Componentwise Polymatroidal Ideals

• Componentwise polymatroidal ideals are monomial ideals in a graded polynomial ring where each degree’s generators satisfy discrete polymatroid exchange axioms.
• They exhibit linear quotients that imply componentwise linearity, enabling explicit computation of Betti numbers through combinatorial methods.
• Recent research links these ideals to homological shifts and asymptotic syzygies, raising open problems in non-pure settings and multi-degree closure behavior.

Searching arXiv for recent and foundational papers on componentwise polymatroidal ideals. Componentwise polymatroidal ideals are monomial ideals in a standard graded polynomial ring $S=K[x_1,\dots,x_n]$ whose homogeneous strands are polymatroidal degree by degree. They were introduced as a way to extend polymatroidal behavior beyond the equigenerated setting, so that one can study non-pure monomial ideals while retaining the discrete-polymatroid combinatorics of exchange axioms. In the subsequent literature, the class has been linked to linear quotients, componentwise linearity, explicit two-variable classifications, and, more recently, the behavior of homological shift ideals and asymptotic syzygies (Bandari et al., 2012, Ficarra, 2023, Ficarra et al., 15 Sep 2025).

1. Definition and basic framework

Let $I\subset S=K[x_1,\dots,x_n]$ be a monomial ideal. For each integer $j\ge 0$, the notation $I_{\langle j\rangle}$ or $I(j)$ is used for the ideal generated by the degree-$j$ part of $I$. The ideal $I$ is called componentwise polymatroidal if, for every $j$, the strand $I_{\langle j\rangle}$ is polymatroidal. Equivalently, the generators in each fixed degree form the set of bases of a discrete polymatroid; in the usual algebraic formulation, each $I\subset S=K[x_1,\dots,x_n]$0 is generated in one degree and satisfies the exchange property (Bandari et al., 2012, Ficarra et al., 15 Sep 2025).

A polymatroidal ideal, in the classical sense, is a monomial ideal generated in a single degree $I\subset S=K[x_1,\dots,x_n]$1 such that for all $I\subset S=K[x_1,\dots,x_n]$2 and every index $I\subset S=K[x_1,\dots,x_n]$3 with $I\subset S=K[x_1,\dots,x_n]$4, there exists an index $I\subset S=K[x_1,\dots,x_n]$5 with $I\subset S=K[x_1,\dots,x_n]$6 and

$I\subset S=K[x_1,\dots,x_n]$7

This is the standard exchange condition. Every polymatroidal ideal is therefore trivially componentwise polymatroidal, but the componentwise notion is designed for ideals generated in several degrees (Bandari et al., 2012, Ficarra, 2023).

The degreewise viewpoint has immediate homological consequences. Each polymatroidal strand has linear quotients and hence a linear resolution; accordingly, componentwise polymatroidal ideals form a natural class of componentwise linear ideals. The converse is not known in general, and that asymmetry is one of the central structural features of the theory (Ficarra et al., 15 Sep 2025).

2. Exchange principles across different degrees

The main combinatorial difficulty is that the pure exchange law is formulated for equigenerated ideals, whereas componentwise polymatroidal ideals are typically non-pure. This led to cross-degree exchange conditions. One such statement, already present in the early development of the subject, is the “non-pure” exchange axiom: if $I\subset S=K[x_1,\dots,x_n]$8 with $I\subset S=K[x_1,\dots,x_n]$9 and $j\ge 0$0, then there exists $j\ge 0$1 with $j\ge 0$2 such that

$j\ge 0$3

This property is proved for componentwise polymatroidal ideals and captures the way exchange can be transported between distinct degrees (Bandari et al., 2012).

Later work places this in a sharper equivalence framework. A theorem cited in the proof of the linear-quotients result states that, for a monomial ideal $j\ge 0$4, componentwise polymatroidality is equivalent to a cross-degree exchange condition of this type, formulated for pairs $j\ge 0$5 with $j\ge 0$6 and $j\ge 0$7. The same circle of ideas also yields a dual exchange statement: if $j\ge 0$8 is componentwise polymatroidal and $j\ge 0$9, then for every $I_{\langle j\rangle}$0 such that $I_{\langle j\rangle}$1, there exists $I_{\langle j\rangle}$2 with $I_{\langle j\rangle}$3 and

$I_{\langle j\rangle}$4

These two cross-degree formulations are key technical tools in the subject (Ficarra, 2023).

