Koszul Simplicial Complex: Algebraic Perspectives

• Koszul simplicial complexes are defined by their associated face rings, where Koszul or K2 properties reveal deep connections between combinatorial topology and homological algebra.
• They generalize the classical Stanley–Reisner correspondence by introducing upper Koszul complexes for monomial ideals, thereby capturing multigraded Betti numbers and extending squarefree methods.
• Applications include using Alexander duality and sequential Cohen–Macaulay criteria to characterize Koszul properties, while factor theorems and spectral sequences provide actionable algebraic insights.

A Koszul simplicial complex is a simplicial complex interpreted through a Koszul-type homological invariant. In one established usage, the relevant object is the Stanley–Reisner face ring $k[\Delta]$, and the question is whether that graded algebra is Koszul or, more generally, $\mathcal K_2$. In another usage, the relevant object is the upper Koszul simplicial complex $K_I^\mu$ attached to a monomial ideal $I$ and a multidegree $\mu$, which extends the Stanley–Reisner correspondence beyond the squarefree case. These two meanings are related by the broader principle that simplicial combinatorics, Alexander duality, and multigraded homological algebra control Koszul-type phenomena in associated rings and complexes (Conner et al., 2011, López-Antón et al., 30 Jul 2025).

1. Terminology and basic objects

For a simplicial complex $\Delta$ on vertex set $[n]=\{1,\dots,n\}$, the Stanley–Reisner ideal is

$$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$

and the face ring is

$k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$

This identifies $\Delta$ with a quotient of the polynomial ring $\mathcal K_2$0, which is Koszul, and it is the starting point for the face-ring interpretation of a Koszul simplicial complex (Conner et al., 2011).

For a monomial ideal $\mathcal K_2$1 and a multidegree $\mathcal K_2$2, the upper Koszul simplicial complex is

$\mathcal K_2$3

This construction depends on $\mathcal K_2$4 and is the simplicial object that extends the Stanley–Reisner correspondence from squarefree monomial ideals to arbitrary monomial ideals. In the squarefree case,

$\mathcal K_2$5

so the construction recovers Alexander duality inside Stanley–Reisner theory (López-Antón et al., 30 Jul 2025).

The two usages are not identical. The first treats $\mathcal K_2$6 as primary and studies the Koszul-type homological algebra of $\mathcal K_2$7. The second treats a monomial ideal $\mathcal K_2$8 as primary and studies the simplicial complexes $\mathcal K_2$9 encoding its multigraded Betti theory. A plausible implication is that the phrase “Koszul simplicial complex” is best understood contextually rather than as a single universally fixed definition.

2. Face rings, Yoneda algebras, and the $K_I^\mu$0 extension

The face-ring formulation is expressed through the Yoneda algebra

$K_I^\mu$1

A graded algebra $K_I^\mu$2 is Koszul if $K_I^\mu$3 is generated by $K_I^\mu$4, equivalently if

$K_I^\mu$5

For a Stanley–Reisner ring, this is a rigid condition: cohomological degree and internal degree match, and the defining ideal must be quadratic (Conner et al., 2011).

The weaker notion used systematically in the literature is the $K_I^\mu$6 property. A graded algebra $K_I^\mu$7 is $K_I^\mu$8 if its Yoneda algebra is generated in cohomological degrees $K_I^\mu$9 and $I$0: $I$1 Every Koszul algebra is $I$2, but not conversely. For simplicial complexes this yields a bifurcation. In the strict sense, a Koszul simplicial complex is one for which $I$3 is Koszul. In the weaker sense, a $I$4 simplicial complex is one for which $I$5 is $I$6 (Conner et al., 2011).

For Stanley–Reisner rings, Fröberg’s theorem gives the strict case: $I$7 is Koszul when $I$8 is generated by quadratic monomials, equivalently when all minimal nonfaces are edges. The $I$9 notion is broader: higher-degree minimal nonfaces are allowed, while the full Ext-algebra is still generated by first and second cohomology. This is the sense in which $\mu$0 is the natural extension of Koszulness to nonquadratic face rings (Conner et al., 2011).

3. Alexander duality and the sequentially Cohen–Macaulay criterion

The decisive combinatorial input is Alexander duality. For a simplicial complex $\mu$1 on $\mu$2, the Alexander dual is

$\mu$3

equivalently, $\mu$4 if and only if $\mu$5. The central sufficient criterion is: $\mu$6 This is the main bridge between the topology of $\mu$7 and the Yoneda generation properties of $\mu$8 (Conner et al., 2011).

