betaFeedback

WorksMathematics

Betti numbers of powers of Ferrers ideals and edge ideals of threshold graphs

Abstract

We provide explicit formulas for the Betti numbers of all powers of Ferrers ideals and edge ideals of threshold graphs.

Introduction

Let λ=(λ1,…,λn)\lambda= (\lambda_1,\ldots,\lambda_n) be a partition, meaning that λ1≥λ2≥⋯≥λn≥1\lambda_1 \ge\lambda_2 \ge\cdots\ge\lambda_n \ge1, and let m=λ1m = \lambda_1. Let R=k[x1,…,xn,y1,…,ym]R = k[x_1,\ldots,x_n,y_1,\ldots,y_m] be a polynomial ring over the field kk. The Ferrers ideal associated with the partition λ=(λ1,…,λn)\lambda= (\lambda_1,\ldots,\lambda_n) is given by

Iλ=(xiyj∣1≤i≤n,1≤j≤λi).I_\lambda= (x_i y_j \mid1 \le i \le n, 1 \le j \le\lambda_i).

For each partition λ\lambda, Corso and Nagel [4] introduced and studied the Ferrers ideal IλI_\lambda and described its minimal free resolution. In this work, we provide a formula for the Betti numbers of all powers of IλI_\lambda. Note that there are very few classes of ideals for which the Betti numbers of all powers are known, including maximal ideals, path ideals of paths [1], and certain path ideals of cycles [2]. To achieve this, we provide an ordering of the minimal generators of IλtI_\lambda^t and show that this ordering yields a linear quotient, from which the formula for the Betti numbers follows. We now introduce these concepts in more detail to state our main results.

Each finitely generated graded RR-module, and in particular R/IλR/I_\lambda, admits a minimal graded free resolution

0←R/Iλ←⨁jR(−j)β0,j(R/Iλ)←⋯←⨁jR(−j)βp,j(R/Iλ)←0,0 \leftarrow R/I_\lambda\leftarrow\bigoplus_j R(-j)^{\beta_{0,j}(R/I_\lambda)} \leftarrow\cdots\leftarrow\bigoplus_j R(-j)^{\beta_{p,j}(R/I_\lambda)} \leftarrow0,

where R(−j)R(-j) denotes the free RR-module obtained by shifting the degrees of RR by jj. The number βi,j(R/Iλ)\beta_{i,j}(R/I_\lambda) is the (i,j)(i,j)-th graded Betti number of R/IλR/I_\lambda. Herzog and Takayama [7] introduced the notion of ideals having linear quotients and proved that, in this case, the resolution can be constructed using iterated mapping cones. In particular, if all generators have the same degree, the ideal has a linear free resolution. We now describe the set of minimal generators of IλtI_\lambda^t.

For each t≥1t \ge1, let

C(λ,t)={(i1,…,it;j1,…,jt)∣1≤i1≤⋯≤it≤n,j1≥j2≥⋯≥jt, ja≤λia for all a=1,…,t}.C(\lambda,t)=\{(i_1,\ldots,i_t;j_1,\ldots,j_t)\mid1\le i_1\le\cdots\le i_t\le n,\quad j_1\ge j_2\ge\cdots\ge j_t,\ j_a\le\lambda_{i_a}\ \text{for all }a=1,\ldots,t\}.

Then we have:

Theorem 1.1. The minimal generators of IλtI_\lambda^t are given by xi1⋯xityj1⋯yjtx_{i_1}\cdots x_{i_t}y_{j_1}\cdots y_{j_t} for (i1,…,it;j1,…,jt)∈C(λ,t)(i_1,\ldots,i_t;j_1,\ldots,j_t)\in C(\lambda,t). Furthermore, when these generators are ordered lexicographically with x1>x2>⋯>xn>y1>⋯>ymx_1>x_2>\cdots>x_n>y_1>\cdots>y_m, IλtI_\lambda^t has a linear quotient with respect to this order.

From this, we deduce the following formula for the Betti numbers:

Theorem 1.2. Let λ=(λ1,…,λn)\lambda=(\lambda_1,\ldots,\lambda_n) be a partition with λ1=m\lambda_1=m. Then for all pp, we have

βpR(Iλt)=∑(i1,…,it;j1,…,jt)∈C(λ,t)(it+j1−2p).\beta_p^R(I_\lambda^t)=\sum_{(i_1,\ldots,i_t;j_1,\ldots,j_t)\in C(\lambda,t)}\binom{i_t+j_1-2}{p}.

The linear quotient property for powers of Ferrers graphs under lexicographic order extends to the class of threshold graphs. This allows us to deduce the Betti numbers of powers of edge ideals of threshold graphs. The remainder of this paper is devoted to proving these formulas and discussing several of their applications.

Linear quotients and Betti numbers

First, we establish the description of the minimal generating set.

Lemma 2.1. Let λ=(λ1,…,λn)\lambda=(\lambda_1,\ldots,\lambda_n) be a partition with λ1=m\lambda_1=m. Then IλtI_\lambda^t is minimally generated by xi1⋯xityj1⋯yjtx_{i_1}\cdots x_{i_t}y_{j_1}\cdots y_{j_t} for (i1,…,it;j1,…,jt)∈C(λ,t)(i_1,\ldots,i_t;j_1,\ldots,j_t)\in C(\lambda,t).