In two variables, the exchange picture becomes particularly rigid. For ideals in $I_{\langle j\rangle}$5, the non-pure exchange property, the non-pure dual exchange property, and componentwise polymatroidality are equivalent. This yields an explicit combinatorial classification and shows that the two-variable case is substantially more tractable than the general multivariable setting (Bandari et al., 2021).

3. Linear quotients and componentwise linearity

A major conjecture in the area asked whether every componentwise polymatroidal ideal has linear quotients. This was posed by Bandari and Herzog and proved positively by Ficarra: if $I_{\langle j\rangle}$6 is componentwise polymatroidal, then $I_{\langle j\rangle}$7 has linear quotients. The proof is by double induction, first on the number of variables and then on $I_{\langle j\rangle}$8, and starts from a decomposition

$I_{\langle j\rangle}$9

with $I(j)$0, where $I(j)$1 remains componentwise polymatroidal in $I(j)$2 and $I(j)$3 is componentwise polymatroidal in the smaller ring $I(j)$4. The critical colon computations use the dual exchange property (Ficarra, 2023).

This theorem settles the earlier conjectural picture and strengthens the homological status of the class. Since ideals with linear quotients are componentwise linear, every componentwise polymatroidal ideal is componentwise linear. Moreover, once a linear-quotients order is fixed, the Betti numbers can be read off through the Eliahou–Kervaire formula from that order. In this sense, the combinatorics of exchange translates directly into explicit control of graded syzygies (Ficarra, 2023).

Before the general theorem, linear quotients had been verified only in specific settings. Bandari and Herzog proved that ideals componentwise of Veronese type have linear quotients and hence are componentwise linear. Bandari and Qureshi later showed linear quotients for two large subclasses: componentwise polymatroidal ideals in $I(j)$5, and componentwise polymatroidal ideals whose graded pieces satisfy the strong exchange property. These partial results anticipated the general theorem by isolating settings where admissible orders can be constructed explicitly (Bandari et al., 2012, Bandari et al., 2021).

4. Special classes and closure behavior

Closure properties for componentwise polymatroidal ideals are subtler than for equigenerated polymatroidal ideals. For polymatroidal ideals generated in one degree, all powers remain polymatroidal. In the componentwise setting, closure under powers is only partial: if $I(j)$6 is componentwise polymatroidal and $I(j)$7 lives in at most two degrees, then every power $I(j)$8 is again componentwise polymatroidal. The proof uses the decomposition

$I(j)$9

and the fact that products of polymatroidal ideals are polymatroidal degreewise (Bandari et al., 2012).

At the same time, powers and products do not preserve the class in general. The literature records explicit counterexamples showing that $j$0 may fail to be polymatroidal even when $j$1 is componentwise polymatroidal, and later work reiterates that powers or products of a componentwise polymatroidal ideal need not remain componentwise polymatroidal (Bandari et al., 2012, Ficarra et al., 15 Sep 2025).

The two-variable case is again exceptional. In $j$2, componentwise polymatroidal ideals admit a complete classification: an ideal is componentwise polymatroidal if and only if it can be written as

$j$3

where $j$4 is a $j$5-tight monomial ideal. Equivalently, if $j$6 with $j$7 in pure lex order, then the degree sequence $j$8, $j$9, has at most one valley. From this description one obtains an explicit admissible order for linear quotients, and one also proves that products of $I$0-tight ideals are $I$1-tight. Consequently, in $I$2 products, and hence powers, of componentwise polymatroidal ideals are again componentwise polymatroidal (Bandari et al., 2021).

The current closure picture can be summarized as follows.

Operation or setting	Status for componentwise polymatroidality	Source
Powers, generators in at most two degrees	Preserved	(Bandari et al., 2012)
Powers in general	Can fail	(Bandari et al., 2012)
Products in general	Can fail	(Ficarra et al., 15 Sep 2025)
Colon by $I$3	Preserved	(Ficarra et al., 15 Sep 2025)
Saturation $I$4	Preserved	(Ficarra et al., 15 Sep 2025)
Products and powers in $I$5	Preserved	(Bandari et al., 2021)

These results show that componentwise polymatroidality is neither as rigid as equigenerated polymatroidality nor as loose as arbitrary componentwise linearity. Its behavior depends strongly on degree distribution, number of variables, and the presence of stronger exchange structures such as the strong exchange property (Bandari et al., 2021).

5. Homological shifts and asymptotic syzygies

Recent work connects componentwise polymatroidal ideals to homological shift ideals. If $I$6 is componentwise polymatroidal, then the first homological shift ideal $I$7 is again componentwise polymatroidal. More precisely,

$I$8

and each strand on the right is polymatroidal. This is Theorem 10.1 in the asymptotic-syzygies study of Ficarra and Lu (Ficarra et al., 15 Sep 2025).

The proof proceeds by decomposing $I$9 and defining a graded object $I$0, where $I$1 is the $I$2-span of the minimal generators of $I$3. One then proves that $I$4 is an ideal and that $I$5. The key input is the bounding-multidegree description for a polymatroidal ideal $I$6,

$I$7

together with the lcm-description of generators of $I$8 as least common multiples of pairs of generators lying in a common homogeneous strand. This mechanism is specific enough to preserve the componentwise polymatroidal structure at the first homological level (Ficarra et al., 15 Sep 2025).

The same paper places this theorem inside a broader asymptotic program for polymatroidal ideals. It proves that the first homological shift algebra $I$9 is generated in degree one as a module over the Rees algebra $j$0, conjectures that $j$1 is generated in degrees $j$2, and establishes that $j$3 has the $j$4st homological strong persistence property. It also conjectures that the sequence $j$5 becomes an increasing chain for $j$6, with the conjecture verified for $j$7 and for many families. Within that framework, the preservation of componentwise polymatroidality by $j$8 identifies a structurally stable first layer in the homological shift hierarchy (Ficarra et al., 15 Sep 2025).

6. Open problems and related directions

Several foundational questions remain open. The converse of componentwise polymatroidal $j$9 componentwise linear is not known in general. Likewise, the general power-closure problem remains unresolved: although powers are known to remain componentwise polymatroidal in two degrees and in $I_{\langle j\rangle}$0, counterexamples show that no unrestricted closure theorem is possible in full generality, and the exact boundary of failure remains a central issue (Bandari et al., 2012, Bandari et al., 2021).

Homological questions are even less settled. One open problem asks whether $I_{\langle j\rangle}$1 is componentwise polymatroidal for every $I_{\langle j\rangle}$2 whenever $I_{\langle j\rangle}$3 is componentwise polymatroidal. No counterexample is known, but only the case $I_{\langle j\rangle}$4 has been proved. A parallel question asks whether the homological shift algebras $I_{\langle j\rangle}$5 remain finitely generated as $I_{\langle j\rangle}$6-modules (Ficarra et al., 15 Sep 2025).

Another line of inquiry concerns exchange properties broader than componentwise polymatroidality. Qureshi and Bandari introduced the non-pure dual exchange property and proved that any ideal satisfying it has linear quotients, hence is componentwise linear. They also showed that this criterion applies to certain ideals of $I_{\langle j\rangle}$7-covers of totally balanced weighted hypergraphs. Since the non-pure dual exchange property is presented as strictly weaker than componentwise polymatroidality, these results place componentwise polymatroidal ideals inside a larger exchange-theoretic landscape of non-pure monomial ideals (Qureshi et al., 2024).

There are also toric and combinatorial directions. For the simplicial multicomplex corresponding to a componentwise polymatroidal ideal, one may ask whether the associated toric rings, including the base ring and Rees algebra, are normal, Cohen–Macaulay, or Koszul. This suggests that the theory is not confined to linear resolutions and Betti tables, but interacts with the broader algebraic geometry of monomial and toric constructions (Ficarra, 2023).

Componentwise polymatroidal ideals thus occupy a precise intermediate position in combinatorial commutative algebra: stronger than general componentwise linearity because they retain exchange structure degree by degree, but more flexible than equigenerated polymatroidal ideals because they admit genuinely non-pure behavior. The modern theory is organized around that tension, with current work moving simultaneously toward sharper structural classifications and toward a homological theory that extends beyond the first shift.