Introduction
Let R = K [ x 1 , … , x n ] = ⨁ d ≥ 0 R d R = \mathbb{K}[x_1,\ldots,x_n] = \bigoplus_{d\ge0} R_d R = K [ x 1 , … , x n ] = ⨁ d ≥ 0 R d be the standard graded polynomial ring over a field K \mathbb{K} K . For a graded ideal I ⊆ R I \subseteq R I ⊆ R , let Ass ( I ) \operatorname{Ass}(I) Ass ( I ) denote the set of all associated prime ideals of I I I . For each p ∈ Ass ( I ) \mathfrak{p} \in\operatorname{Ass}(I) p ∈ Ass ( I ) , the local v v v -number of I I I at p \mathfrak{p} p is defined as
v p ( I ) = min { d ≥ 0 ∣ there exists f ∈ R d such that ( I : f ) = p } . v_{\mathfrak{p}}(I) = \min\{d \ge0 \mid\text{there exists } f \in R_d \text{ such that } (I:f) = \mathfrak{p}\}. v p ( I ) = min { d ≥ 0 ∣ there exists f ∈ R d such that ( I : f ) = p } .
The v v v -number of I I I , denoted by v ( I ) v(I) v ( I ) , is then defined as
v ( I ) = min { v p ( I ) ∣ p ∈ Ass ( I ) } . v(I) = \min\{v_{\mathfrak{p}}(I) \mid\mathfrak{p} \in\operatorname{Ass}(I)\}. v ( I ) = min { v p ( I ) ∣ p ∈ Ass ( I )} .
Originally introduced in [9 ] to analyze the asymptotic behavior of the minimum distance of projective Reed-Muller type codes, the v v v -number serves as a crucial algebraic invariant connecting coding theory, commutative algebra, and combinatorics. This invariant has since spurred considerable interest in commutative algebra and algebraic combinatorics, leading to several significant applications. There has been substantial progress in understanding this invariant across different contexts (see [7 , 8 , 15 , 17 , 21 , 24 ] ). Regarding the asymptotic behavior of v v v -numbers, Ficarra and Sgroi [14 ] and Conca [8 ] independently established that v ( I t ) v(I^t) v ( I t ) is eventually linear in t t t for any graded ideal I I I in a polynomial ring (and more generally, in an N \mathbb{N} N -graded Noetherian domain [8 ] ). We refer the reader to [3 , 12 , 14 , 15 , 24 ] for further developments in this direction. In particular, Hien and Vu in [19 ] gave a combinatorial description of the stabilization index of the v v v -function of powers of edge ideals for arbitrary graphs.
Let Γ \Gamma Γ be a simple graph with vertex set V ( Γ ) = { x 1 , … , x n } V(\Gamma) = \{x_1,\ldots,x_n\} V ( Γ ) = { x 1 , … , x n } and edge set E ( Γ ) E(\Gamma) E ( Γ ) . The edge ideal of Γ \Gamma Γ , introduced in [30 ] , is denoted by I ( Γ ) I(\Gamma) I ( Γ ) and defined as the ideal of the polynomial ring R = K [ x 1 , … , x n ] R = \mathbb{K}[x_1,\ldots,x_n] R = K [ x 1 , … , x n ] over a field K \mathbb{K} K generated by the monomials corresponding to the edges of Γ \Gamma Γ . Explicitly,
I ( Γ ) = ( x i x j ∣ x i x j ∈ E ( Γ ) ) . I(\Gamma) = (x_i x_j \mid x_i x_j \in E(\Gamma)). I ( Γ ) = ( x i x j ∣ x i x j ∈ E ( Γ )) .
2020 Mathematics Subject Classification . 13F20, 13F55, 13A15, 05E40.
Keywords: v v v -number, powers of edge ideal, cubic circulant graph. The v-number of I ( Γ ) I(\Gamma) I ( Γ ) is a fundamental combinatorial invariant of Γ \Gamma Γ that has been used to characterize key topological and structural properties of graphs, such as the family of W 2 W_2 W 2 -graphs. Jaramillo and Villarreal in [22 Theorem 4.5] showed that for a connected graph Γ \Gamma Γ , it is in W 2 W_2 W 2 if and only if v ( I ( Γ ) ) = dim ( R / I ( Γ ) ) \mathrm{v}(I(\Gamma)) = \dim(R/I(\Gamma)) v ( I ( Γ )) = dim ( R / I ( Γ )) . Furthermore, determining exact formulas for v-numbers of powers of edge ideals remains an active and challenging task. Chau et al. in [5 ] showed that for a path P n P_n P n with n ≥ 2 n \ge2 n ≥ 2 vertices and a cycle C n C_n C n with n ≥ 3 n \ge3 n ≥ 3 vertices, the v-number of powers of their edge ideals are
v ( I ( P n ) t ) = 2 t − 1 + max { 0 , ⌊ n − 2 t − 3 4 ⌋ } . \mathrm{v}(I(P_n)^t)=2t-1+\max\left\{0,\left\lfloor\frac{n-2t-3}{4}\right\rfloor\right\}. v ( I ( P n ) t ) = 2 t − 1 + max { 0 , ⌊ 4 n − 2 t − 3 ⌋ } .
v ( I ( C n ) t ) = 2 t − 1 + max { 0 , ⌊ n − 2 t − 2 4 ⌋ } . \mathrm{v}(I(C_n)^t)=2t-1+\max\left\{0,\left\lfloor\frac{n-2t-2}{4}\right\rfloor\right\}. v ( I ( C n ) t ) = 2 t − 1 + max { 0 , ⌊ 4 n − 2 t − 2 ⌋ } .
Circulant graphs have wide applications in coding theory, interconnection networks, and combinatorial designs [6 , 23 , 25 ] . Algebraic studies of edge ideals of circulant graphs have been carried out in [1 , 4 , 20 , 27 , 28 , 29 ] . Recall that for a positive integer n ≥ 3 n \ge3 n ≥ 3 and a subset S ⊆ { 1 , … , ⌊ n / 2 ⌋ } S \subseteq\{1,\ldots,\lfloor n/2\rfloor\} S ⊆ { 1 , … , ⌊ n /2 ⌋} , the circulant graph generated by S S S , denoted by C n ( S ) C_n(S) C n ( S ) , is the graph whose vertex set is V ( C n ( S ) ) = { x 1 , x 2 , … , x n } V(C_n(S))=\{x_1,x_2,\ldots,x_n\} V ( C n ( S )) = { x 1 , x 2 , … , x n } , in which two distinct vertices x i x_i x i and x j x_j x j are adjacent if and only if either ∣ i − j ∣ ∈ S |i-j|\in S ∣ i − j ∣ ∈ S or n − ∣ i − j ∣ ∈ S n-|i-j|\in S n − ∣ i − j ∣ ∈ S . In particular, if S = { 1 , … , ⌊ n / 2 ⌋ } S=\{1,\ldots,\lfloor n/2\rfloor\} S = { 1 , … , ⌊ n /2 ⌋} , then C n ( S ) ≅ K n C_n(S)\cong K_n C n ( S ) ≅ K n , the complete graph on n n n vertices. By Fröberg’s theorem [16 ] , the edge ideal I ( K n ) I(K_n) I ( K n ) has a linear resolution, from which it follows from [12 Theorem 5.1] that v ( I ( K n ) t ) = 2 t − 1 \mathrm{v}(I(K_n)^t)=2t-1 v ( I ( K n ) t ) = 2 t − 1 for all t ≥ 1 t\ge1 t ≥ 1 . On the other hand, if S = { 1 } S=\{1\} S = { 1 } , then C n ( 1 ) ≅ C n C_n(1)\cong C_n C n ( 1 ) ≅ C n , whose v-number formula for powers was completely determined in [5 ] as mentioned above.
Despite recent progress, explicitly computing the v-number for broader classes of circulant graphs remains highly non-trivial due to their intricate topological structures. A cubic circulant graph is a 3-regular circulant graph of the form C 2 n ( S ) C_{2n}(S) C 2 n ( S ) with S = { a , n } S=\{a,n\} S = { a , n } for some integer a a a such that 1 ≤ a < n 1\le a<n 1 ≤ a < n , denoted by C 2 n ( a , n ) C_{2n}(a,n) C 2 n ( a , n ) . Davis and Domke [11 ] proved that cubic circulant graphs can be decomposed into disjoint unions of certain copies of circulant graphs of the form C 2 m ( 1 , m ) C_{2m}(1,m) C 2 m ( 1 , m ) and C 2 m ( 2 , m ) C_{2m}(2,m) C 2 m ( 2 , m ) for odd m > 2 m>2 m > 2 (see Lemma 2.1).
The main goal of this paper is to bridge this gap by providing precise combinatorial and algebraic formulas for the v-number of cubic circulant graphs. Our main result is as follows:
Theorem 1.1. Let Γ = C 2 n ( a , n ) \Gamma=C_{2n}(a,n) Γ = C 2 n ( a , n ) be a cubic circulant graph with 1 ≤ a < n 1\le a<n 1 ≤ a < n , and let I ( Γ ) I(\Gamma) I ( Γ ) be its edge ideal. Let k = gcd ( a , n ) k=\gcd(a,n) k = g cd( a , n ) , m = n k m=\frac{n}{k} m = k n , b = a k b=\frac{a}{k} b = k a , and
ρ = { m , if both m and b are odd , m + 1 , otherwise. \rho=
\begin{cases}
m, & \text{if both }m\text{ and }b\text{ are odd},\\
m+1, & \text{otherwise.}
\end{cases} ρ = { m , m + 1 , if both m and b are odd , otherwise. Then, for all integer t ≥ 1 t\ge1 t ≥ 1 ,
v ( I ( Γ ) t ) = max { 2 t + k − 2 , t − 1 + k ⌊ ρ 3 ⌋ , ⌈ 4 ( t − 1 ) + k ρ 3 ⌉ , ⌈ 3 ( t − 1 ) + k ( ⌊ ρ 2 ⌋ + 1 ) 2 ⌉ } . \mathrm{v}(I(\Gamma)^t)=\max\left\{2t+k-2,\ t-1+k\left\lfloor\frac{\rho}{3}\right\rfloor,\ \left\lceil\frac{4(t-1)+k\rho}{3}\right\rceil,\ \left\lceil\frac{3(t-1)+k\left(\left\lfloor\frac{\rho}{2}\right\rfloor+1\right)}{2}\right\rceil\right\}. v ( I ( Γ ) t ) = max { 2 t + k − 2 , t − 1 + k ⌊ 3 ρ ⌋ , ⌈ 3 4 ( t − 1 ) + k ρ ⌉ , ⌈ 2 3 ( t − 1 ) + k ( ⌊ 2 ρ ⌋ + 1 ) ⌉ } . :::_
The paper is organized as follows: § 2 introduces the basic facts and preliminary definitions that are used throughout the paper. In section 3 and 4, we provide exact formulas for the v v v -numbers of powers of the edge ideals of Γ = C 2 n ( a , n ) \Gamma= C_{2n}(a,n) Γ = C 2 n ( a , n ) with 1 ≤ a < n 1 \leq a < n 1 ≤ a < n .
Preliminaries
In this section, we introduce the notation and recall several basic results that will be used throughout the paper. Throughout this paper, we adopt the following symbols: for two non-negative integers m , n m,n m , n , let
[ m , n ] = { { m , m + 1 , … , n } , if m ≤ n , ∅ , if m > n . [m,n] =
\begin{cases}
\{m,m+1,\ldots,n\}, & \text{if }m \leq n,\\
\varnothing, & \text{if }m > n.
\end{cases} [ m , n ] = { { m , m + 1 , … , n } , ∅ , if m ≤ n , if m > n .
By convention, we write [ n ] [n] [ n ] for [ 1 , n ] [1,n] [ 1 , n ] and ∏ i ∈ A x i = 1 \prod_{i\in A}x_i=1 ∏ i ∈ A x i = 1 if A = ∅ A=\varnothing A = ∅ .
Let Γ \Gamma Γ be a finite simple graph with vertex set V ( Γ ) V(\Gamma) V ( Γ ) and edge set E ( Γ ) E(\Gamma) E ( Γ ) . For a vertex v ∈ V ( Γ ) v \in V(\Gamma) v ∈ V ( Γ ) , its neighborhood is defined by N Γ ( v ) = { u ∈ V ( Γ ) ∣ u v ∈ E ( Γ ) } N_{\Gamma}(v)=\{u \in V(\Gamma) \mid uv \in E(\Gamma)\} N Γ ( v ) = { u ∈ V ( Γ ) ∣ uv ∈ E ( Γ )} . For a subset A ⊆ V ( Γ ) A \subseteq V(\Gamma) A ⊆ V ( Γ ) , its neighborhood is defined by N Γ ( A ) = ⋃ v ∈ A N Γ ( v ) N_{\Gamma}(A)=\bigcup_{v\in A}N_{\Gamma}(v) N Γ ( A ) = ⋃ v ∈ A N Γ ( v ) . A subset C ⊆ V ( Γ ) C \subseteq V(\Gamma) C ⊆ V ( Γ ) is called a vertex cover of Γ \Gamma Γ if every edge of Γ \Gamma Γ has at least one endpoint in C C C . A vertex cover C C C is called minimal if no proper subset of C C C is a vertex cover of Γ \Gamma Γ . The vertex cover number of Γ \Gamma Γ , denoted by τ ( Γ ) \tau(\Gamma) τ ( Γ ) , is defined as
τ ( Γ ) = min { ∣ C ∣ : C is a vertex cover of Γ } . \tau(\Gamma)=\min\{|C|: C\text{ is a vertex cover of }\Gamma\}. τ ( Γ ) = min { ∣ C ∣ : C is a vertex cover of Γ } .
A subset S ⊆ V ( Γ ) S \subseteq V(\Gamma) S ⊆ V ( Γ ) is called an independent set of Γ \Gamma Γ if no two distinct vertices in S S S are adjacent. An independent set S S S of Γ \Gamma Γ is called maximal if every vertex in V ( Γ ) ∖ S V(\Gamma)\setminus S V ( Γ ) ∖ S is adjacent to at least one vertex in S S S . The independence number of Γ \Gamma Γ , denoted by α ( Γ ) \alpha(\Gamma) α ( Γ ) , is defined as
α ( Γ ) = max { ∣ S ∣ ∣ S is an independent set of Γ } . \alpha(\Gamma)=\max\{|S| \mid S\text{ is an independent set of }\Gamma\}. α ( Γ ) = max { ∣ S ∣ ∣ S is an independent set of Γ } .
It is well known that a set C ⊆ V ( Γ ) C \subseteq V(\Gamma) C ⊆ V ( Γ ) is a minimal vertex cover of Γ \Gamma Γ if and only if V ( Γ ) ∖ C V(\Gamma)\setminus C V ( Γ ) ∖ C is a maximal independent set of Γ \Gamma Γ . For any subset A A A of V ( Γ ) V(\Gamma) V ( Γ ) , Γ [ A ] \Gamma[A] Γ [ A ] denotes the induced subgraph of Γ \Gamma Γ on the set A A A , i.e., for any u , v ∈ A u,v \in A u , v ∈ A , u v ∈ E ( Γ [ A ] ) uv \in E(\Gamma[A]) uv ∈ E ( Γ [ A ]) if and only if u v ∈ E ( Γ ) uv \in E(\Gamma) uv ∈ E ( Γ ) .
A walk in Γ \Gamma Γ is a sequence of vertices W = ( x i 0 , x i 1 , … , x i k ) W=(x_{i_0},x_{i_1},\ldots,x_{i_k}) W = ( x i 0 , x i 1 , … , x i k ) such that x i j − 1 x i j ∈ E ( Γ ) x_{i_{j-1}}x_{i_j}\in E(\Gamma) x i j − 1 x i j ∈ E ( Γ ) for each j ∈ [ k ] j\in[k] j ∈ [ k ] . A closed walk is a walk whose initial and terminal vertices coincide, that is, x i 0 = x i k x_{i_0}=x_{i_k} x i 0 = x i k . The length of a walk is the number of its edges. A walk is called odd (resp. even ) if its length is odd (resp. even). A cycle of length k k k in Γ \Gamma Γ is a closed walk C = ( x i 0 , x i 1 , … , x i k = x i 0 ) C=(x_{i_0},x_{i_1},\ldots,x_{i_k}=x_{i_0}) C = ( x i 0 , x i 1 , … , x i k = x i 0 ) such that x i 0 , x i 1 , … , x i k − 1 x_{i_0},x_{i_1},\ldots,x_{i_{k-1}} x i 0 , x i 1 , … , x i k − 1 are pairwise distinct and k ≥ 3 k\geq3 k ≥ 3 .
Let a , n a,n a , n be two positive integers such that n > a n>a n > a . The cubic circulant graph Γ = C 2 n ( a , n ) \Gamma=C_{2n}(a,n) Γ = C 2 n ( a , n ) has vertex set V ( Γ ) = { x i ∣ i ∈ [ 2 n ] } V(\Gamma)=\{x_i\mid i\in[2n]\} V ( Γ ) = { x i ∣ i ∈ [ 2 n ]} , where x i x j x_ix_j x i x j is an edge if and only if ∣ i − j ∣ ∈ { a , n , 2 n − a } |i-j|\in\{a,n,2n-a\} ∣ i − j ∣ ∈ { a , n , 2 n − a } . Note that Γ \Gamma Γ is connected if and only if gcd ( a , n ) = 1 \gcd(a,n)=1 g cd( a , n ) = 1 . Davis and Domke [11 ] proved the following structural result.
Lemma 2.1. Let Γ = C 2 n ( a , n ) \Gamma=C_{2n}(a,n) Γ = C 2 n ( a , n ) with 1 ≤ a < n 1\leq a<n 1 ≤ a < n , and let k = gcd ( a , 2 n ) k=\gcd(a,2n) k = g cd( a , 2 n ) .
(i) If 2 n k \frac{2n}{k} k 2 n is even, then Γ \Gamma Γ is isomorphic to k k k copies of C 2 n k ( 1 , n k ) C_{\frac{2n}{k}}\left(1,\frac{n}{k}\right) C k 2 n ( 1 , k n ) .
(ii) If 2 n k \frac{2n}{k} k 2 n is odd, then Γ \Gamma Γ is isomorphic to k 2 \frac{k}{2} 2 k copies of C 4 n k ( 2 , 2 n k ) C_{\frac{4n}{k}}\left(2,\frac{2n}{k}\right) C k 4 n ( 2 , k 2 n ) .
Lemma 2.2. ([13 Theorem 4.1] and [26 Theorem 4.1(3)] ) Let R 1 = K [ x 1 , … , x m ] R_1=\mathbb{K}[x_1,\ldots,x_m] R 1 = K [ x 1 , … , x m ] , R 2 = K [ x m + 1 , … , x n ] R_2=\mathbb{K}[x_{m+1},\ldots,x_n] R 2 = K [ x m + 1 , … , x n ] and R = K [ x 1 , … , x m , x m + 1 , … , x n ] R=\mathbb{K}[x_1,\ldots,x_m,x_{m+1},\ldots,x_n] R = K [ x 1 , … , x m , x m + 1 , … , x n ] be three polynomial rings over a field K \mathbb{K} K , let I 1 ⊆ R 1 I_1\subseteq R_1 I 1 ⊆ R 1 and I 2 ⊆ R 2 I_2\subseteq R_2 I 2 ⊆ R 2 be two monomial ideals, and let t ≥ 1 t\geq1 t ≥ 1 be an integer. Then
Ass R ( ( I 1 + I 2 ) t ) = ⋃ i = 1 t ⋃ p ∈ Ass R 1 ( I 1 i ) q ∈ Ass R 2 ( I 2 t − i + 1 ) { p + q } . \operatorname{Ass}_{R}\left((I_1+I_2)^t\right)=\bigcup_{i=1}^{t}\ \bigcup_{\substack{\mathfrak{p}\in\operatorname{Ass}_{R_1}(I_1^i)\\ \mathfrak{q}\in\operatorname{Ass}_{R_2}(I_2^{t-i+1})}}\{\mathfrak{p}+\mathfrak{q}\}. Ass R ( ( I 1 + I 2 ) t ) = i = 1 ⋃ t p ∈ Ass R 1 ( I 1 i ) q ∈ Ass R 2 ( I 2 t − i + 1 ) ⋃ { p + q } . Furthermore, if p \mathfrak{p} p and q \mathfrak{q} q are prime ideals in R 1 R_1 R 1 and R 2 R_2 R 2 , respectively, such that p + q ∈ Ass ( ( I 1 + I 2 ) t ) \mathfrak{p}+\mathfrak{q}\in\operatorname{Ass}((I_1+I_2)^t) p + q ∈ Ass (( I 1 + I 2 ) t ) , then
v p + q ( ( I 1 + I 2 ) t ) = min 1 ≤ i ≤ t p ∈ Ass R 1 ( I 1 i ) q ∈ Ass R 2 ( I 2 t − i + 1 ) ( v p ( I 1 i ) + v q ( I 2 t − i + 1 ) ) . v_{\mathfrak{p}+\mathfrak{q}}\left((I_1+I_2)^t\right)=\min_{\substack{1\leq i\leq t\\ \mathfrak{p}\in\operatorname{Ass}_{R_1}(I_1^i)\\ \mathfrak{q}\in\operatorname{Ass}_{R_2}(I_2^{t-i+1})}}\left(v_{\mathfrak{p}}(I_1^i)+v_{\mathfrak{q}}(I_2^{t-i+1})\right). v p + q ( ( I 1 + I 2 ) t ) = 1 ≤ i ≤ t p ∈ Ass R 1 ( I 1 i ) q ∈ Ass R 2 ( I 2 t − i + 1 ) min ( v p ( I 1 i ) + v q ( I 2 t − i + 1 ) ) .