Proof. We have xiayja∈Iλx_{i_a}y_{j_a}\in I_\lambda for all a=1,…,ta=1,\ldots,t; hence, u=xi1⋯xityj1⋯yjt∈Iλtu=x_{i_1}\cdots x_{i_t}y_{j_1}\cdots y_{j_t}\in I_\lambda^t. Conversely, assume that f=(xα1yβ1)⋯(xαtyβt)∈Iλtf=(x_{\alpha_1}y_{\beta_1})\cdots(x_{\alpha_t}y_{\beta_t})\in I_\lambda^t is a generator. We may assume without loss of generality that α1≤⋯≤αt\alpha_1\le\cdots\le\alpha_t. We can repeatedly apply the exchange relation to obtain the canonical representation. Indeed, suppose that we have two factors xiyax_i y_a and xkybx_k y_b with i≤ki\le k and a<ba<b. Then, by definition, xiyb∈Iλx_i y_b\in I_\lambda and xkya∈Iλx_k y_a\in I_\lambda. Thus, we can perform the exchange

(xiya)(xkyb)=(xiyb)(xkya).(x_i y_a)(x_k y_b)=(x_i y_b)(x_k y_a).

Applying this operation finitely many times rearranges the yy-indices into non-increasing order without altering the xx-indices, which completes the proof. □\square

Definition 2.2. Let I=(u1,…,us)I=(u_1,\ldots,u_s) be a monomial ideal with ordered minimal generators u1,…,usu_1,\ldots,u_s. We say that II has linear quotients with respect to this order if, for every j=2,…,sj=2,\ldots,s, the colon ideal

(u1,…,uj−1):uj(u_1,\ldots,u_{j-1}):u_j

is generated by variables.

Before proving the result on linear quotients, let us consider an example for clarity.

Example 2.3. Let λ=(3,2)\lambda=(3,2). Then the minimal generators of Iλ2I_{\lambda}^{2} are ordered lexicographically as follows:

x12y12>x12y1y2>x12y1y3>x12y22>x12y2y3>x12y32>x1x2y12x_{1}^{2}y_{1}^{2}>x_{1}^{2}y_{1}y_{2}>x_{1}^{2}y_{1}y_{3}>x_{1}^{2}y_{2}^{2}>x_{1}^{2}y_{2}y_{3}>x_{1}^{2}y_{3}^{2}>x_{1}x_{2}y_{1}^{2}
>x1x2y1y2>x1x2y1y3>x1x2y22>x1x2y2y3>x22y12>x22y1y2>x22y22.>x_{1}x_{2}y_{1}y_{2}>x_{1}x_{2}y_{1}y_{3}>x_{1}x_{2}y_{2}^{2}>x_{1}x_{2}y_{2}y_{3}>x_{2}^{2}y_{1}^{2}>x_{2}^{2}y_{1}y_{2}>x_{2}^{2}y_{2}^{2}.

With this ordering, we can verify the linear quotients directly; for instance,

(x12y12,x12y1y2):(x12y1y3)=(y1,y2).(x_{1}^{2}y_{1}^{2},x_{1}^{2}y_{1}y_{2}):(x_{1}^{2}y_{1}y_{3})=(y_{1},y_{2}).

Note that while the canonical tuples (i1,…,it;j1,…,jt)∈C(λ,t)(i_{1},\ldots,i_{t};j_{1},\ldots,j_{t})\in C(\lambda,t) write the yy-indices in non-increasing order (j1≥⋯≥jt)(j_{1}\ge\cdots\ge j_{t}), writing the monomial factors with yy-indices appearing in non-decreasing order aligns naturally with the lexicographic ordering.

Lemma 2.4. Let u=xi1⋯xityj1⋯yjtu=x_{i_{1}}\cdots x_{i_{t}}y_{j_{1}}\cdots y_{j_{t}} be a minimal generator of IλtI_{\lambda}^{t} corresponding to the tuple (i1,…,it;j1,…,jt)∈C(λ,t)(i_{1},\ldots,i_{t};j_{1},\ldots,j_{t})\in C(\lambda,t). Let QuQ_{u} denote the ideal generated by all minimal generators preceding uu in the lexicographic order. Setting r=i1r=i_{1} and c=j1c=j_{1}, we have

Qu:u=(x1,…,xr−1,y1,…,yc−1).Q_{u}:u=(x_{1},\ldots,x_{r-1},y_{1},\ldots,y_{c-1}).

Proof. Let r=i1r=i_{1} and c=j1c=j_{1}. First, we prove the inclusion (x1,…,xr−1,y1,…,yc−1)⊆Qu:u(x_{1},\ldots,x_{r-1},y_{1},\ldots,y_{c-1})\subseteq Q_{u}:u. For any a<ra<r, replacing xrx_{r} with xax_{a} yields a valid minimal generator vv in IλtI_{\lambda}^{t} (after sorting the indices into C(λ,t)C(\lambda,t)) that strictly precedes uu in the lexicographic order, meaning xa∈Qu:ux_{a}\in Q_{u}:u. Similarly, for any b<cb<c, replacing ycy_{c} with yby_{b} yields a valid minimal generator vv preceding uu, so yb∈Qu:uy_{b}\in Q_{u}:u.

