Research article

Mittag-Leffler projective synchronization of uncertain fractional-order fuzzy complex valued neural networks with distributed and time-varying delays

  • To study the Mittag-Leffler projective synchronization (MLPS) problem of fractional-order fuzzy neural networks (FOFNNs), in this work we introduced the FOFNNs model. On this basis, we discussed the MLPS of uncertain fractional-order fuzzy complex valued neural networks (FOFCVNNs) with distributed and time-varying delays. Utilizing Banach contraction mapping principle, we proved the existence and uniqueness of the model solution. Moreover, employing the construction of a new hybrid controller, an adaptive hybrid controller, and the fractional-order Razumikhin theorem, algebraic criteria was obtained for implementing MLPS. The algebraic inequality criterion obtained in this article improves and extends the previously published results on MLPS, making it easy to prove and greatly reducing the computational complexity. Finally, different Caputo derivatives of different orders were given, and four numerical examples were provided to fully verify the accuracy of the modified criterion.

    Citation: Yang Xu, Zhouping Yin, Yuanzhi Wang, Qi Liu, Anwarud Din. Mittag-Leffler projective synchronization of uncertain fractional-order fuzzy complex valued neural networks with distributed and time-varying delays[J]. AIMS Mathematics, 2024, 9(9): 25577-25602. doi: 10.3934/math.20241249

    Related Papers:

    [1] Kaixuan Wang, Shixiong Zhang, Yang Cao, Lu Yang . Weakly supervised anomaly detection based on sparsity prior. Electronic Research Archive, 2024, 32(6): 3728-3741. doi: 10.3934/era.2024169
    [2] Peng Zhi, Haoran Zhou, Hang Huang, Rui Zhao, Rui Zhou, Qingguo Zhou . Boundary distribution estimation for precise object detection. Electronic Research Archive, 2023, 31(8): 5025-5038. doi: 10.3934/era.2023257
    [3] Bojian Chen, Wenbin Wu, Zhezhou Li, Tengfei Han, Zhuolei Chen, Weihao Zhang . Attention-guided cross-modal multiple feature aggregation network for RGB-D salient object detection. Electronic Research Archive, 2024, 32(1): 643-669. doi: 10.3934/era.2024031
    [4] Qing Tian, Heng Zhang, Shiyu Xia, Heng Xu, Chuang Ma . Cross-view learning with scatters and manifold exploitation in geodesic space. Electronic Research Archive, 2023, 31(9): 5425-5441. doi: 10.3934/era.2023275
    [5] Yixin Sun, Lei Wu, Peng Chen, Feng Zhang, Lifeng Xu . Using deep learning in pathology image analysis: A novel active learning strategy based on latent representation. Electronic Research Archive, 2023, 31(9): 5340-5361. doi: 10.3934/era.2023271
    [6] Jianjun Huang, Xuhong Huang, Ronghao Kang, Zhihong Chen, Junhan Peng . Improved insulator location and defect detection method based on GhostNet and YOLOv5s networks. Electronic Research Archive, 2024, 32(9): 5249-5267. doi: 10.3934/era.2024242
    [7] Hui-Ching Wu, Yu-Chen Tu, Po-Han Chen, Ming-Hseng Tseng . An interpretable hierarchical semantic convolutional neural network to diagnose melanoma in skin lesions. Electronic Research Archive, 2023, 31(4): 1822-1839. doi: 10.3934/era.2023094
    [8] Jinjiang Liu, Yuqin Li, Wentao Li, Zhenshuang Li, Yihua Lan . Multiscale lung nodule segmentation based on 3D coordinate attention and edge enhancement. Electronic Research Archive, 2024, 32(5): 3016-3037. doi: 10.3934/era.2024138
    [9] Hui Yao, Yaning Fan, Xinyue Wei, Yanhao Liu, Dandan Cao, Zhanping You . Research and optimization of YOLO-based method for automatic pavement defect detection. Electronic Research Archive, 2024, 32(3): 1708-1730. doi: 10.3934/era.2024078
    [10] Zhongnian Li, Jiayu Wang, Qingcong Geng, Xinzheng Xu . Group-based siamese self-supervised learning. Electronic Research Archive, 2024, 32(8): 4913-4925. doi: 10.3934/era.2024226
  • To study the Mittag-Leffler projective synchronization (MLPS) problem of fractional-order fuzzy neural networks (FOFNNs), in this work we introduced the FOFNNs model. On this basis, we discussed the MLPS of uncertain fractional-order fuzzy complex valued neural networks (FOFCVNNs) with distributed and time-varying delays. Utilizing Banach contraction mapping principle, we proved the existence and uniqueness of the model solution. Moreover, employing the construction of a new hybrid controller, an adaptive hybrid controller, and the fractional-order Razumikhin theorem, algebraic criteria was obtained for implementing MLPS. The algebraic inequality criterion obtained in this article improves and extends the previously published results on MLPS, making it easy to prove and greatly reducing the computational complexity. Finally, different Caputo derivatives of different orders were given, and four numerical examples were provided to fully verify the accuracy of the modified criterion.