By the lemma above, we have
Lemma 2.3. Let R 1 = K [ x 1 , … , x m ] R_1=\mathbb{K}[x_1,\ldots,x_m] R 1 = K [ x 1 , … , x m ] and R 2 = K [ x m + 1 , … , x n ] R_2=\mathbb{K}[x_{m+1},\ldots,x_n] R 2 = K [ x m + 1 , … , x n ] be two polynomial rings over a field K \mathbb{K} K , let I 1 ⊆ R 1 I_1\subseteq R_1 I 1 ⊆ R 1 and I 2 ⊆ R 2 I_2\subseteq R_2 I 2 ⊆ R 2 be two monomial ideals, and let t ≥ 1 t\geq1 t ≥ 1 be an integer. Then
v ( ( I 1 + I 2 ) t ) = min 1 ≤ j ≤ t { v ( I 1 j ) + v ( I 2 t + 1 − j ) } . v\left((I_1+I_2)^t\right)=\min_{1\leq j\leq t}\left\{v(I_1^j)+v(I_2^{t+1-j})\right\}. v ( ( I 1 + I 2 ) t ) = 1 ≤ j ≤ t min { v ( I 1 j ) + v ( I 2 t + 1 − j ) } .
Thus, by Lemmas 2.1 and 2.3, determining v ( I ( C 2 n ( a , n ) ) t ) v(I(C_{2n}(a,n))^t) v ( I ( C 2 n ( a , n ) ) t ) for 1 ≤ a < n 1\leq a<n 1 ≤ a < n reduces to considering the cases a = 1 a=1 a = 1 and a = 2 a=2 a = 2 when n ≥ 3 n\geq3 n ≥ 3 is odd.
For a monomial f = ∏ i ∈ [ n ] x i s i ∈ R f=\prod_{i\in[n]}x_i^{s_i}\in R f = ∏ i ∈ [ n ] x i s i ∈ R , we define its degree with respect to x i x_i x i as deg x i ( f ) = s i \deg_{x_i}(f)=s_i deg x i ( f ) = s i , and its total degree as deg ( f ) = ∑ i = 1 n s i \deg(f)=\sum_{i=1}^{n}s_i deg ( f ) = ∑ i = 1 n s i . For a prime ideal p \mathfrak{p} p generated by a subset of the variables, we further define
deg p ( f ) = ∑ x i ∈ p deg x i ( f ) . \deg_{\mathfrak{p}}(f)=\sum_{x_i\in\mathfrak{p}}\deg_{x_i}(f). deg p ( f ) = x i ∈ p ∑ deg x i ( f ) .
Lemma 2.4. ([12 Proposition 2.2] ) Let I ⊂ R I\subset R I ⊂ R be a monomial ideal. Then v p ( I ) ≥ α ( I ) − 1 v_{\mathfrak{p}}(I)\geq\alpha(I)-1 v p ( I ) ≥ α ( I ) − 1 for all p ∈ Ass ( I ) \mathfrak{p}\in\operatorname{Ass}(I) p ∈ Ass ( I ) , where α ( I ) = min { deg ( f ) ∣ f ∈ I ∖ { 0 } } \alpha(I)=\min\{\deg(f)\mid f\in I\setminus\{0\}\} α ( I ) = min { deg ( f ) ∣ f ∈ I ∖ { 0 }} .
Lemma 2.5. ([22 Theorem 3.5] ) Let Γ \Gamma Γ be a simple graph and let I I I be its edge ideal. Let
M Γ = { A ∣ A is a maximal independent set of Γ } , and \mathcal{M}_{\Gamma}=\{A\mid A\text{ is a maximal independent set of }\Gamma\},\quad\text{and} M Γ = { A ∣ A is a maximal independent set of Γ } , and A Γ = { A ∣ A is an independent set of Γ and N Γ ( A ) is a minimal vertex cover of Γ } . \mathcal{A}_{\Gamma}=\{A\mid A\text{ is an independent set of }\Gamma\text{ and }N_{\Gamma}(A)\text{ is a minimal vertex cover of }\Gamma\}. A Γ = { A ∣ A is an independent set of Γ and N Γ ( A ) is a minimal vertex cover of Γ } . Then
M Γ ⊆ A Γ and v ( I ) = min { ∣ A ∣ ∣ A ∈ A Γ } . \mathcal{M}_{\Gamma}\subseteq\mathcal{A}_{\Gamma}\quad\text{and}\quad v(I)=\min\{|A|\mid A\in\mathcal{A}_{\Gamma}\}. M Γ ⊆ A Γ and v ( I ) = min { ∣ A ∣ ∣ A ∈ A Γ } .
Lemma 2.6. ([5 Lemma 2.2] ) Let t t t be a positive integer and let I ⊆ R I \subseteq R I ⊆ R be a nonzero squarefree monomial ideal. Suppose f f f is a monomial such that deg p ( f ) = t − 1 \deg_{\mathfrak p}(f)=t-1 deg p ( f ) = t − 1 and p ⊆ ( I t : f ) \mathfrak p \subseteq(I^t:f) p ⊆ ( I t : f ) for some p ∈ Ass ( I ) \mathfrak p \in\operatorname{Ass}(I) p ∈ Ass ( I ) , then
( I t : f ) = p . (I^t:f)=\mathfrak p. ( I t : f ) = p .
Let Γ \Gamma Γ be a graph and e 1 , … , e t e_1,\ldots,e_t e 1 , … , e t edges of Γ \Gamma Γ . We denote their t t t -fold product by ∏ i = 1 t e i \prod_{i=1}^{t}e_i ∏ i = 1 t e i . To analyze the generators of ( I ( Γ ) t + 1 : ∏ i = 1 t e i ) (I(\Gamma)^{t+1}:\prod_{i=1}^{t}e_i) ( I ( Γ ) t + 1 : ∏ i = 1 t e i ) , we introduce the notion of even-connectedness with respect to t t t -fold products.
Definition 2.7. ([2 Definition 6.2] ) Two vertices u u u and v v v (u u u may be equal to v v v ) are said to be even-connected with respect to a t t t -fold product ∏ i = 1 t e i \prod_{i=1}^{t}e_i ∏ i = 1 t e i if there is a walk W : x 1 , x 2 , … , x 2 k W:x_1,x_2,\ldots,x_{2k} W : x 1 , x 2 , … , x 2 k with k ≥ 2 k\geq2 k ≥ 2 in Γ \Gamma Γ such that
x 1 = u x_1=u x 1 = u and x 2 k = v x_{2k}=v x 2 k = v .
For each ℓ ∈ [ k − 1 ] \ell\in[k-1] ℓ ∈ [ k − 1 ] , x 2 ℓ x 2 ℓ + 1 = e i x_{2\ell}x_{2\ell+1}=e_i x 2 ℓ x 2 ℓ + 1 = e i for some i ∈ [ t ] i\in[t] i ∈ [ t ] .
For all i ∈ [ t ] i\in[t] i ∈ [ t ] ,
∣ { ℓ ∈ [ k − 1 ] ∣ x 2 ℓ x 2 ℓ + 1 = e i } ∣ ≤ ∣ { j ∈ [ t ] ∣ e j = e i } ∣ . \left|\{\ell\in[k-1]\mid x_{2\ell}x_{2\ell+1}=e_i\}\right|
\leq
\left|\{j\in[t]\mid e_j=e_i\}\right|. ∣ { ℓ ∈ [ k − 1 ] ∣ x 2 ℓ x 2 ℓ + 1 = e i } ∣ ≤ ∣ { j ∈ [ t ] ∣ e j = e i } ∣ . For all r ∈ [ 2 k − 1 ] r\in[2k-1] r ∈ [ 2 k − 1 ] , x r x r + 1 x_rx_{r+1} x r x r + 1 is an edge in Γ \Gamma Γ .
If these properties are satisfied, then x 1 , x 2 , … , x 2 k x_1,x_2,\ldots,x_{2k} x 1 , x 2 , … , x 2 k is said to be an even-connection between u u u and v v v with respect to ∏ i = 1 t e i \prod_{i=1}^{t}e_i ∏ i = 1 t e i .
Theorem 2.8. ([2 Theorems 6.1 and 6.7] ) Let t ≥ 1 t\geq1 t ≥ 1 be an integer, and let Γ \Gamma Γ be a graph and e 1 , … , e t e_1,\ldots,e_t e 1 , … , e t be its edges. Then
( I ( Γ ) t + 1 : ∏ i = 1 t e i ) \left(I(\Gamma)^{t+1}:\prod_{i=1}^{t}e_i\right) ( I ( Γ ) t + 1 : i = 1 ∏ t e i ) is generated by monomials of degree two. Moreover, every generator u v uv uv (u u u may be equal to v v v ) of ( I ( Γ ) t + 1 : ∏ i = 1 t e i ) (I(\Gamma)^{t+1}:\prod_{i=1}^{t}e_i) ( I ( Γ ) t + 1 : ∏ i = 1 t e i ) is either an edge of Γ \Gamma Γ or even-connected with respect to ∏ i = 1 t e i \prod_{i=1}^{t}e_i ∏ i = 1 t e i .
For a monomial ideal I I I , let G ( I ) G(I) G ( I ) be the unique minimal set of its monomial generators.
Lemma 2.9. Let t ≥ 2 t\geq2 t ≥ 2 be an integer and let Γ \Gamma Γ be a simple graph. Let f ∈ R f\in R f ∈ R be a monomial such that ( I ( Γ ) t : f ) = p (I(\Gamma)^t:f)=\mathfrak p ( I ( Γ ) t : f ) = p for some p ∈ Ass ( I ( Γ ) ) \mathfrak p\in\operatorname{Ass}(I(\Gamma)) p ∈ Ass ( I ( Γ )) . Then
f ∈ I ( Γ ) t − 1 . f\in I(\Gamma)^{t-1}. f ∈ I ( Γ ) t − 1 .
Proof. Let x ∈ p x\in\mathfrak p x ∈ p , then x f ∈ I ( Γ ) t xf\in I(\Gamma)^t x f ∈ I ( Γ ) t . Thus we can write x f xf x f as x f = u 1 ⋯ u t h xf=u_1\cdots u_th x f = u 1 ⋯ u t h , where each u i ∈ G ( I ( Γ ) ) u_i\in G(I(\Gamma)) u i ∈ G ( I ( Γ )) and h h h is a monomial. If x ∣ h x\mid h x ∣ h , then f = u 1 ⋯ u t ( h / x ) ∈ I ( Γ ) t f=u_1\cdots u_t(h/x)\in I(\Gamma)^t f = u 1 ⋯ u t ( h / x ) ∈ I ( Γ ) t . Otherwise, x ∣ u ℓ x\mid u_\ell x ∣ u ℓ for some ℓ ∈ [ t ] \ell\in[t] ℓ ∈ [ t ] . Thus f = u 1 ⋯ u ℓ − 1 ( u ℓ / x ) u ℓ + 1 ⋯ u t h ∈ I ( Γ ) t − 1 f=u_1\cdots u_{\ell-1}(u_\ell/x)u_{\ell+1}\cdots u_th\in I(\Gamma)^{t-1} f = u 1 ⋯ u ℓ − 1 ( u ℓ / x ) u ℓ + 1 ⋯ u t h ∈ I ( Γ ) t − 1 .
:::
For a monomial f = ∏ i ∈ [ n ] x i s i f = \prod_{i\in[n]} x_i^{s_i} f = ∏ i ∈ [ n ] x i s i , its support is defined by supp ( f ) = { x i ∣ i ∈ [ n ] and s i ≥ 1 } \operatorname{supp}(f) = \{x_i \mid i \in[n] \text{ and } s_i \ge1\} supp ( f ) = { x i ∣ i ∈ [ n ] and s i ≥ 1 } . For a monomial ideal I I I , we set supp ( I ) = ⋃ u ∈ G ( I ) supp ( u ) \operatorname{supp}(I) = \bigcup_{u\in G(I)} \operatorname{supp}(u) supp ( I ) = ⋃ u ∈ G ( I ) supp ( u ) .
Lemma 2.10. Let Γ \Gamma Γ be a simple graph , and let f ∈ R f \in R f ∈ R be a monomial such that ( I ( Γ ) : f ) = p (I(\Gamma) : f) = \mathfrak{p} ( I ( Γ ) : f ) = p for some p ∈ Ass ( I ( Γ ) ) \mathfrak{p} \in\operatorname{Ass}(I(\Gamma)) p ∈ Ass ( I ( Γ )) . Then p = ( x i ∣ x i ∈ N Γ ( supp ( f ) ) ) \mathfrak{p} = (x_i \mid x_i \in N_{\Gamma}(\operatorname{supp}(f))) p = ( x i ∣ x i ∈ N Γ ( supp ( f ))) .
Proof. First, we show that ( x i ∣ x i ∈ N Γ ( supp ( f ) ) ) ⊆ p (x_i \mid x_i \in N_{\Gamma}(\operatorname{supp}(f))) \subseteq\mathfrak{p} ( x i ∣ x i ∈ N Γ ( supp ( f ))) ⊆ p . Let x i ∈ N Γ ( supp ( f ) ) x_i \in N_{\Gamma}(\operatorname{supp}(f)) x i ∈ N Γ ( supp ( f )) , then there exists a x j ∈ supp ( f ) x_j \in\operatorname{supp}(f) x j ∈ supp ( f ) such that x i x j ∈ E ( Γ ) x_i x_j \in E(\Gamma) x i x j ∈ E ( Γ ) . We can write f f f as f = x j g f = x_j g f = x j g for some monomial g g g . Thus x i f = x i ( x j g ) = ( x i x j ) g ∈ I ( Γ ) x_i f = x_i(x_j g) = (x_i x_j)g \in I(\Gamma) x i f = x i ( x j g ) = ( x i x j ) g ∈ I ( Γ ) . Therefore, x i ∈ ( I ( Γ ) : f ) = p x_i \in(I(\Gamma) : f) = \mathfrak{p} x i ∈ ( I ( Γ ) : f ) = p , it follows that ( x i ∣ x i ∈ N Γ ( supp ( f ) ) ) ⊆ p (x_i \mid x_i \in N_{\Gamma}(\operatorname{supp}(f))) \subseteq\mathfrak{p} ( x i ∣ x i ∈ N Γ ( supp ( f ))) ⊆ p .
Next, we show that p ⊆ ( x i ∣ x i ∈ N Γ ( supp ( f ) ) ) \mathfrak{p} \subseteq(x_i \mid x_i \in N_{\Gamma}(\operatorname{supp}(f))) p ⊆ ( x i ∣ x i ∈ N Γ ( supp ( f ))) . Let p = ( x i 1 , x i 2 , … , x i s ) \mathfrak{p} = (x_{i_1}, x_{i_2}, \ldots, x_{i_s}) p = ( x i 1 , x i 2 , … , x i s ) , where s ≥ 1 s \ge1 s ≥ 1 . For any j ∈ [ s ] j \in[s] j ∈ [ s ] , we have x i j ∈ ( I ( Γ ) : f ) x_{i_j} \in(I(\Gamma) : f) x i j ∈ ( I ( Γ ) : f ) , which implies that x i j f ∈ I ( Γ ) x_{i_j}f \in I(\Gamma) x i j f ∈ I ( Γ ) . Thus x u x v ∣ x i j f x_u x_v \mid x_{i_j}f x u x v ∣ x i j f for some x u x v ∈ E ( Γ ) x_u x_v \in E(\Gamma) x u x v ∈ E ( Γ ) . Note that f ∉ I ( Γ ) f \notin I(\Gamma) f ∈ / I ( Γ ) , x u x v ∤ f x_u x_v \nmid f x u x v ∤ f . Thus, x u = x i j x_u = x_{i_j} x u = x i j or x v = x i j x_v = x_{i_j} x v = x i j . Without loss of generality, we can assume x u = x i j x_u = x_{i_j} x u = x i j . Thus x v ∣ f x_v \mid f x v ∣ f , so x v ∈ supp ( f ) x_v \in\operatorname{supp}(f) x v ∈ supp ( f ) . This implies that x i j ∈ N Γ ( x v ) ⊆ N Γ ( supp ( f ) ) x_{i_j} \in N_{\Gamma}(x_v) \subseteq N_{\Gamma}(\operatorname{supp}(f)) x i j ∈ N Γ ( x v ) ⊆ N Γ ( supp ( f )) , since x i j x v = x u x v ∈ E ( Γ ) x_{i_j}x_v = x_u x_v \in E(\Gamma) x i j x v = x u x v ∈ E ( Γ ) . Therefore, p ⊆ ( x i ∣ x i ∈ N Γ ( supp ( f ) ) ) \mathfrak{p} \subseteq(x_i \mid x_i \in N_{\Gamma}(\operatorname{supp}(f))) p ⊆ ( x i ∣ x i ∈ N Γ ( supp ( f ))) .
Lemma 2.11. Let t ≥ 2 t \ge2 t ≥ 2 be an integer and let Γ = C 2 n ( a , n ) \Gamma= C_{2n}(a,n) Γ = C 2 n ( a , n ) be a cubic circulant graph with a ∈ { 1 , 2 } a \in\{1,2\} a ∈ { 1 , 2 } . Suppose that for any edges e 1 , … , e t − 1 ∈ E ( Γ ) e_1,\ldots,e_{t-1} \in E(\Gamma) e 1 , … , e t − 1 ∈ E ( Γ ) , the colon ideal ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) (I(\Gamma)^t : \prod_{i=1}^{t-1} e_i) ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) can be regarded as the edge ideal of a simple graph. Then
v ( I ( Γ ) t ) ≥ 2 t − 2 + ⌈ τ ( Γ ) − 2 t + 2 3 ⌉ . \operatorname{v}(I(\Gamma)^t) \ge2t - 2 + \left\lceil\frac{\tau(\Gamma)-2t+2}{3} \right\rceil. v ( I ( Γ ) t ) ≥ 2 t − 2 + ⌈ 3 τ ( Γ ) − 2 t + 2 ⌉ . Proof. Let I = I ( Γ ) I = I(\Gamma) I = I ( Γ ) and ( I t : f ) = p (I^t : f) = \mathfrak{p} ( I t : f ) = p for some p ∈ Ass ( I t ) \mathfrak{p} \in\operatorname{Ass}(I^t) p ∈ Ass ( I t ) and a monomial f ∈ R f \in R f ∈ R such that deg ( f ) = v ( I t ) \deg(f) = \operatorname{v}(I^t) deg ( f ) = v ( I t ) . By Lemma 2.9, f ∈ I t − 1 f \in I^{t-1} f ∈ I t − 1 . Let f = g ∏ i = 1 t − 1 e i f = g\prod_{i=1}^{t-1} e_i f = g ∏ i = 1 t − 1 e i , where e 1 , e 2 , … , e t − 1 ∈ E ( Γ ) e_1,e_2,\ldots,e_{t-1} \in E(\Gamma) e 1 , e 2 , … , e t − 1 ∈ E ( Γ ) and g g g is a monomial. Since
( I t : f ) = ( I t : g ∏ i = 1 t − 1 e i ) = ( ( I t : ∏ i = 1 t − 1 e i ) : g ) = ( I ( H ) : g ) = p , (I^t : f) = \left(I^t : g\prod_{i=1}^{t-1} e_i\right) = \left(\left(I^t : \prod_{i=1}^{t-1} e_i\right) : g\right) = (I(H) : g) = \mathfrak{p}, ( I t : f ) = ( I t : g i = 1 ∏ t − 1 e i ) = ( ( I t : i = 1 ∏ t − 1 e i ) : g ) = ( I ( H ) : g ) = p , where H H H is a simple graph such that I ( H ) = ( I t : ∏ i = 1 t − 1 e i ) I(H) = \left(I^t : \prod_{i=1}^{t-1} e_i\right) I ( H ) = ( I t : ∏ i = 1 t − 1 e i ) . Then g g g is squarefree and p ∈ Ass ( I ( H ) ) \mathfrak{p} \in\operatorname{Ass}(I(H)) p ∈ Ass ( I ( H )) . Therefore, supp ( p ) \operatorname{supp}(\mathfrak{p}) supp ( p ) is a minimal vertex cover of H H H . It follows that ∣ supp ( p ) ∣ ≥ τ ( H ) |\operatorname{supp}(\mathfrak{p})| \ge\tau(H) ∣ supp ( p ) ∣ ≥ τ ( H ) . On the other hand, by Lemma 2.8, each generator x i x j x_i x_j x i x j of ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) \left(I(\Gamma)^t : \prod_{i=1}^{t-1} e_i\right) ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) is either an edge of Γ \Gamma Γ or even-connected with respect to ∏ i = 1 t − 1 e i \prod_{i=1}^{t-1} e_i ∏ i = 1 t − 1 e i . By Definition 2.7 of even-connected, we know that V ( Γ ) = V ( H ) V(\Gamma) = V(H) V ( Γ ) = V ( H ) and E ( Γ ) ⊆ E ( H ) E(\Gamma) \subseteq E(H) E ( Γ ) ⊆ E ( H ) . Thus ∣ supp ( p ) ∣ ≥ τ ( H ) ≥ τ ( Γ ) |\operatorname{supp}(\mathfrak{p})| \ge\tau(H) \ge\tau(\Gamma) ∣ supp ( p ) ∣ ≥ τ ( H ) ≥ τ ( Γ ) .
Since deg ( f ) = deg ( g ∏ i = 1 t − 1 e i ) \deg(f) = \deg\left(g\prod_{i=1}^{t-1} e_i\right) deg ( f ) = deg ( g ∏ i = 1 t − 1 e i ) and deg ( ∏ i = 1 t − 1 e i ) = 2 ( t − 1 ) \deg\left(\prod_{i=1}^{t-1} e_i\right) = 2(t-1) deg ( ∏ i = 1 t − 1 e i ) = 2 ( t − 1 ) , to prove that deg ( f ) ≥ 2 t − 2 + ⌈ τ ( Γ ) − 2 t + 2 3 ⌉ \deg(f) \ge2t - 2 + \left\lceil\frac{\tau(\Gamma)-2t+2}{3} \right\rceil deg ( f ) ≥ 2 t − 2 + ⌈ 3 τ ( Γ ) − 2 t + 2 ⌉ , it suffices to show that
3 deg ( g ) + 2 ( t − 1 ) ≥ ∣ supp ( p ) ∣ . 3\deg(g) + 2(t-1) \ge|\operatorname{supp}(\mathfrak{p})|. 3 deg ( g ) + 2 ( t − 1 ) ≥ ∣ supp ( p ) ∣.