Conversely, to show that Qu:u⊆(x1,…,xr−1,y1,…,yc−1)Q_{u}:u\subseteq(x_{1},\ldots,x_{r-1},y_{1},\ldots,y_{c-1}), recall the variable ordering x1>⋯>xn>y1>⋯>ymx_{1}>\cdots>x_{n}>y_{1}>\cdots>y_{m}. In the lexicographic order, any preceding generator v<lexuv<_{\mathrm{lex}}u must introduce a strictly larger variable at the earliest possible position where vv and uu differ. Since uu has leading variable components determined by i1=ri_{1}=r and j1=cj_{1}=c, any variable dividing v/gcd⁡(u,v)v/\gcd(u,v) that places vv before uu must belong to the set {x1,…,xr−1}\{x_{1},\ldots,x_{r-1}\} or {y1,…,yc−1}\{y_{1},\ldots,y_{c-1}\}. Consequently, every generator of Qu:uQ_{u}:u is divided by at least one variable from this set, establishing the reverse inclusion and completing the proof. □

Proof of Theorem 1.1. The conclusion follows directly from Lemma 2.1 and Lemma 2.4. □

Proof of Theorem 1.2. The assertion follows by induction from Lemma 2.4 together with [11 Theorem 2.6]. □

Example 2.5. Let λ=(3,2)\lambda=(3,2) and t=4t=4, so that R=k[x1,x2,y1,y2,y3]R=k[x_{1},x_{2},y_{1},y_{2},y_{3}]. Each tuple in C(λ,4)C(\lambda,4) can be written uniquely as

((1,…,1)⏟a,(2,…,2)⏟4−a;(3,…,3)⏟q,(2,…,2)⏟r,(1,…,1)⏟4−q−r),0≤q≤a≤4,0≤r≤4−q−r.\left(\underbrace{(1,\ldots,1)}_{a},\underbrace{(2,\ldots,2)}_{4-a};\underbrace{(3,\ldots,3)}_{q},\underbrace{(2,\ldots,2)}_{r},\underbrace{(1,\ldots,1)}_{4-q-r}\right),\qquad0\le q\le a\le4,\qquad0\le r\le4-q-r.

Here, aa counts the occurrences of 1 among the ii-indices, while qq and rr count the occurrences of 3 and 2 among the jj-indices. Therefore,

∣C(λ,4)∣=∑a=04∑q=0a(5−q)=55.|\mathcal{C}(\lambda,4)|=\sum_{a=0}^{4}\sum_{q=0}^{a}(5-q)=55.

The choices of (a,q,r)(a,q,r) can be grouped according to the value of i4+j1−2i_4+j_1-2 as follows:

Parameter choicesNumber of tuplesi4+j1−2i_4+j_1-2
0≤a≤30\leq a\leq3, q=0q=0, r=0r=041
0≤a≤30\leq a\leq3, q=0q=0, 1≤r≤41\leq r\leq4162
1≤q≤a≤31\leq q\leq a\leq3, 0≤r≤4−q0\leq r\leq4-q203
a=4a=4, q=0q=0, r=0r=010
a=4a=4, q=0q=0, 1≤r≤41\leq r\leq441
a=4a=4, 1≤q≤41\leq q\leq4, 0≤r≤4−q0\leq r\leq4-q102

Table 1.

Thus, the exponent values 0, 1, 2, 3 occur respectively 1, 8, 26, 20 times. By Theorem 1.2,

βpR(Iλ4)=(0p)+8(1p)+26(2p)+20(3p).\beta_p^R(I_\lambda^4)=\binom{0}{p}+8\binom{1}{p}+26\binom{2}{p}+20\binom{3}{p}.

Evaluating this expression gives β0=55\beta_0=55, β1=120\beta_1=120, β2=86\beta_2=86, β3=20\beta_3=20, and βp=0\beta_p=0 for p≥4p\geq4. Hence, the Betti sequence is (55,120,86,20)(55,120,86,20).

Remark 2.6. It is well known that IλtI_\lambda^t has linear resolutions for all t≥1t\geq1 (see, e.g., [8 Theorem 3.2]). The depth of powers of R/IλtR/I_\lambda^t is also well known (see, e.g., [9 Proposition 2.18]). Basser, Diethorn, Miranda, and Stinson-Maas [3 Theorem 3.11] show that powers of edge ideals of cochordal graphs have linear quotients. Theorem 1.1 provides a specific ordering that yields linear quotients for IλtI_\lambda^t, resulting in a more explicit formula for the Betti numbers.

Betti numbers of powers of threshold graphs

Threshold graphs are a special class of cochordal graphs, where Ferrers graphs correspond precisely to bipartite threshold graphs. They have numerous applications (see, e.g., [10]). Corso and Nagel [5] showed that each threshold graph can be obtained by specializing a Ferrers graph. More precisely, by [5 Corollary 4.6], every threshold graph can be characterized by a partition λ=(λ1,…,λn)\lambda=(\lambda_1,\ldots,\lambda_n) with λ1≥λ2≥⋯≥λn>n\lambda_1\geq\lambda_2\geq\cdots\geq\lambda_n>n where {i,j}\{i,j\} with i<ji<j is an edge of the threshold graph TλT_\lambda if and only if j≤λij\leq\lambda_i. Let S=k[z1,…,zm]S=k[z_1,\ldots,z_m] where m=λ1m=\lambda_1. The corresponding edge ideal of TλT_\lambda is given by

I(Tλ)=(zizj∣i<j≤λi).I(T_\lambda)=(z_i z_j\mid i<j\leq\lambda_i).

For a monomial u=z1a1⋯ztatu=z_1^{a_1}\cdots z_t^{a_t} with at>0a_t>0, we denote by last⁡(u)=t\operatorname{last}(u)=t the largest index of a variable dividing uu.

Theorem 3.1. The minimal generators of I(Tλ)tI(T_{\lambda})^{t}, ordered lexicographically with z1>z2>⋯>zmz_{1} > z_{2} > \cdots> z_{m}, form a linear quotient order for I(Tλ)tI(T_{\lambda})^{t}. Furthermore, for a minimal generator u=z1a1⋯zmamu = z_{1}^{a_{1}} \cdots z_{m}^{a_{m}} of I(Tλ)tI(T_{\lambda})^{t}, we have

Qu:u=(zr∣ar<t and r<last⁡(u)),Q_{u} : u = (z_{r} \mid a_{r} < t\ \text{and}\ r < \operatorname{last}(u)),

where QuQ_{u} is the ideal generated by all minimal generators preceding uu in this order.

