M for any M ∈ grmod A. Since A is locally
finite, every M ∈ Tors A is the sum of finite dimensional submodules of M. Since an
123
I. Mori
equivalence functor F commutes with direct limits, F restricts to an equivalence functor

F : Tors A → Tors T , so F induces the following commutative diagram
F
GrMod A −−−−→ GrMod T
∼
=
π A ↓↑ ω A πT ↓↑ ωT
F̄
Tails A −−−−→ Tails T.
∼
=
By [10, Lemma 4.1.5], the exact sequence
0 → m M → M → ω A π A M
in GrMod A induces the following diagram with exact rows
0 −−−−→ F( m M) −−−−→ F(M) −−−−→ F(ω A π A M)

0 −−−−→ n F(M) −−−−→ F(M) −−−−→ ωT πT F(M)

in GrMod T , so F( M) ∼
m = F(M) in GrMod T . It follows that
n
ldim A M := sup R m M = sup F(R m M) = sup R n F(M) =: ldim T F(M)

for any M ∈ grmod A, so F restricts to an equivalence functor grmod≤i A → grmod≤i T
and induces an equivalence functor F̄ : tailsi A → tailsi T for every i ∈ N.
(2) Since A is graded right coherent, grmod T ∼= grmod A is an abelian category, so T
−1
is graded right coherent. Since F(⊕i=0 A(−i)) = T , we have the following commutative
diagram of equivalences
−1
RHomA (⊕i=0 A(−i),−)
Db (tails A) −−−−−−−−−−−−−−→ Db (mod ∇ A)
⏐
⏐ ⏐RHom (T ,−)
Q̄ ⏐ T
[]
(−)ν
Db (tails A[] ) −−−−→ Db (tails T ).
[]
Since F̄ = (−)ν ◦ Q̄ : tails0 A → tails0 T and HomT (T , −) : tails0 T → R(∇ A) are
equivalence functors, the above commutative diagram restricts to the following commutative
diagram of equivalences
−1
HomA (⊕i=0 A(−i),−)
tails0 A −−−−−−−−−−−−−→ R(∇ A)
⏐
⏐ ⏐Hom (T ,−)
Q̄ ⏐ T
[]
(−)ν
tails0 A[] −−−−→ tails0 T.
(3) If A satisfies (CM), then A satisfies (LGK) by Theorem 2.5, so R(∇ A) ∼
= tails0 A =
tails0 A is an abelian category by Lemma 2.8 and Lemma 2.7.

3 2-Dimensional weighted polynomial algebras
As a first step, we will study regular modules over the Beilinson algebra ∇ A of a 2-
dimensional noetherian AS-regular algebra A. It is known [29] that every 2-dimensional
123
noetherian AS-regular algebra A is of the form A ∼ = k x, y/( f ) for some homogeneous

element f ∈ k x, y. Let deg x = a, deg y = b ∈ N+ . If gcd(a, b) = d > 1, then
GrMod A ∼ = (GrMod B)d where B := A(d) = k x, y/( f ) with deg x = a/d, deg y = b/d.
Moreover, if ∇ A ∼ = k Q A and ∇ B = k Q B , then it is easy to see that Q A is the d disjoint union
of Q B , so studying A and ∇ A reduces to studying B and ∇ B. Since gcd(a/d, b/d) = 1,
we may assume that gcd(a, b) = 1. Moreover, since ∇ A ∼ = k A ∼
a+b−1 = ∇k[x, y] with
deg x = a, deg y = b by [24, Theorem 5.4], we will focus on studying a 2-dimensional
weighted polynomial algebra A = k[x, y]. Since ∇ A is the path algebra of an extended
Dynkin quiver A a+b−1 , the behavior of regular modules over ∇ A is well-known. In this sec-
tion, we will recover some of the well-known results using the techniques of noncommutative
algebraic geometry, namely, we will show that isomorphism classes of simple regular mod-
ules over ∇ A are parameterized by | Projnc A|, which is |P1 | together with a few more points,
while regular components of ∇ A are parameterized by | Proj A| = |P1 |. Although these
results are not new, the way of calculating regular modules explicitly using the techniques of
noncommutative algebraic geometry is rather interesting.
3.1 Points of a noncommutative projective scheme
There may be several ways of defining a notion of point of a noncommutative scheme. In

this subsection, we will define the notion of point of a noncommutative projective scheme
following [3,6,29,30] for the sake of this paper.
Definition 3.1 Let A be a graded algebra. We say that M ∈ GrMod A is i-critical if

GKdim M = i, and GKdim M/N < i for any non-zero graded submodule 0 = N ⊂ M.
It is easy to see the following properties of critical modules.
Lemma 3.2 Let A be a graded algebra. If M ∈ GrMod A is i-critical, then every non-zero
graded submodule 0 = N ⊂ M is i-critical.
Lemma 3.3 (cf. [6, Proposition 2.30(i)]) Let A be a right noetherian graded algebra and
M ∈ grmod A. If GKdim M = i, then there exists a graded submodule N ⊂ M such that
M/N is i-critical.
Definition 3.4 Let A be a graded algebra. We say that M, N ∈ GrMod A are equivalent if
πM ∼ = π N in Tails A. A point of Projnc A is an equivalence class of 1-critical modules. In
other words, a point of Projnc A is an isomorphism class of objects in Tails A of the form
π M where M ∈ GrMod A is a 1-critical module. The set of points of Projnc A is denoted by
| Projnc A|.
We denote by |X | the set of closed points of a commutative scheme X . It is easy to see

that if X ⊂ Pn−1 is a projective variety and A = k[x1 , . . . , xn ]/I (X ) is the homogeneous
coordinate ring of X so that X = Proj A, then the map p → π (k[x1 , . . . , xn ]/I ( p)) gives a
bijection from | Proj A| to | Projnc A| (see the next section), so when A is not commutative,
| Projnc A| can be thought of as the set of “closed points” of the noncommutative projective
scheme Projnc A. This notion of point is useful in this paper by the following lemma.
Lemma 3.5 If A is a right noetherian graded algebra, then a point of Projnc A is an isomor-
phism class of simple objects in tails0 A.
123
I. Mori
Proof Let π M ∈ Tails A be a point of Projnc A where M ∈ GrMod A is a 1-critical module.

For any non-zero homogeneous element 0 = m ∈ M, N := m A is a non-zero submodule of
M, so N is a cyclic 1-critical module by Lemma 3.2, in particular, N ∈ grmod≤1 A. Since
GKdim M/N = 0, π M ∼ = π N ∈ tails0 A, so we may assume that M ∈ grmod≤1 A. If
0 = π L ⊂ π M in tails0 A where L ∈ grmod≤1 A, then 0 = L ≥n ⊂ M for some n ∈ Z.
Since GKdim M/L ≥n = 0, it follows that π L ∼ = π L ≥n ∼
= π M, so π M is a simple object in
tails0 A.
Conversely, suppose that π M ∈ tails0 A is a simple object where M ∈ grmod≤1 A. By
Lemma 3.3, there exists a graded submodule N ⊂ M such that M/N is 1-critical. Since π M
is simple and π M/π N ∼ = π(M/N ) = 0, it follows that π N = 0, so π M ∼ = π M/π N ∼ =
π(M/N ) is a point of Projnc A.

Theorem 3.6 Let A be a right noetherian AS-regular algebra of dimension d + 1 satisfying

(CM). Then isomorphism classes of simple d-regular modules over ∇ A are parameterized
by | Projnc A|.
Proof This is an easy consequence of Theorem 2.9 (3) and Lemma 3.5.

In order to make more detailed analysis, we will distinguish three types of points depending
on their multiplicities. For this purpose, we now introduce a rather technical condition on a
graded algebra, which plays an important role in this paper.
Definition 3.7 Let A be a connected graded algebra. We say that A satisfies (H) if there exist
a1 , . . . , an ∈ N+ such that, for every M ∈ grmod A,
f M (t)
HM (t) = n
i=1 (1 − t )
ai
as rational functions for some f M (t) ∈ Z[t, t −1 ].
An algebra satisfying (H) was called geometric in [29]. We will not use this terminology
since a geometric algebra has a different meaning in this paper. It is easy to prove the following
lemma.
Lemma 3.8 The following algebras satisfy (H).

(1) A right noetherian connected graded algebra of finite global dimension.
(2) A finitely generated graded right coherent AS-regular algebra of finite GKdimension.
If A is a right noetherian connected graded algebra satisfying (H), then it is easy to see that
every graded quotient algebra of A also satisfies (H). It follows that every graded quotient
algebra of a right noetherian connected graded algebra of finite global dimension satisfies (H).
In particular, every noetherian commutative connected graded algebra satisfies (H) because
it is a graded quotient algebra of a weighted polynomial algebra. In general, GKdimension
may not be an integer, so the property (H) is essential in this paper by the following lemma.
Lemma 3.9 [31, Corollary 2.2] If A is a connected graded algebra satisfying (H) and M ∈
grmod A, then GKdim M is the order of the pole of HM (t) at t = 1. In particular, GKdim M ∈
N.
Definition 3.10 Let A be a connected graded algebra satisfying (H). We define the multi-
plicity of M ∈ grmod A by e(M) := limt→1− (1 − t)GKdim M HM (t).
123
The above definition of the multiplicity agrees with the usual one in nice cases.
Lemma 3.11 (cf. [6, Proposition 6.4]) Let A be a graded quotient algebra of a right
noetherian connected graded algebra S of finite global dimension such that HS (t) =
1/(1 − t)d . For any M ∈ grmod A with GKdim M = 1, dimk Mi = e(M) for all i 0.
Let A be a finitely generated connected graded right coherent algebra and M, N ∈