By Lemma 2.10, p = ( x i ∣ x i ∈ N H ( supp ( g ) ) ) \mathbf{p}=(x_i\mid x_i\in N_H(\operatorname{supp}(g))) p = ( x i ∣ x i ∈ N H ( supp ( g ))) , so ∣ supp ( p ) ∣ = ∣ N H ( supp ( g ) ) ∣ |\operatorname{supp}(\mathbf{p})|=|N_H(\operatorname{supp}(g))| ∣ supp ( p ) ∣ = ∣ N H ( supp ( g )) ∣ . Note that
N H ( supp ( g ) ) = N Γ ( supp ( g ) ) ∪ ( N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) ) , N_H(\operatorname{supp}(g))=N_\Gamma(\operatorname{supp}(g))\mathbin{\mathop{\cup}}(N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g))), N H ( supp ( g )) = N Γ ( supp ( g )) ∪ ( N H ( supp ( g )) ∖ N Γ ( supp ( g ))) ,
we have
∣ N H ( supp ( g ) ) ∣ = ∣ N Γ ( supp ( g ) ) ∣ + ∣ N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) ∣ . |N_H(\operatorname{supp}(g))|=|N_\Gamma(\operatorname{supp}(g))|+|N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g))|. ∣ N H ( supp ( g )) ∣ = ∣ N Γ ( supp ( g )) ∣ + ∣ N H ( supp ( g )) ∖ N Γ ( supp ( g )) ∣.
Therefore, if we prove that
3 deg ( g ) ≥ ∣ N Γ ( supp ( g ) ) ∣ (1) 3\deg(g)\geq|N_\Gamma(\operatorname{supp}(g))| \tag*{(1)} 3 deg ( g ) ≥ ∣ N Γ ( supp ( g )) ∣ (1)
and
2 ( t − 1 ) ≥ ∣ N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) ∣ , (2) 2(t-1)\geq|N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g))|, \tag*{(2)} 2 ( t − 1 ) ≥ ∣ N H ( supp ( g )) ∖ N Γ ( supp ( g )) ∣ , (2)
then the conclusion follows.
Note that Γ \Gamma Γ is 3-regular, we have ∣ N Γ ( x i ) ∣ = 3 |N_\Gamma(x_i)|=3 ∣ N Γ ( x i ) ∣ = 3 for any i ∈ [ 2 n ] i\in[2n] i ∈ [ 2 n ] . Since g g g is squarefree, it follows that 3 deg ( g ) = 3 ∣ supp ( g ) ∣ ≥ ∣ N Γ ( supp ( g ) ) ∣ 3\deg(g)=3|\operatorname{supp}(g)|\geq|N_\Gamma(\operatorname{supp}(g))| 3 deg ( g ) = 3∣ supp ( g ) ∣ ≥ ∣ N Γ ( supp ( g )) ∣ . That is, (1) holds.
Next, we will prove the relationship (2).
Let d 1 , … , d s d_1,\ldots,d_s d 1 , … , d s be the distinct edges appearing in e 1 , … , e t − 1 e_1,\ldots,e_{t-1} e 1 , … , e t − 1 and let each d i = a i b i d_i=a_i b_i d i = a i b i . Then s ≤ t − 1 s\leq t-1 s ≤ t − 1 . It follows from [18 Lemma 3.5] that
I ( H ) = ( I t : ∏ i = 1 t − 1 e i ) = ( I s + 1 : ∏ i = 1 s d i ) . (3) I(H)=\left(I^t:\prod_{i=1}^{t-1}e_i\right)=\left(I^{s+1}:\prod_{i=1}^{s}d_i\right). \tag*{(3)} I ( H ) = ( I t : i = 1 ∏ t − 1 e i ) = ( I s + 1 : i = 1 ∏ s d i ) . (3)
For each u ∈ N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) u\in N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g)) u ∈ N H ( supp ( g )) ∖ N Γ ( supp ( g )) , there exists a v u ∈ supp ( g ) v_u\in\operatorname{supp}(g) v u ∈ supp ( g ) such that u v u ∈ E ( H ) ∖ E ( Γ ) uv_u\in E(H)\setminus E(\Gamma) u v u ∈ E ( H ) ∖ E ( Γ ) . Thus, by (3) and Lemma 2.8, we can deduce that u u u and v u v_u v u are even-connected path with respect to ∏ i = 1 s d i \prod_{i=1}^{s}d_i ∏ i = 1 s d i in Γ \Gamma Γ . Suppose that this path is π u : u = x u 1 , x u 2 , … , x u 2 k = v u \pi_u:u=x_{u_1},x_{u_2},\ldots,x_{u_{2k}}=v_u π u : u = x u 1 , x u 2 , … , x u 2 k = v u satisfying k k k as small as possible. Since u v u ∉ E ( Γ ) uv_u\notin E(\Gamma) u v u ∈ / E ( Γ ) and k k k is minimus, we have k ≥ 2 k\geq2 k ≥ 2 and u ≠ x u 3 u\neq x_{u_3} u = x u 3 , which implies u ∈ N Γ ( x u 2 ) ∖ { x u 3 } u\in N_\Gamma(x_{u_2})\setminus\{x_{u_3}\} u ∈ N Γ ( x u 2 ) ∖ { x u 3 } .
By the arbitrariness of u u u , for each u ∈ N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) u\in N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g)) u ∈ N H ( supp ( g )) ∖ N Γ ( supp ( g )) , we fix an even-connected path π u : u = x u 1 , x u 2 , … , x u 2 k = v u \pi_u:u=x_{u_1},x_{u_2},\ldots,x_{u_{2k}}=v_u π u : u = x u 1 , x u 2 , … , x u 2 k = v u with respect to ∏ i = 1 s d i \prod_{i=1}^{s}d_i ∏ i = 1 s d i satisfying k k k as small as possible. Since x u 2 x u 3 = a j u b j u x_{u_2}x_{u_3}=a_{j_u}b_{j_u} x u 2 x u 3 = a j u b j u for some j u ∈ [ s ] j_u\in[s] j u ∈ [ s ] , we have either x u 2 = a j u x_{u_2}=a_{j_u} x u 2 = a j u and x u 3 = b j u x_{u_3}=b_{j_u} x u 3 = b j u , or x u 2 = b j u x_{u_2}=b_{j_u} x u 2 = b j u and x u 3 = a j u x_{u_3}=a_{j_u} x u 3 = a j u . Furthermore, u ∈ N Γ ( a j u ) ∖ { b j u } u\in N_\Gamma(a_{j_u})\setminus\{b_{j_u}\} u ∈ N Γ ( a j u ) ∖ { b j u } or u ∈ N Γ ( b j u ) ∖ { a j u } u\in N_\Gamma(b_{j_u})\setminus\{a_{j_u}\} u ∈ N Γ ( b j u ) ∖ { a j u } .
Next, for each j j j in [ s ] [s] [ s ] , we define two sets as follows:
A_j=\left\{u\in N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g))\mathrel{\left|}\text{ the even-connection }\pi_u\text{ satisfies }x_{u_2}=a_j\text{ and }x_{u_3}=b_j\right.\right\},
B_j=\left\{u\in N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g))\mathrel{\left|}\text{ the even-connection }\pi_u\text{ satisfies }x_{u_2}=b_j\text{ and }x_{u_3}=a_j\right.\right\}.
As mentioned earlier, u ∈ N Γ ( a j u ) ∖ { b j u } u\in N_\Gamma(a_{j_u})\setminus\{b_{j_u}\} u ∈ N Γ ( a j u ) ∖ { b j u } or u ∈ N Γ ( b j u ) ∖ { a j u } u\in N_\Gamma(b_{j_u})\setminus\{a_{j_u}\} u ∈ N Γ ( b j u ) ∖ { a j u } . Thus A j ⊆ N Γ ( a j ) ∖ { b j } A_j\subseteq N_\Gamma(a_j)\setminus\{b_j\} A j ⊆ N Γ ( a j ) ∖ { b j } , B j ⊆ N Γ ( b j ) ∖ { a j } B_j\subseteq N_\Gamma(b_j)\setminus\{a_j\} B j ⊆ N Γ ( b j ) ∖ { a j } and
⋃ j ∈ [ s ] ( A j ∪ B j ) = N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) . \bigcup_{j\in[s]}(A_j\cup B_j)=N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g)). j ∈ [ s ] ⋃ ( A j ∪ B j ) = N H ( supp ( g )) ∖ N Γ ( supp ( g )) .
Furthermore, ∣ A j ∣ ≤ 2 |A_j|\leq2 ∣ A j ∣ ≤ 2 and ∣ B j ∣ ≤ 2 |B_j|\leq2 ∣ B j ∣ ≤ 2 , since Γ \Gamma Γ is 3-regular.
Claim: for each j ∈ [ s ] j\in[s] j ∈ [ s ] , at least one of A j A_j A j and B j B_j B j is empty. Suppose on contradiction that A j ≠ ∅ A_j \ne\varnothing A j = ∅ and B j ≠ ∅ B_j \ne\varnothing B j = ∅ for some j ∈ [ s ] j \in[s] j ∈ [ s ] . Then there exist vertices u 1 ∈ A j u_1 \in A_j u 1 ∈ A j , u 2 ∈ B j u_2 \in B_j u 2 ∈ B j and two even-connected paths
π u 1 : u 1 → a j → b j → ⋯ → v u 1 , \pi_{u_1}: u_1 \to a_j \to b_j \to\cdots\to v_{u_1}, π u 1 : u 1 → a j → b j → ⋯ → v u 1 ,
π u 2 : u 2 → b j → a j → ⋯ → v u 2 , \pi_{u_2}: u_2 \to b_j \to a_j \to\cdots\to v_{u_2}, π u 2 : u 2 → b j → a j → ⋯ → v u 2 ,
such that v u 1 , v u 2 ∈ supp ( g ) v_{u_1}, v_{u_2} \in\operatorname{supp}(g) v u 1 , v u 2 ∈ supp ( g ) . Thus, b j v u 1 ∈ E ( Γ ) b_jv_{u_1} \in E(\Gamma) b j v u 1 ∈ E ( Γ ) , or b j v u 1 b_jv_{u_1} b j v u 1 is even-connected with respect to d 1 ⋯ d j − 1 d j ^ d j + 1 ⋯ d s d_1 \cdots d_{j-1}\widehat{d_j}d_{j+1} \cdots d_s d 1 ⋯ d j − 1 d j d j + 1 ⋯ d s , where d j ^ \widehat{d_j} d j denotes the element d j d_j d j omitted from d 1 d 2 ⋯ d s d_1d_2 \cdots d_s d 1 d 2 ⋯ d s . By Lemma 2.8, v u 1 b j ∈ ( I ( Γ ) s : ∏ i ∈ [ s ] ∖ { j } d i ) v_{u_1}b_j \in(I(\Gamma)^s : \prod_{i \in[s]\setminus\{j\}} d_i) v u 1 b j ∈ ( I ( Γ ) s : ∏ i ∈ [ s ] ∖ { j } d i ) . Similarly, v u 2 a j ∈ ( I ( Γ ) s : ∏ i ∈ [ s ] ∖ { j } d i ) v_{u_2}a_j \in(I(\Gamma)^s : \prod_{i \in[s]\setminus\{j\}} d_i) v u 2 a j ∈ ( I ( Γ ) s : ∏ i ∈ [ s ] ∖ { j } d i ) . This implies v u 1 ∈ ( I ( Γ ) s : b j ∏ i ∈ [ s ] ∖ { j } d i ) v_{u_1} \in(I(\Gamma)^s : b_j\prod_{i \in[s]\setminus\{j\}} d_i) v u 1 ∈ ( I ( Γ ) s : b j ∏ i ∈ [ s ] ∖ { j } d i ) and v u 2 ∈ ( I ( Γ ) s : a j ∏ i ∈ [ s ] ∖ { j } d i ) v_{u_2} \in(I(\Gamma)^s : a_j\prod_{i \in[s]\setminus\{j\}} d_i) v u 2 ∈ ( I ( Γ ) s : a j ∏ i ∈ [ s ] ∖ { j } d i ) .
If v u 1 = v u 2 v_{u_1}=v_{u_2} v u 1 = v u 2 , then v u 1 ∈ ( I ( Γ ) s : b j f j ) ∩ ( I ( Γ ) s : a j f j ) v_{u_1} \in(I(\Gamma)^s : b_jf_j) \cap(I(\Gamma)^s : a_jf_j) v u 1 ∈ ( I ( Γ ) s : b j f j ) ∩ ( I ( Γ ) s : a j f j ) , where f j = ∏ i ∈ [ s ] ∖ { j } d i f_j=\prod_{i \in[s]\setminus\{j\}}d_i f j = ∏ i ∈ [ s ] ∖ { j } d i . Note that
( I ( Γ ) s : f j a j ) ∩ ( I ( Γ ) s : f j b j ) = I ( H ) . (4) (I(\Gamma)^s : f_ja_j) \cap(I(\Gamma)^s : f_jb_j)=I(H). \tag*{(4)} ( I ( Γ ) s : f j a j ) ∩ ( I ( Γ ) s : f j b j ) = I ( H ) . (4)
by [18 Lemma 3.7] . Thus v u 1 ∈ I ( H ) v_{u_1} \in I(H) v u 1 ∈ I ( H ) , a contradiction. Therefore, v u 1 ≠ v u 2 v_{u_1} \ne v_{u_2} v u 1 = v u 2 . By the equality (4) , we can deduce that v u 1 v u 2 ∈ I ( H ) v_{u_1}v_{u_2} \in I(H) v u 1 v u 2 ∈ I ( H ) . Furthermore, v u 1 v u 2 ∣ g v_{u_1}v_{u_2} \mid g v u 1 v u 2 ∣ g , since v u 1 , v u 2 ∈ supp ( g ) v_{u_1},v_{u_2} \in\operatorname{supp}(g) v u 1 , v u 2 ∈ supp ( g ) . Therefore, g ∈ I ( H ) g \in I(H) g ∈ I ( H ) , which contradicts the fact that g ∉ I ( H ) g \notin I(H) g ∈ / I ( H ) . Therefore, either A j = ∅ A_j=\varnothing A j = ∅ or B j = ∅ B_j=\varnothing B j = ∅ .
As mentioned earlier, for any j ∈ [ s ] j \in[s] j ∈ [ s ] , we have ∣ A j ∣ ≤ 2 |A_j| \le2 ∣ A j ∣ ≤ 2 , ∣ B j ∣ ≤ 2 |B_j| \le2 ∣ B j ∣ ≤ 2 , and A j = ∅ A_j=\varnothing A j = ∅ or B j = ∅ B_j=\varnothing B j = ∅ . Therefore,
∣ N H ( supp ( g ) ) ∖ N Γ ( supp ( g ) ) ∣ = ∣ ⋃ j ∈ [ s ] ( A j ∪ B j ) ∣ ≤ ∑ j ∈ [ s ] ( ∣ A j ∣ + ∣ B j ∣ ) ≤ 2 s ≤ 2 ( t − 1 ) . \left|N_H(\operatorname{supp}(g))\setminus N_\Gamma(\operatorname{supp}(g))\right|=\left|\bigcup_{j\in[s]}(A_j\cup B_j)\right|\le\sum_{j\in[s]}(|A_j|+|B_j|)\le2s\le2(t-1). ∣ N H ( supp ( g )) ∖ N Γ ( supp ( g )) ∣ = j ∈ [ s ] ⋃ ( A j ∪ B j ) ≤ j ∈ [ s ] ∑ ( ∣ A j ∣ + ∣ B j ∣ ) ≤ 2 s ≤ 2 ( t − 1 ) .
We complete the proof. □
The v-number of the edge ideal of C 2 n ( a , n ) C_{2n}(a,n) C 2 n ( a , n )
In this section, we determine the exact formula for the v-number of the edge ideal of C 2 n ( a , n ) C_{2n}(a,n) C 2 n ( a , n ) , where 1 ≤ a < n 1\le a<n 1 ≤ a < n . By Lemmas 2.1 and 2.3, it suffices to consider the case a = 1 a=1 a = 1 and the case where a = 2 a=2 a = 2 and n ≥ 3 n\ge3 n ≥ 3 is odd. For the sake of convenience, we stipulate that for any positive integer j j j , if j ≡ i ( m o d 2 n ) j\equiv i\pmod{2n} j ≡ i ( mod 2 n ) with i ∈ [ 2 n ] i\in[2n] i ∈ [ 2 n ] , then x j = x i x_j=x_i x j = x i .
The case of C 2 n ( 1 , n ) C_{2n}(1,n) C 2 n ( 1 , n )
We consider the case a = 1 a=1 a = 1 . Suppose the vertex set and the edge set of C 2 n ( 1 , n ) C_{2n}(1,n) C 2 n ( 1 , n ) are V ( C 2 n ( 1 , n ) ) = { x i ∣ i ∈ [ 2 n ] } V(C_{2n}(1,n))=\{x_i\mid i\in[2n]\} V ( C 2 n ( 1 , n )) = { x i ∣ i ∈ [ 2 n ]} and
E ( C 2 n ( 1 , n ) ) = { x i x i + 1 ∣ i ∈ [ 2 n ] } ∪ { x j x j + n ∣ j ∈ [ n ] } , E(C_{2n}(1,n))=\{x_ix_{i+1}\mid i\in[2n]\}\cup\{x_jx_{j+n}\mid j\in[n]\}, E ( C 2 n ( 1 , n )) = { x i x i + 1 ∣ i ∈ [ 2 n ]} ∪ { x j x j + n ∣ j ∈ [ n ]} ,
respectively.
Lemma 3.1. Let Γ = C 2 n ( 1 , n ) \Gamma=C_{2n}(1,n) Γ = C 2 n ( 1 , n ) be a cubic circulant graph, where n ≥ 2 n\ge2 n ≥ 2 . Then
τ ( Γ ) = { n , if n is odd , n + 1 , if n is even . \tau(\Gamma)=
\begin{cases}
n, & \text{if } n \text{ is odd},\\
n+1, & \text{if } n \text{ is even}.
\end{cases} τ ( Γ ) = { n , n + 1 , if n is odd , if n is even .
Proof. Let C C C be a minimal vertex cover of Γ \Gamma Γ . Then C ∩ { x j , x j + n } ≠ ∅ C\cap\{x_j,x_{j+n}\}\ne\varnothing C ∩ { x j , x j + n } = ∅ for all j ∈ [ n ] j\in[n] j ∈ [ n ] . Since the edges x 1 x 1 + n , x 2 x 2 + n , … , x n x 2 n x_1x_{1+n},x_2x_{2+n},\ldots,x_nx_{2n} x 1 x 1 + n , x 2 x 2 + n , … , x n x 2 n are pairwise disjoint, we have ∣ C ∣ ≥ n |C|\ge n ∣ C ∣ ≥ n , so τ ( Γ ) ≥ n \tau(\Gamma)\ge n τ ( Γ ) ≥ n . We consider the following two cases:
If n n n is odd, then C 1 = { x 2 k − 1 ∣ k ∈ [ n ] } C_1=\{x_{2k-1}\mid k\in[n]\} C 1 = { x 2 k − 1 ∣ k ∈ [ n ]} is a vertex cover of Γ \Gamma Γ . Thus, τ ( Γ ) = n \tau(\Gamma)=n τ ( Γ ) = n .
Suppose that n n n is even. We first show that τ ( Γ ) ≥ n + 1 \tau(\Gamma)\ge n+1 τ ( Γ ) ≥ n + 1 . Suppose to the contrary that τ ( Γ ) = n \tau(\Gamma)=n τ ( Γ ) = n . Then there exists a minimal vertex cover C ′ C' C ′ of Γ \Gamma Γ with ∣ C ′ ∣ = n |C'|=n ∣ C ′ ∣ = n . Since C ′ ∩ { x i , x i + 1 } ≠ ∅ C'\cap\{x_i,x_{i+1}\}\ne\varnothing C ′ ∩ { x i , x i + 1 } = ∅ for all i ∈ [ 2 n ] i\in[2n] i ∈ [ 2 n ] , we can deduce that either C ′ = { x 2 k − 1 ∣ k ∈ [ n ] } C'=\{x_{2k-1}\mid k\in[n]\} C ′ = { x 2 k − 1 ∣ k ∈ [ n ]} or C ′ = { x 2 k ∣ k ∈ [ n ] } C'=\{x_{2k}\mid k\in[n]\} C ′ = { x 2 k ∣ k ∈ [ n ]} . If C ′ = { x 2 k − 1 ∣ k ∈ [ n ] } C'=\{x_{2k-1}\mid k\in[n]\} C ′ = { x 2 k − 1 ∣ k ∈ [ n ]} , then C ′ ∩ { x 2 , x 2 + n } = ∅ C'\cap\{x_2,x_{2+n}\}=\varnothing C ′ ∩ { x 2 , x 2 + n } = ∅ , which contradicts x 2 x 2 + n ∈ E ( Γ ) x_2x_{2+n}\in E(\Gamma) x 2 x 2 + n ∈ E ( Γ ) . Therefore, in this case, we have τ ( Γ ) ≥ n + 1 \tau(\Gamma)\ge n+1 τ ( Γ ) ≥ n + 1 . By symmetry, if C ′ = { x 2 k ∣ k ∈ [ n ] } C'=\{x_{2k}\mid k\in[n]\} C ′ = { x 2 k ∣ k ∈ [ n ]} , then we can also obtain τ ( Γ ) ≥ n + 1 \tau(\Gamma)\ge n+1 τ ( Γ ) ≥ n + 1 .