Proof. Let q=last⁡(u)q = \operatorname{last}(u) and let L=(zr∣ar<t and r<q)L = (z_{r} \mid a_{r} < t\ \text{and}\ r < q). First, we claim that L⊆Qu:uL \subseteq Q_{u} : u. Assume that r<qr < q and ar<ta_{r} < t. Choose a factorization of uu into a product of tt edges, u=e1⋯etu = e_{1} \cdots e_{t}. Since q∣uq \mid u, there exists an edge eje_{j} such that zqz_{q} divides eje_{j}. Hence, we may write ej=zpzqe_{j} = z_{p}z_{q} with p<qp < q. Since ej∈I(Tλ)e_{j} \in I(T_{\lambda}), we have q≤λpq \leq\lambda_{p}. There are two cases:

Case 1: p≠rp \ne r. We claim that zpzr∈I(Tλ)z_{p}z_{r} \in I(T_{\lambda}). Indeed, if p<rp < r, then p<r<q≤λpp < r < q \leq\lambda_{p}, hence zpzr∈I(Tλ)z_{p}z_{r} \in I(T_{\lambda}). If r<pr < p, then p<q≤λp≤λrp < q \leq\lambda_{p} \leq\lambda_{r}, hence zpzr∈I(Tλ)z_{p}z_{r} \in I(T_{\lambda}). In either case, we have w=e1⋯ej−1(zpzr)ej+1⋯et∈I(Tλ)tw = e_{1} \cdots e_{j-1}(z_{p}z_{r})e_{j+1} \cdots e_{t} \in I(T_{\lambda})^{t}. Furthermore, w/gcd⁡(w,u)=zrw/\gcd(w,u) = z_{r}. Thus, zr∈Qu:uz_{r} \in Q_{u} : u.

Case 2: p=rp = r. Since ar<ta_{r} < t, there exists an edge eie_{i} such that zrz_{r} does not divide eie_{i}. Write ei=zazbe_{i} = z_{a}z_{b}. Since q=last⁡(u)q = \operatorname{last}(u), we have a,b≤qa,b \leq q. We claim that zrza∈I(Tλ)z_{r}z_{a} \in I(T_{\lambda}). Indeed, since zrzq∈I(Tλ)z_{r}z_{q} \in I(T_{\lambda}), we have r<q≤λrr < q \leq\lambda_{r}. Now, if r<ar < a, then r<a<q≤λrr < a < q \leq\lambda_{r}, hence zrza∈I(Tλ)z_{r}z_{a} \in I(T_{\lambda}). If a<ra < r, then a<r<q≤λr≤λaa < r < q \leq\lambda_{r} \leq\lambda_{a}, hence zrza∈I(Tλ)z_{r}z_{a} \in I(T_{\lambda}). Similarly, zrzb∈I(Tλ)z_{r}z_{b} \in I(T_{\lambda}). Replacing eieje_{i}e_{j} by (zrza)(zrzb)(z_{r}z_{a})(z_{r}z_{b}) yields a new element w∈I(Tλ)tw \in I(T_{\lambda})^{t} such that w/gcd⁡(w,u)=zrw/\gcd(w,u) = z_{r}, which implies zr∈Qu:uz_{r} \in Q_{u} : u.

Now, let vv be any generator preceding uu in the lexicographic order. Write v=z1b1⋯zmbmv = z_{1}^{b_{1}} \cdots z_{m}^{b_{m}}. Let rr be the smallest index such that br≠arb_{r} \ne a_{r}. By definition of the lexicographic order, we have br>arb_{r} > a_{r} and bi=aib_{i} = a_{i} for all i<ri < r. Since vv is a minimal generator of I(Tλ)tI(T_{\lambda})^{t}, we must have br≤tb_{r} \leq t, which implies ar<br≤ta_{r} < b_{r} \leq t. Furthermore, deg⁡(v)=deg⁡(u)\deg(v) = \deg(u), so there must exist an index j>rj > r such that bj<ajb_{j} < a_{j}. In particular, r<j≤last⁡(u)=qr < j \leq\operatorname{last}(u) = q. Consequently, v/gcd⁡(u,v)v/\gcd(u,v) is divisible by zrz_{r}, showing that every generator of Qu:uQ_{u} : u is divisible by a variable zrz_{r} with r<qr < q and ar<ta_{r} < t. The conclusion follows. □

From Theorem 3.1, we see that to find the Betti numbers of the powers I(Tλ)tI(T_{\lambda})^{t}, we need to determine formulas for two key invariants. First, let

Gens⁡t(λ,q)={u∣u is a minimal generator of I(Tλ)t with last⁡(u)=q}\operatorname{Gens}_{t}(\lambda,q) = \{u \mid u\ \text{is a minimal generator of } I(T_{\lambda})^{t}\ \text{with }\operatorname{last}(u)=q\}

and

Gens⁡t∗(λ,q)={u=z1a1⋯zqaq∈Gens⁡t(λ,q)∣ai=t for some i<q}.\operatorname{Gens}_{t}^{*}(\lambda,q) = \{u = z_{1}^{a_{1}} \cdots z_{q}^{a_{q}} \in\operatorname{Gens}_{t}(\lambda,q) \mid a_{i}=t\ \text{for some }i<q\}.