grmod A. Then π M ∼ = π N if and only if M≥n ∼
= N≥n in GrMod A for some n ∈ Z, so
it is easy to see the following lemma.
Lemma 3.12 Let A be a finitely generated connected graded right coherent algebra, and
M, N ∈ grmod A. If π M ∼
= π N , then GKdim M = GKdim N . In addition, if A satisfies
(H) and GKdim M ≥ 1, then e(M) = e(N ).
By the above lemma, the following definition makes sense.
Definition 3.13 (cf. [3]) Let A be a finitely generated connected graded right coherent algebra
satisfying (H).
(1) An ordinary point of Projnc A is an equivalence class of finitely generated 1-critical
modules of multiplicity 1.
(2) A fat point of Projnc A is an equivalence class of finitely generated 1-critical modules of
multiplicity > 1.
(3) A thin point of Projnc A is an equivalence class of finitely generated 1-critical modules
of multiplicity < 1.
If A is a right noetherian connected graded algebra satisfying (H), then every point of
Projnc A is of the form π M where M ∈ grmod≤1 A is cyclic and 1-critical by the proof of
Lemma 3.5, so every point is either ordinary, fat, or thin.
3.2 Points of a weighted projective line
Let A = k[x, y] be a 2-dimensional weighted polynomial algebra. Then it is reasonable

to call Projnc A a weighted projective line. In this subsection, we will find all points of a
weighted projective line. The following lemma is useful in finding all the points of Projnc A.
Lemma 3.14 If A is a right noetherian N-graded algebra satisfying (LGK), then every point
of Projnc A is given by π M where M is a cyclic critical Cohen–Macaulay module of depth
1.
Proof Let π M ∈ Tails A be a point. By the proof of Lemma 3.5, we may assume that
M ∈ grmod≤1 A is a cyclic 1-critical module. Consider the exact sequence
0 → H0m (M) → M → M/ H0m (M) → 0
in grmod A. Since M is 1-critical, if H0m (M) = 0, then GKdim M/ H0m (M) < 1. Since
H0m (M) ∈ tors A, it follows that GKdim H0m (M) = 0, so
GKdim M = max{GKdim H0m (M), GKdim M/ H0m (M)} < 1,
which is a contradiction, hence H0m (M) = 0. Since A satisfies (LGK), 1 ≤ depth M ≤
ldim M = GKdim M = 1, so M is a cyclic critical Cohen–Macaulay module of depth 1.

123
I. Mori
Lemma 3.15 Let A = k[x, y] be a 2-dimensional weighted polynomial algebra with deg x =
a, deg y = b ∈ N+ such that gcd(a, b) = 1. Then M ∈ grmod is a cyclic critical Cohen–
Macaulay module of depth 1 generated in degree 0 if and only if M is isomorphic to A/(x),
A/(y), or A/(βx b − αy a ) for (α, β) ∈ P1 \{(0, 1), (1, 0)}.
Proof Suppose that M ∈ grmod A is a cyclic critical Cohen–Macaulay module of depth 1

generated in degree 0. Since pdM = depth A − depth M = 1 by Auslander–Buchsbaum
formula, the minimal free resolution of M is of the form
0 → ⊕i≥1 A(−i)bi → A → M → 0
where bi ∈ N so that HM (t) = (1 − i≥1 bi t i )/(1 − t a )(1 − t b ). Since A is a commutative

noetherian connected graded algebra, A satisfies (LGK). Since M is Cohen–Macaulay of
depth 1, GKdim M = ldim M = depth M = 1, so i≥1 bi = 1. It follows that there exists
m ∈ N+ such that bm = 1 and bi = 0 for all i = m, so M = A/( f ) for some f ∈ Am .
Clearly, M = A/( f ) is a cyclic Cohen–Macaulay module of depth 1 generated in degree 0.
It is easy to see that M = A/( f ) is critical if and only if f is irreducible, so we will
now find all irreducible homogeneous polynomials in A. Let f ∈ A be a homogeneous
polynomial. If f does not contain a term of the form αy s where s ≥ 1 and 0 = α ∈ k, then
f = xg for some g ∈ A, so f is irreducible if and only if f = x up to units. Similarly, if f
does not contain a term of the form βx t where t ≥ 1 and 0 = β ∈ k, then f = yh for some
h ∈ A, so f is irreducible if and only if f = y up to units. Suppose that f contains both terms
of the form βx t and αy s where s, t ≥ 1 and 0 = α, β ∈ k. If f also contains a term of the
form γ x i y j where i, j ≥ 1 and 0 = γ ∈ k, then ai + bj = at = bs. Since gcd(a, b) = 1,
it follows that a divides both j and s, and b divides both i and t, so f ∈ k[x b , y a ]. Since
k is algebraically closed, if f is irreducible, then f = βx b − αy a for some (α, β) ∈ P1 . If
(α, β) ∈ P1 \{(0, 1), (1, 0)}, then x | f and y | f , so f = βx b − αy a is in fact irreducible.

Let A = k[x, y] with deg x = a, deg y = b ∈ N+ . For p0 = (0, 1), q0 = (1, 0) ∈ P1 ,

define M p0 = A/(x), Mq0 = A/(y) ∈ grmod A. For p = (α, β) ∈ P1 \{ p0 , q0 }, define
M p = A/(βx b − αy a ) ∈ grmod A.
Theorem 3.16 Let A = k[x, y] with deg x = a, deg y = b ∈ N+ such that gcd(a, b) = 1.
Suppose that a, b ≥ 2.
(1) All thin points of Projnc A are π M p0 (−i) where i = 0, . . . , b − 1, and π Mq0 (− j) where
j = 0, . . . , a − 1.
(2) All ordinary points of Projnc A are π M p where p ∈ P1 \{ p0 , q0 }.
(3) There are no fat points of Projnc A.
Proof For i, j, k ∈ Z and p ∈ P1 \{ p0 , q0 }, M p0 (−i), Mq0 (− j), and M p (−k) are 1-critical
modules by Lemma 3.15, so π M p0 (−i), π Mq0 (− j), and π M p (−k) are points of Projnc A.
On the other hand, by Lemma 3.14, every point of Projnc A is given by π M where M is a
cyclic critical Cohen–Macaulay module of depth 1, so every point of Projnc A is one of the
above forms by Lemma 3.15. Since
HM p0 (−i) (t) = t i /(1 − t b ),

HMq0 (− j) (t) = t j /(1 − t a ),
HM p (−k) (t) = t k (1 − t ab )/(1 − t a )(1 − t b ),
123
e(M p0 (−i)) = 1/b < 1, e(Mq0 (− j)) = 1/a < 1, e(M p (−k)) = 1 are distinct, so there are
no isomorphisms among three types of points π M p0 (−i), π Mq0 (− j), π M p (−k) in tails A
by Lemma 3.12.
For p ∈ P1 \{ p0 , q0 }, since there are exact sequences
x
0 → M p (−a) −−−−→ M p → M p /x M p → 0
y
0 → M p (−b) −−−−→ M p → M p /y M p → 0
in grmod A with M p /x M p , M p /y M p ∈ tors A, it follows that π M p (−a) ∼ = π Mp ∼ =
π M p (−b). Since gcd(a, b) = 1, there exist s, t ∈ Z such that 1 = as + bt, so
π M p (−1) = π M p (−as)(−bt) ∼ = π M p , hence π M p (−i) ∼
= π M p for all i ∈ Z.
If x i y j ∈ A(ab) for i, j ∈ N, then ab | ai + bj. Since gcd(a, b) = 1, it follows that a | j,
and b | i, so A(ab) = k[x b , y a ] =: B with deg x b = deg y a = 1. An exact sequence
0 → (βx b − αy a ) → A → M p → 0
in grmod A induces an exact sequence
0 → (βx b − αy a )(ab) → A(ab) = B → M (ab)
p →0
in grmod B, so (M p )(ab) = B/(βx b − αy a ) =: N p . For q ∈ P1 \{ p0 , q0 }, if π A M p ∼

= π A Mq
in tails A, then π B N p ∼
= π B Nq in tails B. Since B is a polynomial algebra generated in
degree 1, we can conclude that p = q (see the proof of Proposition 4.4 in the next section).
Since there is an exact sequence
y
0 → M p0 (−b) −−−−→ M p0 → M p0 /y M p0 → 0
with M p0 /y M p0 ∈ tors A, it follows that π M p0 (−bi) ∼ = π M p0 for all i ∈ Z. Let C =
A(b) . It is easy to see that πC (M p0 (−i))(b) = 0 if and only if b | i by the above Hilbert
series calculation. If π A M p0 (−i) ∼
= π A M p0 in tails A for i ∈ Z, then πC (M p0 (−i))(b) ∼ =
πC (M p0 ) in tails C, so b | i. Combining with the above result, π A M p0 (−i) ∼
(b)
= π A M p0 (−i )
if and only if i ≡ i (mod b). Similarly, π A Mq0 (− j) ∼ = π A Mq0 (− j ) if and only if j ≡

j (mod a), hence the result.

We will leave the reader to modify the above theorem if one of (or both of) a and b is 1.
3.3 Regular modules and regular components
Let A = k[x, y] be a weighted polynomial algebra. In this subsection, we will find mod-
uli spaces parameterizing the isomorphism classes of simple regular modules and regular
components over the Beilinson algebra ∇ A. Moreover, we will give a way of explicitly cal-
culating regular modules over ∇ A corresponding to points of Projnc A found in the previous
subsection.
Lemma 3.17 Let a, b ∈ N+ such that gcd(a, b) = 1. For any integer d ≥ ab, there exist
unique integers 0 ≤ i, 0 ≤ j ≤ a − 1 such that d = ai + bj.
Proof Since gcd(a, b) = 1, there exist i, j ∈ Z such that d = ai + bj. For any k ∈ Z,
a(i − bk) + b( j + ak) = ai + bj = d, so there exist i ∈ Z and 0 ≤ j ≤ a − 1 such that
d = ai + bj. Since bj ≤ b(a − 1) ≤ d, it follow that i ∈ N. If 0 ≤ i , 0 ≤ j ≤ a − 1 such
that d = ai + bj , then a(i − i ) = b( j − j). Since gcd(a, b) = 1, it follows that a | j − j.
Since 0 ≤ j, j ≤ a − 1, it follows that j = j, so i = i.