Finally, let C 2 = { x 1 } ∪ { x 2 i ∣ i ∈ [ n 2 ] } ∪ { x n + 2 i − 1 ∣ i ∈ [ n 2 ] } C_2=\{x_1\}\cup\{x_{2i}\mid i\in[\frac{n}{2}]\}\cup\{x_{n+2i-1}\mid i\in[\frac{n}{2}]\} C 2 = { x 1 } ∪ { x 2 i ∣ i ∈ [ 2 n ]} ∪ { x n + 2 i − 1 ∣ i ∈ [ 2 n ]} , and it is easy to check that ∣ C 2 ∣ = n + 1 |C_2|=n+1 ∣ C 2 ∣ = n + 1 and C 2 C_2 C 2 is a vertex cover of Γ \Gamma Γ . Therefore, τ ( Γ ) = n + 1 \tau(\Gamma)=n+1 τ ( Γ ) = n + 1 . □
Theorem 3.2. Let n ≥ 2 n\ge2 n ≥ 2 be an integer. Let Γ = C 2 n ( 1 , n ) \Gamma=C_{2n}(1,n) Γ = C 2 n ( 1 , n ) be a cubic circulant graph, and let I ( Γ ) I(\Gamma) I ( Γ ) be its edge ideal. Then
v ( I ( Γ ) ) = { ⌈ n 3 ⌉ , if n is odd , ⌈ n + 1 3 ⌉ , if n is even . \mathrm{v}(I(\Gamma))=
\begin{cases}
\left\lceil\frac{n}{3}\right\rceil, & \text{if }n\text{ is odd},\\
\left\lceil\frac{n+1}{3}\right\rceil, & \text{if }n\text{ is even}.
\end{cases} v ( I ( Γ )) = { ⌈ 3 n ⌉ , ⌈ 3 n + 1 ⌉ , if n is odd , if n is even . Proof. By Lemma 2.5, there exists an independent set S S S of Γ \Gamma Γ such that N Γ ( S ) N_\Gamma(S) N Γ ( S ) is a minimal vertex cover of Γ \Gamma Γ and ∣ S ∣ = v ( I ( Γ ) ) |S|=\mathrm{v}(I(\Gamma)) ∣ S ∣ = v ( I ( Γ )) . Thus τ ( Γ ) ≤ ∣ N Γ ( S ) ∣ \tau(\Gamma)\le|N_\Gamma(S)| τ ( Γ ) ≤ ∣ N Γ ( S ) ∣ . Since Γ \Gamma Γ is 3-regular, we have ∣ N Γ ( S ) ∣ ≤ 3 ∣ S ∣ |N_\Gamma(S)|\le3|S| ∣ N Γ ( S ) ∣ ≤ 3∣ S ∣ . Therefore, τ ( Γ ) ≤ 3 ∣ S ∣ \tau(\Gamma)\le3|S| τ ( Γ ) ≤ 3∣ S ∣ . It follows from Lemma 3.1 that v ( I ( Γ ) ) = ∣ S ∣ ≥ ⌈ τ ( Γ ) 3 ⌉ = α n \mathrm{v}(I(\Gamma))=|S|\ge\left\lceil\frac{\tau(\Gamma)}{3}\right\rceil=\alpha_n v ( I ( Γ )) = ∣ S ∣ ≥ ⌈ 3 τ ( Γ ) ⌉ = α n , where
α n = { ⌈ n 3 ⌉ , if n is odd , ⌈ n + 1 3 ⌉ , if n is even . \alpha_n=
\begin{cases}
\left\lceil\frac{n}{3}\right\rceil, & \text{if }n\text{ is odd},\\
\left\lceil\frac{n+1}{3}\right\rceil, & \text{if }n\text{ is even}.
\end{cases} α n = { ⌈ 3 n ⌉ , ⌈ 3 n + 1 ⌉ , if n is odd , if n is even . In the following, we will prove that v ( I ( Γ ) ) ≤ α n \mathrm{v}(I(\Gamma))\le\alpha_n v ( I ( Γ )) ≤ α n . Choose f 2 = f 3 = x 1 f_2=f_3=x_1 f 2 = f 3 = x 1 , f 4 = x 4 x 6 f_4=x_4x_6 f 4 = x 4 x 6 , f 5 = x 1 x 3 f_5=x_1x_3 f 5 = x 1 x 3 , and for any n ≥ 6 n\ge6 n ≥ 6 ,
f n = { x n x n + 2 ( ∏ i ∈ [ k − 1 ] x 6 i + 1 ) ( ∏ j ∈ [ k ] x n + 6 j − 2 ) , if n = 6 k , x 1 x 3 ( ∏ i ∈ [ k ] x 6 i − 1 ) ( ∏ j ∈ [ k − 1 ] x n + 6 j + 2 ) , if n = 6 k + 1 , ( ∏ i ∈ [ k ] x 6 i − 3 ) ( ∏ j ∈ [ k + 1 ] x n + 6 j − 6 ) , if n = 6 k + 2 , ∏ i ∈ [ 2 k + 1 ] x 6 i − 5 , if n = 6 k + 3 , x n ( ∏ i ∈ [ k ] x 6 i − 1 ) ( ∏ j ∈ [ k + 1 ] x n + 6 j − 4 ) , if n = 6 k + 4 , x 1 ( ∏ i ∈ [ k + 1 ] x 6 i − 3 ) ( ∏ j ∈ [ k ] x n + 6 j ) , if n = 6 k + 5 , f_n=
\begin{cases}
x_nx_{n+2}\left(\prod_{i\in[k-1]}x_{6i+1}\right)\left(\prod_{j\in[k]}x_{n+6j-2}\right), & \text{if }n=6k,\\
x_1x_3\left(\prod_{i\in[k]}x_{6i-1}\right)\left(\prod_{j\in[k-1]}x_{n+6j+2}\right), & \text{if }n=6k+1,\\
\left(\prod_{i\in[k]}x_{6i-3}\right)\left(\prod_{j\in[k+1]}x_{n+6j-6}\right), & \text{if }n=6k+2,\\
\prod_{i\in[2k+1]}x_{6i-5}, & \text{if }n=6k+3,\\
x_n\left(\prod_{i\in[k]}x_{6i-1}\right)\left(\prod_{j\in[k+1]}x_{n+6j-4}\right), & \text{if }n=6k+4,\\
x_1\left(\prod_{i\in[k+1]}x_{6i-3}\right)\left(\prod_{j\in[k]}x_{n+6j}\right), & \text{if }n=6k+5,
\end{cases} f n = ⎩ ⎨ ⎧ x n x n + 2 ( ∏ i ∈ [ k − 1 ] x 6 i + 1 ) ( ∏ j ∈ [ k ] x n + 6 j − 2 ) , x 1 x 3 ( ∏ i ∈ [ k ] x 6 i − 1 ) ( ∏ j ∈ [ k − 1 ] x n + 6 j + 2 ) , ( ∏ i ∈ [ k ] x 6 i − 3 ) ( ∏ j ∈ [ k + 1 ] x n + 6 j − 6 ) , ∏ i ∈ [ 2 k + 1 ] x 6 i − 5 , x n ( ∏ i ∈ [ k ] x 6 i − 1 ) ( ∏ j ∈ [ k + 1 ] x n + 6 j − 4 ) , x 1 ( ∏ i ∈ [ k + 1 ] x 6 i − 3 ) ( ∏ j ∈ [ k ] x n + 6 j ) , if n = 6 k , if n = 6 k + 1 , if n = 6 k + 2 , if n = 6 k + 3 , if n = 6 k + 4 , if n = 6 k + 5 ,
where k ≥ 1 k \ge1 k ≥ 1 . Then a simple calculation shows that
( I ( Γ ) : f n ) = { ( x 2 , x 3 , x 4 ) , if n = 2 , ( x 2 i ∣ i ∈ [ n − 2 2 ] ) + ( x n + 2 j − 1 ∣ 0 ≤ j ≤ n 2 ) + ( x 2 n ) , if n ≥ 3 is even , ( x 2 i ∣ i ∈ [ n ] ) , if n ≥ 3 is odd . (I(\Gamma):f_n)=
\begin{cases}
(x_2,x_3,x_4), & \text{if } n=2,\\
(x_{2i}\mid i\in[\frac{n-2}{2}])+(x_{n+2j-1}\mid0\le j\le\frac{n}{2})+(x_{2n}), & \text{if } n\ge3\text{ is even},\\
(x_{2i}\mid i\in[n]), & \text{if } n\ge3\text{ is odd}.
\end{cases} ( I ( Γ ) : f n ) = ⎩ ⎨ ⎧ ( x 2 , x 3 , x 4 ) , ( x 2 i ∣ i ∈ [ 2 n − 2 ]) + ( x n + 2 j − 1 ∣ 0 ≤ j ≤ 2 n ) + ( x 2 n ) , ( x 2 i ∣ i ∈ [ n ]) , if n = 2 , if n ≥ 3 is even , if n ≥ 3 is odd .
By the definition of the v-number, we have v ( I ( Γ ) ) ≤ α n v(I(\Gamma))\le\alpha_n v ( I ( Γ )) ≤ α n .
The case of C 2 n ( 2 , n ) C_{2n}(2,n) C 2 n ( 2 , n )
We will consider the case a = 2 a=2 a = 2 . Suppose the vertex set and the edge set of C 2 n ( 2 , n ) C_{2n}(2,n) C 2 n ( 2 , n ) are V ( C 2 n ( 2 , n ) ) = { x i ∣ i ∈ [ 2 n ] } V(C_{2n}(2,n))=\{x_i\mid i\in[2n]\} V ( C 2 n ( 2 , n )) = { x i ∣ i ∈ [ 2 n ]} and
E ( C 2 n ( 2 , n ) ) = { x i x i + 2 ∣ i ∈ [ 2 n ] } ∪ { x j x j + n ∣ j ∈ [ n ] } , E(C_{2n}(2,n))=\{x_ix_{i+2}\mid i\in[2n]\}\cup\{x_jx_{j+n}\mid j\in[n]\}, E ( C 2 n ( 2 , n )) = { x i x i + 2 ∣ i ∈ [ 2 n ]} ∪ { x j x j + n ∣ j ∈ [ n ]} ,
respectively.
By Lemma 2.1, it is enough to consider the case where n ≥ 3 n\ge3 n ≥ 3 is odd. In this case, we can relabel the vertices of C 2 n ( 2 , n ) C_{2n}(2,n) C 2 n ( 2 , n ) as follows: for each i ∈ [ n ] i\in[n] i ∈ [ n ] , we replace the vertices x 2 i − 1 x_{2i-1} x 2 i − 1 and x n + 2 i − 1 x_{n+2i-1} x n + 2 i − 1 with x i x_i x i and y i y_i y i , respectively. Thus, V ( C 2 n ( 2 , n ) ) = { x i , y i ∣ i ∈ [ n ] } V(C_{2n}(2,n))=\{x_i,y_i\mid i\in[n]\} V ( C 2 n ( 2 , n )) = { x i , y i ∣ i ∈ [ n ]} and
E ( C 2 n ( 2 , n ) ) = { x i x i + 1 , y i y i + 1 , x i y i ∣ i ∈ [ n ] } . (5) E(C_{2n}(2,n))=\{x_ix_{i+1},y_iy_{i+1},x_iy_i\mid i\in[n]\}. \tag*{(5)} E ( C 2 n ( 2 , n )) = { x i x i + 1 , y i y i + 1 , x i y i ∣ i ∈ [ n ]} . (5)
That is, the graph C 2 n ( 2 , n ) C_{2n}(2,n) C 2 n ( 2 , n ) consists of two n n n -cycles formed by the vertex sets { x i ∣ i ∈ [ n ] } \{x_i\mid i\in[n]\} { x i ∣ i ∈ [ n ]} and { y i ∣ i ∈ [ n ] } \{y_i\mid i\in[n]\} { y i ∣ i ∈ [ n ]} with edges x i x i + 1 x_ix_{i+1} x i x i + 1 and y i y i + 1 y_iy_{i+1} y i y i + 1 , respectively. These two cycles are linked by cross-edges of the form x i y i x_iy_i x i y i for each i ∈ [ n ] i\in[n] i ∈ [ n ] . To make it easier to understand, we have provided two examples for n = 3 n=3 n = 3 and n = 5 n=5 n = 5 , see Figure 1 .
Figure 1.
Lemma 3.3. Let n ≥ 3 n\ge3 n ≥ 3 be an odd integer, and let Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n ) be a cubic circulant graph. Then τ ( Γ ) = n + 1 \tau(\Gamma)=n+1 τ ( Γ ) = n + 1 .
Proof. Let the vertex set of Γ \Gamma Γ be V ( Γ ) = { x i , y i ∣ i ∈ [ n ] } V(\Gamma)=\{x_i,y_i\mid i\in[n]\} V ( Γ ) = { x i , y i ∣ i ∈ [ n ]} , and let the edge set be defined as in (5). Since Γ \Gamma Γ contains two vertex-disjoint induced odd cycles C n 1 C_n^1 C n 1 and C n 2 C_n^2 C n 2 with vertex sets { x i ∣ i ∈ [ n ] } \{x_i\mid i\in[n]\} { x i ∣ i ∈ [ n ]} and { y i ∣ i ∈ [ n ] } \{y_i\mid i\in[n]\} { y i ∣ i ∈ [ n ]} , respectively, and each has a vertex cover number of n + 1 2 \frac{n+1}{2} 2 n + 1 , we obtain τ ( Γ ) ≥ τ ( C n 1 ) + τ ( C n 2 ) = n + 1 \tau(\Gamma) \ge\tau(C_n^1) + \tau(C_n^2) = n+1 τ ( Γ ) ≥ τ ( C n 1 ) + τ ( C n 2 ) = n + 1 . Conversely, the set
C = { y 2 i − 1 ∣ i ∈ [ n + 1 2 ] } ∪ { x 2 j ∣ j ∈ [ n − 1 2 ] } ∪ { x n } C=\left\{y_{2i-1}\mid i\in\left[\frac{n+1}{2}\right]\right\}\cup\left\{x_{2j}\mid j\in\left[\frac{n-1}{2}\right]\right\}\cup\{x_n\} C = { y 2 i − 1 ∣ i ∈ [ 2 n + 1 ] } ∪ { x 2 j ∣ j ∈ [ 2 n − 1 ] } ∪ { x n }
forms a vertex cover of Γ \Gamma Γ of size n + 1 n+1 n + 1 . Thus, τ ( Γ ) = n + 1 \tau(\Gamma)=n+1 τ ( Γ ) = n + 1 .
Theorem 3.4. Let n ≥ 3 n\geq3 n ≥ 3 be an odd integer, and let Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n ) be a cubic circulant graph with I ( Γ ) I(\Gamma) I ( Γ ) being its edge ideal. Then
ν ( I ( Γ ) ) = ⌈ n + 1 3 ⌉ . \nu(I(\Gamma))=\left\lceil\frac{n+1}{3}\right\rceil. ν ( I ( Γ )) = ⌈ 3 n + 1 ⌉ . Proof. Let the vertex set of Γ \Gamma Γ be V ( Γ ) = { x i , y i ∣ i ∈ [ n ] } V(\Gamma)=\{x_i,y_i\mid i\in[n]\} V ( Γ ) = { x i , y i ∣ i ∈ [ n ]} . By Lemma 2.5, there exists an independent set S S S of Γ \Gamma Γ such that N Γ ( S ) N_\Gamma(S) N Γ ( S ) is a minimal vertex cover of Γ \Gamma Γ and ∣ S ∣ = ν ( I ( Γ ) ) |S|=\nu(I(\Gamma)) ∣ S ∣ = ν ( I ( Γ )) . Thus, τ ( Γ ) ≤ ∣ N Γ ( S ) ∣ \tau(\Gamma)\leq|N_\Gamma(S)| τ ( Γ ) ≤ ∣ N Γ ( S ) ∣ . Since Γ \Gamma Γ is 3-regular, we have ∣ N Γ ( S ) ∣ ≤ 3 ∣ S ∣ |N_\Gamma(S)|\leq3|S| ∣ N Γ ( S ) ∣ ≤ 3∣ S ∣ . Therefore, τ ( Γ ) ≤ 3 ∣ S ∣ \tau(\Gamma)\leq3|S| τ ( Γ ) ≤ 3∣ S ∣ . It follows from Lemma 3.3 that ν ( I ( Γ ) ) = ∣ S ∣ ≥ ⌈ τ ( Γ ) 3 ⌉ = ⌈ n + 1 3 ⌉ \nu(I(\Gamma))=|S|\geq\left\lceil\frac{\tau(\Gamma)}{3}\right\rceil=\left\lceil\frac{n+1}{3}\right\rceil ν ( I ( Γ )) = ∣ S ∣ ≥ ⌈ 3 τ ( Γ ) ⌉ = ⌈ 3 n + 1 ⌉ .
Next, we prove that ν ( I ( Γ ) ) ≤ ⌈ n + 1 3 ⌉ \nu(I(\Gamma))\leq\left\lceil\frac{n+1}{3}\right\rceil ν ( I ( Γ )) ≤ ⌈ 3 n + 1 ⌉ . Choose f 3 = x 1 y 2 f_3=x_1y_2 f 3 = x 1 y 2 , f 5 = x 1 y 4 f_5=x_1y_4 f 5 = x 1 y 4 , and for any odd integer n ≥ 7 n\geq7 n ≥ 7 ,
f n = { ( ∏ i ∈ [ k ] x 6 i − 5 ) ( ∏ i ∈ [ k ] y 6 i − 2 ) y 6 k , if n = 6 k + 1 , ( ∏ i ∈ [ k + 1 ] x 6 i − 5 ) ( ∏ i ∈ [ k ] y 6 i − 2 ) y 6 k + 2 , if n = 6 k + 3 , ( ∏ i ∈ [ k + 1 ] x 6 i − 5 ) ( ∏ i ∈ [ k + 1 ] y 6 i − 2 ) , if n = 6 k + 5. f_n=
\begin{cases}
\left(\prod_{i\in[k]}x_{6i-5}\right)\left(\prod_{i\in[k]}y_{6i-2}\right)y_{6k}, & \text{if }n=6k+1,\\
\left(\prod_{i\in[k+1]}x_{6i-5}\right)\left(\prod_{i\in[k]}y_{6i-2}\right)y_{6k+2}, & \text{if }n=6k+3,\\
\left(\prod_{i\in[k+1]}x_{6i-5}\right)\left(\prod_{i\in[k+1]}y_{6i-2}\right), & \text{if }n=6k+5.
\end{cases} f n = ⎩ ⎨ ⎧ ( ∏ i ∈ [ k ] x 6 i − 5 ) ( ∏ i ∈ [ k ] y 6 i − 2 ) y 6 k , ( ∏ i ∈ [ k + 1 ] x 6 i − 5 ) ( ∏ i ∈ [ k ] y 6 i − 2 ) y 6 k + 2 , ( ∏ i ∈ [ k + 1 ] x 6 i − 5 ) ( ∏ i ∈ [ k + 1 ] y 6 i − 2 ) , if n = 6 k + 1 , if n = 6 k + 3 , if n = 6 k + 5. where k ≥ 1 k\geq1 k ≥ 1 . Then, by straightforward calculation, we obtain that
( I ( Γ ) : f n ) = ( y 2 i − 1 ∣ i ∈ [ n + 1 2 ] ) + ( x 2 j ∣ j ∈ [ n − 1 2 ] ) + ( x n ) . (I(\Gamma):f_n)=\left(y_{2i-1}\mid i\in\left[\frac{n+1}{2}\right]\right)+\left(x_{2j}\mid j\in\left[\frac{n-1}{2}\right]\right)+(x_n). ( I ( Γ ) : f n ) = ( y 2 i − 1 ∣ i ∈ [ 2 n + 1 ] ) + ( x 2 j ∣ j ∈ [ 2 n − 1 ] ) + ( x n ) . By the definition of the ν \nu ν -number, we have ν ( I ( Γ ) ) ≤ ⌈ n + 1 3 ⌉ \nu(I(\Gamma))\leq\left\lceil\frac{n+1}{3}\right\rceil ν ( I ( Γ )) ≤ ⌈ 3 n + 1 ⌉ .
The case of C 2 n ( a , n ) C_{2n}(a,n) C 2 n ( a , n ) with 1 ≤ a < n 1\leq a<n 1 ≤ a < n
The cases a = 1 a=1 a = 1 and a = 2 a=2 a = 2 for odd n ≥ 3 n\geq3 n ≥ 3 have already been established. Therefore, by Lemmas 2.1 and 2.3, and direct calculation, we can obtain the following theorem, which corresponds to the case where t = 1 t=1 t = 1 in Theorem 1.1:
Theorem 3.5. Let Γ = C 2 n ( a , n ) \Gamma=C_{2n}(a,n) Γ = C 2 n ( a , n ) be a cubic circulant graph, where 1 ≤ a < n 1\leq a<n 1 ≤ a < n , and let k = gcd ( a , 2 n ) k=\gcd(a,2n) k = g cd( a , 2 n ) . Then
ν ( I ( Γ ) ) = { k ⋅ ⌈ n 3 k ⌉ , if n k ≡ 1 ( m o d 2 ) , k ⋅ ⌈ n + k 3 k ⌉ , if n k ≡ 0 ( m o d 2 ) , k 2 ⋅ ⌈ 2 n + k 3 k ⌉ , otherwise. \nu(I(\Gamma))=
\begin{cases}
k\cdot\left\lceil\frac{n}{3k}\right\rceil, & \text{if }\frac{n}{k}\equiv1\pmod{2},\\
k\cdot\left\lceil\frac{n+k}{3k}\right\rceil, & \text{if }\frac{n}{k}\equiv0\pmod{2},\\
\frac{k}{2}\cdot\left\lceil\frac{2n+k}{3k}\right\rceil, & \text{otherwise.}
\end{cases} ν ( I ( Γ )) = ⎩ ⎨ ⎧ k ⋅ ⌈ 3 k n ⌉ , k ⋅ ⌈ 3 k n + k ⌉ , 2 k ⋅ ⌈ 3 k 2 n + k ⌉ , if k n ≡ 1 ( mod 2 ) , if k n ≡ 0 ( mod 2 ) , otherwise.