We then define the cardinalities:

at(λ;q)=∣Gens⁡t(λ,q)∣,a_{t}(\lambda;q) = \lvert\operatorname{Gens}_{t}(\lambda,q)\rvert,
bt(λ;q)=∣Gens⁡t∗(λ,q)∣.b_{t}(\lambda;q) = \lvert\operatorname{Gens}_{t}^{*}(\lambda,q)\rvert.

Theorem 3.2. Let TλT_{\lambda} be the threshold graph corresponding to a partition λ\lambda with λ1≥⋯≥λn≥n\lambda_{1} \ge\cdots\ge\lambda_{n} \ge n. Then

βp(I(Tλ)t)=∑q=2m[at(λ;q)(q−1p)−bt(λ;q)(q−2p−1)].\beta_{p}(I(T_{\lambda})^{t})=\sum_{q=2}^{m}\left[a_{t}(\lambda;q)\binom{q-1}{p}-b_{t}(\lambda;q)\binom{q-2}{p-1}\right].

Proof. By Theorem 3.1 and [11 Theorem 2.6], we have

βp(I(Tλ)t)=∑u∈G(Iλt)(∣C(u)∣p),\beta_{p}(I(T_{\lambda})^{t})=\sum_{u\in G(I_{\lambda}^{t})}\binom{|C(u)|}{p},

where C(u)C(u) is the set of variables in the quotient ideal Qu:uQ_{u}:u. For any minimal generator uu with last⁡(u)=q\operatorname{last}(u)=q, the size of C(u)C(u) satisfies

∣C(u)∣={q−2if u∈Gens⁡t∗(λ,q),q−1otherwise.|C(u)|=\begin{cases} q-2 & \text{if }u\in\operatorname{Gens}_{t}^{*}(\lambda,q),\\ q-1 & \text{otherwise.} \end{cases}

The conclusion follows. □

Corollary 3.3. Let TλT_{\lambda} be the threshold graph corresponding to a partition λ\lambda with λ1=m≥⋯≥λn≥n\lambda_{1}=m\ge\cdots\ge\lambda_{n}\ge n. Then for all t≥2t\ge2,

pd⁡(I(Tλ)t)={m−1if n≥2,m−2if n=1.\operatorname{pd}(I(T_{\lambda})^{t})=\begin{cases} m-1 & \text{if }n\ge2,\\ m-2 & \text{if }n=1. \end{cases}

Proof. If n=1n=1, then I(Tλ)=z1(z2,…,zm)I(T_{\lambda})=z_{1}(z_{2},\ldots,z_{m}), and its powers have projective dimension m−2m-2. We may therefore assume that n≥2n\ge2 and t≥2t\ge2, which implies λ1=m>n≥2\lambda_{1}=m>n\ge2. By the Betti number formula in Theorem 3.2, it suffices to show that there exists a minimal generator uu of I(Tλ)tI(T_{\lambda})^{t} such that ∣C(u)∣=m−1|C(u)|=m-1. To this end, consider the monomial u=(z1zm)(z2z3)t−1u=(z_{1}z_{m})(z_{2}z_{3})^{t-1}. Then u∈Gens⁡t(λ,m)u\in\operatorname{Gens}_{t}(\lambda,m) but u∉Gens⁡t∗(λ,m)u\notin\operatorname{Gens}_{t}^{*}(\lambda,m). Consequently, ∣C(u)∣=m−1|C(u)|=m-1. The conclusion follows. □

Lemma 3.4. Let ρλ(q)=∣{1≤r<q∣λr≥q}∣\rho_{\lambda}(q)=|\{1\le r<q\mid\lambda_{r}\ge q\}|. Then

bt(λ;q)=ρλ(q)(q+t−3t−1).b_{t}(\lambda;q)=\rho_{\lambda}(q)\binom{q+t-3}{t-1}.

Proof. Let u=z1a1⋯zqaqu=z_{1}^{a_{1}}\cdots z_{q}^{a_{q}} be a minimal generator of I(Tλ)tI(T_{\lambda})^{t} such that ar=ta_{r}=t for some index r<qr<q. This condition implies that the edge zrzqz_{r}z_{q} belongs to I(Tλ)I(T_{\lambda}), which is equivalent to λr≥q\lambda_{r}\ge q. Under this condition, we have zrzi∈I(Tλ)z_{r}z_{i}\in I(T_{\lambda}) for all i∈{1,…,q}∖{r}i\in\{1,\ldots,q\}\setminus\{r\}.

There are ρλ(q)\rho_{\lambda}(q) possible choices for the index r<qr<q such that λr≥q\lambda_{r}\ge q. For a fixed rr, the condition ar=ta_{r}=t means that zrtz_{r}^{t} accounts for tt factors of uu. The remaining part is a monomial of degree tt in the variables {z1,…,zq}∖{zr}\{z_{1},\ldots,z_{q}\}\setminus\{z_{r}\}. The number of such monomials is given by (q+t−3t−1)\binom{q+t-3}{t-1}. Summing over all ρλ(q)\rho_{\lambda}(q) possible choices for rr, we obtain

bt(λ;q)=ρλ(q)(q+t−3t−1).b_{t}(\lambda;q)=\rho_{\lambda}(q)\binom{q+t-3}{t-1}.

:::`

This completes the proof. □

To compute at(λ,q)a_t(\lambda,q), we introduce the following notation. For a partition λ\lambda and an integer qq, we define the truncation partition λ[q]\lambda^{[q]} to be the partition given by λi[q]=min⁡{q,λi}\lambda_i^{[q]}=\min\{q,\lambda_i\}, and we let λ‾\overline{\lambda} denote the partition (λ2−1,…,λn−1)(\lambda_2-1,\ldots,\lambda_n-1). For any r,s≥0r,s\geq0, we set

A(λ;r,s)=dim⁡k(I(Tλ)t)2r+s.A(\lambda;r,s)=\dim_k(I(T_\lambda)^t)_{2r+s}.