123
I. Mori
Proposition 3.18 Let A = k[x, y] with deg x = a, deg y = b ∈ N+ such that gcd(a, b) = 1.
∞
(1) For every point p ∈ P1 \{ p0 , q0 }, Hωπ M p (t) = i=−∞ ti .
∞
(2) For any m ∈ Z, Hωπ M p0 (−m) (t) = i=−∞ t bi+m .
∞
(3) For any m ∈ Z, Hωπ Mq0 (−m) (t) = i=−∞ t ai+m .
Proof If M ∈ grmod A is Cohen–Macaulay of depth 1, then

∞

HM (t) = (−1) j HH j (t) = −HH1 (t)
m (M) m (M)
i=0
as rational functions by [13, Proposition 5.5]. Since there exists an exact sequence
0 → H0m (M) = 0 → M → ωπ M → H1m (M) → 0
in GrMod A by [7, Proposition 7.2(2)], Hωπ M (t) = HM (t) + HH1 (M) (t) as Laurant power
m
series.
If M = M p where p ∈ P1 \{ p0 , p1 }, then
1 − t ab 1 + t b + t 2b + · · · + t (a−1)b
HM (t) = =
(1 − t )(1 − t )
a b 1 − ta

a−1 ∞
∞
a−1
= t bj t ai = t ai+bj .
j=0 i=0 i=0 j=0
Since gcd(a, b) = 1, it follows that dimk Mi = 1 for all i ≥ ab by Lemma 3.17, so

∞ i
HM (t) = i=0 t − f M (t) = 1−t
1
− f M (t) for some f M (t) ∈ Z[t]. Since H1m (M) is right
bounded,
∞
1 1/t
HH1 (M) (t) = −H M (t) = − + f M (t) = − + f M (t) = t −i−1 + f M (t),
m 1−t 1/t − 1
i=0
∞
so Hωπ M (t) = HM (t) + HH1 (M) (t) = i=−∞ ti .
m
∞ bi
If M = M p0 , then HM (t) = 1/(1 − t ) = i=0 t . Since H1m (M) is right bounded,
b
∞
1 1/t b
HH1 (M) (t) = −H M (t) = − =− = t −b(i+1) ,
m 1−t b 1/t − 1
b
i=0
∞
so Hωπ M (t) = HM (t) + HH1 (t) = bi
i=−∞ t , hence
m (M)
∞
∞

Hωπ M(−m) (t) = t m t bi = t bi+m
i=−∞ i=−∞
∞
for any m ∈ Z. Similarly, Hωπ Mq0 (−m) (t) = i=−∞ t
ai+m for any m ∈ Z.

If R = k Q is a path algebra of a quiver Q whose underlying graph is extended Dynkin,

then it is known that two simple regular modules M, N over R are in the same regular
component if and only if N = ν1i (M) for some i ∈ Z [28]. We can interpret this condition
using noncommutative algebraic geometry.
Theorem 3.19 Let A = k[x, y] with deg x = a, deg y = b ∈ N+ such that gcd(a, b) = 1.
123
(1) The isomorphism classes of simple regular modules over ∇ A are parameterized by
| Projnc A| = |P1 | { p1 , . . . , pb−1 , q1 , . . . , qa−1 }.
(2) Regular components of ∇ A are parameterized by | Proj A| = |P1 |.
Proof (1) By Theorem 3.6, the isomorphism classes of simple regular modules over ∇ A
are parameterized by | Projnc A|, which consist of π M p where p ∈ P1 \{ p0 , q0 }, π M p0 (−i)
where 0 ≤ i ≤ b − 1, and π Mq0 (− j) where 0 ≤ j ≤ a − 1 by Theorem 3.16. By writing
π M pi := π M p0 (−i), π Mq j := π Mq0 (− j) for i, j ∈ Z by abuse of notation, we obtain the
result.
(2) Since A is symmetric, M ⊗L ∼
A KA = M(−), so we have the following commutative
diagram
−1
Db (tails A) −−−−−−−−−−−−−−→ Db (mod ∇ A)
⏐ ⏐
⏐
(−)
⏐ν
1
−1
Db (tails A) −−−−−−−−−−−−−−→ Db (mod ∇ A)
by the proof of [17, Theorem 4.12]. It follows that simple regular modules over ∇ A cor-
responding to points M, N of Projnc A are in the same regular component if and only if
N ∼= M(−i) for some i ∈ Z. By the proof of Theorem 3.16, M p0 (−i) ∼ = M p0 (−i ) if
∼
and only if i ≡ i (mod b), and Mq0 (− j) = Mq0 (− j ) if and only if i ≡ i (mod a).

Since gcd(a, ) = gcd(b, ) = 1, for any i, j ∈ Z, there exist i , j ∈ Z such that

M p0 (−i) ∼= M p0 (−i ) and Mq0 (− j) ∼
= Mq0 (− j ), so M pi = M p0 (−i) are in the
same component for any i ∈ Z, and Mq j = Mq0 (− j) are in the same component for any
j ∈ Z, hence the result.

Example 3.20 The behavior of regular modules over a path algebra of an extended Dynkin
quiver is well-known [28]. The following is a noncommutative algebraic geometric interpre-
4 . If A = k[x, y] with deg x = 2, deg y = 3 so that the Gorenstein parameter of
tation for k A
A is 5, then ∇ A is isomorphic to the path algebra of the following quiver by [24, Theorem
5.4]
x /4
@2 ^== y
x
=
0 NNN p 1
NN ppp
y NN' wppp x
3
We now explicitly calculate the simple regular module over ∇ A corresponding to each
point of Projnc A. Recall that the simple regular module corresponding to a point π M of
Projnc A is given by
HomA (⊕i=0
4
A(−i), π M) ∼
= ⊕i=0
4
HomA (A, π M(i)) ∼
= ⊕i=0
4
(ωπ M)i .
Let M = A/(βx 3 − αy 2 ) ∈ grmod A where 0 = α, β ∈ k. By the proof of Theorem 3.16,
π M(2) ∼
= π M in tails A, so we may compute
HomA (⊕i=0
4
A(−i), π M) ∼
= HomA (⊕i=0
4
A(−i), π M(2))
= ⊕i=0 (ωπ M(2))i ∼
∼ 4
= ⊕i=0
4
(ωπ M)i+2
∼
= ⊕ (ωπ M)i .
6
i=2
123
I. Mori
Fix bases M2 = kx, M3 = ky, M4 = kx 2 , M5 = kx y, M6 = kx 3 . By the Hilbert series

calculation in Proposition 3.18, we see that (ωπ M)i = Mi for 2 ≤ i ≤ 6. With this particular
bases, the maps
x
M2 = kx −−−−→ M4 = kx 2
x
M3 = ky −−−−→ M5 = kx y
x
M4 = kx 2 −−−−→ M6 = x 3
y
M2 = kx −−−−→ M5 = kx y
are the identities. Since y 2 = (β/α)x 3 ∈ M6 , the map
y
M3 = ky −−−−→ M6 = kx 3
is given by the multiplication by β/α.
Let M = (A/(x))(−1) ∈ grmod A. By the Hilbert series calculation in Proposition 3.18,

k if i = 1, 4,
Mi = (ωπ M)i =
0 if i = 0, 2, 3,
and the map M1 = k1 −−−y−→ M4 = ky is the identity (in this case, an arbitrary shift is not
allowed).
Using similar calculations for the other points, we have the correspondence in the next
page.
A regular component of π A/(βx 3 − αy 2 ) contains a unique simple regular module and
is a homogeneous stable tube. A regular component of π A/(x) contains three simple regular
modules and is a stable tube of rank 3. A regular component of π A/(y) contains two simple
regular modules and is a stable tube of rank 2.
4 3-Dimensional quantum polynomial algebras
In this last section, we will give some detailed study on 2-regular modules over Beilison alge-
bras of 3-dimensional quantum polynomial algebras of Type S, which are the path algebras of
the 2-dimensional Beilinson quiver with quantum relations. Although these algebras are the
simplest examples of 2-representation infinite algebras, the behavior of 2-regular modules
largely depends on the quantum relations.
4.1 Geometric algebras
In this subsection, we will recall from [5] and [3] some geometric techniques to study 3-
dimensional quantum polynomial algebras in noncommutative algebraic geometry.
Definition 4.1 Let A be a graded algebra finitely generated in degree 1. We say that M ∈
GrMod A is a point module if M is cyclic and HM (t) = 1/(1 − t).
Let A be a graded algebra. If M ∈ GrMod A is a point module, then it is easy to see that
M is 1-critical and e(M) = 1, so π M is an ordinary point of Projnc A. We will see that there
exists a bijection between the set of isomorphism classes of point modules over A and the
set of ordinary points of Projnc A if A is a geometric quantum polynomial algebra defined
below.
123
Points of Projnc A Simple regular modules over ∇ A
π A/(βx 3 − αy 2 )
1 /k
@k ^==β/α

1
=
NNN
ppp
k k
NNN pp
1 N& k xpp 1
π A/(x) 0 0
k NNN 0
NNN
1 N&
k
π(A/(x))(−1) 0 k ^==1
=
0 k
π(A/(x))(−2) k 0
0 0
π A/(y)
1 /k
@
k