The v-number of powers of the edge ideals of C 2 n ( a , n ) C_{2n}(a,n) C 2 n ( a , n )
In this section, we will provide the exact formula for the v-numbers of powers of the edge ideal of Γ = C 2 n ( a , n ) \Gamma= C_{2n}(a,n) Γ = C 2 n ( a , n ) for all t ≥ 2 t \ge2 t ≥ 2 . By Lemmas 2.1 and 2.3, we only need to consider the case where a = 1 a = 1 a = 1 and the case where a = 2 a = 2 a = 2 and n ≥ 3 n \ge3 n ≥ 3 is odd. So, for simplicity, we shall focus solely on these two cases.
The case of Γ = C 2 n ( 1 , n ) \Gamma= C_{2n}(1,n) Γ = C 2 n ( 1 , n )
In this subsection, we consider the case where a = 1 a = 1 a = 1 . When n = 2 n = 2 n = 2 , Γ \Gamma Γ is a complete graph with 4 vertices and so its edge ideal I ( Γ ) I(\Gamma) I ( Γ ) has a linear resolution by [10 Theorem 5.5] . It follows from [12 Theorem 5.1] that v ( I ( Γ ) t ) = 2 t − 1 \mathop{\mathrm{v}}(I(\Gamma)^t) = 2t - 1 v ( I ( Γ ) t ) = 2 t − 1 for all t ≥ 1 t \ge1 t ≥ 1 . In the following, we assume that n ≥ 3 n \ge3 n ≥ 3 .
Theorem 4.1. Let Γ = C 2 n ( 1 , n ) \Gamma= C_{2n}(1,n) Γ = C 2 n ( 1 , n ) be a cubic circulant graph, and let I ( Γ ) I(\Gamma) I ( Γ ) be its edge ideal. Then
v ( I ( Γ ) t ) = 2 t − 1 for all t ≥ ⌈ n 2 ⌉ . \mathop{\mathrm{v}}(I(\Gamma)^t) = 2t - 1 \qquad\text{for all } t \ge\left\lceil\frac{n}{2} \right\rceil. v ( I ( Γ ) t ) = 2 t − 1 for all t ≥ ⌈ 2 n ⌉ . Proof. Let I = I ( Γ ) I = I(\Gamma) I = I ( Γ ) and μ = ⌊ n 2 ⌋ \mu= \left\lfloor\frac{n}{2} \right\rfloor μ = ⌊ 2 n ⌋ . By Lemma 2.4, we have v ( I t ) ≥ 2 t − 1 \mathop{\mathrm{v}}(I^t) \ge2t - 1 v ( I t ) ≥ 2 t − 1 . It remains to prove v ( I t ) ≤ 2 t − 1 \mathop{\mathrm{v}}(I^t) \le2t - 1 v ( I t ) ≤ 2 t − 1 . To this end, we choose f = ( x n x n + 1 ) t − μ ( ∏ i = 0 2 μ − 2 x n + i ) f = (x_nx_{n+1})^{t-\mu}\left(\prod_{i=0}^{2\mu-2}x_{n+i}\right) f = ( x n x n + 1 ) t − μ ( ∏ i = 0 2 μ − 2 x n + i ) and
p = { ( x 2 i ∣ i ∈ [ n − 2 2 ] ) + ( x n + 2 j − 1 ∣ 0 ≤ j ≤ n 2 ) + ( x 2 n ) , if n is even , ( x 2 i ∣ i ∈ [ n ] ) , if n is odd . \mathfrak{p} =
\begin{cases}
\left(x_{2i}\mid i\in\left[\frac{n-2}{2}\right]\right)+\left(x_{n+2j-1}\mid0\le j\le\frac{n}{2}\right)+(x_{2n}), & \text{if } n\text{ is even},\\
\left(x_{2i}\mid i\in[n]\right), & \text{if } n\text{ is odd}.
\end{cases} p = { ( x 2 i ∣ i ∈ [ 2 n − 2 ] ) + ( x n + 2 j − 1 ∣ 0 ≤ j ≤ 2 n ) + ( x 2 n ) , ( x 2 i ∣ i ∈ [ n ] ) , if n is even , if n is odd . Then deg ( f ) = 2 t − 1 \deg(f)=2t-1 deg ( f ) = 2 t − 1 and p ∈ Ass ( I ) \mathfrak{p}\in\operatorname{Ass}(I) p ∈ Ass ( I ) by Theorem 3.2. Let A = { x n + 2 ℓ ∣ 0 ≤ ℓ ≤ μ − 1 } A=\{x_{n+2\ell}\mid0\le\ell\le\mu-1\} A = { x n + 2 ℓ ∣ 0 ≤ ℓ ≤ μ − 1 } , and then it is easy to check that p = ( x i ∣ x i ∈ N Γ ( A ) ) \mathfrak{p}=(x_i\mid x_i\in N_{\Gamma}(A)) p = ( x i ∣ x i ∈ N Γ ( A )) . We will prove that ( I t : f ) = p (I^t:f)=\mathfrak{p} ( I t : f ) = p , thus p ∈ Ass ( I t ) \mathfrak{p}\in\operatorname{Ass}(I^t) p ∈ Ass ( I t ) and the result holds.
We shall use Lemma 2.6 to prove ( I t : f ) = p (I^t:f)=\mathfrak{p} ( I t : f ) = p , so we must verify the conditions of Lemma 2.6. Since supp ( p ) ∩ supp ( f ) = { x n + 2 i − 1 ∣ i ∈ [ μ − 1 ] } \operatorname{supp}(\mathfrak{p})\cap\operatorname{supp}(f)=\{x_{n+2i-1}\mid i\in[\mu-1]\} supp ( p ) ∩ supp ( f ) = { x n + 2 i − 1 ∣ i ∈ [ μ − 1 ]} , we have
deg p ( f ) = ∑ i ∈ [ μ − 1 ] deg x n + 2 i − 1 ( f ) = t − 1. \deg_{\mathfrak{p}}(f)=\sum_{i\in[\mu-1]}\deg_{x_{n+2i-1}}(f)=t-1. deg p ( f ) = i ∈ [ μ − 1 ] ∑ deg x n + 2 i − 1 ( f ) = t − 1. For any x i ∈ p x_i\in\mathfrak{p} x i ∈ p , by the expression of p \mathfrak{p} p , we have x i ∈ N Γ ( x n + 2 ℓ ) x_i\in N_{\Gamma}(x_{n+2\ell}) x i ∈ N Γ ( x n + 2 ℓ ) , where 0 ≤ ℓ ≤ μ − 1 0\le\ell\le\mu-1 0 ≤ ℓ ≤ μ − 1 . Thus
x i f = ( x i x n + 2 ℓ ) ( ∏ r = 0 ℓ − 1 x n + 2 r x n + 2 r + 1 ) ( ∏ j = ℓ + 1 μ − 1 x n + 2 j − 1 x n + 2 j ) ( x n x n + 1 ) t − μ ∈ I t . x_if=(x_ix_{n+2\ell})\left(\prod_{r=0}^{\ell-1}x_{n+2r}x_{n+2r+1}\right)\left(\prod_{j=\ell+1}^{\mu-1}x_{n+2j-1}x_{n+2j}\right)(x_nx_{n+1})^{t-\mu}\in I^t. x i f = ( x i x n + 2 ℓ ) ( r = 0 ∏ ℓ − 1 x n + 2 r x n + 2 r + 1 ) j = ℓ + 1 ∏ μ − 1 x n + 2 j − 1 x n + 2 j ( x n x n + 1 ) t − μ ∈ I t . Hence p ⊆ ( I t : f ) \mathfrak{p}\subseteq(I^t:f) p ⊆ ( I t : f ) . Therefore, by Lemma 2.6, ( I t : f ) = p (I^t:f)=\mathfrak{p} ( I t : f ) = p .
Remark 4.2. From the proof of the above theorem, we observe that if we choose U = ∅ U=\varnothing U = ∅ and V = { x n + 2 ℓ ∣ 0 ≤ ℓ ≤ ⌊ n / 2 ⌋ − 1 } V=\{x_{n+2\ell}\mid0\le\ell\le\lfloor n/2\rfloor-1\} V = { x n + 2 ℓ ∣ 0 ≤ ℓ ≤ ⌊ n /2 ⌋ − 1 } , then we can check that ( U , V ) (U,V) ( U , V ) is an admissible pair as defined in [19 Definition 1.1] . Thus, it follows from [19 Theorem 1.2] that vstab ( I ( Γ ) ) ≤ ⌊ n / 2 ⌋ \operatorname{vstab}(I(\Gamma))\le\lfloor n/2\rfloor vstab ( I ( Γ )) ≤ ⌊ n /2 ⌋ . Therefore, by [19 Lemma 2.2] , for all t ≥ ⌊ n / 2 ⌋ t\ge\lfloor n/2\rfloor t ≥ ⌊ n /2 ⌋ , we have v ( I ( Γ ) t ) = 2 t − 1 \mathop{\mathrm{v}}(I(\Gamma)^t)=2t-1 v ( I ( Γ ) t ) = 2 t − 1 .
However, in [19 ] , Hien and Vu did not provide any techniques for finding admissible pairs. For general graphs, finding admissible pairs and computing λ ( U , V ) \lambda(U,V) λ ( U , V ) in [19 Theorem 1.2] are both quite challenging; therefore, while their conclusions have significant theoretical value, they are challenging to implement in practice. It was only after completing the proof of the above theorem and studying the associated prime ideal p \mathfrak{p} p that we discovered the method to choose U U U and V V V .
In the following, we will consider cases where n ≥ 6 n \ge6 n ≥ 6 and 2 ≤ t < ⌊ n 2 ⌋ 2 \le t < \left\lfloor\frac{n}{2} \right\rfloor 2 ≤ t < ⌊ 2 n ⌋ .
Lemma 4.4. Let Γ = C 2 n ( 1 , n ) \Gamma= C_{2n}(1,n) Γ = C 2 n ( 1 , n ) be a cubic circulant graph, and e 1 , … , e t − 1 e_1,\ldots,e_{t-1} e 1 , … , e t − 1 be its t − 1 t-1 t − 1 edges. Suppose I ( Γ ) I(\Gamma) I ( Γ ) is the edge ideal of Γ \Gamma Γ . Then ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) \left(I(\Gamma)^t : \prod_{i=1}^{t-1} e_i\right) ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) can be regarded as the edge ideal of a simple graph.
Proof. By Theorem 2.8 , the colon ideal ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) \left(I(\Gamma)^t : \prod_{i=1}^{t-1} e_i\right) ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) is generated by monomials of degree two. We only need to show that x i 2 ∉ G ( ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) ) x_i^2 \notin\mathcal{G}\left(\left(I(\Gamma)^t : \prod_{i=1}^{t-1} e_i\right)\right) x i 2 ∈ / G ( ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) ) for any i ∈ [ 2 n ] i \in[2n] i ∈ [ 2 n ] .
Suppose for a contradiction that x i 2 ∈ ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) x_i^2 \in\left(I(\Gamma)^t : \prod_{i=1}^{t-1} e_i\right) x i 2 ∈ ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) for some i ∈ [ 2 n ] i \in[2n] i ∈ [ 2 n ] . Since Γ \Gamma Γ contains no loops, we have x i 2 ∉ E ( Γ ) x_i^2 \notin E(\Gamma) x i 2 ∈ / E ( Γ ) . By Theorem 2.8 , x i x_i x i is even-connected to itself with respect to ∏ i = 1 t − 1 e i \prod_{i=1}^{t-1} e_i ∏ i = 1 t − 1 e i . Thus, there exists a walk W : x i = x ℓ 1 , x ℓ 2 , … , x ℓ 2 k = x i W : x_i = x_{\ell_1}, x_{\ell_2}, \ldots, x_{\ell_{2k}} = x_i W : x i = x ℓ 1 , x ℓ 2 , … , x ℓ 2 k = x i of length 2 k − 1 2k-1 2 k − 1 satisfying Definition 2.7 . Hence, W W W contains an odd cycle. Furthermore, by (2) and (3) of Definition 2.7 , we have k ≤ t k \le t k ≤ t .
We consider the following two cases depending on the parity of n n n :
(I) If n n n is odd, then Γ \Gamma Γ is bipartite with bipartition V ( Γ ) = A ⊔ B V(\Gamma) = A \sqcup B V ( Γ ) = A ⊔ B , where A = { x 2 i ∣ i ∈ [ n ] } A = \{x_{2i} \mid i \in[n]\} A = { x 2 i ∣ i ∈ [ n ]} and B = { x 2 i − 1 ∣ i ∈ [ n ] } B = \{x_{2i-1} \mid i \in[n]\} B = { x 2 i − 1 ∣ i ∈ [ n ]} . Thus Γ \Gamma Γ contains no odd cycles, which contradicts the fact that W W W contains an odd cycle.
(II) Suppose n n n is even. We will prove that k ≥ n + 2 2 k \ge\frac{n+2}{2} k ≥ 2 n + 2 . This will imply t ≥ k ≥ n + 2 2 t \ge k \ge\frac{n+2}{2} t ≥ k ≥ 2 n + 2 , which contradicts the fact that t < ⌊ n 2 ⌋ t < \left\lfloor\frac{n}{2} \right\rfloor t < ⌊ 2 n ⌋ .
To this end, we consider an odd cycle C C C of length m m m in W W W with vertex set V ( C ) = { x v i ∣ i ∈ [ m ] } V(C) = \{x_{v_i} \mid i \in[m]\} V ( C ) = { x v i ∣ i ∈ [ m ]} and edge set E ( C ) = { x v i x v i + 1 ∣ i ∈ [ m ] } E(C) = \{x_{v_i}x_{v_{i+1}} \mid i \in[m]\} E ( C ) = { x v i x v i + 1 ∣ i ∈ [ m ]} , where x v m + 1 = x v 1 x_{v_{m+1}} = x_{v_1} x v m + 1 = x v 1 . For each i ∈ [ m ] i \in[m] i ∈ [ m ] , let v i + 1 − v i ≡ d i ( m o d 2 n ) v_{i+1} - v_i \equiv d_i \pmod{2n} v i + 1 − v i ≡ d i ( mod 2 n ) , where d i ∈ { 1 , − 1 , n } d_i \in\{1,-1,n\} d i ∈ { 1 , − 1 , n } . When we traverse once along C C C , we have ∑ i ∈ [ m ] ( v i + 1 − v i ) ≡ 0 ( m o d 2 n ) \sum_{i \in[m]}(v_{i+1}-v_i) \equiv0 \pmod{2n} ∑ i ∈ [ m ] ( v i + 1 − v i ) ≡ 0 ( mod 2 n ) , that is, ∑ i ∈ [ m ] d i ≡ 0 ( m o d 2 n ) \sum_{i \in[m]} d_i \equiv0 \pmod{2n} ∑ i ∈ [ m ] d i ≡ 0 ( mod 2 n ) .
Let a = ∣ { i ∈ [ m ] ∣ d i = 1 } ∣ a = |\{i \in[m] \mid d_i = 1\}| a = ∣ { i ∈ [ m ] ∣ d i = 1 } ∣ , b = ∣ { i ∈ [ m ] ∣ d i = − 1 } ∣ b = |\{i \in[m] \mid d_i = -1\}| b = ∣ { i ∈ [ m ] ∣ d i = − 1 } ∣ and c = ∣ { i ∈ [ m ] ∣ d i = n } ∣ c = |\{i \in[m] \mid d_i = n\}| c = ∣ { i ∈ [ m ] ∣ d i = n } ∣ . Then ∑ i ∈ [ m ] d i = a − b + c n \sum_{i \in[m]} d_i = a-b+cn ∑ i ∈ [ m ] d i = a − b + c n and m = a + b + c m=a+b+c m = a + b + c . Therefore, a − b + c n = 2 n δ a-b+cn=2n\delta a − b + c n = 2 n δ for some integer δ \delta δ . Thus ∣ a − b ∣ = n ∣ 2 δ − c ∣ |a-b|=n|2\delta-c| ∣ a − b ∣ = n ∣2 δ − c ∣ , which is even since n n n is even. It follows that a + b a+b a + b is even. Since m m m is odd, c c c must be odd, which implies ∣ 2 δ − c ∣ ≥ 1 |2\delta-c| \ge1 ∣2 δ − c ∣ ≥ 1 is an odd integer. Consequently,
m = ( a + b ) + c ≥ ∣ a − b ∣ + c = ∣ n ( 2 δ − c ) ∣ + c ≥ n + 1. m=(a+b)+c \ge|a-b|+c = |n(2\delta-c)|+c \ge n+1. m = ( a + b ) + c ≥ ∣ a − b ∣ + c = ∣ n ( 2 δ − c ) ∣ + c ≥ n + 1.
Since W W W contains an odd cycle C C C of length m m m , we have 2 k − 1 ≥ m ≥ n + 1 2k-1 \ge m \ge n+1 2 k − 1 ≥ m ≥ n + 1 , yielding k ≥ n + 2 2 k \ge\frac{n+2}{2} k ≥ 2 n + 2 .
Lemma 4.5. Let t , n t,n t , n be two integers such that 2 ≤ t < ⌊ n 2 ⌋ 2 \le t < \left\lfloor\frac{n}{2} \right\rfloor 2 ≤ t < ⌊ 2 n ⌋ and n ≥ 6 n \ge6 n ≥ 6 . Let Γ = C 2 n ( 1 , n ) \Gamma= C_{2n}(1,n) Γ = C 2 n ( 1 , n ) be a cubic circulant graph and I ( Γ ) \mathcal{I}(\Gamma) I ( Γ ) its edge ideal. Then
v ( I ( Γ ) t ) ≥ 2 t − 2 + Θ n , t , where Θ n , t = { ⌈ n − 2 t + 2 3 ⌉ , if n is odd , ⌈ n − 2 t + 3 3 ⌉ , if n is even . \operatorname{v}(\mathcal{I}(\Gamma)^t) \ge2t-2+\Theta_{n,t},\quad\text{where }\Theta_{n,t}=
\begin{cases}
\left\lceil\frac{n-2t+2}{3}\right\rceil, & \text{if }n\text{ is odd},\\
\left\lceil\frac{n-2t+3}{3}\right\rceil, & \text{if }n\text{ is even}.
\end{cases} v ( I ( Γ ) t ) ≥ 2 t − 2 + Θ n , t , where Θ n , t = { ⌈ 3 n − 2 t + 2 ⌉ , ⌈ 3 n − 2 t + 3 ⌉ , if n is odd , if n is even . Proof. Let I = I ( Γ ) I=\mathcal{I}(\Gamma) I = I ( Γ ) and let e 1 , … , e t − 1 e_1,\ldots,e_{t-1} e 1 , … , e t − 1 be t − 1 t-1 t − 1 edges of Γ \Gamma Γ . By Lemma 4.4 , the colon ideal ( I t : ∏ i = 1 t − 1 e i ) (I^t:\prod_{i=1}^{t-1}e_i) ( I t : ∏ i = 1 t − 1 e i ) can be regarded as the edge ideal of a simple graph. The conclusion then follows from Lemmas 2.11 and 3.1 .
Theorem 4.6. Let t , n t,n t , n be two integers such that 2 ≤ t < ⌊ n 2 ⌋ 2 \le t < \left\lfloor\frac{n}{2} \right\rfloor 2 ≤ t < ⌊ 2 n ⌋ and n ≥ 6 n \ge6 n ≥ 6 . Let Γ = C 2 n ( 1 , n ) \Gamma= C_{2n}(1,n) Γ = C 2 n ( 1 , n ) be a cubic circulant graph and I ( Γ ) \mathcal{I}(\Gamma) I ( Γ ) its edge ideal. Then
v ( I ( Γ ) t ) = 2 t − 2 + Θ n , t , \operatorname{v}(\mathcal{I}(\Gamma)^t)=2t-2+\Theta_{n,t}, v ( I ( Γ ) t ) = 2 t − 2 + Θ n , t , where Θ n , t \Theta_{n,t} Θ n , t is defined as in Lemma 4.5 .
Proof. Let I = I ( Γ ) I=\mathcal{I}(\Gamma) I = I ( Γ ) and m = n − 2 t + 2 m=n-2t+2 m = n − 2 t + 2 . By Lemma 4.5 , v ( I t ) ≥ 2 ( t − 1 ) + Θ n , t \operatorname{v}(I^t)\ge2(t-1)+\Theta_{n,t} v ( I t ) ≥ 2 ( t − 1 ) + Θ n , t . It remains to establish the reverse inequality.
Let
f t , n = ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) f ~ m , f_{t,n}=\left(\prod_{i\in[t-1]}x_{2i}x_{2i+1}\right)\widetilde{f}_m, f t , n = i ∈ [ t − 1 ] ∏ x 2 i x 2 i + 1 f m ,
where f ~ 4 = x 1 x n − 1 \widetilde{f}_4=x_1x_{n-1} f 4 = x 1 x n − 1 , f ~ 5 = x 1 x 2 t + 1 \widetilde{f}_5=x_1x_{2t+1} f 5 = x 1 x 2 t + 1 , and for m ≥ 6 m\ge6 m ≥ 6 ,
f ~ m = { x 1 x 2 t + 1 ( ∏ i ∈ [ k ] x 6 i + 2 t − 3 ) ( ∏ j ∈ [ k − 1 ] x n + 2 t + 6 j ) , if m = 6 k or m = 6 k + 1 , x 1 x n − 1 ( ∏ i ∈ [ k − 1 ] x 6 i + 2 t − 1 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 4 ) , if m = 6 k + 2 , x 1 x n − 2 ( ∏ i ∈ [ k − 1 ] x 6 i + 2 t − 1 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 4 ) , if m = 6 k + 3 , x 1 x n − 3 x n − 1 ( ∏ i ∈ [ k − 1 ] x 6 i + 2 t − 1 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 4 ) , if m = 6 k + 4 , x 1 ( ∏ i ∈ [ k + 1 ] x 6 i + 2 t − 5 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 2 ) , if m = 6 k + 5. \widetilde{f}_m=
\begin{cases}
x_1x_{2t+1}\left(\prod_{i\in[k]}x_{6i+2t-3}\right)\left(\prod_{j\in[k-1]}x_{n+2t+6j}\right), & \text{if }m=6k\text{ or }m=6k+1,\\
x_1x_{n-1}\left(\prod_{i\in[k-1]}x_{6i+2t-1}\right)\left(\prod_{j\in[k]}x_{n+2t+6j-4}\right), & \text{if }m=6k+2,\\
x_1x_{n-2}\left(\prod_{i\in[k-1]}x_{6i+2t-1}\right)\left(\prod_{j\in[k]}x_{n+2t+6j-4}\right), & \text{if }m=6k+3,\\
x_1x_{n-3}x_{n-1}\left(\prod_{i\in[k-1]}x_{6i+2t-1}\right)\left(\prod_{j\in[k]}x_{n+2t+6j-4}\right), & \text{if }m=6k+4,\\
x_1\left(\prod_{i\in[k+1]}x_{6i+2t-5}\right)\left(\prod_{j\in[k]}x_{n+2t+6j-2}\right), & \text{if }m=6k+5.
\end{cases} f m = ⎩ ⎨ ⎧ x 1 x 2 t + 1 ( ∏ i ∈ [ k ] x 6 i + 2 t − 3 ) ( ∏ j ∈ [ k − 1 ] x n + 2 t + 6 j ) , x 1 x n − 1 ( ∏ i ∈ [ k − 1 ] x 6 i + 2 t − 1 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 4 ) , x 1 x n − 2 ( ∏ i ∈ [ k − 1 ] x 6 i + 2 t − 1 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 4 ) , x 1 x n − 3 x n − 1 ( ∏ i ∈ [ k − 1 ] x 6 i + 2 t − 1 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 4 ) , x 1 ( ∏ i ∈ [ k + 1 ] x 6 i + 2 t − 5 ) ( ∏ j ∈ [ k ] x n + 2 t + 6 j − 2 ) , if m = 6 k or m = 6 k + 1 , if m = 6 k + 2 , if m = 6 k + 3 , if m = 6 k + 4 , if m = 6 k + 5.
where k ≥ 1 k\ge1 k ≥ 1 . Then deg ( f t , n ) = deg ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) + deg ( f ~ m ) = 2 ( t − 1 ) + Θ n , t \deg(f_{t,n})=\deg\left(\prod_{i\in[t-1]}x_{2i}x_{2i+1}\right)+\deg(\widetilde{f}_m)=2(t-1)+\Theta_{n,t} deg ( f t , n ) = deg ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) + deg ( f m ) = 2 ( t − 1 ) + Θ n , t .
Consider the ideal
p = { ( x 2 i ∣ i ∈ [ n ] ) , if n is odd , ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ n 2 ] ) + ( x 2 n ) , if n is even . \mathfrak{p}=
\begin{cases}
(x_{2i}\mid i\in[n]), & \text{if }n\text{ is odd},\\
(x_{2i},x_{n+2i-1}\mid i\in\left[\frac{n}{2}\right])+(x_{2n}), & \text{if }n\text{ is even}.
\end{cases} p = { ( x 2 i ∣ i ∈ [ n ]) , ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ 2 n ] ) + ( x 2 n ) , if n is odd , if n is even .
It is straightforward to verify that supp ( p ) \operatorname{supp}(\mathfrak{p}) supp ( p ) is a vertex cover of Γ \Gamma Γ . By Lemma 3.1 , it is a minimal vertex cover. Hence, p ∈ Ass ( I ) \mathfrak{p}\in\operatorname{Ass}(I) p ∈ Ass ( I ) . Moreover, since supp ( f ~ m ) ∩ supp ( p ) = ∅ \operatorname{supp}(\widetilde{f}_m)\cap\operatorname{supp}(\mathfrak{p})=\varnothing supp ( f m ) ∩ supp ( p ) = ∅ , we have
deg p ( f t , n ) = deg p ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) = t − 1. \deg_{\mathfrak p}(f_{t,n})=\deg_{\mathfrak p}\left(\prod_{i\in[t-1]}x_{2i}x_{2i+1}\right)=t-1. deg p ( f t , n ) = deg p i ∈ [ t − 1 ] ∏ x 2 i x 2 i + 1 = t − 1.
By Lemma 2.6, it suffices to prove that
p ⊆ ( I t : f t , n ) . \mathfrak p\subseteq(I^t:f_{t,n}). p ⊆ ( I t : f t , n ) .
For ℓ ∈ [ t ] \ell\in[t] ℓ ∈ [ t ] , we have
x 2 ℓ f t , n = ( ∏ i ∈ [ ℓ ] x 2 i − 1 x 2 i ) ( ∏ j ∈ [ ℓ , t − 1 ] x 2 j x 2 j + 1 ) ( f ~ m x 1 ) ∈ I t , x_{2\ell}f_{t,n}=\left(\prod_{i\in[\ell]}x_{2i-1}x_{2i}\right)\left(\prod_{j\in[\ell,t-1]}x_{2j}x_{2j+1}\right)\left(\frac{\widetilde f_m}{x_1}\right)\in I^t, x 2 ℓ f t , n = i ∈ [ ℓ ] ∏ x 2 i − 1 x 2 i j ∈ [ ℓ , t − 1 ] ∏ x 2 j x 2 j + 1 ( x 1 f m ) ∈ I t ,
x n + 2 ℓ − 1 f t , n = ( x 2 ℓ − 1 x n + 2 ℓ − 1 ) ( ∏ i ∈ [ ℓ − 1 ] x 2 i − 1 x 2 i ) ( ∏ j ∈ [ ℓ , t − 1 ] x 2 j x 2 j + 1 ) ( f ~ m x 1 ) ∈ I t . x_{n+2\ell-1}f_{t,n}=(x_{2\ell-1}x_{n+2\ell-1})\left(\prod_{i\in[\ell-1]}x_{2i-1}x_{2i}\right)\left(\prod_{j\in[\ell,t-1]}x_{2j}x_{2j+1}\right)\left(\frac{\widetilde f_m}{x_1}\right)\in I^t. x n + 2 ℓ − 1 f t , n = ( x 2 ℓ − 1 x n + 2 ℓ − 1 ) i ∈ [ ℓ − 1 ] ∏ x 2 i − 1 x 2 i j ∈ [ ℓ , t − 1 ] ∏ x 2 j x 2 j + 1 ( x 1 f m ) ∈ I t .
Moreover, x 2 n f t , n = ( x 1 x 2 n ) ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) ( f ~ m x 1 ) ∈ I t x_{2n}f_{t,n}=(x_1x_{2n})\left(\prod_{i\in[t-1]}x_{2i}x_{2i+1}\right)\left(\frac{\widetilde f_m}{x_1}\right)\in I^t x 2 n f t , n = ( x 1 x 2 n ) ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) ( x 1 f m ) ∈ I t . Consequently,
( x 2 i , x n + 2 i − 1 ∣ i ∈ [ t ] ) + ( x 2 n ) ⊆ ( I t : f t , n ) . (x_{2i},x_{n+2i-1}\mid i\in[t])+(x_{2n})\subseteq(I^t:f_{t,n}). ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ t ]) + ( x 2 n ) ⊆ ( I t : f t , n ) .
We consider the following two cases depending on the parity of n n n :
Case 1. Suppose that n n n is odd. Then m m m is odd and
p = ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ t ] ) + ( x 2 n ) + p 1 , \mathfrak p=(x_{2i},x_{n+2i-1}\mid i\in[t])+(x_{2n})+\mathfrak p_1, p = ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ t ]) + ( x 2 n ) + p 1 ,
where
p 1 = ( x 2 i ∣ i ∈ [ t + 1 , n − 1 2 ] ) + ( x n + 2 j + 1 ∣ j ∈ [ t , n − 3 2 ] ) . \mathfrak p_1=(x_{2i}\mid i\in[t+1,\tfrac{n-1}{2}])+(x_{n+2j+1}\mid j\in[t,\tfrac{n-3}{2}]). p 1 = ( x 2 i ∣ i ∈ [ t + 1 , 2 n − 1 ]) + ( x n + 2 j + 1 ∣ j ∈ [ t , 2 n − 3 ]) .
Since
f t , n f ~ m = ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ∈ I t − 1 , \frac{f_{t,n}}{\widetilde f_m}=\prod_{i\in[t-1]}x_{2i}x_{2i+1}\in I^{t-1}, f m f t , n = i ∈ [ t − 1 ] ∏ x 2 i x 2 i + 1 ∈ I t − 1 ,
it remains to show that
x f ~ m ∈ I for each x ∈ G ( p 1 ) . x\widetilde f_m\in I\qquad\text{for each }x\in\mathcal G(\mathfrak p_1). x f m ∈ I for each x ∈ G ( p 1 ) .
Case 2. Suppose that n n n is even. Then m m m is even, and
p = ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ t ] ) + ( x 2 n ) + p 2 , \mathfrak p=(x_{2i},x_{n+2i-1}\mid i\in[t])+(x_{2n})+\mathfrak p_2, p = ( x 2 i , x n + 2 i − 1 ∣ i ∈ [ t ]) + ( x 2 n ) + p 2 ,
where
p 2 = ( x 2 i ∣ i ∈ [ t + 1 , n 2 ] ) + ( x n + 2 j + 1 ∣ j ∈ [ t , n − 2 2 ] ) . \mathfrak p_2=(x_{2i}\mid i\in[t+1,\tfrac n2])+(x_{n+2j+1}\mid j\in[t,\tfrac{n-2}{2}]). p 2 = ( x 2 i ∣ i ∈ [ t + 1 , 2 n ]) + ( x n + 2 j + 1 ∣ j ∈ [ t , 2 n − 2 ]) .
As in Case 1, it suffices to prove that
x f ~ m ∈ I for each x ∈ G ( p 2 ) . x\widetilde f_m\in I\qquad\text{for each }x\in\mathcal G(\mathfrak p_2). x f m ∈ I for each x ∈ G ( p 2 ) .
In both cases, the required containment follows by substituting the corresponding expression for f ~ m \widetilde f_m f m and factoring x f ~ m x\widetilde f_m x f m as an edge monomial times a monomial. More precisely, one checks the possible residues of m m m modulo 6: for odd n n n , these are m = 5 , 6 k + 1 , 6 k + 3 , 6 k + 5 m=5,6k+1,6k+3,6k+5 m = 5 , 6 k + 1 , 6 k + 3 , 6 k + 5 ; for even n n n , they are m = 4 , 6 k , 6 k + 2 , 6 k + 4 m=4,6k,6k+2,6k+4 m = 4 , 6 k , 6 k + 2 , 6 k + 4 . For the sake of completeness, we shall verify each of these eight cases separately: (i) If m = 5 m=5 m = 5 , then n = 2 t + 3 n=2t+3 n = 2 t + 3 and p 1 = ( x 2 t + 2 , x 4 t + 4 ) \mathbf{p}_1=(x_{2t+2},x_{4t+4}) p 1 = ( x 2 t + 2 , x 4 t + 4 ) . A simple calculation shows that x u f ~ m = ( x u x 2 t + 1 ) ( f ~ m x 2 t + 1 ) ∈ I x_u\widetilde{f}_m=(x_ux_{2t+1})\left(\frac{\widetilde{f}_m}{x_{2t+1}}\right)\in I x u f m = ( x u x 2 t + 1 ) ( x 2 t + 1 f m ) ∈ I for x u ∈ G ( p 1 ) x_u\in\mathcal{G}(\mathbf{p}_1) x u ∈ G ( p 1 ) .
(ii) If m = 6 k + 1 m=6k+1 m = 6 k + 1 , then n = 2 t + 6 k − 1 n=2t+6k-1 n = 2 t + 6 k − 1 and p 1 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k − 1 ] ) + ( x 6 k + 2 t + 2 j ∣ j ∈ [ t , t + 3 k − 2 ] ) \mathbf{p}_1=(x_{2i}\mid i\in[t+1,t+3k-1])+(x_{6k+2t+2j}\mid j\in[t,t+3k-2]) p 1 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k − 1 ]) + ( x 6 k + 2 t + 2 j ∣ j ∈ [ t , t + 3 k − 2 ]) . There are three cases: (a) if 0 ≤ ℓ ≤ k − 1 0\leq\ell\leq k-1 0 ≤ ℓ ≤ k − 1 and u ∈ { 2 t + 2 , 2 t + 4 , 4 t + 6 k + 2 } u\in\{2t+2,2t+4,4t+6k+2\} u ∈ { 2 t + 2 , 2 t + 4 , 4 t + 6 k + 2 } , then x 6 ℓ + u f ~ m = ( x 6 ℓ + u x 6 ℓ + 2 t + 3 ) ( f ~ m x 6 ℓ + 2 t + 3 ) ∈ I x_{6\ell+u}\widetilde{f}_m=(x_{6\ell+u}x_{6\ell+2t+3})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t+3}}\right)\in I x 6 ℓ + u f m = ( x 6 ℓ + u x 6 ℓ + 2 t + 3 ) ( x 6 ℓ + 2 t + 3 f m ) ∈ I ; (b) if j ∈ [ k − 1 ] j\in[k-1] j ∈ [ k − 1 ] and v ∈ { 2 t , 4 t + 6 k − 2 , 4 t + 6 k } v\in\{2t,4t+6k-2,4t+6k\} v ∈ { 2 t , 4 t + 6 k − 2 , 4 t + 6 k } , then x 6 j + v f ~ m = ( x 6 j + v x 6 j + 4 t + 6 k − 1 ) ( f ~ m x 6 j + 4 t + 6 k − 1 ) ∈ I x_{6j+v}\widetilde{f}_m=(x_{6j+v}x_{6j+4t+6k-1})\left(\frac{\widetilde{f}_m}{x_{6j+4t+6k-1}}\right)\in I x 6 j + v f m = ( x 6 j + v x 6 j + 4 t + 6 k − 1 ) ( x 6 j + 4 t + 6 k − 1 f m ) ∈ I ; (c) x 4 t + 6 k f ~ m = ( x 4 t + 6 k x 2 t + 1 ) ( f ~ m x 2 t + 1 ) ∈ I x_{4t+6k}\widetilde{f}_m=(x_{4t+6k}x_{2t+1})\left(\frac{\widetilde{f}_m}{x_{2t+1}}\right)\in I x 4 t + 6 k f m = ( x 4 t + 6 k x 2 t + 1 ) ( x 2 t + 1 f m ) ∈ I .
(iii) If m = 6 k + 3 m=6k+3 m = 6 k + 3 , then n = 2 t + 6 k + 1 n=2t+6k+1 n = 2 t + 6 k + 1 and p 1 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k ] ) + ( x 6 k + 2 t + 2 j + 2 ∣ j ∈ [ t , t + 3 k − 1 ] ) \mathbf{p}_1=(x_{2i}\mid i\in[t+1,t+3k])+(x_{6k+2t+2j+2}\mid j\in[t,t+3k-1]) p 1 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k ]) + ( x 6 k + 2 t + 2 j + 2 ∣ j ∈ [ t , t + 3 k − 1 ]) . There are two cases: (a) if ℓ ∈ [ k ] \ell\in[k] ℓ ∈ [ k ] and u ∈ { 2 t − 2 , 2 t , 4 t + 6 k } u\in\{2t-2,2t,4t+6k\} u ∈ { 2 t − 2 , 2 t , 4 t + 6 k } , then x 6 ℓ + u f ~ m = ( x 6 ℓ + u x 6 ℓ + 2 t − 1 ) ( f ~ m x 6 ℓ + 2 t − 1 ) ∈ I x_{6\ell+u}\widetilde{f}_m=(x_{6\ell+u}x_{6\ell+2t-1})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t-1}}\right)\in I x 6 ℓ + u f m = ( x 6 ℓ + u x 6 ℓ + 2 t − 1 ) ( x 6 ℓ + 2 t − 1 f m ) ∈ I ; (b) if j ∈ [ k ] j\in[k] j ∈ [ k ] and v ∈ { 2 t − 4 , 4 t + 6 k − 4 , 4 t + 6 k − 2 } v\in\{2t-4,4t+6k-4,4t+6k-2\} v ∈ { 2 t − 4 , 4 t + 6 k − 4 , 4 t + 6 k − 2 } , then x 6 j + v f ~ m = ( x 6 j + v x 6 j + 4 t + 6 k − 3 ) ( f ~ m x 6 j + 4 t + 6 k − 3 ) ∈ I x_{6j+v}\widetilde{f}_m=(x_{6j+v}x_{6j+4t+6k-3})\left(\frac{\widetilde{f}_m}{x_{6j+4t+6k-3}}\right)\in I x 6 j + v f m = ( x 6 j + v x 6 j + 4 t + 6 k − 3 ) ( x 6 j + 4 t + 6 k − 3 f m ) ∈ I .
(iv) If m = 6 k + 5 m=6k+5 m = 6 k + 5 , then n = 2 t + 6 k + 3 n=2t+6k+3 n = 2 t + 6 k + 3 and p 1 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k + 1 ] ) + ( x 6 k + 2 t + 2 j + 4 ∣ j ∈ [ t , t + 3 k ] ) \mathbf{p}_1=(x_{2i}\mid i\in[t+1,t+3k+1])+(x_{6k+2t+2j+4}\mid j\in[t,t+3k]) p 1 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k + 1 ]) + ( x 6 k + 2 t + 2 j + 4 ∣ j ∈ [ t , t + 3 k ]) . There are three cases: (a) if u ∈ { 2 t + 2 , 4 t + 6 k + 4 } u\in\{2t+2,4t+6k+4\} u ∈ { 2 t + 2 , 4 t + 6 k + 4 } , then x u f ~ m = ( x u x 2 t + 1 ) ( f ~ m x 2 t + 1 ) ∈ I x_u\widetilde{f}_m=(x_ux_{2t+1})\left(\frac{\widetilde{f}_m}{x_{2t+1}}\right)\in I x u f m = ( x u x 2 t + 1 ) ( x 2 t + 1 f m ) ∈ I ; (b) if 2 ≤ ℓ ≤ k + 1 2\leq\ell\leq k+1 2 ≤ ℓ ≤ k + 1 and v ∈ { 2 t − 6 , 2 t − 4 , 4 t + 6 k − 2 } v\in\{2t-6,2t-4,4t+6k-2\} v ∈ { 2 t − 6 , 2 t − 4 , 4 t + 6 k − 2 } , then x 6 ℓ + v f ~ m = ( x 6 ℓ + v x 6 ℓ + 2 t − 5 ) ( f ~ m x 6 ℓ + 2 t − 5 ) ∈ I x_{6\ell+v}\widetilde{f}_m=(x_{6\ell+v}x_{6\ell+2t-5})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t-5}}\right)\in I x 6 ℓ + v f m = ( x 6 ℓ + v x 6 ℓ + 2 t − 5 ) ( x 6 ℓ + 2 t − 5 f m ) ∈ I ; (c) if j ∈ [ k ] j\in[k] j ∈ [ k ] and w ∈ { 2 t − 2 , 4 t + 6 k , 4 t + 6 k + 2 } w\in\{2t-2,4t+6k,4t+6k+2\} w ∈ { 2 t − 2 , 4 t + 6 k , 4 t + 6 k + 2 } , we have x 6 j + w f ~ m = ( x 6 j + w x 6 j + 4 t + 6 k + 1 ) ( f ~ m x 6 j + 4 t + 6 k + 1 ) ∈ I x_{6j+w}\widetilde{f}_m=(x_{6j+w}x_{6j+4t+6k+1})\left(\frac{\widetilde{f}_m}{x_{6j+4t+6k+1}}\right)\in I x 6 j + w f m = ( x 6 j + w x 6 j + 4 t + 6 k + 1 ) ( x 6 j + 4 t + 6 k + 1 f m ) ∈ I .
(v) If m = 4 m=4 m = 4 , then n = 2 t + 2 n=2t+2 n = 2 t + 2 and p 2 = ( x 2 t + 2 , x 4 t + 3 ) \mathbf{p}_2=(x_{2t+2},x_{4t+3}) p 2 = ( x 2 t + 2 , x 4 t + 3 ) . A simple calculation shows that x u f ~ m = ( x u x 2 t + 1 ) ( f ~ m x 2 t + 1 ) ∈ I x_u\widetilde{f}_m=(x_ux_{2t+1})\left(\frac{\widetilde{f}_m}{x_{2t+1}}\right)\in I x u f m = ( x u x 2 t + 1 ) ( x 2 t + 1 f m ) ∈ I for x u ∈ G ( p 2 ) x_u\in\mathcal{G}(\mathbf{p}_2) x u ∈ G ( p 2 ) .