In particular, A(λ;t,0)=μ(I(Tλ)t)A(\lambda;t,0)=\mu(I(T_\lambda)^t), which is the number of minimal generators of I(Tλ)tI(T_\lambda)^t.

First, we have the following lemma:

Lemma 3.5. With the notation above, we have

at(λ,q)=A(λ[q];t,0)−A(λ[q−1];t,0).a_t(\lambda,q)=A(\lambda^{[q]};t,0)-A(\lambda^{[q-1]};t,0).

Proof. Any generator of Iλ[q]tI_{\lambda^{[q]}}^t that does not belong to Iλ[q−1]tI_{\lambda^{[q-1]}}^t is divisible by zqz_q. The conclusion then follows directly from the definitions. □

Lemma 3.6. Assume that 2≤q≤n+12\leq q\leq n+1. Then

at(λ;q)=(2t+q−2q−1)−(q−1)(t+q−3q−1)−(t+q−2q−1).a_t(\lambda;q)=\binom{2t+q-2}{q-1}-(q-1)\binom{t+q-3}{q-1}-\binom{t+q-2}{q-1}.

Proof. When 2≤q≤n+12\leq q\leq n+1, the truncation λ[q]\lambda^{[q]} gives rise to a complete graph on qq vertices. Thus, at(λ;q)a_t(\lambda;q) equals the number of tuples of nonnegative integers (a1,…,aq)(a_1,\ldots,a_q) satisfying ai≤ta_i\leq t for all ii, aq≥1a_q\geq1, and ∑i=1qai=2t\sum_{i=1}^{q}a_i=2t.

Setting bq=aq−1b_q=a_q-1, this condition corresponds to tuples of nonnegative integers (a1,…,aq−1,bq)(a_1,\ldots,a_{q-1},b_q) such that ∑i=1q−1ai+bq=2t−1\sum_{i=1}^{q-1}a_i+b_q=2t-1, with the constraints ai≤ta_i\leq t for 1≤i≤q−11\leq i\leq q-1 and bq≤t−1b_q\leq t-1.

Without these upper-bound constraints, the number of such tuples—equal to the number of degree 2t−12t-1 monomials in qq variables—is given by the first term, (2t+q−2q−1)\binom{2t+q-2}{q-1}.

If any of the variables aia_i (1≤i≤q−11\leq i\leq q-1) violates the condition by satisfying ai≥t+1a_i\geq t+1, we can factor out zit+1z_i^{t+1}, leaving a monomial of degree t−2t-2 in qq variables. Summing over the q−1q-1 choices for such an index yields the second term, (q−1)(t+q−3q−1)(q-1)\binom{t+q-3}{q-1}.

Similarly, if the boundary variable satisfies bq≥tb_q\geq t, the remaining terms form a monomial of degree t−1t-1 in qq variables, giving the third term, (t+q−2q−1)\binom{t+q-2}{q-1}.

Since no two upper-bound violations can occur simultaneously, we obtain the desired formula. □

Lemma 3.7. Assume that λ1=⋯=λa=m>λa+1\lambda_1=\cdots=\lambda_a=m>\lambda_{a+1} with a≥2a\geq2. Then

A(λ;t,s)=(m+t+s−1m−1)−(m+t−2m−1)+∑r=1tA(λ‾;r,t+s−r).A(\lambda;t,s)=\binom{m+t+s-1}{m-1}-\binom{m+t-2}{m-1}+\sum_{r=1}^{t}A(\overline{\lambda};r,t+s-r).

Proof. For simplicity of notation, let J=IλJ=I_{\lambda} and K=IλˉK=I_{\bar{\lambda}}. Then we have J=z1(z2,…,zm)+KJ=z_1(z_2,\ldots,z_m)+K. Consequently, the tt-th power of JJ admits the expansion

Jt=∑r=0tz1t−r(z2,…,zm)t−rKr.J^t=\sum_{r=0}^{t}z_1^{t-r}(z_2,\ldots,z_m)^{t-r}K^r.

For any monomial u∈(Jt)2t+su\in(J^t)_{2t+s}, we can write u=z1dvu=z_1^dv for some integer d≥0d\ge0. When d≤t−1d\le t-1, setting r=t−dr=t-d, the remaining part vv belongs to (z2,…,zm)t−rKr(z_2,\ldots,z_m)^{t-r}K^r, which corresponds to the graded component of KrK^r in degree 2t+s−d=t+s+r=2r+(t+s−r)2t+s-d=t+s+r=2r+(t+s-r). This yields the contribution A(λˉ;r,t+s−r)A(\bar{\lambda};r,t+s-r).

When d≥td\ge t, the monomials arise entirely from the term (z2,…,zm)t(z_2,\ldots,z_m)^t. The number of such monomials equals the number of monomials of degree tt through t+st+s in a polynomial ring in m−1m-1 variables. The total number of such monomials is given by

∑d=tt+s(d+m−2m−2)=(m+t+s−1m−1)−(m+t−2m−1).\sum_{d=t}^{t+s}\binom{d+m-2}{m-2}=\binom{m+t+s-1}{m-1}-\binom{m+t-2}{m-1}.