The mechanism factors through the ideal $\mu$9. Eagon–Reiner states that $\Delta$0 is Cohen–Macaulay if and only if $\Delta$1 has a linear free resolution over $\Delta$2. Herzog–Hibi states that $\Delta$3 is sequentially Cohen–Macaulay if and only if $\Delta$4 has a componentwise linear resolution over $\Delta$5. Over a Koszul algebra $\Delta$6, a module with componentwise linear resolution is a $\Delta$7-module. Since $\Delta$8 is Koszul and commutative, one obtains the chain of implications

$\Delta$9

The commutative hypothesis is important because the technical requirement that $[n]=\{1,\dots,n\}$0 act trivially on $[n]=\{1,\dots,n\}$1 is automatic in this setting (Conner et al., 2011).

This criterion is sufficient, not necessary. That distinction is structural rather than incidental: the paper explicitly emphasizes that $[n]=\{1,\dots,n\}$2 face rings exist outside the sequentially Cohen–Macaulay Alexander-dual regime (Conner et al., 2011).

4. Factor theorems, spectral sequences, and limitations of the criterion

The underlying algebraic theorem is a factor theorem for quotients of Koszul algebras. If $[n]=\{1,\dots,n\}$3 is Koszul, $[n]=\{1,\dots,n\}$4 is a graded ideal, $[n]=\{1,\dots,n\}$5, $[n]=\{1,\dots,n\}$6 acts trivially on $[n]=\{1,\dots,n\}$7, and $[n]=\{1,\dots,n\}$8 is a $[n]=\{1,\dots,n\}$9 $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$0-module, then $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$1 is a $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$2 algebra. Its proof uses the change-of-rings spectral sequence

$$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$3

which simplifies under the triviality hypothesis to

$$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$4

as $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$5-$$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$6-bimodules. The strategy is to rule out new generators in $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$7 for $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$8 by contradiction with the purity of $$I_\Delta=\big\langle x_{i_1}\cdots x_{i_r}\ :\ \{i_1,\dots,i_r\}\notin \Delta \big\rangle \subset S=k[x_1,\dots,x_n],$$9 (Conner et al., 2011).

The criterion is not reversible. One example uses

$k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$0

Here $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$1 is disconnected, so $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$2 is not sequentially Cohen–Macaulay, yet a minimal free resolution and the matrix criterion show that $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$3 is a $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$4 module; consequently $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$5 is $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$6. Another example uses

$k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$7

In this case $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$8 is $k[\Delta]=k[x_1,\dots,x_n]/I_\Delta.$9, but $\Delta$0 fails the matrix criterion for $\Delta$1-modules. Thus the converse of the factor theorem fails: a quotient may be $\Delta$2 although the defining ideal is not a $\Delta$3-module (Conner et al., 2011).

The paper also shows that weaker topological conditions on $\Delta$4 do not suffice. In particular, Buchsbaum dual complexes are too weak: for a family with $\Delta$5 Buchsbaum, the ring $\Delta$6 is not $\Delta$7, detected by

$\Delta$8

which is too large to be generated by $\Delta$9 and $\mathcal K_2$00. A common misconception is therefore that any mild dual-complex regularity should imply a Koszul-type property of the face ring; the available results do not support that conclusion (Conner et al., 2011).

5. Upper Koszul simplicial complexes of monomial ideals

For a monomial ideal $\mathcal K_2$01, the upper Koszul simplicial complex $\mathcal K_2$02 encodes multigraded Betti numbers through the formula

$\mathcal K_2$03

This is the Hochster-type bridge used throughout the modern theory. In the squarefree case, the identity

$\mathcal K_2$04

recovers the classical Alexander-dual interpretation, and the corresponding Betti formula becomes

$\mathcal K_2$05

Thus upper Koszul complexes generalize the squarefree Stanley–Reisner world rather than replacing it (López-Antón et al., 30 Jul 2025).