(vi) If m = 6 k m=6k m = 6 k , then n = 2 t + 6 k − 2 n=2t+6k-2 n = 2 t + 6 k − 2 and p 2 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k − 1 ] ) + ( x 6 k + 2 t + 2 j − 1 ∣ j ∈ [ t , t + 3 k − 2 ] ) \mathbf{p}_2=(x_{2i}\mid i\in[t+1,t+3k-1])+(x_{6k+2t+2j-1}\mid j\in[t,t+3k-2]) p 2 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k − 1 ]) + ( x 6 k + 2 t + 2 j − 1 ∣ j ∈ [ t , t + 3 k − 2 ]) . There are three cases: (a) if 0 ≤ ℓ ≤ k − 1 0\leq\ell\leq k-1 0 ≤ ℓ ≤ k − 1 and u ∈ { 2 t + 2 , 2 t + 4 , 4 t + 6 k + 1 } u\in\{2t+2,2t+4,4t+6k+1\} u ∈ { 2 t + 2 , 2 t + 4 , 4 t + 6 k + 1 } , then x 6 ℓ + u f ~ m = ( x 6 ℓ + u x 6 ℓ + 2 t + 3 ) ( f ~ m x 6 ℓ + 2 t + 3 ) ∈ I x_{6\ell+u}\widetilde{f}_m=(x_{6\ell+u}x_{6\ell+2t+3})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t+3}}\right)\in I x 6 ℓ + u f m = ( x 6 ℓ + u x 6 ℓ + 2 t + 3 ) ( x 6 ℓ + 2 t + 3 f m ) ∈ I ; (b) if j ∈ [ k − 1 ] j\in[k-1] j ∈ [ k − 1 ] and v ∈ { 2 t , 4 t + 6 k − 3 , 4 t + 6 k − 1 } v\in\{2t,4t+6k-3,4t+6k-1\} v ∈ { 2 t , 4 t + 6 k − 3 , 4 t + 6 k − 1 } , then x 6 j + v f ~ m = ( x 6 j + v x 6 j + 4 t + 6 k − 2 ) ( f ~ m x 6 j + 4 t + 6 k − 2 ) ∈ I x_{6j+v}\widetilde{f}_m=(x_{6j+v}x_{6j+4t+6k-2})\left(\frac{\widetilde{f}_m}{x_{6j+4t+6k-2}}\right)\in I x 6 j + v f m = ( x 6 j + v x 6 j + 4 t + 6 k − 2 ) ( x 6 j + 4 t + 6 k − 2 f m ) ∈ I ; (c) x 4 t + 6 k − 1 f ~ m = ( x 4 t + 6 k − 1 x 2 t + 1 ) ( f ~ m x 2 t + 1 ) ∈ I x_{4t+6k-1}\widetilde{f}_m=(x_{4t+6k-1}x_{2t+1})\left(\frac{\widetilde{f}_m}{x_{2t+1}}\right)\in I x 4 t + 6 k − 1 f m = ( x 4 t + 6 k − 1 x 2 t + 1 ) ( x 2 t + 1 f m ) ∈ I .
(vii) If m = 6 k + 2 m=6k+2 m = 6 k + 2 , then n = 2 t + 6 k n=2t+6k n = 2 t + 6 k and p 2 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k ] ) + ( x 6 k + 2 t + 2 j + 1 ∣ j ∈ [ t , t + 3 k − 1 ] ) \mathbf{p}_2=(x_{2i}\mid i\in[t+1,t+3k])+(x_{6k+2t+2j+1}\mid j\in[t,t+3k-1]) p 2 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k ]) + ( x 6 k + 2 t + 2 j + 1 ∣ j ∈ [ t , t + 3 k − 1 ]) . There are two cases: (a) if ℓ ∈ [ k ] \ell\in[k] ℓ ∈ [ k ] and u ∈ { 2 t − 2 , 2 t , 4 t + 6 k − 1 } u\in\{2t-2,2t,4t+6k-1\} u ∈ { 2 t − 2 , 2 t , 4 t + 6 k − 1 } , then x 6 ℓ + u f ~ m = ( x 6 ℓ + u x 6 ℓ + 2 t − 1 ) ( f ~ m x 6 ℓ + 2 t − 1 ) ∈ I x_{6\ell+u}\widetilde{f}_m=(x_{6\ell+u}x_{6\ell+2t-1})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t-1}}\right)\in I x 6 ℓ + u f m = ( x 6 ℓ + u x 6 ℓ + 2 t − 1 ) ( x 6 ℓ + 2 t − 1 f m ) ∈ I ; (b) if j ∈ [ k ] j\in[k] j ∈ [ k ] and v ∈ { 2 t − 4 , 4 t + 6 k − 5 , 4 t + 6 k − 3 } v\in\{2t-4,4t+6k-5,4t+6k-3\} v ∈ { 2 t − 4 , 4 t + 6 k − 5 , 4 t + 6 k − 3 } , then x 6 j + v f ~ m = ( x 6 j + v x 6 j + 4 t + 6 k − 4 ) ( f ~ m x 6 j + 4 t + 6 k − 4 ) ∈ I x_{6j+v}\widetilde{f}_m=(x_{6j+v}x_{6j+4t+6k-4})\left(\frac{\widetilde{f}_m}{x_{6j+4t+6k-4}}\right)\in I x 6 j + v f m = ( x 6 j + v x 6 j + 4 t + 6 k − 4 ) ( x 6 j + 4 t + 6 k − 4 f m ) ∈ I .
(viii) If m = 6 k + 4 m=6k+4 m = 6 k + 4 , then n = 2 t + 6 k + 2 n=2t+6k+2 n = 2 t + 6 k + 2 and p 2 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k + 1 ] ) + ( x 6 k + 2 t + 2 j + 3 ∣ j ∈ [ t , t + 3 k ] ) \mathbf{p}_2=(x_{2i}\mid i\in[t+1,t+3k+1])+(x_{6k+2t+2j+3}\mid j\in[t,t+3k]) p 2 = ( x 2 i ∣ i ∈ [ t + 1 , t + 3 k + 1 ]) + ( x 6 k + 2 t + 2 j + 3 ∣ j ∈ [ t , t + 3 k ]) . There are three cases: (a) if ℓ ∈ [ k ] \ell\in[k] ℓ ∈ [ k ] and u ∈ { 2 t − 2 , 2 t , 4 t + 6 k + 1 } u\in\{2t-2,2t,4t+6k+1\} u ∈ { 2 t − 2 , 2 t , 4 t + 6 k + 1 } , then x 6 ℓ + u f ~ m = ( x 6 ℓ + u x 6 ℓ + 2 t − 1 ) ( f ~ m x 6 ℓ + 2 t − 1 ) ∈ I x_{6\ell+u}\widetilde{f}_m=(x_{6\ell+u}x_{6\ell+2t-1})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t-1}}\right)\in I x 6 ℓ + u f m = ( x 6 ℓ + u x 6 ℓ + 2 t − 1 ) ( x 6 ℓ + 2 t − 1 f m ) ∈ I ; (b) if j ∈ [ k ] j\in[k] j ∈ [ k ] and v ∈ { 2 t − 4 , 4 t + 6 k − 3 , 4 t + 6 k − 1 } v\in\{2t-4,4t+6k-3,4t+6k-1\} v ∈ { 2 t − 4 , 4 t + 6 k − 3 , 4 t + 6 k − 1 } , then x 6 j + v f ~ m = ( x 6 j + v x 6 j + 4 t + 6 k − 2 ) ( f ~ m x 6 j + 4 t + 6 k − 2 ) ∈ I x_{6j+v}\widetilde{f}_m=(x_{6j+v}x_{6j+4t+6k-2})\left(\frac{\widetilde{f}_m}{x_{6j+4t+6k-2}}\right)\in I x 6 j + v f m = ( x 6 j + v x 6 j + 4 t + 6 k − 2 ) ( x 6 j + 4 t + 6 k − 2 f m ) ∈ I ; (c) if w ∈ { 2 t + 6 k + 2 , 4 t + 12 k + 3 } w\in\{2t+6k+2,4t+12k+3\} w ∈ { 2 t + 6 k + 2 , 4 t + 12 k + 3 } , then x w f ~ m = ( x w x 2 t + 6 k + 1 ) ( f ~ m x 2 t + 6 k + 1 ) ∈ I x_w\widetilde{f}_m=(x_wx_{2t+6k+1})\left(\frac{\widetilde{f}_m}{x_{2t+6k+1}}\right)\in I x w f m = ( x w x 2 t + 6 k + 1 ) ( x 2 t + 6 k + 1 f m ) ∈ I .
Consequently,
p 1 ⊆ ( I : f ~ m ) or p 2 ⊆ ( I : f ~ m ) , \mathbf{p}_1\subseteq(I:\widetilde{f}_m)\quad\text{or}\quad\mathbf{p}_2\subseteq(I:\widetilde{f}_m), p 1 ⊆ ( I : f m ) or p 2 ⊆ ( I : f m ) ,
according to the parity of n n n . It follows that
p ⊆ ( I t : f t , n ) . (6) \mathbf{p}\subseteq(I^t:f_{t,n}). \tag*{(6)} p ⊆ ( I t : f t , n ) . (6)
By Lemma 2.6 , we conclude that
( I t : f t , n ) = p . (I^t:f_{t,n})=\mathfrak{p}. ( I t : f t , n ) = p .
Therefore, f t , n f_{t,n} f t , n yields
v ( I t ) ≤ deg ( f t , n ) = 2 ( t − 1 ) + Θ n , t . \operatorname{v}(I^t)\leq\deg(f_{t,n})=2(t-1)+\Theta_{n,t}. v ( I t ) ≤ deg ( f t , n ) = 2 ( t − 1 ) + Θ n , t .
Combining this with the lower bound gives
v ( I t ) = 2 t − 2 + Θ n , t . \operatorname{v}(I^t)=2t-2+\Theta_{n,t}. v ( I t ) = 2 t − 2 + Θ n , t .
This completes the proof. □
The case of Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n )
In this subsection, we consider the cases where a = 2 a=2 a = 2 and n ≥ 3 n\geq3 n ≥ 3 is odd.
Theorem 4.7. Let n ≥ 3 n\geq3 n ≥ 3 be odd, and let Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n ) be a cubic circulant graph and I ( Γ ) I(\Gamma) I ( Γ ) its edge ideal. Then
v ( I ( Γ ) t ) = 2 t − 1 for all t ≥ n + 1 2 . \operatorname{v}(I(\Gamma)^t)=2t-1 \qquad\text{for all }t\geq\frac{n+1}{2}. v ( I ( Γ ) t ) = 2 t − 1 for all t ≥ 2 n + 1 . Proof. Let I = I ( Γ ) I=I(\Gamma) I = I ( Γ ) and η = n + 1 2 \eta=\frac{n+1}{2} η = 2 n + 1 . As mentioned earlier, for each i ∈ [ n ] i\in[n] i ∈ [ n ] , we relabel the vertices x 2 i − 1 x_{2i-1} x 2 i − 1 and x n + 2 i − 1 x_{n+2i-1} x n + 2 i − 1 as x i x_i x i and y i y_i y i , respectively. By Lemma 2.4, we have v ( I t ) ≥ 2 t − 1 \operatorname{v}(I^t)\geq2t-1 v ( I t ) ≥ 2 t − 1 . It remains to prove v ( I t ) ≤ 2 t − 1 \operatorname{v}(I^t)\leq2t-1 v ( I t ) ≤ 2 t − 1 . To this end, we choose f = ( y n − 1 y n ) t − η ( ∏ i ∈ [ n ] y i ) f=(y_{n-1}y_n)^{t-\eta}\left(\prod_{i\in[n]}y_i\right) f = ( y n − 1 y n ) t − η ( ∏ i ∈ [ n ] y i ) and m = ( x i , y i ∣ i ∈ [ n ] ) \mathfrak{m}=(x_i,y_i\mid i\in[n]) m = ( x i , y i ∣ i ∈ [ n ]) . We will prove that ( I t : f ) = m (I^t:f)=\mathfrak{m} ( I t : f ) = m , the result then follows. Clearly, ( I t : f ) ⊆ m (I^t:f)\subseteq\mathfrak{m} ( I t : f ) ⊆ m . For any i ∈ [ n ] i\in[n] i ∈ [ n ] , we have
x i f = ( x i y i ) ( y n − 1 y n ) t − η ( ∏ j ∈ [ n − 1 2 ] y i + 2 j − 1 y i + 2 j ) ∈ I t , x_i f=(x_i y_i)(y_{n-1}y_n)^{t-\eta}\left(\prod_{j\in[\frac{n-1}{2}]}y_{i+2j-1}y_{i+2j}\right)\in I^t, x i f = ( x i y i ) ( y n − 1 y n ) t − η j ∈ [ 2 n − 1 ] ∏ y i + 2 j − 1 y i + 2 j ∈ I t , y i f = ( y i y i + 1 ) ( y n − 1 y n ) t − η ( ∏ j ∈ [ n − 1 2 ] y i + 2 j y i + 2 j + 1 ) ∈ I t . y_i f=(y_i y_{i+1})(y_{n-1}y_n)^{t-\eta}\left(\prod_{j\in[\frac{n-1}{2}]}y_{i+2j}y_{i+2j+1}\right)\in I^t. y i f = ( y i y i + 1 ) ( y n − 1 y n ) t − η j ∈ [ 2 n − 1 ] ∏ y i + 2 j y i + 2 j + 1 ∈ I t . Therefore, m ⊆ ( I t : f ) \mathfrak{m}\subseteq(I^t:f) m ⊆ ( I t : f ) . □
Remark 4.8. From the proof of the above theorem, we observe that if we choose U = ∅ U=\varnothing U = ∅ and V = { y 2 i − 1 ∣ i ∈ [ n + 1 2 ] } ∪ { x n } V=\{y_{2i-1}\mid i\in[\frac{n+1}{2}]\}\cup\{x_n\} V = { y 2 i − 1 ∣ i ∈ [ 2 n + 1 ]} ∪ { x n } , then we can check that ( U , V ) (U,V) ( U , V ) is an admissible pair as defined in [19 Definition 1.1] . Thus, it follows from [19 Theorem 1.2] that v s t a b ( I ( Γ ) ) ≤ n + 1 2 \operatorname{v}_{\mathrm{stab}}(I(\Gamma))\leq\frac{n+1}{2} v stab ( I ( Γ )) ≤ 2 n + 1 . Therefore, by [19 Lemma 2.2] , for all t ≥ n + 1 2 t\geq\frac{n+1}{2} t ≥ 2 n + 1 , we have v ( I ( Γ ) t ) = 2 t − 1 \operatorname{v}(I(\Gamma)^t)=2t-1 v ( I ( Γ ) t ) = 2 t − 1 .
Remark 4.9. When n = 3 n=3 n = 3 , by Theorems 3.4 and 4.7 , we have
v ( I ( C 2 n ( 2 , n ) ) t ) = { 2 , if t = 1 , 2 t − 1 , otherwise. \operatorname{v}(I(C_{2n}(2,n))^t)=
\begin{cases}
2, & \text{if }t=1,\\
2t-1, & \text{otherwise.}
\end{cases} v ( I ( C 2 n ( 2 , n ) ) t ) = { 2 , 2 t − 1 , if t = 1 , otherwise. In the following, we will consider cases where n ≥ 5 n\geq5 n ≥ 5 is odd and 2 ≤ t < n + 1 2 2\leq t<\frac{n+1}{2} 2 ≤ t < 2 n + 1 .
Lemma 4.10. Let Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n ) be a cubic circulant graph, and e 1 , … , e t − 1 e_1,\ldots,e_{t-1} e 1 , … , e t − 1 be its t − 1 t-1 t − 1 edges. Suppose I ( Γ ) I(\Gamma) I ( Γ ) is the edge ideal of Γ \Gamma Γ . Then ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) (I(\Gamma)^t:\prod_{i=1}^{t-1}e_i) ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) can be regarded as the edge ideal of a simple graph.
Proof. By Theorem 2.8 , the colon ideal ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) \left(I(\Gamma)^t:\prod_{i=1}^{t-1}e_i\right) ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) is generated by monomials of degree two. We only need to show that x i 2 ∉ G ( ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) ) x_i^2\notin\mathcal{G}\left(\left(I(\Gamma)^t:\prod_{i=1}^{t-1}e_i\right)\right) x i 2 ∈ / G ( ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) ) for any i ∈ [ 2 n ] i\in[2n] i ∈ [ 2 n ] .
Suppose for a contradiction that x i 2 ∈ ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) x_i^2\in\left(I(\Gamma)^t:\prod_{i=1}^{t-1}e_i\right) x i 2 ∈ ( I ( Γ ) t : ∏ i = 1 t − 1 e i ) for some i ∈ [ 2 n ] i\in[2n] i ∈ [ 2 n ] . Since Γ \Gamma Γ contains no loops, we have x i 2 ∉ E ( Γ ) x_i^2\notin E(\Gamma) x i 2 ∈ / E ( Γ ) . By Theorem 2.8 , x i x_i x i is even-connected to itself with respect to ∏ i = 1 t − 1 e i \prod_{i=1}^{t-1}e_i ∏ i = 1 t − 1 e i . Thus, there exists a walk W : x i = x ℓ 1 , x ℓ 2 , … , x ℓ 2 k = x i W:x_i=x_{\ell_1},x_{\ell_2},\ldots,x_{\ell_{2k}}=x_i W : x i = x ℓ 1 , x ℓ 2 , … , x ℓ 2 k = x i of length 2 k − 1 2k-1 2 k − 1 satisfying Definition 2.7 . Hence, W W W contains an odd cycle. Furthermore, by (2) and (3) of Definition 2.7 , we have k ≤ t k\leq t k ≤ t . As in the case of Lemma 4.4 where n n n is even, we can prove that k ≥ n + 1 2 k\geq\frac{n+1}{2} k ≥ 2 n + 1 . Thus t ≥ k ≥ n + 1 2 t\geq k\geq\frac{n+1}{2} t ≥ k ≥ 2 n + 1 , which contradicts the fact that t < n + 1 2 t<\frac{n+1}{2} t < 2 n + 1 . □
The following lemma can be shown by similar arguments as in Lemma 4.5 , we omit its proof.
Lemma 4.11. Let t , n t,n t , n be two integers such that 2 ≤ t < n + 1 2 2\leq t<\frac{n+1}{2} 2 ≤ t < 2 n + 1 and n ≥ 5 n\geq5 n ≥ 5 is odd. Let Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n ) be a cubic circulant graph and I ( Γ ) I(\Gamma) I ( Γ ) its edge ideal. Then
v ( I ( Γ ) t ) ≥ 2 t − 1 + ⌊ n − 2 t 3 ⌋ . \operatorname{v}(I(\Gamma)^t)\geq2t-1+\left\lfloor\frac{n-2t}{3}\right\rfloor. v ( I ( Γ ) t ) ≥ 2 t − 1 + ⌊ 3 n − 2 t ⌋ .
Theorem 4.12. Let t , n t,n t , n be two integers such that 2 ≤ t < n + 1 2 2\leq t<\frac{n+1}{2} 2 ≤ t < 2 n + 1 and n ≥ 5 n\geq5 n ≥ 5 is odd. Let Γ = C 2 n ( 2 , n ) \Gamma=C_{2n}(2,n) Γ = C 2 n ( 2 , n ) be a cubic circulant graph and I ( Γ ) I(\Gamma) I ( Γ ) its edge ideal. Then
v ( I ( Γ ) t ) = 2 t − 1 + ⌊ n − 2 t 3 ⌋ . \operatorname{v}(I(\Gamma)^t)=2t-1+\left\lfloor\frac{n-2t}{3}\right\rfloor. v ( I ( Γ ) t ) = 2 t − 1 + ⌊ 3 n − 2 t ⌋ .
Proof. Let I = I ( Γ ) I=I(\Gamma) I = I ( Γ ) . By Lemma 4.11 , v ( I t ) ≥ 2 t − 1 + ⌊ n − 2 t 3 ⌋ \operatorname{v}(I^t)\geq2t-1+\left\lfloor\frac{n-2t}{3}\right\rfloor v ( I t ) ≥ 2 t − 1 + ⌊ 3 n − 2 t ⌋ . It remains to prove the reverse inequality. Let m = n − 2 t m=n-2t m = n − 2 t . Since n n n is odd and t < n + 1 2 t<\frac{n+1}{2} t < 2 n + 1 , the integer m m m is positive and odd. Let
f t , n = x 2 t − 1 ( ∏ i ∈ [ t − 1 ] x 2 i − 1 x 2 i ) f ~ m . f_{t,n}=x_{2t-1}\left(\prod_{i\in[t-1]}x_{2i-1}x_{2i}\right)\widetilde{f}_m. f t , n = x 2 t − 1 i ∈ [ t − 1 ] ∏ x 2 i − 1 x 2 i f m .
where f ~ 1 = y 2 t \widetilde{f}_1=y_{2t} f 1 = y 2 t , and for m ≥ 3 m\geq3 m ≥ 3 ,
f ~ m = { ( ∏ i ∈ [ 2 , k ] x 2 t + 6 i − 7 ) ( ∏ j ∈ [ k ] y 2 t + 6 j − 4 ) , if m = 6 k − 3 , ( ∏ i ∈ [ 2 , k ] x 2 t + 6 i − 7 ) ( ∏ j ∈ [ k ] y 2 t + 6 j − 4 ) y 2 t + 6 k − 2 , if m = 6 k − 1 , ( ∏ i ∈ [ 2 , k + 1 ] x 2 t + 6 i − 7 ) ( ∏ j ∈ [ k ] y 2 t + 6 j − 4 ) y 2 t + 6 k , if m = 6 k + 1 , \widetilde{f}_m=
\begin{cases}
\left(\prod_{i\in[2,k]}x_{2t+6i-7}\right)\left(\prod_{j\in[k]}y_{2t+6j-4}\right), & \text{if }m=6k-3,\\
\left(\prod_{i\in[2,k]}x_{2t+6i-7}\right)\left(\prod_{j\in[k]}y_{2t+6j-4}\right)y_{2t+6k-2}, & \text{if }m=6k-1,\\
\left(\prod_{i\in[2,k+1]}x_{2t+6i-7}\right)\left(\prod_{j\in[k]}y_{2t+6j-4}\right)y_{2t+6k}, & \text{if }m=6k+1,
\end{cases} f m = ⎩ ⎨ ⎧ ( ∏ i ∈ [ 2 , k ] x 2 t + 6 i − 7 ) ( ∏ j ∈ [ k ] y 2 t + 6 j − 4 ) , ( ∏ i ∈ [ 2 , k ] x 2 t + 6 i − 7 ) ( ∏ j ∈ [ k ] y 2 t + 6 j − 4 ) y 2 t + 6 k − 2 , ( ∏ i ∈ [ 2 , k + 1 ] x 2 t + 6 i − 7 ) ( ∏ j ∈ [ k ] y 2 t + 6 j − 4 ) y 2 t + 6 k , if m = 6 k − 3 , if m = 6 k − 1 , if m = 6 k + 1 ,
where k ≥ 1 k\geq1 k ≥ 1 . By construction,
deg ( f t , n ) = 2 t − 1 + ⌊ m 3 ⌋ = 2 t − 1 + ⌊ n − 2 t 3 ⌋ . \deg(f_{t,n})=2t-1+\left\lfloor\frac{m}{3}\right\rfloor=2t-1+\left\lfloor\frac{n-2t}{3}\right\rfloor. deg ( f t , n ) = 2 t − 1 + ⌊ 3 m ⌋ = 2 t − 1 + ⌊ 3 n − 2 t ⌋ .
Consider the ideal
p = ( y 2 i − 1 ∣ i ∈ [ n + 1 2 ] ) + ( x 2 j ∣ j ∈ [ n − 1 2 ] ) + ( x n ) . \mathfrak{p}=(y_{2i-1}\mid i\in[\frac{n+1}{2}])+(x_{2j}\mid j\in[\frac{n-1}{2}])+(x_n). p = ( y 2 i − 1 ∣ i ∈ [ 2 n + 1 ]) + ( x 2 j ∣ j ∈ [ 2 n − 1 ]) + ( x n ) .
By the proof of Theorem 3.4 , we have p ∈ Ass ( I ) \mathfrak{p}\in\operatorname{Ass}(I) p ∈ Ass ( I ) . Moreover, supp ( f ~ m ) ∩ supp ( p ) = ∅ \operatorname{supp}(\widetilde{f}_m)\cap\operatorname{supp}(\mathfrak{p})=\varnothing supp ( f m ) ∩ supp ( p ) = ∅ , and hence
deg p ( f t , n ) = t − 1. \deg_{\mathfrak{p}}(f_{t,n})=t-1. deg p ( f t , n ) = t − 1.
By Lemma 2.6 , it suffices to show that p ⊆ ( I t : f t , n ) \mathfrak{p}\subseteq(I^t:f_{t,n}) p ⊆ ( I t : f t , n ) . For ℓ ∈ [ t ] \ell\in[t] ℓ ∈ [ t ] , we have
x 2 ℓ f t , n = ( ∏ i ∈ [ ℓ ] x 2 i − 1 x 2 i ) ( ∏ j ∈ [ ℓ , t − 1 ] x 2 j x 2 j + 1 ) f ~ m ∈ I t , x_{2\ell}f_{t,n}=\left(\prod_{i\in[\ell]}x_{2i-1}x_{2i}\right)\left(\prod_{j\in[\ell,t-1]}x_{2j}x_{2j+1}\right)\widetilde{f}_m\in I^t, x 2 ℓ f t , n = i ∈ [ ℓ ] ∏ x 2 i − 1 x 2 i j ∈ [ ℓ , t − 1 ] ∏ x 2 j x 2 j + 1 f m ∈ I t ,
y 2 ℓ − 1 f t , n = ( x 2 ℓ − 1 y 2 ℓ − 1 ) ( ∏ i ∈ [ ℓ − 1 ] x 2 i − 1 x 2 i ) ( ∏ j ∈ [ ℓ , t − 1 ] x 2 j x 2 j + 1 ) f ~ m ∈ I t . y_{2\ell-1}f_{t,n}=(x_{2\ell-1}y_{2\ell-1})\left(\prod_{i\in[\ell-1]}x_{2i-1}x_{2i}\right)\left(\prod_{j\in[\ell,t-1]}x_{2j}x_{2j+1}\right)\widetilde{f}_m\in I^t. y 2 ℓ − 1 f t , n = ( x 2 ℓ − 1 y 2 ℓ − 1 ) i ∈ [ ℓ − 1 ] ∏ x 2 i − 1 x 2 i j ∈ [ ℓ , t − 1 ] ∏ x 2 j x 2 j + 1 f m ∈ I t .
Moreover, we have
x n f t , n = ( x n x 1 ) ( ∏ i ∈ [ t − 1 ] x 2 i x 2 i + 1 ) f ~ m ∈ I t . x_nf_{t,n}=(x_nx_1)\left(\prod_{i\in[t-1]}x_{2i}x_{2i+1}\right)\widetilde{f}_m\in I^t. x n f t , n = ( x n x 1 ) i ∈ [ t − 1 ] ∏ x 2 i x 2 i + 1 f m ∈ I t .
Therefore,
( x 2 i , y 2 i − 1 ∣ i ∈ [ t ] ) + ( x n ) ⊆ ( I t : f t , n ) . (x_{2i},y_{2i-1}\mid i\in[t])+(x_n)\subseteq(I^t:f_{t,n}). ( x 2 i , y 2 i − 1 ∣ i ∈ [ t ]) + ( x n ) ⊆ ( I t : f t , n ) .
Write
p = ( x 2 i , y 2 i − 1 ∣ i ∈ [ t ] ) + ( x n ) + p 1 , \mathfrak{p}=(x_{2i},y_{2i-1}\mid i\in[t])+(x_n)+\mathfrak{p}_1, p = ( x 2 i , y 2 i − 1 ∣ i ∈ [ t ]) + ( x n ) + p 1 ,
where
p 1 = ( x 2 j ∣ j ∈ [ t + 1 , n − 1 2 ] ) + ( y 2 i − 1 ∣ i ∈ [ t + 1 , n + 1 2 ] ) . \mathfrak{p}_1=(x_{2j}\mid j\in[t+1,\frac{n-1}{2}])+(y_{2i-1}\mid i\in[t+1,\frac{n+1}{2}]). p 1 = ( x 2 j ∣ j ∈ [ t + 1 , 2 n − 1 ]) + ( y 2 i − 1 ∣ i ∈ [ t + 1 , 2 n + 1 ]) .
Since
x 2 t − 1 ∏ i ∈ [ t − 1 ] x 2 i − 1 x 2 i ∈ I t − 1 , x_{2t-1}\prod_{i\in[t-1]}x_{2i-1}x_{2i}\in I^{t-1}, x 2 t − 1 i ∈ [ t − 1 ] ∏ x 2 i − 1 x 2 i ∈ I t − 1 ,
it remains to verify that u f ~ m ∈ I u\widetilde{f}_m\in I u f m ∈ I for every u ∈ G ( p 1 ) u\in G(\mathfrak{p}_1) u ∈ G ( p 1 ) . We consider the three possible forms of m m m :
m = 6 k − 3 , m = 6 k − 1 , m = 6 k + 1 , m=6k-3,\qquad m=6k-1,\qquad m=6k+1, m = 6 k − 3 , m = 6 k − 1 , m = 6 k + 1 ,
together with the special case m = 1 m=1 m = 1 . For m = 1 m=1 m = 1 , the assertion follows immediately from f ~ 1 = y 2 t \widetilde{f}_1=y_{2t} f 1 = y 2 t . For the other three cases, substituting the corresponding expression for f ~ m \widetilde{f}_m f m shows that every generator u u u of p 1 \mathfrak{p}_1 p 1 can be paired with a variable appearing in f ~ m \widetilde{f}_m f m such that their product is an edge monomial of Γ \Gamma Γ . For the sake of completeness, we shall verify each of these three cases separately:
(1) If m = 6 k − 3 m=6k-3 m = 6 k − 3 , then n = 2 t + 6 k − 3 n=2t+6k-3 n = 2 t + 6 k − 3 and p 1 = ( x 2 t + 2 i ∣ i ∈ [ 3 k − 2 ] ) + ( y 2 t + 2 j − 1 ∣ j ∈ [ 3 k − 1 ] ) \mathfrak{p}_1=(x_{2t+2i}\mid i\in[3k-2])+(y_{2t+2j-1}\mid j\in[3k-1]) p 1 = ( x 2 t + 2 i ∣ i ∈ [ 3 k − 2 ]) + ( y 2 t + 2 j − 1 ∣ j ∈ [ 3 k − 1 ]) . There are two cases: (i) if u ∈ { x 6 ℓ + 2 t + 2 , y 6 ℓ + 2 t + 1 , y 6 ℓ + 2 t + 3 ∣ 0 ≤ ℓ ≤ k − 1 } u\in\{x_{6\ell+2t+2},y_{6\ell+2t+1},y_{6\ell+2t+3}\mid0\leq\ell\leq k-1\} u ∈ { x 6 ℓ + 2 t + 2 , y 6 ℓ + 2 t + 1 , y 6 ℓ + 2 t + 3 ∣ 0 ≤ ℓ ≤ k − 1 } , then
u f ~ m = ( u y 6 ℓ + 2 t + 2 ) ( f ~ m y 6 ℓ + 2 t + 2 ) ∈ I ; u\widetilde{f}_m=(uy_{6\ell+2t+2})\left(\frac{\widetilde{f}_m}{y_{6\ell+2t+2}}\right)\in I; u f m = ( u y 6 ℓ + 2 t + 2 ) ( y 6 ℓ + 2 t + 2 f m ) ∈ I ;
(ii) if v ∈ { x 6 ℓ + 2 t + 4 , x 6 ℓ + 2 t + 6 , y 6 ℓ + 2 t + 5 ∣ 0 ≤ ℓ ≤ k − 2 } v\in\{x_{6\ell+2t+4},x_{6\ell+2t+6},y_{6\ell+2t+5}\mid0\leq\ell\leq k-2\} v ∈ { x 6 ℓ + 2 t + 4 , x 6 ℓ + 2 t + 6 , y 6 ℓ + 2 t + 5 ∣ 0 ≤ ℓ ≤ k − 2 } , then
v f ~ m = ( v x 6 ℓ + 2 t + 5 ) ( f ~ m x 6 ℓ + 2 t + 5 ) ∈ I . v\widetilde{f}_m=(vx_{6\ell+2t+5})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t+5}}\right)\in I. v f m = ( v x 6 ℓ + 2 t + 5 ) ( x 6 ℓ + 2 t + 5 f m ) ∈ I .
(2) If m = 6 k − 1 m=6k-1 m = 6 k − 1 , then n = 2 t + 6 k − 1 n=2t+6k-1 n = 2 t + 6 k − 1 and p 1 = ( x 2 t + 2 i ∣ i ∈ [ 3 k − 1 ] ) + ( y 2 t + 2 j − 1 ∣ j ∈ [ 3 k ] ) \mathfrak{p}_1=(x_{2t+2i}\mid i\in[3k-1])+(y_{2t+2j-1}\mid j\in[3k]) p 1 = ( x 2 t + 2 i ∣ i ∈ [ 3 k − 1 ]) + ( y 2 t + 2 j − 1 ∣ j ∈ [ 3 k ]) . There are three cases: (i) if u ∈ { x 6 ℓ + 2 t + 2 , y 6 ℓ + 2 t + 1 , y 6 ℓ + 2 t + 3 ∣ 0 ≤ ℓ ≤ k − 1 } u\in\{x_{6\ell+2t+2},y_{6\ell+2t+1},y_{6\ell+2t+3}\mid0\leq\ell\leq k-1\} u ∈ { x 6 ℓ + 2 t + 2 , y 6 ℓ + 2 t + 1 , y 6 ℓ + 2 t + 3 ∣ 0 ≤ ℓ ≤ k − 1 } , then
u f ~ m = ( u y 6 ℓ + 2 t + 2 ) ( f ~ m y 6 ℓ + 2 t + 2 ) ∈ I ; u\widetilde{f}_m=(uy_{6\ell+2t+2})\left(\frac{\widetilde{f}_m}{y_{6\ell+2t+2}}\right)\in I; u f m = ( u y 6 ℓ + 2 t + 2 ) ( y 6 ℓ + 2 t + 2 f m ) ∈ I ;
(ii) if v ∈ { x 6 ℓ + 2 t + 4 , x 6 ℓ + 2 t + 6 , y 6 ℓ + 2 t + 5 ∣ 0 ≤ ℓ ≤ k − 2 } v\in\{x_{6\ell+2t+4},x_{6\ell+2t+6},y_{6\ell+2t+5}\mid0\leq\ell\leq k-2\} v ∈ { x 6 ℓ + 2 t + 4 , x 6 ℓ + 2 t + 6 , y 6 ℓ + 2 t + 5 ∣ 0 ≤ ℓ ≤ k − 2 } , then
v f ~ m = ( v x 6 ℓ + 2 t + 5 ) ( f ~ m x 6 ℓ + 2 t + 5 ) ∈ I ; v\widetilde{f}_m=(vx_{6\ell+2t+5})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t+5}}\right)\in I; v f m = ( v x 6 ℓ + 2 t + 5 ) ( x 6 ℓ + 2 t + 5 f m ) ∈ I ;
(iii) if w ∈ { x n − 1 , y n − 2 , y n } w\in\{x_{n-1},y_{n-2},y_n\} w ∈ { x n − 1 , y n − 2 , y n } , then
w f ~ m = ( w y n − 1 ) ( f ~ m y n − 1 ) ∈ I . w\widetilde{f}_m=(wy_{n-1})\left(\frac{\widetilde{f}_m}{y_{n-1}}\right)\in I. w f m = ( w y n − 1 ) ( y n − 1 f m ) ∈ I .
(3) If m = 6 k + 1 m = 6k + 1 m = 6 k + 1 , then n = 2 t + 6 k + 1 n = 2t + 6k + 1 n = 2 t + 6 k + 1 and p 1 = ( x 2 t + 2 i ∣ i ∈ [ 3 k ] ) + ( y 2 t + 2 j − 1 ∣ j ∈ [ 3 k + 1 ] ) \mathfrak{p}_1 = (x_{2t+2i} \mid i \in[3k]) + (y_{2t+2j-1} \mid j \in[3k+1]) p 1 = ( x 2 t + 2 i ∣ i ∈ [ 3 k ]) + ( y 2 t + 2 j − 1 ∣ j ∈ [ 3 k + 1 ]) . There are three cases: (i) if u ∈ { x 6 ℓ + 2 t + 2 , y 6 ℓ + 2 t + 1 , y 6 ℓ + 2 t + 3 ∣ 0 ≤ ℓ ≤ k − 1 } u \in\{x_{6\ell+2t+2}, y_{6\ell+2t+1}, y_{6\ell+2t+3} \mid0 \le\ell\le k-1\} u ∈ { x 6 ℓ + 2 t + 2 , y 6 ℓ + 2 t + 1 , y 6 ℓ + 2 t + 3 ∣ 0 ≤ ℓ ≤ k − 1 } , then u f ~ m = ( u y 6 ℓ + 2 t + 2 ) ( f ~ m y 6 ℓ + 2 t + 2 ) ∈ I u\widetilde{f}_m = (uy_{6\ell+2t+2})\left(\frac{\widetilde{f}_m}{y_{6\ell+2t+2}}\right) \in I u f m = ( u y 6 ℓ + 2 t + 2 ) ( y 6 ℓ + 2 t + 2 f m ) ∈ I ; (ii) if v ∈ { x 6 ℓ + 2 t + 4 , x 6 ℓ + 2 t + 6 , y 6 ℓ + 2 t + 5 ∣ 0 ≤ ℓ ≤ k − 1 } v \in\{x_{6\ell+2t+4}, x_{6\ell+2t+6}, y_{6\ell+2t+5} \mid0 \le\ell\le k-1\} v ∈ { x 6 ℓ + 2 t + 4 , x 6 ℓ + 2 t + 6 , y 6 ℓ + 2 t + 5 ∣ 0 ≤ ℓ ≤ k − 1 } , then v f ~ m = ( v x 6 ℓ + 2 t + 5 ) ( f ~ m x 6 ℓ + 2 t + 5 ) ∈ I v\widetilde{f}_m = (vx_{6\ell+2t+5})\left(\frac{\widetilde{f}_m}{x_{6\ell+2t+5}}\right) \in I v f m = ( v x 6 ℓ + 2 t + 5 ) ( x 6 ℓ + 2 t + 5 f m ) ∈ I . (iii) if w ∈ { x n − 1 , y n − 2 , y n } w \in\{x_{n-1}, y_{n-2}, y_n\} w ∈ { x n − 1 , y n − 2 , y n } , then w f ~ m = ( w y n − 1 ) f ~ m y n − 1 ∈ I w\widetilde{f}_m = (wy_{n-1})\frac{\widetilde{f}_m}{y_{n-1}} \in I w f m = ( w y n − 1 ) y n − 1 f m ∈ I .
Consequently, u f ~ m ∈ I u\widetilde{f}_m \in I u f m ∈ I for every u ∈ G ( p 1 ) u \in\mathcal{G}(\mathfrak{p}_1) u ∈ G ( p 1 ) , and therefore
p 1 ⊆ ( I : f ~ m ) . \mathfrak{p}_1 \subseteq(I : \widetilde{f}_m). p 1 ⊆ ( I : f m ) .
It follows that p ⊆ ( I t : f t , n ) \mathfrak{p} \subseteq(I^t : f_{t,n}) p ⊆ ( I t : f t , n ) . By Lemma 2.6,
( I t : f t , n ) = p . (I^t : f_{t,n}) = \mathfrak{p}. ( I t : f t , n ) = p .
Hence
ν ( I t ) ≤ deg ( f t , n ) = 2 t − 1 + ⌈ n − 2 t 3 ⌉ . \nu(I^t) \leq\deg(f_{t,n}) = 2t - 1 + \left\lceil\frac{n-2t}{3} \right\rceil. ν ( I t ) ≤ deg ( f t , n ) = 2 t − 1 + ⌈ 3 n − 2 t ⌉ .
Combining this with the lower bound yields
ν ( I t ) = 2 t − 1 + ⌈ n − 2 t 3 ⌉ . \nu(I^t) = 2t - 1 + \left\lceil\frac{n-2t}{3} \right\rceil. ν ( I t ) = 2 t − 1 + ⌈ 3 n − 2 t ⌉ .
This completes the proof. □
The case of Γ = C 2 n ( a , n ) \Gamma= C_{2n}(a,n) Γ = C 2 n ( a , n ) with 1 ≤ a < n 1 \leq a < n 1 ≤ a < n
The cases a = 1 a = 1 a = 1 and a = 2 a = 2 a = 2 for odd n ≥ 3 n \geq3 n ≥ 3 have already been established. Therefore, by Lemma 2.3 and Theorems 4.1, 4.6, 4.7, and 4.12, and direct calculation, we can obtain the following theorem, which corresponds to the case where t ≥ 2 t \geq2 t ≥ 2 in Theorem 1.1:
Theorem 4.13. Let Γ = C 2 n ( a , n ) \Gamma= C_{2n}(a,n) Γ = C 2 n ( a , n ) be a cubic circulant graph with 1 ≤ a < n 1 \leq a < n 1 ≤ a < n , and let I ( Γ ) I(\Gamma) I ( Γ ) be its edge ideal. Let k = gcd ( a , n ) k = \gcd(a,n) k = g cd( a , n ) , m = n k m = \frac{n}{k} m = k n , b = a k b = \frac{a}{k} b = k a and
ρ = { m , if both m and b are odd , m + 1 , otherwise . \rho=
\begin{cases}
m, & \text{if both } m \text{ and } b \text{ are odd},\\
m+1, & \text{otherwise}.
\end{cases} ρ = { m , m + 1 , if both m and b are odd , otherwise . Then, for all integer t ≥ 1 t \geq1 t ≥ 1 ,
ν ( I ( Γ ) t ) = max { 2 t + k − 2 , t − 1 + k ⌈ ρ 3 ⌉ , ⌈ 4 ( t − 1 ) + k ρ 3 ⌉ , ⌈ 3 ( t − 1 ) + k ( ⌊ ρ 2 ⌋ + 1 ) 2 ⌉ } . \nu(I(\Gamma)^t) = \max\left\{2t+k-2,\ t-1+k\left\lceil\frac{\rho}{3}\right\rceil,\ \left\lceil\frac{4(t-1)+k\rho}{3}\right\rceil,\ \left\lceil\frac{3(t-1)+k\left(\left\lfloor\frac{\rho}{2}\right\rfloor+1\right)}{2}\right\rceil\right\}. ν ( I ( Γ ) t ) = max { 2 t + k − 2 , t − 1 + k ⌈ 3 ρ ⌉ , ⌈ 3 4 ( t − 1 ) + k ρ ⌉ , ⌈ 2 3 ( t − 1 ) + k ( ⌊ 2 ρ ⌋ + 1 ) ⌉ } .
Acknowledgment: This research is supported by the Natural Science Foundation of Jiangsu Province (No. BK20221353) and the National Natural Science Foundation of China (No.12471246). The first author carried out part of this work while visiting the International Center for Research in Mathematics and Postgraduate Training (ICRTM), affiliated with the Institute of Mathematics at the Vietnam Academy of Science and Technology, through the ICRTM–TWAS Visiting Program. She expresses her sincere gratitude for the financial support and warm hospitality received during this period.
School of Mathematical Sciences, Soochow University, Suzhou, Jiangsu, 215006, P. R. China
Email address: zhuguangjun@suda.edu.cn (Corresponding author:Guangjun Zhu), lijiaxinworking@163.com(Jiaxin Li), 1246683863@qq.com(Yulin Hu)