Combining these two cases yields the desired formula. The conclusion follows. □

Lemma 3.8. Assume that λ1=m>λ2\lambda_1=m>\lambda_2, and let e=m−λ2e=m-\lambda_2. Then

A(λ;t,s)=(m+t+s−1m−1)−(m+t−2m−1)+∑r=1t∑v=0t+s−r(e+t+s−r−v−1e−1)A(λˉ;r,v).A(\lambda;t,s)=\binom{m+t+s-1}{m-1}-\binom{m+t-2}{m-1}+\sum_{r=1}^{t}\sum_{v=0}^{t+s-r}\binom{e+t+s-r-v-1}{e-1}A(\bar{\lambda};r,v).

Proof. As in the proof of Lemma 3.7, the case where d≥td\ge t yields the standard binomial difference term. When d<td<t, we must account for the graded components of KrK^r within the polynomial ring in variables z2,…,zmz_2,\ldots,z_m. Since KK is supported only on the variables z2,…,zλ2z_2,\ldots,z_{\lambda_2}, the remaining e=m−λ2e=m-\lambda_2 variables lie outside the support of KK. The inner sum over vv correctly accounts for the contribution of these ee additional variables. □

Corollary 3.9. Let KmK_m denote the complete graph on mm vertices. Then for all 0≤p≤m−10\le p\le m-1, we have

βp(I(Km)t)=(2t+m−12t+p)(2t+p−1p)−m(t+m−2m−1)(m−1p).\beta_p(I(K_m)^t)=\binom{2t+m-1}{2t+p}\binom{2t+p-1}{p}-m\binom{t+m-2}{m-1}\binom{m-1}{p}.

Proof. The complete graph KmK_m corresponds to the threshold graph with n=m−1n=m-1 and λ1=⋯=λn=m\lambda_1=\cdots=\lambda_n=m. By Theorem 3.2, Lemma 3.4, and Lemma 3.6, we have

βp(I(Tλ)t)=∑q=2m[at(λ;q)(q−1p)−bt(λ;q)(q−2p−1)].\beta_p(I(T_{\lambda})^t)=\sum_{q=2}^{m}\left[a_t(\lambda;q)\binom{q-1}{p}-b_t(\lambda;q)\binom{q-2}{p-1}\right].

where at(λ;q)=(2t+q−2q−1)−(q−1)(t+q−3q−1)−(t+q−2q−1)a_t(\lambda;q)=\binom{2t+q-2}{q-1}-(q-1)\binom{t+q-3}{q-1}-\binom{t+q-2}{q-1} and bt(λ;q)=(q−1)(t+q−3q−1)b_t(\lambda;q)=(q-1)\binom{t+q-3}{q-1}. Setting k=q−1k=q-1, we can rewrite the Betti number summation as

βp(I(Tλ)t)=∑k=1m−1(2t+k−1k)(kp)−∑k=1m−1Ep(k),\beta_p(I(T_\lambda)^t)=\sum_{k=1}^{m-1}\binom{2t+k-1}{k}\binom{k}{p}-\sum_{k=1}^{m-1}E_p(k),

where

Ep(k)=[k(t+k−2k)+(t+k−1k)](kp)+k(t+k−2k−1)(k−1p−1).E_p(k)=\left[k\binom{t+k-2}{k}+\binom{t+k-1}{k}\right]\binom{k}{p}+k\binom{t+k-2}{k-1}\binom{k-1}{p-1}.

Using the binomial identity

(2t+k−1k)(kp)=(2t+p−1p)(2t+k−1k−p),\binom{2t+k-1}{k}\binom{k}{p}=\binom{2t+p-1}{p}\binom{2t+k-1}{k-p},

we obtain

∑k=1m−1(2t+k−1k)(kp)=(2t+p−1p)(2t+m−12t+p)−δp,0,\sum_{k=1}^{m-1}\binom{2t+k-1}{k}\binom{k}{p}=\binom{2t+p-1}{p}\binom{2t+m-1}{2t+p}-\delta_{p,0},

where δp,0=1\delta_{p,0}=1 if p=0p=0 and 00 otherwise. Furthermore, using the binomial identities

(kp)=(k−1p)+(k−1p−1)and(t+k−1k)=(t+k−2k)+(t+k−2k−1),\binom{k}{p}=\binom{k-1}{p}+\binom{k-1}{p-1} \quad\text{and}\quad \binom{t+k-1}{k}=\binom{t+k-2}{k}+\binom{t+k-2}{k-1},

we can express Ep(k)E_p(k) as a telescoping difference Ep(k)=Fp(k+1)−Fp(k)E_p(k)=F_p(k+1)-F_p(k), where

Fp(k)=(k+1)(t+k−1k)(kp).F_p(k)=(k+1)\binom{t+k-1}{k}\binom{k}{p}.

Summing over kk from 1 to m−1m-1, the terms telescope, allowing us to deduce the final formula. The conclusion follows.

Remark 3.10. Theorem 3.2 and Lemmas 3.4–3.8 allow us to compute the Betti numbers of arbitrary threshold graphs. Dung and Vu [6] associated a type sequence to any chordal graph and gave a formula for the Betti numbers of its edge ideal in terms of this type sequence. However, we note that the type sequence alone is not sufficient to determine the Betti numbers of powers of its edge ideal. For example, K2,2K_{2,2} and a triangle with a pendant edge both have the same type sequence (2,2)(2,2), but their second powers do not share the same Betti numbers.

We do not know whether the linear quotient property with respect to the lexicographic ordering holds for powers of edge ideals of arbitrary specializations. Nonetheless, we can show that the projective dimension of powers of any arbitrary specialization is equal to mm if it contains a triangle. Note that if the specialization is bipartite, then its edge ideal is isomorphic to a Ferrers ideal.