1
k 0
π(A/(y))(−1) 0 0
0 pk
ppppp
k
xp 1
Note that if A is not generated in degree 1, then the above notion of point module is not
very useful, so, in this subsection, we will assume that A is finitely generated in degree 1. In
this case, we can express A as a graded quotient algebra of the tensor algebra A = T (V )/I
where V is a finite dimensional vector space and I is a homogeneous two-sided ideal of
T (V ). Using this expression, for a point p ∈ P(V ∗ ), we define M p = A/ p ⊥ A ∈ GrMod A
where p ⊥ = { f ∈ V = A1 | f ( p) = 0}. Note that, for p, q ∈ P(V ∗ ), M p ∼ = Mq if and only
if p = q.
It is easy to see that if X ⊂ Pn−1 is a projective variety (over an arbitrary field k) and
A = k[x1 , . . . , xn ]/I (X ) is the homogeneous coordinate ring of X so that X = Proj A,
then the map p → M p gives a bijection from | Proj A| to the set of isomorphism classes of
point modules over A. Moreover, if k is an algebraically closed field (which is a standard
assumption of this paper), then the map p → π M p gives a bijection from | Proj A| to
| Projnc A|. It follows that there is no thin point nor fat point of Projnc A.
123
I. Mori
Definition 4.2 [19, Definition 4.3] Let A = T (V )/(R) be a quadratic algebra where R ⊂
V ⊗k V is a subspace and (R) is the homogeneous two-sided ideal generated by R.
(1) We say that A satisfies (G1) if there exist a closed subscheme E ⊂ P(V ∗ ) and an
automorphism σ ∈ Aut k E such that
{( p, q) ∈ P(V ∗ ) × P(V ∗ ) | f ( p, q) = 0 for all f ∈ R}
= {( p, σ ( p)) ∈ P(V ∗ ) × P(V ∗ ) | p ∈ E}.
If this is the case, we write P(A) = (E, σ ).
(2) We say that A satisfies (G2) if there exist a closed subscheme E ⊂ P(V ∗ ) and an
automorphism σ ∈ Aut k E such that
R = { f ∈ V ⊗k V | f ( p, σ ( p)) = 0 for all p ∈ E}.
If this is the case, we write A = A(E, σ ).
(3) We say that A is geometric if A satisfies both (G1) and (G2) by the same pair (E, σ ),
that is, A = A(P(A)).
Example 4.3 All skew polynomial algebras, and all 3-dimensional quantum polynomial alge-
bras are geometric. Moreover, all known examples of 4-dimensional quantum polynomial
algebras satisfy (G1). In particular, every 4-dimensional Auslander-regular quantum poly-
nomial algebra satisfying (CM) satisfies (G1) by [27, Theorem 1.4].
If A is a quadratic algebra satisfying (G1) with P(A) = (E, σ ), then the assignment
p → M p gives a bijection from |E| to the set of isomorphism classes of point modules over
A, and Mσ ( p) ∼
= M p (1)≥0 . By this reason, E is called the point scheme of A.
For a reasonably nice class of algebras, we expect that two point modules are equivalent
if and only if they are isomorphic. Unfortunately, this is false in general. For example, if
A = k x, y/(x y), then there are non-isomorphic point modules M, N ∈ GrMod A such
that π M ∼ = π N (cf. [20, Example 4.4]).
Proposition 4.4 Let A be a right noetherian quadratic algebra of finite global dimension
with H A (t) = 1/(1 − t)d . If A satisfies (G1) with P(A) = (E, σ ), then the functor π :
grmod A → tails A induces a bijection from the set of isomorphism classes of point modules
over A to the set of ordinary points of Projnc A. In particular, ordinary points of Projnc A are
parameterized by |E|.
Proof Let P be the set of the isomorphism classes of point modules over A and O the set of
ordinary points of Projnc A. If M is a point module, then it is easy to see that M is 1-critical
and e(M) = 1, so π M is an ordinary point of Projnc A, so the functor π : grmod A → tails A
induces a map π : P → O. If M p , Mq ∈ grmod A are point modules where p, q ∈ E such
that π M p ∼
= π Mq in tails A, then p = q by [26, Lemma 3.4], so M p ∼ = Mq , hence the map
π : P → O is injective.
If π M is an ordinary point of Projnc A where M ∈ grmod≤1 A is a 1-critical module with
e(M) = 1, then there exists n ∈ N such that dimk Mi = 1 for all i ≥ n by Lemma 3.11. For
0 = m ∈ Mn , 0 = m A ⊂ M≥n ⊂ M is a non-zero graded submodule, so M≥n /m A is finite
dimensional because M≥n is 1-critical by Lemma 3.2. Since HM(n)≥0 (t) = 1/(1−t) and A is
generated in degree 1, M(n)≥0 = m A, so M(n)≥0 is a point module so that M(n)≥0 ∼ = Mp
for some p ∈ E. Since Mσ −n ( p) is a point module such that
(Mσ −n ( p) )≥n (n) ∼
= Mσ −n ( p) (n)≥0 ∼
= Mp ∼
= M(n)≥0 ∼
= M≥n (n),
π Mσ −n ( p) ∼
= π M, so π : P → O is surjective.

123
We now pay attention to thin points and fat points. The following lemma says that there
is no thin point in a quantum projective space.
Lemma 4.5 (cf. [6, Proposition 2.21(iii)]) If A is a connected graded right coherent algebra
of finite global dimension satisfying (H), and 0 = M ∈ grmod A, then e(M)/e(A) ∈ N+ . In
particular, if A is a graded quotient algebra of a quantum polynomial algebra, then there is
no thin point in Projnc A.
The existence of fat points in a quantum projective space is more subtle.
Definition 4.6 For a geometric algebra A = A(E, σ ) where E ⊂ P(V ∗ ), we define

||σ || := inf{i ∈ N+ | σ i = τ | E for some τ ∈ Aut k P(V ∗ )}.
Note that if A = A(E, σ ) is a geometric algebra and L = O E (1) is the very ample
invertible sheaf corresponding to the embedding E ⊂ P(V ∗ ), then
||σ || = inf{i ∈ N+ | (σ i )∗ L ∼
= L}.
Lemma 4.7 [3, Theorem 3.4(i)] Let A = A(E, σ ) be a 3-dimensional quantum polynomial
algebra. If M ∈ GrMod A is a 1-critical module, then either e(M) = 1 or e(M) = ||σ ||.
Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra. By the above lemma,

Projnc A has a fat point if and only if 1 < ||σ || < ∞. It is easy to see that ||σ || = 1 if and
only if E = P2 if and only if Projnc A ∼ = P2 , that is, tails A ∼
= coh P2 . On the other hand, by
[6, Theorem 7.1], A is finite over its center if and only if |σ | < ∞. Since ||σ || ≤ |σ |, if A is
finite over its center and E = P2 , then there are fat points in Projnc A. In general, ||σ || = |σ |,
so there may be fat points in Projnc A even if A is not finite over its center.
Theorem 4.8 Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra. If ||σ || =

1 or ||σ || = ∞, then the isomorphism classes of simple 2-regular modules over ∇ A are
parameterized by |E|.
Proof This follows from Theorem 2.9, Lemma 3.5 and Proposition 4.4.

4.2 d-Representation tame
Let R be a finite dimensional hereditary algebra of infinite representation type, that is, a 1-
representation infinite algebra. Here, infinite representation type means that there are infinitely
many isomorphism classes of indecomposable R-modules. In the study of such an algebra
R, understanding regular modules is essential. For this reason, infinite representation type
is divided further into two types, roughly speaking, tame representation type where the
class of regular modules can be controlled, and wild representation type where it cannot be
controlled. It is natural to extend the notion of tame representation type to d-representation
infinite algebras.
Definition 4.9 [11, Definition 6.10] We say that a d-representation infinite algebra R is d-
representation tame if R is right noetherian and finite over its center.
We will determine when a 2-dimensional quantum Beilinson algebra is 2-representation

tame using the techniques of noncommutative algebraic geometry. The proof of the following
lemma is due to Minamoto.
123
I. Mori
Lemma 4.10 For a graded algebra A and r ∈ N+ , Z (A[r ] ) ∼

= Z (A)(r ) as graded algebras.
Proof (Minamoto) Since we have a commutative diagram of equivalence functors

Q
GrMod A −−−−→ GrMod A[r ]
⏐ ⏐
⏐
(ri)
⏐(i)

Q
GrMod A −−−−→ GrMod A[r ]
for any i ∈ Z,
Z (A[r ] )i ∼
= Nat(IdGrMod A[r] , (i))
∼
= Nat(IdGrMod A , (ri))
∼
= Z (A)ri .
It is easy to see that the above isomorphisms are compatible with multiplications, so Z (A[r ] ) ∼
=
Z (A)(r ) as graded algebras.

Let A be a graded algebra and r ∈ N+ . If A is right noetherian, then it is known that A(r )
is right noetherian and the exact functor (−)(r ) : GrMod A → GrMod A(r ) restricts to the
exact functor (−)(r ) : grmod A → grmod A(r ) .
Proposition 4.11 Let A be a right noetherian graded algebra and r ∈ N+ . If Z (A) is

noetherian and A is finite over its center Z (A), then A[r ] is right noetherian and finite over
its center.
Proof Since A is right noetherian, the functor grmod A → grmod A(r ) ; M → M (r ) is

well-defined. It follows that A(i)(r ) ∈ grmod A(r ) for all i ∈ Z, so
⎛ ⎞
A(r ) A(1)(r ) · · · A(r − 1)(r )
⎜ A(−1)(r ) A(r ) · · · A(r − 2)(r ) ⎟
⎜ ⎟
A[r ] = ⎜ .. .. .. .. ⎟
⎝ . . . . ⎠
(r )
A(−r + 1) A(−r + 2) · · · (r ) A (r )

∼
= (r )
A( j − i) ∈ grmod A . (r )
0≤i, j≤r −1
Since Z (A) is noetherian, the functor grmod Z (A) → grmod Z (A)(r ) ; M → M (r ) is well-
defined. Since A ∈ grmod Z (A), we have A(r ) ∈ grmod Z (A)(r ) , so A[r ] is finite over
Z (A)(r ) = Z (A[r ] ) by Lemma 4.10. Since Z (A[r ] ) ∼
= Z (A)(r ) is noetherian, A[r ] is right
noetherian.

Let A be an AS-Gorenstein algebra of dimension d and of Gorenstein parameter , and

K A := D Hdm (A) the canonical module of A. Recall that K A ∼ = Aν −1 (−) as graded A-
A bimodules for some graded algebra automorphism ν ∈ Aut k A, called the generalized
Nakayama automorphism, and we say that A is symmetric if ν = id. The main purpose of
this last section is to study the class of algebras defined below.
Definition 4.12 We say that a 3-dimensional quantum polynomial algebra A = A(E, σ ) is

of Type S if E ⊂ P2 is a triangle, and, in this case, we call ∇ A a 2-dimensional quantum
Beilinson algebra of Type S.
123
The Type S is further divided into three types, namely,

• Type S1 : σ fixes all components of E.
• Type S2 : σ switched two of components of E.
• Type S3 : σ circulates three components of E.
Since every 3-dimensional quantum polynomial algebra of Type S is graded Morita equivalent
to that of Type S1 by [26, Lemma 3.8], we will focus on 3-dimensional quantum polynomial
algebras of Type S1 . The following lemma lists properties of an 3-dimensional quantum
polynomial algebra of Type S1 , which are needed in this paper.
Lemma 4.13 (cf. [19]) A 3-dimensional quantum polynomial algebra A is of Type S1 if and
only if A ∼
= k x, y, z/(yz − αzy, zx − βx z, x y − γ yx) for some α, β, γ ∈ k, αβγ = 0, 1.
Let A = A(E, σ ) = k x, y, z/(yz −αzy, zx −βx z, x y −γ yx) be a 3-dimensional quantum
polynomial algebra of Type S1 , and ν ∈ Aut k A the generalized Nakayama automorphism
of A. Then the following hold:
• E = V (x yz) = V (x) ∪ V (y) ∪ V (z) ⊂ P2 .
• σ |V (x) (0, b, c) = (0, b, αc), σ |V (y) (a, 0, c) = (βa, 0, c), σ |V (z) (a, b, 0) = (a, γ b, 0).
• |σ | = lcm{|α|, |β|, |γ |}.
• ||σ || ⎛
= |αβγ |. ⎞
γβ −1 0 0
• ν = ⎝ 0 αγ −1 0 ⎠ ∈ GL(3, k) = GL(A1 ).
0 0 βα −1
• A is symmetric if and only if α = β = γ .
• Let A = k x, y, z/(yz −α zy, zx −β x z, x y −γ yx) be another 3-dimensional quantum
polynomial algebra of Type S1 . Then GrMod A ∼ = GrMod A if and only if α β γ =
(αβγ ) . ±1
Note that if A is a 3-dimensional quantum polynomial algebra of Type S1 as above, then

∇ A is isomorphic to the path algebra k Q/I of the 2-dimensional Beilinson quiver
x1
/ x2
/
y1
Q= • /• y2
/•
z1
/ z2
/
with quantum relations
I = (y1 z 2 − αz 1 y2 , z 1 x2 − βx1 z 2 , x1 y2 − γ y1 x2 ) αβγ = 0, 1
For d-dimensional quantum polynomial algebras A, we expect that GrMod A ∼
A , =
∼
GrMod A if and only if Projnc A = Projnc A . This was proved to be true for many 3-
dimensional quantum polynomial algebras [19, Theorem 5.2]. By brevity, we will make the
following definition.
Definition 4.14 Let A be a d-dimensional quantum polynomial algebra. We say that Projnc A
is finite over its center if there exists a d-dimensional quantum polynomial algebra A finite
over its center such that GrMod A ∼ = GrMod A .
Definition 4.15 Let A be a right noetherian graded domain. We define the function field of
Projnc A by
k(Projnc A) := Q gr (A)0 = {ab−1 | a, b ∈ Ai for some i, b = 0}
where Q gr (A) is the total graded right quotient ring of A.
123
I. Mori
Despite the terminology, k(Projnc A) is a division algebra over k.
Lemma 4.16 Let A, A be d-dimensional quantum polynomial algebras satisfying (G1) with
P(A) = (E, σ ), P(A ) = (E , σ ). If GrMod A ∼
= GrMod A , then the following hold.
(1) ||σ || = ||σ ||.
(2) Projnc A is finite over its center if and only if Projnc A is finite over its center.
(3) k(Projnc A) ∼ = k(Projnc A ) as k-algebras.
(4) ∼
∇ A = ∇ A as graded algebras.
Proof (1) By [26, Lemma 2.5], there exists a sequence of automorphisms τi ∈ Aut k Pd−1
such that σ = τi+1 σ τi−1 . If ||σ || = n < ∞, then (σ )n = τn σ n τ0−1 extends to an automor-
phism of Pd−1 , so ||σ || ≤ n = ||σ ||. By symmetry, ||σ || = ||σ ||.
(2) Clear from the definition.
(3) Since GKdim A = d, it follows that grmod A = grmod≤d A. It is well-known (cf. [6,
Proposition 2.30(v)]) that the functor − ⊗ A Q gr (A) : grmod A → grmod Q gr (A) induces
an equivalence of categories
tailsd−1 A := grmod A/ grmod≤d−1 A ∼
= grmod Q gr (A).
Since A is generated in degree 1, Q gr (A) is strongly graded, so the functor (−)0 :
grmod Q gr (A) → mod k(Projnc A) is an equivalence functor. Since any equivalence functor
GrMod A → GrMod A restricts to an equivalence functor grmod≤i A → grmod≤i A for
every i ∈ N by [34], we have
mod k(Projnc A) ∼
= tailsd−1 A := grmod≤d A/ grmod≤d−1 A
∼ grmod≤d A / grmod≤d−1 A =: tailsd−1 A ∼
= = mod k(Proj A ).
nc
Since k(Projnc A), k(Projnc A ) are division algebras over k, k(Projnc A) ∼

= k(Projnc A ) as
k-algebras.
(4) [17, Theorem 4.17].

Theorem 4.17 Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra of Type

S. Then the following are equivalent:
(1) ||σ || < ∞.
(2) Projnc A is finite over its center.
(3) k(Projnc A) is finite over its center.
(4) ∇ A is 2-representation tame.
Proof By Lemma 4.16, we may replace A by an algebra which is graded Morita equivalent
to A. Since every 3-dimensional quantum polynomial algebra of Type S is graded Morita
equivalent to that of Type S1 by [26, Lemma 3.8], we may assume that
A = k x, y, z/(yz − αzy, zx − βx z, x y − γ yx)
for some α, β, γ ∈ k, αβγ = 0, 1 by Lemma 4.13.
(1) ⇒ (2): By Lemma 4.13, if ||σ || = |αβγ | < ∞, then there exists
A = k x, y, z/(yz − zy, zx − x z, x y − αβγ yx) = A(E, σ )
such that GrMod A ∼ = GrMod A . Since |σ | = |αβγ | = ||σ || < ∞, A is finite over its
center by [6, Theorem 7.1], hence Projnc A is finite over its center.
123
(2) ⇒ (1): Conversely, if Projnc A is finite over its center, then there exists a 3-dimensional
quantum polynomial algebra A = A(E , σ ) finite over its center such that GrMod A ∼ =
GrMod A , so ||σ || = ||σ || ≤ |σ | < ∞ by Lemma 4.16.
(1) ⇐⇒ (3): Since k(Projnc A) ∼ = Q(k x, y/(x y − αβγ yx)) the total right quotient ring
of k x, y/(x y − αβγ yx), ||σ || = |αβγ | < ∞ if and only if k(Projnc A) is finite over its
center by [2, Proposition 3.2].
(1)√⇐⇒ (4): If A = k x, y, z/(yz − δzy, zx − δx z, x y − δyx) = A(E, σ ) where
δ = 3 αβγ , then A is a 3-dimensional symmetric quantum polynomial algebra such that
GrMod A ∼ = GrMod A by Lemma 4.13, so ∇ A ∼ = ∇ A ∼ = (A )[3] by Lemma 4.16 (4)
and Theorem 1.6 (2).
If ||σ || < ∞, then |σ | = |δ 3 | = |αβγ | = ||σ || < ∞, so A is finite over its center by
[6, Theorem 7.1]. By [18, Theorem 3.3], Z (A ) = k[x s , y s , z s , x yz] is noetherian where
s := |δ| < ∞, so ∇ A ∼ = (A )[3] is right noetherian and finite over its center by Proposition
4.11, hence ∇ A is 2-representation tame.
Conversely, if ||σ || = ∞, then |δ| = ∞, so Z (A ) = k[x yz] by the proof of [18, Theorem
3.3], hence Z (∇ A ) = Z ((A )[3] ) = Z (A )(3) = k[x yz](3) = k[w] with deg w =
1 by Lemma 4.10. Since GKdim ∇ A = GKdim A = 3 and GKdim Z (∇ A ) =
GKdim k[w] = 1, it follows that ∇ A ∼ = ∇ A is not finite over its center, hence ∇ A is
not 2-representation tame.

By Lemma 4.13 and Theorem 4.17, A = k x, y, z/(yz − αzy, zx − βx z, x y − γ yx) is
finite over its center if and only if α, β, γ are all roots of unity, and Projnc A is finite over its
center if and only if αβγ is a root of unity. So it could be possible that A is not finite over its
center but ∇ A is 2-representation tame.
4.3 2-Regular modules
In this subsection, we will find a moduli space parameterizing the isomorphism classes of
simple 2-regular modules over a 2-dimensional quantum Beilinson algebra of Type S. In
order to do this, we need the notion of point for a noncommutative affine scheme as well.
Definition 4.18 Let R be an algebra. A point of Specnc R is an isomorphism class of finite
dimensional simple R-modules. The set of points of Specnc R is denoted by | Specnc R|. A
point M ∈ mod R is called ordinary if dimk M = 1, and is called fat if dimk M > 1.
If R is a commutative algebra (over an algebraically closed field), then every simple R-
module is 1-dimensional, that is, every point of Specnc R is ordinary, and the maps m → R/m
and M → Ann R M give bijections inverses to each other between | Spec R| and | Specnc R|.
Lemma 4.19 (cf. [30]) Let R = k x, y/(x y − αyx) where 0 = α ∈ k.
(1) All ordinary points of Specnc R are given by
N(a,0) := R/(x − a)R + y R,
N(0,b) := R/x R + (y − b)R,
where a, b ∈ k, so ordinary points of Specnc R are parameterized by |V (x y)| |A2 |.
(2) If |α| = ∞, then there are no fat points.
(3) If |α| = n < ∞, then all fat points of Specnc R are given by
√
n √
N(a,b) := R/(x n − a)R + ( bx + n −a y)R
where 0 = a, b ∈ k, so fat points of Specnc R are parameterized by |A2 \V (x y)|.
123
I. Mori
Proof (1) and (2) were shown in [30], so we will omit the proof.
(3) If |α| = n < ∞, then all fat points are given by the modules N(a,b) where 0 = a, b ∈ k
with the following properties [30]:
• dimk N(a,b) = n.
• Ann R (N(a,b) ) = (x n − a, y n − b).
• For c, d ∈ k, there is a surjective map R/(cx +dy) → N(a,b) if and only if cn a +d n b = 0.
It follows that there exists a surjective map
√
n √
R/(x n − a)R + (y n − b)R + ( bx + n −a y)R → N(a,b) .
√ √
It is easy to see that dimk (R/(x n − a)R + (y n − b)R + ( n bx + n −a y)R) ≤ n, so
√ √ √ √
N(a,b) ∼
n n
= R/(x n −a)R +(y n − b)R +( bx + −a y)R = R/(x n − a)R +( bx + −a y)R
n n
√ √
where 0 = a, b ∈ k. By [30], N(a,b) does not depend on the choice of n −a, n b ∈ k. It is
easy to see that N(a,b) ∼= N(c,d) if and only if (a, b) = (c, d), so fat points of Specnc R are
parameterized by |A2 \V (x y)|.

In summary, if |α| = ∞, then there is a bijection between | Specnc

R| and |V (x y)|, and if
|α| < ∞, then there is a bijection between | Specnc R| and |V (x y)| |A2 \V (x y)| = |A2 |.
If A is a commutative N-graded algebra and u ∈ A is a homogeneous element of positive
degree, then Proj A ∼= Proj A/(u) Spec A[u −1 ]0 . A similar decomposition is known to hold
in the noncommutative setting (cf. [6,29,30]). Let A, B be graded algebras. A graded algebra
homomorphism f : A → B defines an adjoint pair of functors
f ∗ : GrMod A → GrMod B ; M → M ⊗ A B
f ∗ : GrMod B → GrMod A ; N → Hom B ( A B, N ) = N A
so that Hom B ( f ∗ M, N ) ∼
= Hom A (M, f ∗ N ). Recall that an element u ∈ A is called normal-
izing if u A = Au.
Theorem 4.20 Let A be a right noetherian connected graded algebra satisfying (H), u ∈ A
a homogeneous regular normalizing element of positive degree, f : A → A/(u) the natural
surjection, and g : A → A[u −1 ] the natural embedding. If A[u −1 ] is strongly graded, then
the functors
Tails A/(u) → Tails A; π A/(u) M → π A f ∗ M
Mod A[u −1 ]0 → Tails A; N → π A g∗ (N ⊗ A[u −1 ]0 A[u −1 ])
give a bijection from | Projnc A/(u)| | Specnc A[u −1 ]0 | to | Projnc A|. Moreover, in this
correspondence, the fat points correspond to fat points.
Theorem 4.21 Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra of Type

S and R = ∇ A the quantum Beilinson algebra of A.
(1) If R is not 2-representation tame, then the isomorphism classes of simple 2-regular
R-modules are parameterized by |E| |P2 |.
(2) If R is 2-representation tame, then the isomorphism classes of simple 2-regular R-
modules are parameterized by |P2 |.
Proof We may assume that A = k x, y, z/(yz − αzy, zx − βx z, x y − γ yx) = A(E, σ )

where α, β, γ ∈ k, αβγ = 0, 1.
123
If R is not 2-representation tame, then ||σ || = ∞ by Theorem 4.17, so the isomorphism

classes of simple 2-regular R-modules are parameterized by |E| |P2 | by Theorem 4.8.
If R is 2-representation tame, then ||σ || = |αβγ | < ∞ by Theorem 4.17. Since A is
generated in degree 1, A[z −1 ] is strongly graded, so

| Projnc A| = | Projnc A/(z)| | Specnc A[z −1 ]0 |

= | Projnc k x, y/(x y − γ yx)| | Specnc k x, y/(x y − αβγ yx)|

= |P 1 | |A 2 |
= |P 2 |
by Theorem 4.20 and Lemma 4.19, hence the isomorphism classes of simple 2-regular R-
modules are parameterized by |P2 |.

In the above theorem, the isomorphism classes of simple 2-regular modules over ∇ A
are parameterized (controlled) by algebraic varieties in both cases, so it may be reasonable
to call ∇ A 2-representation tame in both cases. That is, it may be a better definition for a
d-representation infinite algebra R to be d-representation tame if R is right noetherian as
a ring, dropping the requirement that R is finite over its center.
4.4 2-Regular components
Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra with the generalized

Nakayama automorphism ν ∈ Aut k A, and R = ∇ A the quantum Beilinson algebra of A. At
this time, we do not know what should be a right definition of 2-regular components of R,
but we expect that the results in this subsection may give some clues. Note that the dual of
the linear map ν| A1 : A1 = V → A1 = V induces an automorphism ν ∗ : P(V ∗ ) → P(V ∗ ),
which restricts to an automorphism ν ∗ ∈ Aut k E by abuse of notation [21].
Proposition 4.22 If A = A(E, σ ) is a 3-dimensional quantum polynomial algebra with the

generalized Nakayama automorphism ν ∈ Aut k A, and M p , Mq are point modules over A,
then the following are equivalent:
(1) RHomA (M p , Mq ) = 0.
(2) Ext1A (M p , Mq ) = 0.
(3) Either q = p or q = (ν ∗ )−1 σ −3 ( p).
Proof By [1, Lemma 2.16], we know RHom A (M p , Nq ) = 0 if and only if Ext1A (M p , Mq ) =

0 if and only if either q = p or q = (ν ∗ )−1 σ −3 ( p). By the proof of [22, Proposition 2.8],
ExtiA (M p , Mq ) ∼= ExtiA (M p , (ωMq )≥0 ) ∼
= ExtiA (M p , Mq ), hence the result.

Lemma 4.23 Let A be a right noetherian connected graded algebra satisfying (H), u ∈ A
a homogeneous regular normalizing element of positive degree, and g : A → A[u −1 ] the
natural embedding.
(1) For N ∈ GrMod A[u −1 ], ω A π A g∗ N ∼
= g∗ N in GrMod A and g ∗ ω A π A g∗ N ∼
= N in
−1
GrMod A[u ].
(2) For M ∈ GrMod A, if GKdim M = 1 and Ann M (u) := {m ∈ M | mu = 0} = 0, then
ωAπA M ∼
= g∗ g ∗ M in GrMod A and π A M ∼ = π A g∗ g ∗ M in Tails A.
123
I. Mori
Proof (1) If T ∈ Tors A, then g ∗ T = 0. Since both g∗ : GrMod A[u −1 ] → GrMod A

and g ∗ : GrMod A → GrMod A[u −1 ] are exact functors, for N ∈ GrMod A[u −1 ],
RHom A (T, g∗ N ) ∼
= RHom A[u −1 ] (g ∗ T, N ) = 0, so
Him (g∗ N ) = lim ExtiA (A/A≥n , g∗ N ) = 0

n→∞
for all i ∈ Z. Since A is right noetherian connected graded, we have an exact sequence
0 → H0m (g∗ N ) → g∗ N → ω A π A g∗ N → H1m (g∗ N ) → 0
by [7, Proposition 7.2(2)], so ω A π A g∗ N ∼ = g∗ N in GrMod A. Since g ∗ g∗ ∼
= IdGrMod A[u −1 ] ,
it follows that g ω A π A g∗ N ∼
∗
= g g∗ N ∼
∗
= N in GrMod A[u −1 ].
·u
(2) Let u ∈ Ad . Since Ann M (u) = 0, for each i, the map Mi −−− −→ Mi+d is injective,
so {dimk Mi+ jd } j∈N is an increasing sequence. Since A satisfies (H) and M ∈ grmod A,
there exist polynomials p0 (t), . . . , pr (t) ∈ Z[t] of degree < GKdim M such that dimk Mn =
pi (n) for n 0 with n (mod r ) = i by the proof of [31, Corollary 2.2]. Since GKdim M = 1,
pi (t) are constant functions, so there exists e ∈ N such that dimk Mi ≤ e for all i ∈ Z. It
follows that there exists i 0 such that, for any i ≥ i 0 and any j ∈ N, Mi −−− ·u j
−→ Mi+ jd is
surjective. For any i ≥ i 0 and any element mu − j ∈ M[u −1 ]i , there exists m ∈ Mi such that
m = m u j , so mu − j = m ∈ Mi , hence (g∗ g ∗ M)≥i0 = M[u −1 ]≥i0 ∼ = M≥i0 in GrMod A.
It follows that π A g∗ g ∗ M ∼
= π A M in Tails A, so ω A π A M ∼= ω A π A g∗ g ∗ M ∼= g∗ g ∗ M in
GrMod A by (1).

Proposition 4.24 Let A be a 3-dimensional quantum polynomial algebra of Type S. For any
fat points M, N of Projnc A, the following are equivalent:
(1) RHomA (M, N ) = 0.
(2) Ext1A (M, N ) = 0.
(3) N ∼
= M.
Proof (3) ⇒ (2): If M ∼

= N , then HomA (M, N ) = 0. By [25, Theorem 8.6],
dimk HomA (M, N ) − dimk Ext1A (M, N ) + dimk Ext2A (M, N ) = 0,
so Ext1A (M, N ) = 0.
(2) ⇒ (1): Trivial.
(1) ⇒ (3): Let A be a 3-dimensional quantum polynomial algebra with an equivalence
functor F : grmod A → grmod A . Since H F(M) (t) = HM (t) for any M ∈ grmod A
by [34], F preserves the GKdimension and the multiplicity. It follows that F induces an
equivalence functor F̄ : tails0 A → tails0 A , so F gives a bijection from the set of fat points
of Projnc A to the set of fat points of Projnc A by Lemma 3.5, hence we may assume that
A = k x, y, z/(yz − αzy, zx − βx z, x y − γ yx) where α, β, γ ∈ k, αβγ = 0, 1 by Lemma
4.13.
If |αβγ | = ∞ so that R is not 2-representation tame by Theorem 4.17, then there is no
fat point of Projnc A, so we may assume that n := |αβγ | < ∞. Let M = π M, N = π N be
fat points of Projnc A such that M N . Since there is no fat point of Projnc A/(z), every
fat point of Projnc A is a homogenization of a fat point of Specnc A[z −1 ]0 by Theorem 4.20,
so we may assume that
√
n √ √
n √
M = A/(x n − az n )A + ( bx + n −a y)A, N = A/(x n − cz n )A + ( d x + n −cy)A
for 0 = a, b, c, d ∈ k such that (a, b) = (c, d) by Lemma 4.19.
123
Let g : A → A[z −1 ] be the natural embedding and consider the following composition
of functors
−⊗ A[z −1 ] A[z −1 ] π A g∗
h ∗ : Mod A[z −1 ]0 −−−−−−−−−−→ GrMod A[z −1 ] −−−−→
0
Tails A
∼
=
g∗ ω A (−)0
h ∗ : Tails A −−−−→ GrMod A[z −1 ] −−−−→ Mod A[z −1 ]0 .
∼
=
Since (π A , ω A ) and (g∗ , g ∗ ) are adjoint pairs of functors, it follows that (h ∗ , h ∗ ) is also an
adjoint pair of functors. Since π A , g∗ are exact, h ∗ is also exact. Note that g ∗ is exact and ω A is
left exact. For N ∈ Tails A, Ri ω A N ∈ Tors A for i ≥ 1, so Ri (g ∗ ω A )N ∼ = g ∗ (Ri ω A N ) =
−1 ∗
R ω A N ⊗ A A[z ] = 0 for i ≥ 1, so h is also exact.
i
Write
A z := A[z −1 ]0 ∼
= k x, y/(x y − αβγ yx),
√ √
Mz := (g M)0 = M[z −1 ]0 ∼
∗ n
= A z /(x n − a)A z + ( bx + −a y)A z ,
n
√ √
Nz := (g ∗ N )0 = N [z −1 ]0 ∼
n
= A z /(x n − c)A z + ( d x + −cy)A z .
n
Since GKdim M = GKdim N = 1 and Ann M (z) = Ann N (z) = 0,
h ∗ Mz ∼
= π A g∗ ((g ∗ M)0 ⊗ Az A[z −1 ]) ∼
= π A g∗ g ∗ M ∼
= π A M = M,
∼ π A g∗ ((g N )0 ⊗ A A[z ]) ∼
h ∗ Nz = ∗ −1
= π A g∗ g N ∼
∗
= πA N = N ,
z
by Lemma 4.23 (2). Since
h ∗ h ∗ Nz ∼
= (g ∗ ω A π A N )0 ∼
= (g ∗ g∗ g ∗ N )0 ∼
= (g ∗ N )0 = Nz
by Lemma 4.23 (2) again,
RHomA (M, N ) ∼
= RHomA (h ∗ Mz , h ∗ Nz )
∼
= RHom A (M, h ∗ h ∗ Nz )
z
∼
= RHom Az (Mz , Nz ),
so it is enough to show that RHom Az (Mz , Nz ) = 0. Since x n − az n ∈ Z (A), it is easy to see

that
√ √
− n bx − n −a y
·
x n − az n
0 → A(−n − 1) −−−−−−−−−−−−−−→ A(−n) ⊕ A(−1)
√ √
x − az
n n n
bx + −a y
n ·
−−−−−−−−−−−−−−−−−−−→ A → M → 0
is the minimal free resolution of M ∈ grmod A, which induces a free resolution of Mz ∈
mod A z
√ √
− n bx − n −a y
·
xn − a
0 → Az −−−−−−−−−−−−−−→ Az ⊕ Az
√ √
x − a bx + −a y ·
n n n
−−−−−−−−−−−−−−−−−−→ A z → Mz → 0,
123
I. Mori
so ExtiAz (Mz , Nz ) are homologies of

√ √
√ √ − n bx − n −a y
·
− a bx + −a y
· xn
n n
xn − a
Nz −−−−−−−−−−−−−−−−−−→ Nz ⊕ Nz −−−−−−−−−−−−−−→ Nz .
φ ψ
n−1
For i=0 ai x
i ∈ Nz ,

n−1
n−1
n−1
n−1
ai x i (x n − a) = ai (x n − a)x i = ai (c − a)x i = (c − a) ai x i ,
i=0 i=0 i=0 i=0
so if a = c, then φ is injective and ψ is surjective. Since

√
n √
Nz = A z /(y n − d)A z + ( d x + n −cy)A z ,
if b = d, then we can similarly show that φ is injective and ψ is surjective. It follows that if
(a, b) = (c, d), then
0 → Nz −−−−→ Nz ⊕ Nz −−−−→ Nz → 0
φ ψ
is an exact sequence by the dimension calculation, so ExtiAz (Mz , Nz ) = 0 for all i ∈ Z.

Proposition 4.25 Let A be a 3-dimensional quantum polynomial algebra of Type S. For any
fat point M of Projnc A, we have M ⊗A KA ∼ = M.
Proof Let A be a 3-dimensional quantum polynomial algebra with an equivalence functor

F : grmod A → grmod A . Since
F̄
tails A −−−−→ tails A
⏐ ⏐
−⊗A KA
⏐ ⏐−⊗ K
A A
F
tails A −−−−→ tails A
commutes by the proof of [26, Theorem 3.5], we may assume that
A = k x, y, z/(yz − δzy, zx − δx z, x y − δyx) = A(E, σ )
is a symmetric 3-dimensional quantum polynomial algebra by Lemma 4.13. If ||σ || = ∞,
then there is no fat point of Projnc A, so we may assume that n := ||σ || = |δ 3 | < ∞ so
that δ 3n = 1. As in the proof of Proposition√4.24, every
√ fat point of Projnc A is of the form
M = π M where M = A/(x n − az n )A + ( bx + n −a y)A, 0 = a, b ∈ k. Note that
n
z 3 (x n − az n ) = (δ 3n x n − az n )z 3 = (x n − az n )z 3 ,
√ √ √ √
z 3 ( bx + n −a y) = ( bδ 3 x + n −aδ −3 y)z 3
n n
√ √
in A. Since M does not depend on the choice of n b, n −a ∈ k and |δ ±3 | = n, the map
φ : A(−3) → A defined by φ(u) = z 3 u induces a graded right A-module homomorphism
√ √
φ̄ : M(−3) → (A/(x n − az n )A + ( bδ 3 x + n −aδ −3 y)A) ∼
n
= M.
Since φ̄ = 0 and M, M(−3) are simple objects in tails0 A by Lemma 3.5, M(−3) ∼
= M in
= M(−3) ∼
tails0 A. Since A is symmetric, M ⊗LA K A ∼ = M.

As a summary, we have the following result.
123
Theorem 4.26 Let A be a 3-dimensional quantum polynomial algebra of Type S, R = ∇ A

the quantum Beilinson algebra of A, and M, N simple 2-regular modules over R. If both of
M, N corespond to ordinary points or fat points of Projnc A, then the following are equivalent:
(1) RHom R (M, N ) = 0.
(2) Ext1R (M, N ) = 0.
(3) N ∼
= M or N ∼ = ν2 (M).
Proof Since we have the commutative diagram of equivalence functors
RHomA (⊕i=0
2 A(−i),−)
Db (tails A) −−−−−−−−−−−−−−→ Db (mod ∇ A)
⏐ ⏐
⏐ ⏐ν
A KA
−⊗L 2
RHomA (⊕i=0
2 A(−i),−)
Db (tails A) −−−−−−−−−−−−−−→ Db (mod ∇ A)
by the proof of [17, Theorem 4.12], it is enough to show that the conditions
(1) RHomA (M, N ) = 0.
(2) Ext1A (M, N ) = 0.
(3) N ∼
= M or N ∼ = M ⊗A KA .
are equivalent for ordinary points M, N or fat points M, N of Projnc A.
Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra with the generalized
Nakayama automorphism ν ∈ Aut k A. Any ordinary point of Projnc A is of the form M p
where p ∈ E by Proposition 4.4, and M p ⊗A KA ∼ = M(ν ∗ )−1 σ −3 ( p) by the proof of [26,
Theorem 3.5], so if M, N are ordinary points of Projnc A, then the result follows from
Proposition 4.22. If M, N are fat points of Projnc A, then the result follows from Proposition
4.24 and Proposition 4.25.

Recall that if R = k Q is a path algebra of a quiver Q whose underlying graph is extended
Dynkin, then it is known that two simple regular modules M, N over R are in the same
regular component if and only if N = ν1i (M) for some i ∈ Z [28]. By the above theorem,
we ask the following question.
Question 4.27 Let A = A(E, σ ) be a 3-dimensional quantum polynomial algebra, and
R = ∇ A the quantum Beilinson algebra of A. Can we define a suitable notion of 2-regular
component of R having the following properties?
(1) If R is not 2-representation tame, then 2-regular components of R are parameterized by
|E|/ ν ∗ σ 3 .
(2) If R is 2-representation
tame, then 2-regular components of R are parameterized by
|E|/ ν ∗ σ 3 |P2 \E|.
(3) In either case, two simple 2-regular modules M, N over R are in the same 2-regular
component if and only if N ∼ = ν2i (M) for some i ∈ Z.
Acknowledgments We thank O. Iyama and H. Minamoto for many discussions concerning this work. We
would like to mention that this paper is largely inspired by unpublished lecture notes [29,30] by S. P. Smith.
References
1. Ajitabh, K.: Modules over elliptic algebras and quantum planes. Proc. Lond. Math. Soc. 72, 567–587
(1996)
123
I. Mori
2. Alev, J., Dumas, F.: On the quotient field of some quantum algebras. J. Algebra 170, 229–265 (1994)
3. Artin, M.: Geometry of quantum planes. In: Azumaya Algebras, Actions, and Modules (Bloomington,
IN, 1990), 1–15, Contemp. Math., vol. 124. American Mathematical Society, Providence, RI (1992)
4. Artin, M., Schelter, W.: Graded algebras of global dimension 3. Adv. Math. 66, 171–216 (1987)
5. Artin, M., Tate, J., Van den Bergh, M.: Some algebras associated to automorphisms of elliptic curves. In:
The Grothendieck Festschrift, vol. 1, Progr. Math., vol. 86, pp. 33–85. Birkhauser, Boston (1990)
6. Artin, M., Tate, J., Van den Bergh, M.: Modules over regular algebras of dimension 3. Invent. Math. 106,
335–388 (1991)
7. Artin, M., Zhang, J.J.: Noncommutative projective schemes. Adv. Math. 109, 228–287 (1994)
8. Assem, I., Simson, D., Skowronski, A.: Elements of the representation theory of associative algebras,
vol. 1. In: Techniques of Representation Theory. London Mathematical Society Student Texts, vol. 65.
Cambridge University Press, Cambridge (2006)
9. Baer, D., Geigle, W., Lenzing, H.: The preprojective algebra of a tame hereditary Artin algebra. Comm.
Algebra 15, 425–457 (1987)
10. Bondal, A., Van den Bergh, M.: Generators and representability of functors in commutative and noncom-
mutative geometry. Mosc. Math. J. 3, 1–36 (2003)
11. Herschend, M., Iyama, O., Oppermann, S.: n-Representation infinite algebras. Adv. Math. 252, 292–342
(2014)
12. Jørgensen, P.: Local cohomology for non-commutative graded algebras. Comm. Algebra 25, 575–591
(1997)
13. Jørgensen, P., Zhang, J.J.: Gourmet’s guide to Gorensteinness. Adv. Math. 151, 313–345 (2000)
14. Keller, B.: Deformed Calabi–Yau completions, with an appendix by Michel Van den Bergh. J. Reine
Angew. Math. 654, 125–180 (2011)
15. Levasseur, T., Stafford, J.T.: The quantum coordinate ring of the special linear group. J. Pure Appl. Algebra
86, 181–186 (1993)
16. Minamoto, H.: Ampleness of two-sided tilting complexes. Int. Math. Res. Not. 1, 67–101 (2012)
17. Minamoto, H., Mori, I.: The structure of AS-Gorenstein algebras. Adv. Math. 226, 4061–4095 (2011)
18. Mori, I.: The center of some quantum projective planes. J. Algebra 204, 15–31 (1998)
19. Mori, I.: Noncommutative projective schemes and point schemes. In: Algebras, Rings and Their Repre-
sentations, pp. 215–239. World Scientific, Hackensack (2006)
20. Mori, I.: Co-point modules over Koszul algebras. J. Lond. Math. Soc. 74, 639–656 (2006)
21. Mori, I.: Co-point modules over Frobenius Koszul algebras. Comm. Algebra 36, 4659–4677 (2008)
22. Mori, I.: Asymmetry of Ext-groups. J. Algebra 322, 2235–2250 (2009)
23. Mori, I.: B-construction and C-construction. Comm. Algebra 41, 2071–2091 (2013)
24. Mori, I.: McKay type correspondence for AS-regular algebras. J. Lond. Math. Soc. 88, 97–117 (2013)
25. Mori, I., Smith, S.P.: Bézout’s theorem for non-commutative projective spaces. J. Pure Appl. Algebra
157, 279–299 (2001)
26. Mori, I., Ueyama, K.: Graded Morita equivalences for geometric AS-regular algebras. Glasg. Math. J.
55, 241–257 (2013)
27. Shelton, B., Vancliff, M.: Embedding a quantum rank three quadric in a quantum P3 . Comm. Algebra
27, 2877–2904 (1999)
28. Simson, D., Skowronski, A.: Elements of the representation theory of associative algebras, vol. 2. Tubes
and concealed algebras of Euclidean type. In: London Mathematical Society Student Texts, vol. 71.
Cambridge University Press, Cambridge (2007)
29. Smith, S.P.: Noncommutative Algebraic Geometry, Lecture Notes. University of Washington, Seattle
(1994)
30. Smith, S.P.: Noncommutative Algebraic Geometry, Lecture Notes. University of Washington, Seattle
(1999)
31. Stephenson, D.R., Zhang, J.J.: Growth of graded noetherian rings. Proc. Am. Math. Soc. 125, 1593–1605
(1997)
32. Vancliff, M., Van Rompay, K.: Embedding a quantum nonsingular quadric in a quantum P3 . J. Algebra
195, 93–129 (1997)
33. Van den Bergh, M.: Existence theorems for dualizing complexes over non-commutative graded and filtered
rings. J. Algebra 195, 662–679 (1997)
34. Zhang, J.J.: Twisted graded algebras and equivalences of graded categories. Proc. Lond. Math. Soc. 72,
281–311 (1996)
123

Mori 2015

Uploaded by

Copyright:

Available Formats

Mori 2015

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mori 2015

Uploaded by

Copyright:

Available Formats

Math. Z.

DOI 10.1007/s00209-015-1407-x Mathematische Zeitschrift

Regular modules over 2-dimensional quantum Beilinson

Received: 5 November 2014 / Accepted: 13 November 2014

Abstract In the study of a finite dimensional hereditary algebra of infinite representation

Keywords Regular modules · Beilinson algebras · AS-regular algebras · Representation

Mathematics Subject Classification 16G60 · 16S38 · 16S36 · 16W50 · 14A22

3.2 Points of a weighted projective line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

In representation theory of finite dimensional algebras, hereditary algebras and preprojective

1.1 Notations and terminologies

If A is N-graded, then we define the left exact functor m : GrMod A → GrMod A by

Let M ∈ GrMod A. An element m ∈ M is called torsion if m A ∈ GrMod A is right

1.2 Classes of algebras to study

We now define several classes of algebras which

(−)θ : GrMod A → GrMod Aθ . Since M ∈ Tors A if and only if M θ ∈ Tors Aθ , it induces

In this paper, if A is a quantum polynomial algebra, then we call ∇ A a quantum Beilinson

(1) The Beilinson algebra ∇ A is a d-representation infinite algebra.

gives an equivalence of triangulated categories.

2 The category of d-regular modules

2.1 The conditions (CM) and (LGK)

Let A be a graded algebra and M ∈ GrMod A. By the definition of GKdimension, if M

Theorem 2.5 Let A be either

Proof If A is of dimension d and of Gorenstein parameter , then D R m (A) ∼

2.2 The category of d-regular modules

in grmod A, GKdim M = max{GKdim L , GKdim N }, so M ∈ grmod≤i A if and only if

Proof Since grmod≤i+1 A is a Serre subcategory of an abelian category grmod A, it

tails0 A := grmod≤1 A/ tors A = grmod≤1 A/ grmod≤0 A =: tails0 A.

restricts to an equivalence functor

(3) If A satisfies (CM), then R(∇ A) ∼

Proof Let  be the Gorenstein parameter of A, ν ∈ Aut k A the generalized Nakayama

equivalence functor F commutes with direct limits, F restricts to an equivalence functor

0 −−−−→ n F(M) −−−−→ F(M) −−−−→ ωT πT F(M)

ldim A M := sup R m M = sup F(R m M) = sup R n F(M) =: ldim T F(M)

3 2-Dimensional weighted polynomial algebras

noetherian AS-regular algebra A is of the form A ∼ = k x, y/( f ) for some homogeneous

3.1 Points of a noncommutative projective scheme

There may be several ways of defining a notion of point of a noncommutative scheme. In

Definition 3.1 Let A be a graded algebra. We say that M ∈ GrMod A is i-critical if

It is easy to see the following properties of critical modules.

We denote by |X | the set of closed points of a commutative scheme X . It is easy to see

Proof Let π M ∈ Tails A be a point of Projnc A where M ∈ GrMod A is a 1-critical module.

Theorem 3.6 Let A be a right noetherian AS-regular algebra of dimension d + 1 satisfying

as rational functions for some f M (t) ∈ Z[t, t −1 ].

Lemma 3.8 The following algebras satisfy (H).

Let A be a finitely generated connected graded right coherent algebra and M, N ∈

By the above lemma, the following definition makes sense.

3.2 Points of a weighted projective line

Let A = k[x, y] be a 2-dimensional weighted polynomial algebra. Then it is reasonable

Proof Suppose that M ∈ grmod A is a cyclic critical Cohen–Macaulay module of depth 1

where bi ∈ N so that HM (t) = (1 − i≥1 bi t i )/(1 − t a )(1 − t b ). Since A is a commutative

Let A = k[x, y] with deg x = a, deg y = b ∈ N+ . For p0 = (0, 1), q0 = (1, 0) ∈ P1 ,

HM p0 (−i) (t) = t i /(1 − t b ),

in grmod B, so (M p )(ab) = B/(βx b − αy a ) =: N p . For q ∈ P1 \{ p0 , q0 }, if π A M p ∼

3.3 Regular modules and regular components

Proof If M ∈ grmod A is Cohen–Macaulay of depth 1, then

Since gcd(a, b) = 1, it follows that dimk Mi = 1 for all i ≥ ab by Lemma 3.17, so

If R = k Q is a path algebra of a quiver Q whose underlying graph is extended Dynkin,

Since gcd(a, ) = gcd(b, ) = 1, for any i, j ∈ Z, there exist i  , j  ∈ Z such that

Fix bases M2 = kx, M3 = ky, M4 = kx 2 , M5 = kx y, M6 = kx 3 . By the Hilbert series

Proof If A is of dimension d and of Gorenstein parameter , then D R m (A) ∼

Proof Let be the Gorenstein parameter of A, ν ∈ Aut k A the generalized Nakayama

noetherian AS-regular algebra A is of the form A ∼ = k x, y/( f ) for some homogeneous

Since gcd(a, ) = gcd(b, ) = 1, for any i, j ∈ Z, there exist i , j ∈ Z such that

Proof Since A is right noetherian, the functor grmod A → grmod A(r ) ; M → M (r ) is

Let A be an AS-Gorenstein algebra of dimension d and of Gorenstein parameter , and

Since k(Projnc A), k(Projnc A ) are division algebras over k, k(Projnc A) ∼

Proof We may assume that A = k x, y, z/(yz − αzy, zx − βx z, x y − γ yx) = A(E, σ )

is an exact sequence by the dimension calculation, so ExtiAz (Mz , Nz ) = 0 for all i ∈ Z.

Proof Let A be a 3-dimensional quantum polynomial algebra with an equivalence functor