Polarization makes this relationship explicit at the simplicial level. If $\mathcal K_2$06 is the polarization of $\mathcal K_2$07, the paper constructs an expanded Koszul complex $\mathcal K_2$08 and proves

$\mathcal K_2$09

It also proves that $\mathcal K_2$10 has the same homology as $\mathcal K_2$11, in fact by a collapse onto a subcomplex canonically isomorphic to $\mathcal K_2$12. This gives a geometric explanation of the equality of multigraded Betti numbers under polarization (López-Antón et al., 30 Jul 2025).

Depolarization runs in the opposite direction and yields homology-preserving compression. For a simplicial complex $\mathcal K_2$13, the associated Koszul ideal is

$\mathcal K_2$14

If $\mathcal K_2$15 is a depolarization of $\mathcal K_2$16, then

$\mathcal K_2$17

The paper describes this reduction as a non-elementary collapse and uses it as a preprocessing step for algorithms on simplicial complexes, particularly for Alexander dual computation (López-Antón et al., 30 Jul 2025).

6. Higher Koszul modules and resonance schemes of simplicial complexes

A different but related construction starts from the exterior Stanley–Reisner algebra

$\mathcal K_2$18

Using the BGG complex over

$\mathcal K_2$19

one defines higher Koszul modules

$\mathcal K_2$20

These modules are $\mathcal K_2$21-graded square-free $\mathcal K_2$22-modules, and their square-free multigraded pieces satisfy

$\mathcal K_2$23

for square-free $\mathcal K_2$24. This identifies the higher Koszul modules directly with the reduced homology of induced subcomplexes (Aprodu et al., 2023).

The support resonance scheme of $\mathcal K_2$25 is reduced. More precisely, the support locus decomposes as

$\mathcal K_2$26

This is sharper than a mere set-theoretic description: reducedness follows from the fact that annihilators of square-free $\mathcal K_2$27-modules are square-free monomial ideals (Aprodu et al., 2023).

The Hilbert series is also determined combinatorially: $\mathcal K_2$28 This leads to a resonance–Hilbert-series relationship generalizing the graph case associated with Chen ranks of right-angled Artin groups. A plausible implication is that higher Koszul modules provide a third meaning of “Koszul simplicial complex,” not via $\mathcal K_2$29 itself and not via $\mathcal K_2$30, but via a family of square-free modules canonically attached to $\mathcal K_2$31 (Aprodu et al., 2023).

7. Related algebraic models and broader usage

The phrase also appears indirectly in constructions where a simplicial complex controls the Koszul property of an associated algebra. For every pure flag simplicial complex $\mathcal K_2$32, one can associate a standard graded Gorenstein algebra $\mathcal K_2$33 such that

$\mathcal K_2$34

and, more generally, the residue field has a $\mathcal K_2$35-step linear $\mathcal K_2$36-resolution if and only if $\mathcal K_2$37 satisfies Serre’s condition $\mathcal K_2$38. The same construction satisfies

$\mathcal K_2$39

This does not define $\mathcal K_2$40 itself as a Koszul complex, but it gives a precise algebraic model in which Cohen–Macaulayness of $\mathcal K_2$41 is equivalent to Koszulness of a canonical Gorenstein algebra (D'Alì et al., 2021).

A parallel phenomenon occurs for $\mathcal K_2$42-flag sortable simplicial complexes $\mathcal K_2$43. Their associated toric rings

$\mathcal K_2$44

and the Rees algebras of the facet ideals $\mathcal K_2$45 are shown to be Koszul, normal Cohen-Macaulay domains. The proof proceeds through sorting orders, quadratic Gröbner bases, and the $\mathcal K_2$46-exchange property, rather than through Yoneda algebras or upper Koszul complexes. Here again, the simplicial complex is “Koszul” only through an attached graded algebra (Ficarra et al., 2024).

Accordingly, the most precise encyclopedic interpretation is plural. In the strict Stanley–Reisner sense, a Koszul simplicial complex is one whose face ring is Koszul. In the broader face-ring sense developed via $\mathcal K_2$47, it is one whose face ring has Yoneda algebra generated in cohomological degrees $\mathcal K_2$48 and $\mathcal K_2$49. In the monomial-ideal sense, it is the upper Koszul simplicial complex $\mathcal K_2$50, whose homology computes multigraded Betti numbers. The shared theme is that simplicial data, especially Alexander duality and induced-subcomplex homology, governs Koszul-type algebraic behavior across several distinct but compatible frameworks (Conner et al., 2011, López-Antón et al., 30 Jul 2025).