Proposition 3.11. Let λ=(λ1,…,λn)\lambda=(\lambda_1,\ldots,\lambda_n) be a partition and μ=(μ1,…,μn)\mu=(\mu_1,\ldots,\mu_n) satisfy μ1≤⋯≤μn<λn\mu_1\leq\cdots\leq\mu_n<\lambda_n. Let I=(zizj∣1≤i≤n,μi<j≤λi)I=(z_i z_j\mid1\leq i\leq n,\mu_i<j\leq\lambda_i) be the edge ideal of the specialization graph Gλ,μG_{\lambda,\mu}. Assume that II contains a triangle, or equivalently, μ1<n\mu_1<n. Then pd⁡(S/It)=m\operatorname{pd}(S/I^t)=m for all t≥2t\geq2, where m=λ1m=\lambda_1 denotes the number of vertices.

Proof. First, we show that II is non-bipartite if and only if μ1<n\mu_{1}<n. Indeed, if μ1≥n\mu_{1}\ge n, then the underlying graph Gλ,μG_{\lambda,\mu} is bipartite with bipartition {1,…,n}∪{μ1+1,…,λn}\{1,\ldots,n\}\cup\{\mu_{1}+1,\ldots,\lambda_{n}\}. Conversely, if μ1<n\mu_{1}<n, setting j=μ1+1j=\mu_{1}+1 and k=λnk=\lambda_{n}, we see that {1,j,k}\{1,j,k\} forms a triangle.

Next, we claim that the triangle {1,j,k}\{1,j,k\} dominates Gλ,μG_{\lambda,\mu}. To see this, let vv be any vertex:

  • If v>μ1v>\mu_{1}, then {1,v}\{1,v\} is an edge by definition.

  • If v≤μ1v\le\mu_{1}, then μv≤μn<λn=k≤λv\mu_{v}\le\mu_{n}<\lambda_{n}=k\le\lambda_{v}, which implies that {v,k}\{v,k\} is an edge.

Since Gλ,μG_{\lambda,\mu} contains a dominating triangle, it follows that pd⁡(S/It)=m\operatorname{pd}(S/I^{t})=m for all t≥2t\ge2. The conclusion follows. □

Remark 3.12. The conclusion of Proposition 3.11 does not hold for arbitrary cochordal graphs containing a triangle. Indeed, let

I=(x1x2,x1x3,x1x4,x2x3,x2x4,x3x4,x1y1,x2y2,x3y3,x4y4)I=(x_{1}x_{2},x_{1}x_{3},x_{1}x_{4},x_{2}x_{3},x_{2}x_{4},x_{3}x_{4},x_{1}y_{1},x_{2}y_{2},x_{3}y_{3},x_{4}y_{4})

be the edge ideal of the whisker graph on the complete graph K4K_{4}. Then pd⁡(S/I2)=7<8\operatorname{pd}(S/I^{2})=7<8.

Acknowledgment. Thanh Vu is partially supported by the Vietnam National Foundation for Science and Technology Development under the grant number NAFOS-TED/101/10104/2026/1279.

References

Faculty of Mathematics and Informatics, Hanoi University of Science and Technology, 1 Dai Co Viet, Bach Mai, Hanoi, Vietnam. Email address: hoang.dotr [email protected]

Institute of Mathematics, VAST, 18 Hoang Quoc Viet, Hanoi, Vietnam Email address: [email protected]

References

  1. [1][BCV1] S. Balanescu, M. Cimpoeas, and T. Vu, Betti numbers of powers of path ideals of cycles, Journal of Algebraic Combinatorics 61 (2025), 44. 1arxiv.org/abs/2404.17880
  2. [2][BCV2] S. Balanescu, M. Cimpoeas, and T. Vu, Graded Betti numbers of powers of path ideals of paths, Journal of Commutative Algebra 18 (2026), 129–138. 1arxiv.org/abs/2405.04747
  3. [3][BDMS] E. Basser, R. Diethorn, R. Miranda, and M. Stinson-Maas, Powers of edge ideals with linear quotients, arXiv:2412.03468 4arxiv.org/abs/2412.03468
  4. [4][CN1] A. Corso and U. Nagel, Monomial and toric ideals associated to Ferrers graphs, Trans. Am. Math. Soc. 361 (2009), 1371–1395. 1DOI
  5. [5][CN2] A. Corso and U. Nagel, Specializations of Ferrers ideals, J Algebra. Comb. 28 (2008), 425–437. 4arxiv.org/abs/math/0703695
  6. [6][DV] L. X. Dung and T. Vu, Cochordal zero divisor graphs and Betti numbers of their edge ideals, Comm. Alg. 54 (2026), 736–744. 9arxiv.org/abs/2411.05251
  7. [7][HT] J. Herzog and Y. Takayama, Resolutions by mapping cones, Homology Homotopy Appl. 4 (2002), 277–294. 1DOI
  8. [8][HHZ] J. Herzog, T. Hibi and Zheng, Monomial ideals whose powers have a linear resolution, Math. Scand. 95 (2004), 23–32. 4DOI
  9. [9][HV] D. T. Hoang and T. Vu, Depth and regularity of tableau ideals, Adv. Appl. Math. 169 (2025), 102913. 4arxiv.org/abs/2311.07778
  10. [10][MP] N. V. R. Mahadev and U. N. Peled, Threshold Graphs and Related Topics, Annals Discrete Mathematics, vol. 56. North-Holland, Amsterdam (1995) 4
  11. [11][SV] L. Sharifan and M. Varbaro, Graded Betti numbers of ideals with linear quotients, Le Matematische 68 (2008), 257–265. 3, 6arxiv.org/abs/0901.3088

Paper details

vlnc:W_q3y226fy
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents