INTRODUCTION TO STANLEY–REISNER RINGS

RYOTA OKAZAKI

(Non)-commutative Algebra and Topology,

Shinshu University,
Feb. 19–20-th, 2014

Abstract. This is the lecture note for the author’s talk in “(Non)-Commutative algebra
and Topology” at Shinshu University. This note (talk) is based on books on combinatorial
commutative algebra such as [10, 37, 52].

Contents
0. Notations and conventions 1
1. Face enumeration and Stanley–Reisner rings 2
1.1. Simplicial complexes and f-vectors (historical background) 2
1.2. Stanley–Reisner rings and h-vectors 6
1.3. Cohen–Macaulay-ness and Gorenstein-ness 8
1.4. More about h-vectors 12
2. Algebraic properties of Stanley–Reisner rings 14
2.1. Shellability and CM-ness 14
2.2. Alexander duality 18
3. Further developments 21
3.1. Simplicial posets 21
3.2. Squarefree modules and Alexander duality 22
3.3. Sheaves associated with squarefree modules 23
3.4. Toric face rings 23
Appendix A. A brief review of graded algebras and modules 24
References 25

0. Notations and conventions

In this note, unless otherwise said, all the rings are assumed to be commutative with the
identity element and the symbols below has the fixed meaning.
• n := a fixed positive integer.
• [n] := {1, . . . , n}.
• ∆ := a simplicial complex on [n].
• |∆| := a geometric realization of ∆.
• F (∆) := the set of the facets of ∆.
• He i (∆; k) (resp. H e i (∆; k) :=) the i-th reduced homology (resp. cohomology) of ∆ with
coefficients in k.
• χ
e(∆) := the reduced Euler characteristic of ∆.
This work was supported by JSPS KAKENHI Grant Number 24740013.
1
2 RYOTA OKAZAKI

• ∂P := the boundary of a polytope P .

• k := a field.
• S := k[x1 , . . . , xn ].
• m := (x1 , . . . , xn ).
• d := dim ∆ + 1.
• ei := i-th unit vector.
∑
• eF := i∈F ei for F ⊆ [n].
• For a ∈ Zn , supp(a) := {i ∈ [n] | ai > 0}.
• Z≥0 := {r ∈ Z | r ≥ 0}.
• ⪯ denotes the partial order on Zn defined by
a⪯b ⇐⇒ ai ≤ bi for all i.
• For a, b ∈ Zn , set [a, b] := {c ∈ Zn | a ⪯ c ⪯ b}.
• M (g) := the module whose degree is shifted by g.
• C • [r] := the complex translated by r.
• modG S := the category consisting of finitely generated G-graded S-modules and degree-
preserving S-homomorphisms.

1. Face enumeration and Stanley–Reisner rings

The theory of Stanley-Reisner rings has its origin at three famous theorems, called Upper
Bound Theorem, g-Theorem, and Hochster’s formula.

1.1. Simplicial complexes and f-vectors (historical background). Throughout this note,
n denotes a fixed positive number n, and set [n] := {1, . . . , n}.
Definition 1.1. Let 2[n] be the power set of [n].
(1) A subset ∆ ⊆ 2[n] is called a (finite abstract) simplicial complex1 (on [n]) if
F ⊆ G ⊆ [n] , G ∈ ∆ =⇒ F ∈∆
(2) An element F of ∆ is called a face of ∆, and its dimension, dim F , is defined by
dim F := #F − 1.
(3) The maximal faces of ∆ (w.r.t. ⊆) are called facets. The set of all the facets is denoted
by F (∆)
(4) The dimension dim ∆ of ∆ is then defined by
dim ∆ := max {dim F | F ∈ ∆} .
(5) For a simplicial complex ∆, a geometric simplicial complex |∆| can be constructed up
to homeomorphism, which is called a geometric realization of ∆.
(6) For F1 , . . . , Fr ⊆ 2[n] , set
⟨F1 , . . . , Fr ⟩ = {G ⊆ [n] | G ⊆ Fi for some i}

Remark 1.2. Whenever ∆ ̸= ∅, the complex ∆ has a unique face ∅ ∈ 2[n] of dimension −1.
Henceforth ∆ denotes a simplicial complex on [n] with ∆ ̸= ∅ of dimension d − 1, unless
otherwise said.
1All the simplicial complex ∆ may admit ghost vertices, that is, there may exist k ∈ [n] such that {k} ̸∈ ∆.
 INTRODUCTION TO STANLEY–REISNER RINGS 3

Definition 1.3. The vector f (∆) := (f−1 (∆), f0 (∆), . . . , fd−1 (∆)), where
fi (∆) := # {F ∈ ∆ | dim F = i} ,
is called the f -vector of ∆.
Remark 1.4.
f−1 (∆) ̸= 0 ⇐⇒ f−1 (∆) = 1 ⇐⇒ ∆ ̸= ∅.
Problem. Characterize vectors of integers which appear as f -vectors of some special class of
∆. For example, how about those for simplicial spheres, i.e., simplicial complexes ∆ such that
|∆| ≈ Sd−1 , where Sd−1 denotes a (d − 1)-dimensional sphere.
In the study of f -vectors, h-vectors are frequently considered instead of f -vectors themselves.
Definition 1.5. The vector h(∆) := (h0 (∆), . . . , hd (∆)) such that
∑
d ∑
d
hi (∆)t = i
fi−1 (∆)ti (1 − t)d−i
i=0 i=0

is called the h-vector of ∆.

Unless there’s no fear of confusion, hi (∆) and fi (∆) are simply denoted by hi and fi .
Lemma 1.6. For ∆ of dim d − 1,
∑
i ( ) i+1 (
∑ )
i−k d−k d−k
hi = (−1) fk−1 , fi = hk .
k=0
i−k k=0
i + 1 − k

In particular,
∑
d
h0 = f−1 = 1, h1 = f0 − d, hd = (−1) d−1
e(∆),
χ hi = fd−1 ,
i=0
∑d−1
e(∆) :=
where χ i
i=−1 (−1) fi = χ(∆) − 1.
As Lemma 1.6 shows,

knowing h(∆) is equivalent to knowing f (∆).

h-vectors can be easily computed by the following triangle like Pascal’s.

1 f0
1 f0 − 1 f1
. .. x y
..
.
. .. y−x fd−1
1 h1 h2 · · · ··· ··· hd

Example 1.7. (1) ∆ := ⟨{i, i + 1 | i = 1, . . . , n − 1} ∪ {{n, 1}}⟩. The geometric realiza-

tion |∆| is then an n-gon as in Figure 1.
4 RYOTA OKAZAKI

1 n

f (∆) = (1, n, n).

Moreover
1 n 2
1 n−1 n
1 n−2 1
Thus h(∆) = (1, n − 2, 1). 3 4
Figure 1. n-gon

(2) ∆ := ⟨{1, 2, 3} , {3, 4, 5}⟩. |∆| is as in Figure 2.

1 4

f (∆) = (1, 5, 6, 2), h(∆) = (1, 2, −1, 0). 3

2 5
Figure 2.

(3) ∆ := ⟨{1, 2, 3} , {1, 3, 4} , {1, 3, 5}⟩. |∆| = Figure 3.

1 4

f (∆) = (1, 5, 7, 3), h(∆) = (1, 2, 0, 0).

2 3 5
Figure 3.

(4) ∆ := ⟨{1, 2, 3} , {1, 3, 4} , {1, 4, 5} , {1, 2, 5} , {2, 3, 6} , {3, 4, 6} , {2, 5, 6} , {4, 5, 6}⟩. |∆| =
Figure 4.
 INTRODUCTION TO STANLEY–REISNER RINGS 5

5 4

f (∆) = (1, 6, 12, 8), h(∆) = (1, 3, 3, 1). 2 3

6
Figure 4. octahedron

(5) ∆ := ⟨{1, 2, 4} , {2, 4, 5} , {2, 3, 5} , {3, 5, 6} , {1, 3, 6} , {1, 4, 6}⟩. |∆| = Figure 5.

4 6
f (∆) = (1, 6, 12, 6), h(∆) = (1, 3, 3, −1).
5

2 3
Figure 5. cylinder

h-vectors first appear in Sommerville’s paper [47], where he did not give a name to them,
to describe a relation among components of f -vectors of boundary of polytopes, which is the
generalization of [14].
Recall that a convex polytope is called a simplicial polytope if each faces are simplex. Let
P(n, d) be the set of simplicial polytopes of dimension d with n vertices. Clearly, for P ∈
P(n, d),
∂P ≈ Sd−1 ,

where ∂P denotes the boundary complex of P . For ∆ = ∂P with P ∈ P(n, d) (more pre-
cisely when ∆ is the boundary complex of P ), Dehn and Sommerville discovered the following
beautiful equations.

Theorem 1.8 (Dehn–Sommerville equation for simplicial polytopes [14, 47]). For ∆ = ∂P
with P ∈ P(n, d) (hence dim ∆ = d − 1),

hi = hd−i

for all i.
6 RYOTA OKAZAKI

In the theorem, h0 = hd implies χ e(∆) = (−1)d−1 , which is just Euler’s polyhedron formula.
Actually the equations hold for simplicial spheres (more generally Eulerian complex), which
will be proved later.
Let {C(n, d) be a polytope given } as the convex hull of n (≥ d + 1) distinct points on the
curve (t, t , . . . , t ) ∈ R | t ∈ R . Such polytopes are called cyclic polytopes. It is well-known
2 d d

that C(n, d) ∈ P(n, d) and its combinatorial data (e.g. the f -vectors and the face poset of its
boundary complex (See Subsection 3.1 for the definition)) does not depend on the choice of
n-distinct points.
In 1957, T. S. Motzkin [38] claimed that ∂C(n, d) has a maximal f -vector, i.e.
fi (∂P ) ≤ fi (∂C(n, d)) (*)
holds for all i and P ∈ P(n, d).
Later on, McMullen proved the claim above by showing
Theorem 1.9 (Upper Bound Theorem (apprev. UBT) McMullen 1970 [34]). For ∆ = ∂P
with P ∈ P(n, d) and any i, ( )
n−d+i−1
hi ≤ .
i
Indeed, Dehn-Sommerville equation and Upper Bound Theorem imply Motzkin’s inequality,
since ( )
n−d+i−1
hi (∂C(n, d)) =
i
for all i with 0 ≤ i ≤ ⌊d/2⌋ (see [10, 52]).
Remark 1.10. One can also naturally define the notion of f -vector for convex polytopes that is
not necessarily simplicial. Clearly the inequality (*) still holds for such polytopes.
It is natural to ask whether UBT holds for simplicial spheres or not, but unfortunately
McMullen’s proof cannot be applied to them; his proof depends on the shellability of convex
polytopes (see Definition 2.1 and Theorem 2.4) and there is a simplicial sphere that is not
shellable (e.g. triangulations of non-PL spheres such as double suspension of Poincaré homology
3-sphere).
Stanley showed UBT for simplicial spheres proving that the corresponding Stanley–Reisner
ring is Cohen–Macaulay.
1.2. Stanley–Reisner rings and h-vectors. From now to the end, k denotes a field, let
S := k[x1 , . . . , xn ] be a polynomial ring over k, and set m := (x1 , . . . , xn ). The following
symbols will be used.
• Alphabets a := (a1 , . . . , an ), b := (b1 , . . . , bn ), . . . in the bold font with an underline
denotes elements of Zn .
∑
• ei denotes the i-th unit vector, and for F ⊆ [n], set eF := i∈F ei .
• ⪯ denotes the partial order on Zn defined by
a⪯b ⇐⇒ ai ≤ bi for all i.
• For a, b ∈ Zn , set [a, b] := {c ∈ Zn | a ⪯ c ⪯ b}.
∑ ∏
• For a ∈ Zn , set |a| := ni=1 ai and xa := ni=1 xai i .
∏
• For F ⊆ [n], set xF := xeF = i∈F xi .
• S has a natural structure of a Zn -graded ring with the grading given by
{
k · xa for a ∈ Zn≥0
Sa :=
0 otherwise,
 INTRODUCTION TO STANLEY–REISNER RINGS 7

where Z≥0 is the set of non-negative integers. Note that

– S0 = k, where 0 = (0, . . . , 0).
– Sei = k · xi .
• Also, S has a natural structure of Z-graded ring:
⊕
Si := Sa .
|a|=i

See Appendix A for basics of graded rings and modules. As is stated below Definition A.1, any
Zn -graded S-module is tacitly regarded as a Z-graded one with the natural Z-grading.
For G = Zn or Z, let modG S be the category consisting of finitely generated graded S-
modules and of degree-preserving S-homomorphisms.
Remark 1.11. The ideal m is clearly a maximal ideal of S, and moreover it is also the greatest
ideal among all the ideal belonging to modG S. In other words, S is graded local. By this
structure, S is equipped with many properties that a usual noetherian local ring has. See
[10, 21, 22]
Definition 1.12. For M ∈ modZ S,
∑
∞
HM (t) := (dimk Mi )ti ∈ Z[[t, t−1 ]]
i=−∞

is called the Hilbert Series of M .

The following is a classical result in commutative algebra. See [10, 31] for details.
Proposition 1.13. For M ∈ modZ S with d := dim M , where dim M denotes the Krull dimen-
sion of M , there exists a polynomial QM (t) ∈ Z[t, t−1 ] with QM (1) ̸= 0 such that
QM (t)
HM (t) = .
(1 − t)d
Definition 1.14. For not necessarily empty ∆ on [n], set
I∆ := (xF | F ⊆ [n] , F ̸∈ ∆) ⊆ S, k[∆] := S/I∆ .
The ideal I∆ is called the Stanley–Reisner ideal of ∆, and k[∆] the Stanley–Reisner ring of ∆.
Clearly I∆ , k[∆] ∈ modZn S. Note that I∆ is set to be (0) when ∆ = 2[n] , and I∆ = S when
∆ = ∅.
Recall that an ideal of S is called monomial if it is generated by monomials in S. A monomial
of the form xF with F ⊆ [n] is called a squarefree monomial, and an ideal of S is said to be
squarefree if it is generated by squarefree monomial ideals. Clearly I∆ is squarefree for any ∆.
Moreover there is the following one-to-one corresponding:
{ } { } { }
simplicial complexes squarefree monomial radical monomial
↔ =
on [n] ideals of S ideals of S
For a ∈ Zn , define
supp(a) := {i ∈ [n] | ai > 0} ,
and call it the support of a. For F ⊆ [n], set
pF := (xi | i ∈ [n] \ F ), k[F ] := S/pF .
Lemma 1.15. Let ∆1 , ∆2 be simplicial complexes on [n].
(1) I∆1 ∩∆2 = I∆1 + I∆2 and I∆1 ∪∆2 = I∆1 ∩ I∆2 .
8 RYOTA OKAZAKI
∩
(2) I∆ = F ∈F (∆) pF , and hence
dim k[∆] = max {#F | F ∈ F (∆)} = dim ∆ + 1.
(3) There is one-to-one corresponding
F (∆) ∋ F ←→ pF ∈ Ass(k[∆]),
where Ass(−) denotes the set of associated prime ideals (see Definition A.2 for the
definition).
(4) As Zn -graded k-vector spaces,
⊕
k[∆] = xF · k[F ].
F ∈∆

In particular, setting d := dim k[∆] = dim ∆ + 1, one obtains

∑ t#F
Hk[∆] (t) =
F ∈∆
(1 − t)#F

1 ∑d
1 ∑d
= fi−1 (∆)t (1 − t) =
i d−i
hi (∆)ti .
(1 − t)d i=0 (1 − t)d i=0

Proof. (1): an easy exercise. (2): an easy consequence of the fact that ∆ = ⟨F | F ∈ F (∆)⟩.
(3): use Proposition A.3 in conjunction with the fact that in the present case, the monomial m
in Proposition A.3 can be chosen to be xG for some G ∈ ∆. (4): follows from the fact that
0 ̸= xa ∈ k[∆] ⇐⇒ supp(a) ∈ ∆.
□
Thus,

we can apply some results in commutative algebra, to the study of h-vectors!

1.3. Cohen–Macaulay-ness and Gorenstein-ness.

Definition 1.16. Let M ∈ modZ S, and θ := θ1 , . . . , θd of homogeneous elements (w.r.t. Z-
grading) in S.
(1) The sequence θ is called a homogeneous system of parameters (abbrev. h.s.o.p) if d =
dim M and l(M/(θ)M ) < ∞, or equivalently (M/(θ)M )i = 0 for all i ≫ 0.
(2) θ is a (homogeneous) regular sequence on M in m if
(a) each θi is in m,
(b) M/(θ)M ̸= 0, and
(c) each θi is a non-zero divisor on M/(θ1 , . . . , θi−1 )M , where (θ1 , . . . , θi−1 ) is set to be
(0) when i = 1.
(3) The maximal length of homogeneous regular sequence on M in m is called the depth of
S, and denoted by depth M .
Lemma 1.17.
(1) Every M ∈ modZ S admits a h.s.o.p., and it is always algebraically independent over
the field k.
(2) Whenever #k = ∞, one always can be chosen a h.s.o.p. from S1 . Such a h.s.o.p. is
called a linear s.o.p. (abbrev. l.s.o.p.).
 INTRODUCTION TO STANLEY–REISNER RINGS 9

(3) depth M ≤ dim M (whenever M ̸= 0).

Sketch of Proof. The first two assertions follow from a classical result, so-called “Prime Avoid-
ance”, in commutative ring theory. The last one is an immediate consequence of the fact that
any homogeneous regular sequence is a part of some h.s.o.p. (see [10, 31] for details). □
Definition 1.18. An S-module M ∈ modZ S is said to be Cohen-Macaulay if it satisfies the
following conditions equivalent to each other:
(1) Any h.s.o.p θ = θ1 , . . . , θd is a regular sequence on M .
(2) There exists a h.s.o.p which is regular on M .
(3) depth M = dim M .
Remark 1.19. Formally, CM-ness and depth is defined for finitely generated modules over a
noetherian local ring, and a finitely generated module M over a noetherian ring A is said to
be CM if the localization Mp at p is a CM Rp -module for any p ∈ Spec(A). It is not trivial
but well-known that the conditions in Definition 1.18 is equivalent to say that Mp is CM for
all p ∈ Spec(S) and also that Mp is CM for all maximal ideal p of S. Moreover it follows that
depth M = depthSm Mm . See [10, 21, 22] for details.
In commutative algebra, especially in the theory of Cohen-Macaulay modules or Gorenstein
modules (Definition 1.24), canonical modules play a very important role to give a duality.
Definition 1.20. Let I be an ideal of S with I ∈ modG S, where G = Zn or Z, and set
R := S/I. Assume R is Cohen-Macaulay of dimension d. An R-module M ∈ modG R is said
to be a G-graded canonical module of R if
{
0 for i ̸= d,
ExtiR (k, M ) ∼
=
k for i = d,
in modG R.
Remark 1.21. A G-graded canonical module of R is necessarily CM as an S-module and it is
a maximal CM as an R-module (i.e., a CM R-module of dimension dim R). Moreover it is
indecomposable.
The notion of a canonical module also defined for a (not necessarily graded) Cohen-Macaulay
ring. If M ∈ modG R is a canonical module in the sense of Definition 1.20, then the shifted
modules M (g) where g ∈ G (see Definition A.1) are canonical in the non-graded sense, while not
in the sense of Definition 1.20. It is also noteworthy that with the hypothesis of Definition 1.20,
R necessarily admits a G-graded canonical module (see Proposition 1.23).
If R = S/I with I ∈ modG R is CM of dimension d and M is a Cohen-Macaulay R-module
of dimension t, then there exists a natural isomorphism
M −→ Extd−t d−t
R (ExtR (M, ωR ), ωR )

(see [10, 31]).

Set 1 := (1, . . . , 1) ∈ Zn and ωS := x1 · S = S(−1). It is well-known that ωS is a Zn -graded
canonical module of S.
Theorem 1.22 (Grothendieck vanishing and non-vanishing (cf. [10, 31])). For M ∈ modG S,
it follows that {
= 0 if i < depthS M or i > dimS M
Extn−i
S (M, ωS )
̸= 0 if i = depthS M, dimS M
The bottom non-vanishing Ext module is very important particularly in the CM case.
10 RYOTA OKAZAKI

Proposition 1.23 (cf. [10, 21, 22]). Assume R is CM of dimension d. Then ωR := Extn−d
S (R, ωS )
is a G-graded canonical module of R.
Definition 1.24. Let I be an ideal of S with I ∈ modZ S, and set R := S/I.
(1) The ring R is said to be Gorenstein (abbrev. Gor) if R is CM and R ∼ = ωR (−a(R)) for
some a(R) ∈ Z. The integer a(R) is called the a-invariant (or Gorenstein parameter)
of R.
(2) ∆ is said to be CM (resp. Gor) (over k) if so is k[∆].
(3) Finally, ∆ is said to be Gorenstein* (abbrev. Gor*) if ∆ is Gor and a(k[∆]) = 0.
Remark 1.25. It is well-know that CM-ness, Gor-ness, and Gor*-ness does not depend on an
extension of the base field k (under the assumption Definition 1.18 and 1.24). There several
characterization for CM-ness and Gor-ness. See [10, 31] for details.
Definition 1.26.
(1) For ∆ and F ∈ ∆, the link lk∆ F of ∆ with respect to F is defined to be the complex
lk∆ F := {G ∈ ∆ | F ∩ G = ∅, F ∪ G ∈ ∆} .
Note that lk∆ ∅ = ∆.
(2) ∆ is said to be pure if dim F = dim ∆ for all F ∈ F (∆).
e(lk∆ F ) = (−1)dim lk∆ F for all F ∈ ∆.
(3) ∆ is said to be Eulerian if it is pure and χ
Proposition 1.27. Set d := dim ∆ + 1 and R := k[∆]. The following hold.
( )#F
−1
∑ t
d
(1) (−1) HR (t ) = F ∈∆ (−1) d−dim F
e(lk∆ F )
χ . In particular, if ∆ is Eulerian,
1−t
then
(−1)d HR (t−1 ) = HR (t).
(2) Assume ∆ is CM over k. Then
(a) ∆ is pure.
(b) For every l.s.o.p. θ,

(1 − t)d HR (t) = HR/θR (t).

(c) HωR (t) = (−1)d HR (t−1 ).

Sketch of Proof. (1) follows from a simple calculation in conjunction with the fact that HR (t) =
∑
F ∈∆ (t/(1 − t)) and that t/(1 − t) is transformed into − (1 + t/(1 − t)) by the substitution
#F

t → t−1 .
(2) (a) It is well known that all the associated prime ideals of CM module have the same
codimension, which in conjunction with Lemma 1.15 implies that all F ∈ F (∆) have the same
cardinality.
(b) Set θ i := θ1 , . . . , θi for all i with 1 ≤ i ≤ d = dim k[∆] and θ 0 := 0. By the definition of
CM-ness,
( ) θi
0 −→ M/(θ i−1 )M (−1) − → M/(θ i−1 )M −→ M/(θ i )M −→ 0
is exact, which implies (1 − t)HM/(θi−1 )M (t) = HM/(θi )M (t).
(c) Needs some arguments. See [10, 31]. □
Corollary 1.28. Set d := dim ∆ + 1.
 INTRODUCTION TO STANLEY–REISNER RINGS 11

(1) (R. P. Stanley 1975 [48]) For a CM ∆,

( )
n−d+i−1
0 ≤ hi ≤
i
for all i. In particular, UBT holds for CM ∆.
(2) Dehn–Sommerville equation (hi = hd−i ) holds for any Eulerian ∆.
Proof. (1) Choose a l.s.o.p. θ of k[∆], set R := k[∆]/(θ), and let y1 , . . . , yn−d be a basis of
R1 (recall that θ is algebraically (hence linearly) independent over k). Clearly, the following
inequalities hold
( )
n−d−i+1
0 ≤ dimk Ri ≤ dimk (k[y1 , . . . , yn−d ])i = .
i
∑d
hd−i
(2) Immediate consequence of Proposition 1.27 and (−1)d HR (t−1 ) = i=0 d . □
(1 − t)
So the natural question is which complexes are CM or Eulerian? As for CM-ness the problem
is immediately settled by Hochster’s formula.
Theorem 1.29 (Hochster’s formula for Ext). For ∆, i ∈ Z, and a ∈ Zn with F := supp(a),
it follows that
{
Extn−i ∼ Hi−#F −1 (lk∆ F ; k) if ai ≥ 0 for all i,
S (k[∆], ωS )a =
0 o.w.

Sketch of Proof. Set d := dim ∆ + 1. It is known [10] that Extn−i

S (k[∆], ωS ) is isomorphic (in
modZn S) to the i-th homology of the following complex
0 −→ Dd −→ · · · −→ D1 −→ D0 −→ 0
such that
⊕
• Di = F ∈∆, #F =i k[F ] and
• Di → Di−1 is just the sum of the natural map k[F ] → k[F ′ ] with a suitable sign ±1,
where F ′ ⊆ F .
It is easy to verify that (D• )a ∼
=Ce• (lk∆ supp(a); k)[−#F − 1], where [−] means a homological
translation, if a ̸∈ Z≥0 and otherwise (D• )a = 0.
n
□
Remark 1.30. Hochster’s formula is usually (and quite often) introduced as the formula for the
local cohomologies. By the graded local duality, the above formula is equivalent to the usual
one. See [10, 52] for the usual formula.
Corollary 1.31.
(1) (Reisner 1976 [41]) ∆ is CM over k if and only if He i (lk∆ F ; k) = 0 for all F ∈ ∆ and
i < dim lk∆ F (= dim ∆ − #F , actually).
(2) ∆ is Gor* over k if and only if
{
k if i = dim lk∆ F ,
Hi (lk∆ F ; k) ∼
=
0 otherwise
for all F ∈ ∆.
Proof. (1) Immediate consequence of Hochster’s formula.
(2) This follows from Hochster’s formula and the fact that k[∆] ∼
= M in modZn k[∆] for any
indecomposable M ∈ modZn k[∆] with k[∆] ∼ = M as Zn -graded k-vector spaces (See [10]). □
12 RYOTA OKAZAKI

Note that homologies of links are isomorphic to local homologies.

Proposition 1.32 (Munkres [39, Lemma 3.3]). Set X := |∆|. For p ∈ X, a simplex σ of X
with p ∈ relint σ, and a face F ∈ ∆ corresponding to σ,
Hi (X, X \ {p} ; k) ∼
=He i−#F −1 (lk∆ F ; k)
for all i.
Consequently, the following hold.
Corollary 1.33.
(1) (Munkres and Stanley) The following hold.
(a) ∆ is CM over if and only if Hi (X; k) ∼
= Hi (X, X \ {p} ; k) = 0 for all p ∈ X and
i < dim X.
(b) ∆ is Gor* over k if and only if
{
k if i = dim X,
Hi (X; k) ∼
= Hi (X, X \ {p}) ∼
=
0 otherwise.
(c) In particular, CM-ness and Gor*-ness of ∆ depends only on X and k.
(2) ∆ is Gor* if and only if ∆ is CM and Eulerian.
(3) A simplicial sphere is Gor*. In particular, Dehn–Sommerville equation and UBT holds
for simplicial spheres.
(4) If |∆| is a homology manifold (with or without boundary) and its homology vanishes
expect for the top homology, then ∆ is CM.
Remark 1.34. CM-ness depends on the characteristic of k. For example, let ∆ be a triangulation
of RP 2 . Since RP 2 is a manifold, it follows that ∆ is CM over k when char k ̸= 2 and not CM
when char k = 2.
1.4. More about h-vectors.
Lemma 1.35. Let r be a positive integer. For any k ∈ Z with k > 0, there exist unique integers
k(r) > k(d − 1) > · · · > k(1) ≥ 0 such that
∑r ( )
k(i)
k= . (1.1)
i=1
i

Proof. An easy exercise. □

The representation in (1.1) is called the r-th Macaulay representation (abbrev. M-rep.). For
a positive integer k with the M-rep.
∑r ( )
k(i)
k= ,
i=1
i
set
∑r ( )
⟨r⟩ k(i) + 1
k = .
i=1
i+1

Theorem 1.36 (Macaulay and Stanley). For (h0 , . . . , hd ) ∈ Zd+1 , the following are equivalent.
∑
(1) There exists an ideal I of S with I ∈ modZ S such that HS/I (t) = di=0 hi ti (and hence
S/I is finite-dimensional over k).
 INTRODUCTION TO STANLEY–REISNER RINGS 13

(2) There exists an ideal I of S with I ∈ modZ S such that S/I is Cohen-Macaulay of
dimension d and ∑d
hi t i
HS/I (t) = i=0 d .
(1 − t)
⟨i⟩
(3) h0 = 1 and 0 ≤ hi+1 ≤ hi for all i ≥ 1.
Definition 1.37. A vector satisfying the conditions in Theorem 1.36 called an M-vector.
In 1971, McMullen [35] conjectured that for an vector h := (h0 , . . . , hd ) ∈ Zd+1 the following
are equivalent:
∪
(1) There exists P ∈ n P(n, d) with h(∂P ) = h.
(2) Set gi := hi − hi−1 . It follows that
(a) h0 = 1,
(b) hi = hd−i for all i, and
(c) (g0 , . . . , g⌊d/2⌋ ) is an M -vector.
This conjecture, called g-conjecture, has already been proved to be true by Billera, Lee, and
Stanley.
Theorem 1.38 (g-theorem). McMullen’s conjecture is true.
Billera–Lee [3, 4] proved (2) ⇒ (1), and Stanley [50] proved the inverse making use of the
theory of toric variety and the hard Lefschetz theorem (See also [19, 52]). Now the following
conjecture naturally comes up.
Conjecture 1.39 ((new) g-conjecture). The same characterization of h-vectors holds for sim-
plicial spheres.
This conjecture is still open. Clearly, the only problem is whether the h-vector of any
simplicial sphere satisfies the above three conditions (only the condition (c) remains to be
proven) or not.
More generally, it is natural to try to find the characterization of h-vectors for (homological
or topological) manifolds (with or without boundary). A manifold is not necessarily CM, while
it satisfies parts of the conditions for CM-ness.
Definition 1.40.
(1) k[∆] with dim k[∆] = d is said to be Buchsbaum (abbrev. Bbm) if Extn−i
S (k[∆], ωS )a = 0
for all i ̸= d and a ̸= 0.
(2) ∆ is said to be Bbm over k if k[∆] is Bbm.
Remark 1.41. In general, Bbm-ness is more subtle and the definition above is valid almost only
for Stanley-Reisner rings.
By Hochster’s formula for Ext and Munkres’ isomorphism, the following is clear.
Theorem 1.42 (Miyazaki and Schenzel). Let dim ∆ = d − 1 and set X := |∆|. The following
are equivalent:
(1) ∆ is Bbm over k.
(2) e i (lk∆ F ; k) = 0 for all F ̸= ∅ and i < d − #F − 1.
H
(3) e i (X, X \ {p}) = 0 for all p ∈ Z and i < dim X.
H
In particular, if X is a homology manifold (with or without ∂), then ∆ is Bbm.
14 RYOTA OKAZAKI

Now we obtain the following implications.

homology mfd
simplicial sphere +3 whose homologies are vanished +3 homology mfd
expect for the top one

  
Gor* +3 CM +3 Bbm

As a corollary of the characterization of h-vectors of Hilbert series of CM graded k-algebra due

to Macaulay and Stanley, the following holds.
Theorem 1.43 (Stanley). For h = (h0 , . . . , hd ) ∈ Zn+1 , the following are equivalent.
(1) h = h(∆) for some CM ∆ of dimension d.
(2) h is an M -vector.
As for Bbm complexes, it is still an open problem to characterize their h-vectors. (Note that
components of h-vector of a Bbm complex could be negative (e.g. the cylinder above)).
Problem 1.44 (Hibi [26]). Give a (combinatorial) characterization of h-vectors of Bbm ∆.
There are some results on h-vectors of Bbm ∆. See [44, 53] for details.
2. Algebraic properties of Stanley–Reisner rings
2.1. Shellability and CM-ness. As is stated in the previous section, McMullen proved UBT
by the shellability of simplicial polytopes. Let us recall the definition of shellability. Recall
that F (∆) denotes the set of all the facets of ∆.
Definition 2.1. A simplicial complex ∆ is said to be shellable if there is a linear ordering,
called shelling, F1 , . . . , Fr on F (∆) with the following equivalent conditions:
(1) ⟨Fi ⟩ ∩ ⟨F1 , . . . , Fi−1 ⟩ is pure (actually pure shellable) of dimension dim Fi − 1 for each i.
⨿
(2) ⟨Fi ⟩ \ ⟨F1 , . . . , Fi−1 ⟩ has a unique minimal face Gi for each i. Hence ∆ = ri=1 [Gi , Fi ],
where G1 := ∅ and [Gi , Fi ] := {G ⊆ [n] | Gi ⊆ G ⊆ Fi }.
Remark 2.2. Each ⟨Fi ⟩ ∩ ⟨F1 , . . . , Fi−1 ⟩ is homeomorphic to a sphere or a ball. Moreover it is
well-known [6] that any shellable complex is contractible or has a homotopy type of wedge of
spheres (of various dimension).
Example 2.3. In Example 1.7, the simplicial complexes in Figures 2 and 5 are not shellable,
and the other complexes are pure and shellable. The complex in the following Figure 6 is not
pure but shellable.

1 4

2 5

Figure 6.
 INTRODUCTION TO STANLEY–REISNER RINGS 15

The ordering {1, 2, 3} , {3, 4} , {4, 5} , {3, 5} is a shelling of this complex, and one obtains the
decomposition
⨿ ⨿ ⨿
[∅, {1, 2, 3}] [{4} , {3, 4}] [{5} , {4, 5}] [{3, 5} , {3, 5}] .

McMullen used the following fact to prove UBT for simplicial polytopes.
Theorem 2.4 (Bruggesser–Mani [7]). Every boundary complex of a simplicial polytope is pure
and shellable.
Recall that, on the other hand, Stanley’s proof of UBT uses CM-ness the boundary complex.
So it is natural to expect that there is a relation between pure-shellability and CM-ness.
Note that for two simplicial complex ∆, ∆′ ,
∆ ⊆ ∆′ ⇐⇒ I∆ ⊇ I∆′ .
Proposition 2.5. A pure shellable ∆ is CM over any field k.
Sketch of Proof. Let F1 , . . . , Fr be a shelling of ∆ and set ∆i := ⟨F1 , . . . , Fi ⟩. Since there exists
the following exact sequence
0 −→ k[∆i ] −→ k[∆i−1 ] ⊕ k[∆i ] −→ k[∆i−1 ∩ ∆i ] −→ 0
induced from the assertion (1) of Lemma 1.15 and since each k[∆i−1 ∩ ∆i ] is CM of dimension
d − 1 (which is easy to verify), the desired assertion follows from the induction on r and the
long exact sequence of Ext modules induced from the short exact sequence above. □

In view of k[∆], a shelling can be considered as a filtration of k[∆]. There is another proof
in this point of view. (Compare this with Theorem 2.7).

Another Proof. Let F1 , . . . , Fr be a shelling of ∆, G1 := ∅, and G2 , . . . , Gr unique minimal

faces. Set ∆r−i := ⟨F1 , . . . , Fi ⟩ for i = 1, . . . , r. The filtration
∅ = ∆r ⊂ ∆r−1 ⊂ · · · ⊂ ∆1 ⊂ ∆0 = ∆
induces the one
I∆ = I∆0 ⊂ I∆1 ⊂ I∆2 ⊂ · · · ⊂ I∆r−1 ⊂ I∆r = S,
and hence
0 ⊂ I∆1 /I∆ ⊂ I∆2 /I∆ ⊂ · · · ⊂ I∆r−1 /I∆ ⊂ k[∆]. (2.1)
By the definition of shelling, I∆i+1 /I∆i ∼
= (S/pFi+1 )(−eGi+1 ) for i = 0, . . . , r − 1. Now use the
long exact sequence of Ext and the fact that each S/pFi+1 is CM of dim dim k[∆] (since ∆ is
pure). □
Remark 2.6. The inverse of Proposition 2.5 is false. For example, every triangulation of a ball
or a sphere is CM over any k, while there exists a unshellable triangulation of a ball and a
sphere (see the end of Subsection 1.1 and [43, 57]).
As for shellability, Dress gives the following algebraic criterion.
Theorem 2.7 (Dress [15]). A simplicial complex ∆ is shellable if and only if M := k[∆] is
clean, i.e., there is a filtration
0 = Φ0 (M ) ⊊ Φ1 (M ) ⊊ · · · ⊊ Φr (M ) = M
such that Φi (M )/Φi−1 ∼
= S/p for some minimal prime ideal p of M .
16 RYOTA OKAZAKI

Now it is natural to quest a suitable notion to fit the “???” below.

∆ : pure shellable +3 ∆ : CM

 
∆ : shellable +3 ∆ : ???

The required notion is a sequential Cohen–Macaulay-ness.

Definition 2.8 (cf. [45, 52]). Let M ∈ modZn S.

(1) M is said to be sequentially Cohen–Macaulay (abbrev. SCM) if there exists a filtration,
called a Cohen–Macaulay filtration

0 = Φ0 (M ) ⊊ Φ1 (M ) ⊊ · · · ⊊ Φr (M ) = M

with Φi (M ) ∈ modZn S such that

(a) dim(Φi (M )/Φi−1 (M )) < dim(Φi+1 (M )/Φi (M )) for each i, and
(b) Each quotient Φi (M )/Φi−1 (M ) is CM.
(2) A simplicial complex ∆ is said to be SCM over k if so is k[∆].
Remark 2.9. A Cohen-Macaulay filtration, if exists, is unique, and moreover

dim(Φi (M )) = dim (Φi (M )/Φi−1 (M )) (2.2)

for all i. In [45], Schenzel considers SCM modules over a (not necessarily graded) noetherian
ring. There, he calls SCM modules Cohen-Macaulay filtered modules and studies algebraic
aspects of SCM modules in view of commutative ring theory.

Proposition 2.10. For ∆,

(1) ∆ is CM over k if and only if it is pure and SCM over k.
(2) A shellable ∆ is SCM over any field k.
Proof. (1) The implication ⇒ is clear. For the inverse, note that if follows from Lemma 1.15
that all the associated prime ideals of k[∆] has the same codimension, which implies that the
length of Cohen-Macaulay filtration of k[∆] must be less than or equal to 1.
(2) As is well-known [6], each shelling F1 , . . . , Fr can be arranged so that dim F1 ≥ dim F2 ≥
· · · ≥ Fr . The filtration as in (2.1) induced from the arranged shelling is then a CM filtration.
□

Definition 2.11. For ∆ and i ∈ Z, set

∪
∆[i] := ⟨F | F ∈ ∆, dim F = i⟩ , ∆⟨i⟩ := ∆[k]
k≥i

∆[i] is called the pure i-skeleton of ∆.

Example 2.12. Let ∆ be the simplicial complex in Figure 6. The complexes ∆[2] and ∆[1] are
as follows.
 INTRODUCTION TO STANLEY–REISNER RINGS 17

1 4 1 4

3 3

2 5 2 5

Figure 7. ∆[2] Figure 8. ∆[1]

Set d = dim ∆ + 1, and
Φj (k[∆]) := I∆⟨j⟩ /I∆
for j = −1, 0, . . . , d. One obtains the following filtration.
0 = Φ−1 (k[∆]) ⊆ Φ0 (k[∆]) ⊆ · · · ⊆ Φd−1 (k[∆]) ⊆ Φd (k[∆]) = k[∆] (2.3)
Theorem 2.13. For ∆ above, the following are equivalent.
(1) ∆ is SCM over k.
(2) (Stanley) Each Φi (k[∆])/Φi−1 (k[∆]) is equal to 0 or CM over k.
(3) (Duval) Each ∆[i] is empty or CM over k, for all i.
(4) (Peskine and Stanley) Each Extn−iS (k[∆], ωS ) is 0 or CM of dimension i.

Sketch of Proof. See [16] for the equivalence (2) ⇔ (3).

The implication (2) ⇒ (4) is an easy consequence of the following: the first is the fact that
dim (Φi (k[∆])/Φi−1 (k[∆])) = i whenever Φi (k[∆]) ̸= Φi−1 (k[∆]), and the second is the long
exact sequences of Ext modules induced from each short exact sequences given by the filtration
(2.3).
(4) ⇒ (1): Set DS• := ωS [n], where [n] denotes the translation of complex. The assertion can
be shown by the spectral sequence
E2p,q := Extn+p (Extn−q (k[∆], ωS ), ωS ) ⇒ H p+q (R HomS (R Hom(k[∆], DS• ), DS• ))
and the fact that R Hom(R Hom(k[∆], DS• ), DS• ) is quasi-isomorphic to the complex where the
0-th component is k[∆] and the others are 0.
(1) ⇒ (2): Set R := k[∆] and let
0 = Ψ0 (R) ⊊ Ψ1 (R) ⊊ · · · ⊊ Ψr (R) = R
be a CM filtration. For i = 1, . . . , r, set di := dim Ψi (R)/Ψi−1 (R) = dim Ψi (R). The assertion
immediately holds when one shows that the filtration coincides with the filtration given by
skipping all the Φi (R) with Φi (R) = Φi−1 (R) in (2.3). The last assertion follows from the
following facts:
• dim Φi (R) ≤ i.
• dim M = dim Φi+1 (R) for any M ∈ modZn S with Φi (R) ⊊ M ⊆ Φi+1 (R).
• each Ψi (R) is the largest Zn -graded submodule of R with dimension less than or equal
to
di := dim Ψi (R)/Ψi−1 (R) = dim Ψi (R),
and is equal to Φdi (R).
□
18 RYOTA OKAZAKI

2.2. Alexander duality.

Definition 2.14. For ∆ on [n], define
∆∨ := {F ⊆ [n] | F c := [n] \ F ̸∈ ∆} .
The simplicial complex ∆∨ is called the Alexander dual of ∆ with respect to [n]. Clearly,
I∆∨ := (xF c | F ∈ F (∆)).
Theorem 2.15 ((Combinatorial) Alexander duality (cf. [17])). For all i,
e i−2 (∆∨ ; k) ∼
H =He n−i−1 (∆; k).

Sketch of Proof. Let Γ := 2[n] be the (n − 1)-simplex. From the one-to-one corresponding
∆ ∋ F 7→ F c ∈ Γ \ ∆∨ (and suitable choice of orientations), we get the isomorphism of
complexes ( )
∼
e• (∆; k) −
C
=
→ C e• (Γ; k)/C
e• (∆∨ ; k) [n − 2].
Since Γ is contractible, it follows that
Hn−i−1 (∆; k) ∼= H i−3 (Γ, ∆∨ ; k) ∼
= H i−2 (∆∨ ; k).
□
Corollary 2.16 (cf. [17]). For F ⊆ [n] and i,
He i−2 (lk∆∨ F ; k) ∼
=He n−#F −i−1 (∆F c ; k).

Sketch of Proof. Note that lk∆∨ F = (∆F c )∨ as simplicial complexes on F c , where ∨ in the right
hand denotes the Alexander dual with respect to F c . Applying Theorem 2.15, one obtains the
desired isomorphism. □
Definition 2.17. For i ∈ Z and g ∈ G, the non-negative integer
βi,g := βi,g (M ) := dimk TorSi (M, k)g
is called the (i, g)-th graded Betti number of M .
Remark 2.18. As is well-known, S has a finite global dimension n, and hence βi,g = 0 for all
i > n. Moreover each βi,g can be characterized as follows: there exists a minimal graded free
resolution of M such that
⊕ ⊕ ⊕
P• : · · · −→ S(−g)βi,g −→ · · · S(−g)β1,g −→ S(−g)β0,g −→ 0,
g∈G g g

with Hi (P• ) = 0 for i ̸= 0 and H0 (P• ) ∼

= M.
Definition 2.19. Let M ∈ modZ S.
(1) The integer indeg M := inf {i ∈ Z | Mi ̸= 0} is called the initial degree of M .
{ }
(2) Set regS M := sup r ∈ Z | βi,i+r (M ) ̸= 0 ∃ i . (Note that indeg M ≤ regS M < ∞
whenever M ̸= 0). The integer regS M is called the Castelnuovo-Mumford regularity of
M.
(3) M is said to have a q-linear resolution if indeg M = regS M , or equivalently M has a
graded free resolution of the following form
· · · −→ S(−q − 2)β2,q+2 −→ S(−q − 1)β1,q+1 −→ S(−q)β0,q −→ 0.

Betti numbers are important in commutative ring theory. For example, the following hold.
Proposition 2.20 (cf. [10, 31]). For M ∈ modZ S with M ̸= 0, it follows that
{ }
(1) projdimS M := max i ∈ Z | ∃ j s.t. βi,j ̸= 0 (≤ n = gldim S).
 INTRODUCTION TO STANLEY–REISNER RINGS 19

(2) (Auslander–Buchsbaum formula) projdimS M = n − depth M . In particular, M is CM

if and only if projdimS M = n − dim M (recall that dim S = n), or equivalently βi,j = 0
for all i > n − dim M .
Let θ := θ1 , . . . , θd be the sequence of the variables of S, and consider the exterior algebra
K• (θ; S) := S ⟨y1 , . . . , yd | d(yi ) = θi ⟩ , (2.4)
where yi is a S-free basis of homological degree 1.
Definition 2.21. With the above notation,
(1) K• (θ; S) is called the Koszul complex of S with respect to θ.
(2) For an S-module M , the Koszul complex of M with respect to θ is defined to be
K• (θ; M ) := K• (θ; S) ⊗S M.

The following is well-known.

Proposition 2.22 (cf. [10, 31]). Let x := x1 , . . . , xn be the sequence of variables of S. The
Koszul complex K• (x; S) then gives a Zn -graded minimal S-free resolution of k, where each
degree (in the sense of Zn -grading) of yi in (2.4) is set to be ei .
Theorem 2.23 (Hochster (cf. [10, 52]). For a ∈ Zn with F := supp(a), it follows that
{
e #F −i−1 (∆F ; k) if a ∈ [0, 1] := {b ∈ Zn | 0 ⪯ b ⪯ 1} ,
dimk H
βi,a (k[∆]) =
0 o.w.

Proof. For simplicity, set R := k[∆]. By Proposition 2.22, ⊕the Koszul complex K• := K• (x; S)
is a minimal Z -graded free resolution
n
∧ of k. Let V := i=1 k · vi be the Z -graded k-vector
n n
∧
space with basis vi of degree ei , and V be the exterior algebra of V over k. Regarding V
as just a Zn -graded k-vector space, not a differential graded algebra, one can describe K• as
follows. For σ = {i1 , . . . , is } ⊆ [n] with i1 < · · · < is , let vσ denote vi1 ∧ · · · ∧ vis . Clearly
∧
Ki = S ⊗k i V and the differential maps of K• is described (with a suitable choice of sign ±)
as
∑s
Ks ∋ z ⊗ vσ 7→ ±(zxi ) ⊗ vσ\{ij } ∈ Ks−1 ,
j=1

where σ = {i1 , . . . , is } with 1 ≤ i1 < · · · < is ≤ n. It thus follows that

∧
TorSi (R, k) ∼
= H −i (R ⊗S K• ) ∼= Hi (R ⊗k V)
in modZn S.
It is an easy exercise to verify that for a ∈ [0, 1] with supp(a) = F , the assignment
ei (∆F ; k) = Homk (C
C ei (∆F ; k), k) ∋ y ∗ 7→ ±xG ⊗ vF \G ∈ (K#F −i−1 )a ,
G

where yG denotes the basis of C ei (∆F ; k) corresponding to G ∈ ∆F and y ∗ its dual basis, gives
G
rise to the isomorphism C e• (∆F ; k)[#F − 1] → (K• )a .
If a ̸∈ Zn≥0 , then (K• )a = 0. In the remaining case where a ⪰ 0 and a ̸⪯ 1, it then follows
′
that a′ := esupp(a) ≺ a. Setting u := xa−a ∈ m, one can easily verify that the multiplication
map u· : (K• )a′ → (K• )a is isomorphism of complexes. Therefore H• (K• )a = 0 since H• (K• )
is annihilated by m. □
Corollary 2.24 (Eagon–Reinser [17]). For ∆ with dim ∆ = d − 1, ∆ is CM over k if and only
if I∆∨ has a (n − d)-linear resolution.
20 RYOTA OKAZAKI

Proof. Note that indeg I∆∨ = n − d. On the other hand, from Hochster’s formulas (Theo-
rems 1.29 and 2.23), Alexander duality, and βi,a (I∆∨ ) = βi−1,a (k[∆∨ ]) for all i, it follows that
∆ is CM over k ⇐⇒ regS I∆∨ ≤ n − d.
□
Now it is natural to ask about the blank below.
∆ : CM ks +3 I∆∨ : has a lin. res.

 
∆ : SCM ks +3 I∆∨ : ???.

Definition 2.25. Let I be an ideal of S with I ∈ modZ S.

(1) For i ∈ Z, set I⟨i⟩ = (f | f ∈ Ii ).
(2) I is said to be component-wise linear if each I⟨i⟩ has a i-linear resolution.
Theorem 2.26 (Herzog–Hibi [24]). k[∆] is SCM if and only if I∆∨ is component-wise linear.
Sketch of Proof. For a simplicial complex Γ and i ∈ Z with 0 ≤ k ≤ n, set
(IΓ )[k] := (xF | F ̸∈ Γ, #F = k) .
It is known [24, Proposition 1.5] that IΓ is component-wise linear if and only if (IΓ )[k] has a
linear resolution for all k with 0 ≤ k ≤ n.
The assertion is now an immediate consequence from the fact above, Eagon-Reiner’s formula,
Duval’s criterion for SCM-ness, and the following fact
(I∆∨ )[k] = I(∆[n−k−1] )∨
for k = 1, . . . , n, which can be easily verified. □
We will complete this section to mention the result on componentwise linear ideals by Herzog,
Reiner and Welker. First let us recall the definition of Golod rings. See [2, 23] for details.
Let I be an ideal of S with I ∈ modZ S, and set R := S/I. Note that R is graded local with
the unique maximal ideal m = (x1 , . . . , xn ), where the maximal ideal m of S is identified its
image by the natural surjective homomorphism S ↠ R. Clearly k ∼ = R/m ∼ = Rm /mRm .
Definition 2.27. Let M ∈ modZ R.
(1) The formal power series
∑
∞

i (k, M ))t ∈ Z[[t]]

dimk (TorR
R i
PM (t) :=
i=0

is called the Poincaré series of M .

(2) The Poincaré series of R is, by definition, PkR (t).
It is well-known that the following inequality holds.
Proposition 2.28 (Serre (cf. [2, 23])). Let R, M be as above. There is the following inequality
S
PM (t)
R
PM (t) ⩽ , (2.5)
1 − t(PR (t) − 1)
S

where ⩽ means that each coefficients of the formal power series PM

R
are greater than or equal
to those of the rational function in the left hand.
 INTRODUCTION TO STANLEY–REISNER RINGS 21

Remark 2.29. (1) In [2, 23], the same inequality as (2.5) is proved under the condition that S
is a regular local ring, R a noetherian local ring, and S ↠ R a homomorphism of local rings.
The inequality (2.5) is an immediate consequence of this equality, since
TorS (M, k) ∼
• = TorSm (Mm , Sm /mSm ), TorR (M, k) ∼
• • = TorRm (Mm , Rm /mRm )
•

for any M ∈ modZ R.

S
(2) Since S is regular, both of PM (t) and PRS (t) is just a polynomial of degree less than or
equal to n.
Definition 2.30. With the notation above, the k-algebra R is said to be Golod if the local ring
Rm is Gold, or equivalently the equality holds in (2.5) for the residue field k, that is,
PkS (t)
PkR (t) = . (2.6)
1 − t(PRS (t) − 1)
By Proposition 2.22, one can easily verify that
∑n
S
Pk (t) = dimk Hi (K• (x; k))ti = (1 + t)n .
i=0
It thus follows that R is Golod if and only if
(1 + t)n
PkR (t) = .
1 − t(PRS (t) − 1)
Clearly a Golod ring gives a typical example of those whose Poincaré series is rational. Whether
a Poincaré series is always rational or not was a problem posed by Serre and Kaplansky ([46]).
This problem has already been resolved negatively in Anick’s famous paper [1].
The name “Golod” comes from Golod’s characterization for rings satisfying (2.6). Recall that
Koszul complex K• (x; R) has a structure of DGA, and hence one can define Massey products
(also called Massey operation) on H• (K• (x; R)). See [23, 28, 32, 33] for the definition of Massey
products.
Theorem 2.31 (Golod [20] (cf. [2, 23])). With the above notation, R is Golod if and only if
all the Massey products (Massey operation) on H• (K• (x; R)) = TorS• (k, R) vanishes.
Herzog, Reiner, and Welker proved the following making use of this characterization.
Theorem 2.32 (Herzog–Reiner–Welker, [25]). For a componentwise linear ideal I, the Z-graded
k-algebra S/I is Golod.

3. Further developments
3.1. Simplicial posets. In this subsection, all the posets (i.e. partially ordered sets) are
assumed to be finite. For a simplicial complex ∆, ∆ itself together with the inclusion ⊆ forms
the poset, called the face poset. Let P be a poset with the order ≤. Recall that
(1) P is called a Boolean lattice (or Boolean algebra) if it is isomorphic to a face poset of a
simplex.
(2) P is said to be simplicial if it satisfies the following conditions:
(a) P has the least element{ 0̂. }
(b) each intervals [0̂, p] := p′ ∈ P | 0̂ ≤ p′ ≤ p are a Boolean lattice for all p ∈ P .
Clearly, a face poset of a simplicial complex is simplicial. More generally it is well-known [5]
that every simplicial poset appears as the face poset of a regular CW complex XP , i.e., the set
of the closed cells of X with the order given by inclusion (while the face poset of an arbitrary
regular CW complex is not necessarily simplicial).
22 RYOTA OKAZAKI

The notion of f -vectors and h-vectors can be generalized to simplicial posets as follows. Let
P be a simplicial poset. For p ∈ P , the interval [0̂, p] is then isomorphic to the face poset of
a (r − 1)-simplex. The integer r is called the rank of p, denoted by rk p; hence rk p is just the
maximal length of the chains in P that ends at p. The rank of P , denoted by rk P , is defined
to be max {rk p | p ∈ P }. Set d := rk P . The vector f (P ) := (f−1 , . . . , fd−1 ), where

fi := # {p ∈ P | rk p = i + 1} ,

is called the f -vector of P , and the h-vector of P is then defined to be h(P ) := (h0 , . . . , hd ),
where
∑d ∑d
i
hi t = fi−1 ti (1 − t)d−i .
i=0 i=0

A simplicial poset is called a simplicial cell sphere if (the underlying space of) its associated
regular CW complex XP is homeomorphic to a sphere. For a simplicial cell sphere, the following
characterization of h-vectors are well-known.

Theorem 3.1 (Masuda [29] and Stanley [51]). A vector (h0 , . . . , hd ) ∈ Zd+1 is the h-vector of
a simplicial cell sphere if and only if it satisfies the following conditions:
(1) h0 = 1 and hi = hd−i for all i,
(2) hi ≥ 0 for all i, and
∑
(3) if hi = 0 for some i with 1 ≤ i ≤ d − 1, then the sum di=0 hi is even.
The “if” part is proved by Stanley [51] and he conjectured that the “only if” part also holds.
Later on, Masuda [29] proved it making use of the theory of toric topology [12, 30].

3.2. Squarefree modules and Alexander duality.

3.2.1. Squarefree modules and Alexander duality. In [54], Yanagawa introduced the notion,
called a squarefree module, as a generalization of Stanley-Reisner rings, and extended the theory
of Stanley-Reisner rings to squarefree modules. The full subcategory Sq S of modZn S consisting
of squarefree S-modules is abelian and contains, as objects, basic modules related with Stanley–
Reisner rings such as k[∆], I∆ , I∆ /I∆′ where ∆ ⊆ ∆′ , and ωS . In [36], Miller generalized
squarefree modules to the notion, called positively a-determined S-modules. As the class of
squarefree modules contains Stanley-Reisner rings (hence residue ring by squarefree monomial
ideals), that of positively a-determined modules does residue rings of S by monomial ideals
generated by monomials of degree ⪯ a. On other ∧ hand, Römer [42] defined the notion of
squarefree modules over the exterior algebra E := S1 over k with respect to S1 .
By the notions above, Miller [36] and Römer [42] independently succeeded to discover the
duality A on Sq S, called the Alexander duality functor, which plays the role of Alexander
duality. See their paper in loc. cit. for the precise definitions by them (or see below). It is
noteworthy that A sends k[∆] to I∆∨ , not to k[∆∨ ]. Many of facts on Alexander duality still
hold for A. For example, Theorems 2.24 and 2.26 hold for squarefree modules and A .
Here we shall give another description of the Alexander duality functor following [56]. Let
A be the incidence algebra over k associated with the subposet P := [0, 1] of Zn (See the
paper in loc. sit. for the definition). The algebra A is a finite-dimensional associative k-
algebra, and the category mod A of finitely generated left A-modules is equivalent to Sq S
(hence Sq S has enough injectives, enough projectives, and finite global dimension). Through
this equivalence, A coincides with HomA (−, k) (precisely, the composition of HomA (−, k) and
the functor induced from the ring isomorphism A → Aop ).
 INTRODUCTION TO STANLEY–REISNER RINGS 23

3.2.2. Relation with Koszul duality. As is shown in [55, 56], the category Db (Sq S) can be
regarded as a triangulated full subcategory of Db (modZn S), and R HomS (−, ωS ) induces the
functor D : Db (Sq S)op → Db (Sq S). Let A : Db (Sq S)op → Db (Sq S) denote the functor
induced from the Alexander duality functor by abuse of notation. The composition D ◦ A op
and A ◦D op then correspond to the functors giving the Koszul duality Db (mod A) ∼
= Db (mod A! )
(Since A and A are finite-dimensional over k, Koszul duality induces this equivalence). See
!

[56] for details.

3.2.3. Relation with Auslander–Reiten translate and Nakayama functor. In [8], Brun and Fløystad
investigated the composition A ◦ D op and inferred that A ◦ D op corresponds to the Nakayama
functor on Db (mod A) and hence to Auslander–Reiten translate on Db (mod AP ). (Actually they
proved that there is the corresponding similar to above for positively a-determined modules.
See [8] for details).

3.3. Sheaves associated with squarefree modules. In [55], Yanagawa introduced a sheaf
M + associated with a squarefree modules M ∈ Sq S. These sheaves are those on the (geometric)
(n − 1)-simplex X and with values in the category of k-vector spaces. For example, k[∆]+ =
j∗ k|∆| , where k|∆| denotes the constant sheaf on |∆| with stalk k and j : |∆| → X is the natural
embedding. In particular, S + is just the constant sheaf kX . The construction of M + gives an
exact functor from Sq S to the category Sh(X) of kX -modules, and through this functor, local
duality (for M ∈ Sq S with M0 = 0) corresponds to Poincaré–Verdier duality. Moreover through
the functor Db (Sq S) → Db (Sh(X)) induced from (−)+ , the complex ωS [n − 1] corresponds to
a dualizing complex on X. For M ∈ Sq S and a ∈ Zn≥0 , it follows that
{
H i−1 (X, M + ) if a = 0
ExtS (M, ωS )a ∼
n−i
= i−1 ∗ +
Hc (Uσ , j M ) otherwise

for all i ≥ 1, where σ denotes the face of X corresponding to supp(a), Hci−1 the (i − 1)-th
cohomology with compact support, Uσ the (open) star of σ, and j the embedding Uσ → X.
When M = k[∆], it follows that Hci−1 (Uσ , j ∗ M + ) ∼
= H i−1 (|∆| , |∆| \ {p} ; k) for any p in the
interior of σ. The above isomorphism (in conjunction with Proposition 1.32) thus gives a
generalization of Theorem 1.29 (except for Extn (k[∆], ωS )). See [55] for details.

3.4. Toric face rings. A Stanley–Reisner ring is one of main subject in combinatorial com-
mutative algebra; another main subject is an affine semigroup ring. For a given affine monoid
C, i.e., a finitely generated additive submonoid of ZN for some positive integer N , its affine
semigroup ring k[C] is a k-algebra with the k-basis C whose multiplication is induced from
the addition in C. For example, the polynomial ring S is an affine semigroup ring associated
with Zn≥0 . An affine semigroup ring k[C] has a connection with the cone generated by C (in
R ⊗Z ZC). See [9, 10, 37] for details.
A toric face ring is a common generalization of Stanley-Reisner rings and affine semigroup
rings. Roughly speaking, a toric face ring is an k-algebra given by gluing affine semigroup
rings along a given rational pointed fan (see [9, 11, 27] for the definition). For example, a
Stanley-Reisner ring k[∆] can be constructed by gluing each polynomial ring k[F ] with F ∈ ∆
along the fan consisting of the cones generated by a face of ∆.
In [11, 27], the following facts on Stanley-Reisner rings can be generalized to toric face rings.
• Hochster’s formula for local cohomologies (Remark 1.30) and Tor modules (Theorem 2.23).
• The relation among CM-ness, Gor*-ness, and Eulerian-ness.
• Relation among shellability, CM, and SCM.
24 RYOTA OKAZAKI

With an additional condition, in [40], Bbm-ness, CM-ness, and Gor*-ness for toric face rings
are shown to depend only on k and the regular CW complex associated with the given fan (as
for CM-ness, this assertion also follows from the results in [13]).

Appendix A. A brief review of graded algebras and modules

In this section, we will recall basics on graded modules. See [10, 21, 22] for details.
Definition A.1. Let G = Zn or Z, and R be a commutative ring.
(1) The ring R is said to be G-graded if
⊕
(a) R has a decomposition R = g∈G Rg as Z-modules such that
(b) Rg · Rh ⊆ Rg+h .
(2) An R-module M is said to be G-graded if
⊕
(a) M has a decomposition M = g∈G Mg as Z-modules such that
(b) Rg · Mh ⊆ Mg+h for all g, h ∈ G.
(3) For a G-graded module M and g ∈ G, M (g) denotes the G-graded module such that
M = M (g) as underlying S-modules and the grading of M is given by M (g)h := Mg+h .
Any Zn -graded module M has the natural structure of a Z-graded module by setting
⊕
Mi := Ma .
|a|=i

Thus all the Z -graded module M is tacitly regarded as a Z-graded one.

n

Let us recall the definition of associated prime ideals.

Definition A.2. Let R be a noetherian ring and M a finitely generated R-module.
(1) A prime ideal p ∈ Spec(R) is said to be an associated prime ideal if
p = Ann(m) := {r ∈ R | rm = 0}
for some m ∈ M .
(2) The set of associated prime ideals of M is denoted by AssR (M ).
Henceforth let R be a noetherian commutative graded ring, and ModG R denotes the cat-
egory consisting of G-graded R-modules and of degree preserving R-homomorphisms, i.e., R-
homomorphisms f : M → N with M, N G-graded such that f (Mg ) ⊆ Ng . The full subcategory
of ModG R consisting of finitely generated G-graded R-modules is denoted by modG R. It is
well-known that ModG R has enough injectives and projectives (see [10]).
From an easy observation, the following hold.
Lemma A.3.
(1) For any M ∈ modG R and p ∈ AssR (M ), it follows that p ∈ modG R and p = Ann(m)
for some m ∈ Mg and g ∈ G.
(2) For each M ∈ modG R, there exists a complex
P• : · · · −→ Pi −→ · · · −→ P1 −→ P0 −→ 0
such that
(a) Pi is finite direct sums of finitely many copies of R(g) for various g (hence it is a
finitely generated free R-module if one forgets the grading),
(b) each differential map is degree-preserving, and
(c) H0 (P• ) ∼
= M and Hi (P• ) = 0 for i ̸= 0.
 INTRODUCTION TO STANLEY–REISNER RINGS 25

(3) For M, N ∈ modG R (actually N is not needed to be finitely generated),

⊕
HomS (M, N ) = HomModG R (M, N (g)).
g∈G

(4) In particular, with the natural structure of an R-module, HomS (M, N ) ∈ modG R (when-
ever M, N ∈ modG R), where
HomS (M, N )g := HomModG R (M, N (g)).
(5) For M, N ∈ ModG R, M ⊗R N ∈ ModG R, where
∑
(M ⊗R N )g := Z · (y ⊗ z) ⊂ M ⊗R N.
y∈Mh , z∈Nh′
h+h′ =g

Corollary A.4. With the grading induced from results in Lemma A.3, it follows that ExtiR (M, N ) ∈
modG R and TorR i (M, N ) ∈ modG R for M, N ∈ modG R.
⊕
Proof. If one define ∗ HomS (M, N ) := g∈G HomModG R (M, N (g)) for M, N ∈ ModG R, then
one gets the bifunctor
∗
HomS (−, −) : (ModG R)op × ModG R → ModG R.
As the usual Hom functor, ∗ HomS (M, −) and ∗ HomS (−, N ) are left exact. One can thus
define ∗ ExtS (−, −). Lemma A.3 tells us that ∗ ExtS (M, N ) ∼
= ExtS (M, N ) for M ∈ modG R
and N ∈ ModG R.
Similarly one can verify the assertion on Tor by Lemma A.3. □

References
[1] D. J. Anick, A counterexample to a conjecture of Serre, Ann. Math. 115 (1982), 1–33.
[2] L. L. Avramov, Infinite free resolutions, in: Six lectures on commutative algebra, Birkhäuser, 1998.
[3] L. J. Billera and C. W. Lee, Sufficiency of McMullen’s conditions for f -vectors of simplicial polytopes,
Bull. Amer. Math. Soc. (New Series) 2 (1980),
[4] L. J. Billera and C. W. Lee, A proof of the sufficiency of McMullen’s conditions for f -vectors of
simplicial convex polytopes, J. Combin. Theory Ser. A 31 (1981), 237–255.
[5] A. Björner, Posets, regular CW complexes and Bruhat order, European J. Combin. 5 (1984), 7–16.
[6] A. Björner and M. L. Wachs, Shellable nonpure complexes and posets. I, Trans. Amer. Math. Soc.
348 (1996), 1299–1327.
[7] H. Bruggesser and P. Mani, Shellable decompositions of cells and spheres, Math. Scand. 29 (1971),
197–205.
[8] M. Brun and G. Fløystad, The Auslander–Reiten translate on monomial rings, Adv. in Math. 226
(2011), 952–991.
[9] W. Bruns and J. Gubeladze, Polytopes, Rings, and K-theory, Springer, 2009.
[10] W. Bruns and J. Herzog, Cohen–Macaulay rings, rev. ed., Cambridge University Press, 1998.
[11] W. Bruns, R. Koch, and T. Römer, Gröbner bases and Betti numbers of monoidal complexes, Michigan
Math J. 57 (2008), 71–91.
[12] V. M. Buchstaber and T. E. Panov, Torus actions and their applications in topology and combinatorics,
Univ. Lect. Ser. 24, Amer. Math. Soc., 2002.
[13] Z. Caijun, Cohen-Macaulay section rings, Trans. Amer. Math. Soc. 349 (1997), 4659–4667.
[14] M. Dehn, Dier Eulersche Formel im zusammenhang mit dem inhalt in der nichit-Euklidischen geome-
trie, Math. Ann. 61 (1906), 561–586.
[15] A. Dress, A new algebraic criterion for shellability, Beiträge zur Alg. und Geom. 34 (1993), 45–55.
[16] A. M. Duval, Algebraic shifting and sequentially Cohen-Macaulay simplicial complexes, Elec. J. Com-
bin. 3 (1996), #R21.
[17] J. A. Eagon and V. Reiner, Resolutions of Stanley-Reisner rings and Alexander duality, J. Pure Appl.
Alg. 130 (1998), 265–272.
[18] D. Eisenbud and S. Goto, Linear free resolutions and minimal multiplicity, J. Alg. 88 (1984), 89–133.
26 RYOTA OKAZAKI

[19] W. Fulton, An introduction to toric varieties, Ann. Math. Stud. 113, Princeton University Press,
1993.
[20] E. S. Golod, On the homologies of certain local rings, Soviet Math. Dokl. 3 (1962), 745–748 (translated
from: Dokl. Akad. Nauk. SSSR 144 (1962), 479–482).
[21] S. Goto and K. Watanabe, On graded rings, I, J. Math. Soc. Japan 30 (1978), 179–213.
[22] S. Goto and K. Watanabe, On graded rings, II (Zn -graded rings), Tokyo J. Math. 1 (1978), 237–261.
[23] T. H. Gulliksen and G. Levin, Homology of local rings, Queen’s Papers Pure Appl. Math. 20, Queen’s
University, 1969.
[24] J. Herzog and T. Hibi, Componentwise linear ideals, Nagoya Math. J. 153 (1999), 141–153.
[25] J. Herzog, V. Reiner, and V. Welker, Componentwise linear ideals and Golod rings, Michigan Math.
J. 46 (1999), 211–223.
[26] T. Hibi, Algebraic combinatorics on convex polytopes, Carslaw Publications, 1992.
[27] B. Ichim and T. Römer, On toric face rings, J. Pure Appl. Algebra 210 (2007), 249–266.
[28] D. Kraines, Massey higher products, Trans. Amer. Math. Soc. 124 (1966), 431–449.
[29] M. Masuda, h-vectors of Gorenstein* simplicial posets, Adv. Math. 194 (2005), 332–344.
[30] M. Masuda and T. Panov, On the cohomology of torus manifolds, Osaka J. Math. 43 (2006), 711–746.
[31] H. Matsumura, Commutative ring theory, Cambridge University Press, 1989.
[32] J. P. May, Matric Massey products, J. Algebra 12 (1969), 533–568.
[33] J. McCleary, A user’s guide to spectral sequences, 2nd ed., Cambridge University Press, 2000.
[34] P. McMullen, The maximum numbers of faces of a convex polytope, Mathematika 17 (1970), 179–184.
[35] P. McMullen, The numbers of faces of simplicial polytopes, Israel J. Math. 9 (1971), 559–570.
[36] E. Miller, The Alexander duality functors and local duality with monomial support, J. Algebra 231,
180–234.
[37] E. Miller and B. Sturmfels, Combinatorial commutative algebra, Springer, 2005.
[38] T. S. Motzkin, Comonotone curves and polyhedra, Bull. Amer. Math. Soc. 63 (1957), 35.
[39] J. R. Munkres, Topological results in combinatorics, Michigan Math. J. 31 (1984), 113–128.
[40] R. Okazaki and K. Yanagawa, Dualizing complex of a toric face ring, Nagoya Math. J. 196 (2009),
87–116.
[41] G. A. Reisner, Cohen–Macaulay quotients of polynomial rings, Adv. in Math. 21 (1976), 30–49.
[42] T. Römer, Generalized Alexander duality and applications, Osaka J. Math. 38 (2001), 469–485.
[43] M. E. Rudin, An unshellable triangulation of a tetrahedron, Bull. Amer. Math. Soc. 64 (1958), 90–91.
[44] P. Schenzel, On the number of faces of simplicial complexes and the purity of Forbenius, Math. Z.
178 (1975), 369–386.
[45] P. Schenzel, On the dimension filtration and Cohen-Macaulay filtered modules, in: Commutative al-
gebra and algebraic geometry, Lect. Notes in Pure and Appl. Math., Dekker, 1999, 245–246.
[46] J. P. Serre, Algébre Locale, Multiplicités, Lect. Notes in Math. 11, Springer, 1965.
[47] D. M. Y. Sommerville, The relations connecting the angle-sums and volume of a polytope in space of
n dimensions, Proc. R. Soc. Lond. A 115 (1927), 103–119.
[48] R. P. Stanley, The upper bound conjecture and Cohen-Macaulay rings, Stud. Appl. Math. 54 (1975),
135–142.
[49] R. P. Stanley, Hilbert functions of graded algebras, Adv. in Math. 28 (1978), 57–83.
[50] R. P. Stanley, The number of faces of a simplicial convex polytope, Adv. in Math. 35 (1980), 236–238.
[51] R. P. Stanley, f -vectors and h-vectors of simplicial posets, J. Pure and Appl. Algebra 71 (1991),
319–331.
[52] R. P. Stanley, Combinatorics and commutative algebra, 2nd ed., Birkhäuser, 1996.
[53] J. Stückrad and W. Vogel, Buchsbaum rings and applications: an interaction between algebra, geom-
etry and topology, Springer, 1986.
[54] K. Yanagawa, Alexander duality for Stanley–Reisner rings and squarefree Nn -graded modules, J. Al-
gebra 225 (2000), 630–645.
[55] K. Yanagawa, Stanley–Reisner rings, sheaves, and Poincaré–Verdier duality, Math. Res. Lett. 10
(2003), 635–650.
[56] K. Yanagawa, Derived category of squarefree modules and local cohomology with monomial ideal sup-
port, J. Math. Soc. Japan 56 (2004), 289–308.
[57] G. Ziegler, Shelling polyhedral 3-balls and 4-polytopes, Discr. Comput. Geom. 19 (1998), 159–174.

Faculty of Eduation, Fukuoka University of Education, Munakata, Fukuoka 811-4192, Japan

E-mail address: [email protected]