    In this paper, we consider connected simple and finite graphs, and refer to Bondy and Murty [4] for notation and terminologies used but not defined here.

    Let G be a connected graph with vertex set V(G) and edge set E(G), |V(G)|=n and |E(G)|=m. We denote Gv and Guv the graph obtained from G by deleting a vertex vV(G), or an edge uvE(G), respectively. Similarly, G+uv is obtained from G by adding an edge uvE(G), where u,vV(G). An edge uv of a graph G is called a cut edge if the graph Guv is disconnected. For a vertex uV(G), its degree dG(u) is equal to the number of vertices in G adjacent to u; the neighborhood of u is denoted by NG(u), or N(u) for short. The symbols Δ(G) and δ(G) represent the maximum degree and the minimum degree of G. We use Tn, Cn, Pn and Sn to denote the tree, cycle, path and star of order n, respectively.

    Gutman[13] defined a new vertex-degree-based graph invariant, called Sombor index. Precisely, for a graph G, it is denoted by SO(G) and is defined as

    SO(G)=uvE(G)dG(u)2+dG(v)2.

    He proved that for any tree T with n3 vertices,

    22(n3)+25SO(T)(n1)n22n+2,

    with left side of equality if and only if TPn, and with the right side of equality if and only if TSn. Chen etal. [5] determined the extremal values of the Sombor index of trees with some given parameters, including matching number, pendant vertices, diameter, segment number, branching number, etc. The corresponding extremal trees are characterized completely. Deng etal. [11] obtained a sharp upper bound for the Sombor index among all molecular trees with fixed numbers of vertices, and characterize those molecular trees achieving the extremal value. Cruz etal. [8] determined the extremal values of Sombor indices over trees with at most three branch vertices. Li etal. [19] give sharp bounds for the Sombor index of trees with a given diameter. Das and Gutman [9] present bounds on the Sombor index of trees in terms of order, independence number, and number of pendent vertices, and characterize the extremal cases. In addition, analogous results for quasi-trees are established. Sun and Du [30] present the maximum and minimum Sombor indices of trees with fixed domination number, and identified the corresponding extremal trees. Zhou etal. [36] determined the graph with minimum Sombor index among all trees with given number of vertices and maximum degree, respectively, among all unicyclic graphs with given number of vertices and maximum degree.

    Cruz and Rada [7] investigate the Sombor indices of unicyclic and bicyclic graphs. Let U(n,p,q,r)G(n,1), where pqr0 and p+q+r=n3, be a unicyclic graph obtained from 3-cycle C3 with V(C3)={u,v,w}, adding p, q and r pendent vertices to the vertices u, v and w, respectively. They showed that for a unicyclic graph with n3 vertices,

    22nSO(G)(n3)(n1)2+1+2(n1)2+22+22.

    The lower and upper bound is uniquely attained by GCn and GU(n,n3,0,0), respectively. Alidadi etal. [2] gave the minimum Sombor index for unicyclic graphs with the diameter D2.

    Aashtab etal. [1] studied the structure of a graph with minimum Sombor index among all graphs with fixed order and fixed size. It is shown that in every graph with minimum Sombor index the difference between minimum and maximum degrees is at most 1. Cruz etal. [6] characterize the graphs extremal with respect to the Sombor index over the following families of graphs: (connected) chemical graphs, chemical trees, and hexagonal systems. Liu etal. [21] determined the minimum Sombor indices of tetracyclic (chemical) graphs. Das and Shang [10] present some lower and upper bounds on the Sombor index of graph G in terms of graph parameters (clique number, chromatic number, number of pendant vertices, etc.) and characterize the extremal graphs. For the Sombor index of a connected graph with given order, Horoldagva and Xu [16] presented sharp upper and lower bounds when its girth is fixed, a lower bound if its maximum degree is given and an upper bound in terms of given number of pendent vertices or pendent edges, respectively. In [29], Shang observe power-law and small-word effect for the simplicial networks and examine the effectiveness of the approximation method for Sombor index through computational experiments.

    Relations between the Sombor index and some other well-known degree-based descriptors [14,24,25,33]. A number of application of Sombor index in chemistry were reported in [3,20,27]. Besides, the relationship between the energy and Sombor index of a graph G is studied in [12,15,26,28,31,32].

    Some variations of Sombor index, for instance, the reduced Sombor index, average Sombor index, are investigated. Redžepović [27] examined the predictive and discriminative potentials of Sombor index, the reduced Sombor index, average Sombor index. Liu etal. [23] obtained some bounds for reduced Sombor index of graphs with given several parameters (such as maximum degree, minimum degree, matching number, chromatic number, independence number, clique number), some special graphs (such as unicyclic grahs, bipartite graphs, graphs with no triangles, graphs with no Kr+1 and the Nordhaus-Gaddum-type results). A conjecture related to the chromatic number in the above paper was verified to be true by Wang and Wu [34]. Liu etal. [22] ordered the chemical trees, chemical unicyclic graphs, chemical bicyclic graphs and chemical tricyclic graphs with respect to Sombor index and reduced Sombor index. Furthermore, they determined the first fourteen minimum chemical trees, the first four minimum chemical unicyclic graphs, the first three minimum chemical bicyclic graphs, the first seven minimum chemical tricyclic graphs. Finally, the applications of reduced Sombor index to octane isomers were given. Wang and Wu [35] investigated the reduced Sombor index and the exponential reduced Sombor index of a molecular tree solving a conjecture [23] and an open problem [11].

    A vertex of degree 1 is said to be a pendant vertex. Further, an edge is said to be a pendant edge if one of its end vertices is a pendant vertex. A connected graph that has no cut vertices is called a block, the blocks of G which correspond to leaves of its block tree are referred to as its end blocks. A cactus is a connected graph in which every block is either an edge or a cycle. Let G(n,k) be the family of all cacti with n vertices and k cycles. Clearly, |E(G)|=n+k1 for any GG(n,k). Note that G(n,0) is the set of all trees and G(n,1) is the set of all unicyclic graphs. Gutman [13] characterized the tree with extremal value Sombor index.

    It is our main concern in this paper to study the extremal value problem of Sombor index on G(n,k), k2. In this paper, we will determine the maximum Sombor index of graphs among GG(n,k), and also characterize the corresponding extremal graphs. Later, we will determine the minimum Sombor index of graphs with given conditions among GG(n,k), and also characterize the corresponding extremal graphs.

    Let G0(n,k)G(n,k) be a bundle of k triangles with n2k1 pendent vertices attached to the common vertex, as illustrated in Figure 1.

    Figure 1.  G0(n,k).

    By a simple computation, we have SO(G0(n,k))=(n2k1)(n1)2+1 +2k(n1)2+22+22k. We will see that G0(n,k) has the maximum Sombor index among G(n,k).

    Theorem 1.1. Let k1 and n3. For any GG(n,k),

    SO(G)(n2k1)(n1)2+1+2k(n1)2+22+22k,

    with equality if and only if GG0(n,k).

    Let C(n,k) denote the set of the elements G of G(n,k) with the following properties:

    (1) δ(G)=2 and Δ(G)=3;

    (2) a vertex is a cut vertex if and only if it has degree 3, and there are exactly 2k2 cut vertices;

    (3) at least k32 internal cycles with all three degrees are triangles;

    (4) at most one vertex not belong to any cycle;

    (5) the three degree vertices on the cycle are adjacent.

    Generally speaking, if k is even, an element of C(n,k) obtained from a tree T of order k with each vertex having degree 1 or 3 by replacing each vertex of degree 3 with a triangle and replacing each vertex of degree 1 with a cycle. If k is odd, an element of C(n,k) obtained from a tree T of order k with exactly a vertex having degree 2 by replacing two (adjacent) vertices of degree 3 with a cycle, and other vertices having degree 1 or 3 by replacing each vertex of degree 3 with a triangle, replacing each vertex of degree 1 with a cycle or an element of C(n,k) obtained from a tree T of order k with each vertex having degree 1 or 3 by retention one vertex of degree 3, and by replacing other vertices of degree 3 with a triangle and replacing each vertex of degree 1 with a cycle.

    Three elements of C(n,k) are shown in terms of the parity of k in Figure 2.

    Figure 2.  Three cacti in C(n,k) when n3k.

    Theorem 1.2. Let k2 and n6k3. For any GG(n,k), we have

    SO(G)22n+52(k22)+213(kk2+1),

    with equality if and only if GC(n,k).

    The proofs of Theorems 1.1 and 1.2 are given Sections 2 and 3, respectively.

    In this section, we will determine the maximum value of the Sombor index of cacti with n vertices and k cycles, and characterize the corresponding extremal graph. We start with several known results, which will be used in the proof of Theorem 1.1.

    Lemma 2.1 (Horoldagva and Xu [16]). If uv is a non-pendent cut edge in a connected graph G, then SO(G)>SO(G), where G is the graph obtained by the contraction of uv onto the vertex u and adding a pendent vertex v to u.

    In 1932, Karamata proved an interesting result, which is now known as the majorization inequality or Karamata's inequailty. Let A=(a1,a2,,an) and B=(b1,b2,,bn) be non-increasing two sequences on an interval I of real numbers such that a1+a2++an=b1+b2++bn. If a1+a2++aib1+b2++bi for all 1in1 then we say that A majorizes B.

    Lemma 2.2 (Karamata[18]). Let f:IR be a strictly convex function. Let A=(a1,a2,,an) and B=(b1,b2,,bn) be non-increasing sequences on I. If A majorizes B then f(a1)+f(a2)++f(an)f(b1)+f(b2)++f(bn) with equality if and only if ai=bi for all 1in.

    Lemma 2.3. Let G be a connected graph with a cycle Cp=v1v2vpv1 (p4) such that GE(Cp) has exactely p components G1,G2,,Gp, where Gi is the component of GE(Cp) containing vi for each i{1,2,,p}, as shown in Figure 3. If G=G{vp1vp,uvp| uNGp(vp)}+{v1vp1,uv1| uNGp(vp)}, then SO(G)>SO(G).

    Figure 3.  The graph G.

    Proof. Let NG(v1){v2,vp}={x1,x2,,xs}, NG(vp){v1,vp1}={y1,y2,,yt}, where s=dG(v1)2 and t=dG(vp)2. Hence, by the definition of SO(G), we obtain

    SO(G)SO(G)=si=1dG(xi)2+dG(v1)2si=1dG(xi)2+dG(v1)2+tj=1dG(yj)2+dG(v1)2tj=1dG(yj)2+dG(vp)2+dG(v1)2+dG(v2)2dG(v1)2+dG(v2)2+dG(v1)2+dG(vp1)2dG(vp)2+dG(vp1)2+dG(v1)2+12dG(v1)2+dG(vp)2=si=1dG(xi)2+(s+t+3)2si=1dG(xi)2+(s+2)2+tj=1dG(yj)2+(s+t+3)2tj=1dG(yj)2+(t+2)2+(s+t+3)2+dG(v2)2(s+2)2+dG(v2)2+(s+t+3)2+dG(vp1)2(t+2)2+dG(vp1)2+(s+t+3)2+12(s+2)2+(t+2)2>(s+t+3)2+12(s+2)2+(t+2)2.

    Since s0,t0, we have [(s+t+3)2+12][(s+2)2+(t+2)2]=2st+4s+2t+2>0, implying that SO(G)>SO(G).

    Lemma 2.4. Let n and k be two nonnegative integers with n2k+1. If GG(n,k) has a triangle v1v2v3v1 with dG(v3)dG(v2)3 as shown in Figure 4, then SO(G)>SO(G), where G=G{v2u| uNG2(v2)}+{v3u| uNG2(v2)}.

    Figure 4.  The graphs G and G.

    Proof. Let NG(v3){v2}={v1,x1,x2,,xs}, NG(v2){v3}={v1,y1,y2,,yt}, then s=dG(v3)2, t=dG(v2)2. Defined according to G, dG(v3)=s+t+2. Assume that dG(v1)=dG(v1)=p, (s+t+2)2+22+(s+2)2+(t+2)2=q.

    SO(G)SO(G)=si=1(dG(v3)2+dG(xi)2dG(v3)2+dG(xi)2)+tj=1(dG(v3)2+dG(yj)2dG(v2)2+dG(yj)2)+(dG(v3)2+dG(v1)2dG(v3)2+dG(v1)2)+(dG(v2)2+dG(v3)2dG(v2)2+dG(v3)2)+(dG(v1)2+dG(v2)2dG(v1)2+dG(v2)2)=si=1(s+t+2)2+dG(xi)2si=1(s+2)2+dG(xi)2+tj=1(s+t+2)2+dG(yj)2tj=1(t+2)2+dG(yj)2+(s+t+2)2+dG(v1)2(s+2)2+dG(v1)2+(s+t+2)2+22(s+2)2+(t+2)2+dG(v1)2+22dG(v1)2+(t+2)2>p2+(s+t+2)2+p2+22p2+(s+2)2p2+(t+2)2+(s+t+2)2+22(s+2)2+(t+2)2=p(1+(s+t+2p)2+1+(2p)21+(s+2p)21+(t+2p)2)+2stq. (2.1)

    Let us consider a function f(x)=1+x2 and it is easy to see that this function is strictly convex for x[0,+). Since A={s+t+2p, 2p} majorizes B={s+2p, t+2p}, By Karamata's inequality, f(s+t+2p)+f(2p)>f(s+2p)+f(t+2p). Combining this with (2.1), it follows that SO(G)>SO(G).

    Now, we are ready to present the proof of Theorem 1.1.

    Proof of Theorem 1.1:

    Let G be a cactus with the maximum Sombor index value among G(n,k). By Lemma 2.1, each cut edge of G is pendent. By Lemma 2.3, every cycles of G is a triangle. Furthermore, by Lemma 2.4, GG0(n,k). Thus, the maximum Sombor index of cacti among G(n,k) is

    SO(G)=(n2k1)(n1)2+1+2k(n1)2+22+22k.

    In this section, we determine the minimum Sombor index of graphs in G(n,k), and characterize the corresponding extremal graphs.

    First, we introduce some additional notations. For a graph G, Vi(G)={vV(G)| d(v)=i}, ni=|Vi(G)|, Ei,j(G)={uvE(G)| d(u)=i, d(v)=j} and ei,j(G)=|Ei,j(G)|. Obviously, ei,j(G)=ej,i(G). If there is no confusion, ei,j(G) is simply denoted by ei,j. For any simple graph G of order n, we have

    n=n1+n2++nn1, (3.1)

    and

    {2e1,1+e1,2++e1,n1=n1e2,1+2e2,2++e2,n1=2n2en1,1+en1,2++2en1,n1=(n1)nn1 (3.2)

    Let

    Ln,n={(i,j)| i,jN,1ijn1}.

    it follows easily from (3.1) and (3.2) that

    n=(i,j)Ln,ni+jijei,j, (3.3)

    Let GG(n,k) with n2k+1 and k1. It implies that e1,1(G)=0 and ei,j(G)=0 for any 1ijn1 with i+j>n+k. Let Lkn,n={(i,j)Ln,n:i+jn+k}, Lkn,n=Lkn,n{(2,2),(2,3),(3,3)}. By a simple calculation we obtain the following result.

    Lemma 3.1. For any graph GG(n,k) (k1),

    SO(G)=22n+(613102)(k1)+(52213)e3,3+(i,j)Lkn,ng(i,j)ei,j,

    where

    g(i,j)=i2+j2(122613)i+jij+(102613). (3.4)

    Proof. For any GG(n,k),

    n=(i,j)Lkn,ni+jijei,j, (3.5)
    n+k1=(i,j)Lkn,nei,j. (3.6)

    Relations (3.5) and (3.6) can be rewritten as

    5e2,3+4e3,3=6n6e2,26(i,j)Lkn,ni+jijei,j,
    e2,3+e3,3=n+k1e2,2(i,j)Lkn,nei,j.

    Combining the above, we have

    e2,3=6k62e3,3+(i,j)Lkn,n(6i+jij6)ei,j, (3.7)
    e2,2=n5k+5+e3,3(i,j)Lkn,n(6i+jij5)ei,j. (3.8)

    g(2,2)=g(2,3)=0, g(3,3)=(52213)<0. Thus,

    SO(G)=13e2,3+32e3,3+22e2,2+(i,j)Lkn,ni2+j2ei,j=22n+(613102)(k1)+(52213)e3,3+(i,j)Lkn,ng(i,j)ei,j. (3.9)

    Lemma 3.2 (Chen, Li, Wang [5]). Let f(x,y)=x2+y2 and h(x,y)=f(x,y)f(x1,y), where x,y1. If x,y1, then h(x,y) strictly decreases with y for fixed x and increases with x for fixed y.

    Since f(x+k,y)f(x,y)=ki=1[f(x+i,y)f(x+i1,y)]=ki=1h(x+i,y) for any kZ+, we have the following corollary.

    Corollary 3.1. If x,y1 and kZ+, then f(x+k,y)f(x,y) strictly decreases with y for fixed x and increases with x for fixed y.

    Let Pl=u0u1ul, l1 be a path of G with d(u0)3, d(ui)=2 for 1il1 when l>1. We call Pl an internal path if d(ul)3, and a pendent path if d(ul)=1.

    Lemma 3.3. Let G be a cactus graph of order n4. If there exists two edges uu,v1v2E(G) such that d(u)=1 and min{dG(v1), dG(v2)}2. Let G=Guuv1v2+uv1+uv2, then SO(G)<SO(G).

    Proof. Let NG(u)={u,w1,w2,,wt1}, where t=d(u)(t2). By the assumption, dG(u)=2 and dG(u)=t1. Since G is a cactus graph, then d(wi)nt+1(i=1,,t1), d(vj)nt+1(j=1,2), t<n1. Thus,

    SO(G)SO(G)=[t1i=1f(t,d(wi))+f(t,1)+f(d(v1),d(v2))][t1i=1f(t1,d(wi))+f(2,d(v1))+f(2,d(v2))]=t1i=1h(t,d(wi))+f(t,1)f(2,d(v1))+[f(d(v1),d(v2))f(2,d(v2))](t1)h(t,nt+1)+f(t,1)f(2,d(v1))+f(d(v1),nt+1)f(2,nt+1)(t1)h(t,nt+1)+f(t,1)+f(nt+1,nt+1)2f(2,nt+1)h(2,nt+1)+f(2,1)+f(nt+1,nt+1)2f(2,nt+1)=22+(nt+1)212+(nt+1)2+5+(nt+1)2222+(nt+1)2=5+(nt+1)(212+(1nt+1)2(2nt+1)2+12)>5+2(212+(12)2(22)2+12)=0.

    A repeated application of the above lemma result in the following consequence.

    Corollary 3.2. If G is a cactus has a pendent path Pl=u0u1ul with d(ul)=1 and v1v2E(G), min{dG(v1), dG(v2)}2, then SO(G)<SO(G), where G=Gu0u1v1v2+u1v1+ulv2.

    The following result is immediate from the above corollary.

    Corollary 3.3. Let k1. If G is a cactus has the minimum Sombor index among G(n,k), then δ(G)2.

    The following result is due to Jiang and Lu [17], which is a key lemma in the proof of Theorem 1.2.

    Lemma 3.4 (Jiang and Lu [17]). Let k and n be two integers with k2 and n6k4. If GG(n,k) with δ(G)2, then there exists a path x1x2x3x4 of length 3 in G such that d(x2)=d(x3)=2 and x1x4.

    Lemma 3.5. Let k and n be two integers with k2 and n6k3. If G has the minimum Sombor index among G(n,k), then Δ(G)3.

    Proof. By contradiction, suppose that Δ(G)4. Let vV(G) with dG(v)=Δ(G). Since n6k3, by Lemma 3.4, there exists a path x1x2x3x4 in G such that dG(x2)=dG(x3)=2 and x1x4. Let G1=Gx1x2x2x3+x1x3. Clearly, G1G(n1,k) and dG1(u)=dG(u) for all uV(G1).

    Since n16k4, by Lemma 3.4, there exists a path y1y2y3y4 in G1 such that dG1(y2)=dG1(y3)=2 and y1y4. Let G2=G1y1y2y2y3+y1y3. Then G2G(n2,k) and dG2(u)=dG1(u) for all uV(G2). Since dG(v)4, v{x2,y2} and dG(v)=dG1(v)=dG2(v).

    For convenience, let t=Δ(G). Let NG2(v)={w1,,wt}. Assume that w1,w2, v are in the same block if v is contained in a cycle in G2. Let G=G2vw1vw2+vx2+x2y2+y2w1+y2w2. Then GG(n,k), dG(v)=t1, dG(x2)=2, dG(y2)=3 and dG(u)=dG2(u)=dG(u)2 for all uV(G2){v}. Next, by showing SO(G)<SO(G), we arrive at a contradiction.

    By the construction above, SO(G1)=SO(G)22+22=SO(G)22 and SO(G2)=SO(G1)22+22=SO(G)42. Thus,

    SO(G)SO(G)=SO(G2)SO(G)+42=2i=1[f(dG(v),dG(wi))f(dG(y2),dG(wi))]+ti=3[f(dG(v),dG(wi))f(dG(v),dG(wi))]+42f(dG(v),dG(x2))f(dG(x2),dG(y2))=2i=1[f(t,dG(wi))f(3,dG(wi))]+ti=3[f(t,dG(wi))f(t1,dG(wi))]+42f(t1,2)f(2,3)2[f(t,t)f(3,t)]+(t2)[f(t,t)f(t1,t)]+42f(t1,2)f(2,3)=2t2232+t2(t2)(t1)2+t2(t1)2+22+4213.

    Hence, to show SO(G)<SO(G), it suffices to show that f(t)>0 for t4, where f(t)=2t2232+t2(t2)(t1)2+t2(t1)2+22+4213. One can see that for any t4,

    f(t)=t(222t2+321(t1)2+2221+(11t)2)+5tt2+(t1)2t2+(t1)2+1(t1)2+222t2+(t1)2t(222511385)+5tt2+(t1)2t2+(t1)2+1(t1)2+222t2+(t1)2=(222113)t+5t5t2+(t1)2+3t2+(t1)2t2+(t1)2+1(t1)2+22=(222113)t+51+(1+1t1)2(t1)1+(1+1t1)2+3t2+(t1)2+1(t1)2+22>4(222113)+35>0.

    Hence, f(t) is an increasing function with respect to t[4,n1], implying f(t)f(4)=20241310>0. This contradicts the minimality of G.

    Lemma 3.6. Let k2 and n6k3. If G has the minimum Sombor index in G(n,k), then it does not exist a path v1v2vl(l3) in G such that dG(v1)=dG(vl)=3 and dG(vi)=2 (i=2,,l1), where v1 and vl are not adjacent. Thus,

    (1) if a cycle C is not an end block of G, then all vertices of it have degree three in G, or it contains exactly two (adjacent) vertices of degree three.

    (2) any vertices of degree two lie on a cycle.

    Proof. Suppose that there exist a path Pl=v1v2vlG as given in the assumption of the lemma. From Corollary 3.3 and Lemmas 3.5, we have 2d(v)3 of any vertex v in G. By Lemma 3.1, in equation (3.9), Lkn,n=,

    SO(G)=22n+(613102)(k1)+(52213)e3,3(G).

    Let Cs be an end block of G, w1,w2V(Cs), w1w2E(Cs), dG(w1)=dG(w2)=2. Let G=Gv1v2vl1vl+v1vlw1w2+w1v2+w2vl1. Clearly,

    SO(G)=22n+(613102)(k1)+(52213)(e3,3(G)+1).

    Thus, SO(G)<SO(G), a contradiction.

    Thus, (1) and (2) is immediate.

    Lemma 3.7. Let k2 and n6k3. Let GG(n,k) has the minimum Sombor index, then e3,3(G)5k24=2k+k24, the equality holds if and only if GC(n,k).

    Proof. By Corollary 3.3 and Lemma 3.5, we have 2d(v)3 of any vertex v in G.

    Let n3 be the number of vertices with degree 3 not belongs to a cycle, c1 the number of end blocks, and c2 the number of cycles with exactly two (adjacent) vertices of degree three in G. Clearly n30 and c20. By Lemma 3.6 (1), there are kc1c2 remaining cycles, denoted by C1,,Ckc1c2, all vertices of which have degree three. Let di be the length of the cycle Ci.

    Let Tk+n3 be the tree obtained from contracting each cycle of G into a vertex. By the hand-shaking lemma, we have

    3n3+d1+d2++dkc1c2+2c2+c1=2(k+n31) (3.10)

    Since di3 for each i{1,,kc1c2}, by (3.10), we have

    2kn3c12c22=d1+d2++dkc1c23(kc1c2), (3.11)

    relation (3.11) can be rewritten as

    (kc1c2)+kn3c223(kc1c2), (3.12)

    implying that

    kc1c2k22n3+c22. (3.13)

    On the other hand, by Lemma 3.6,

    e3,3(G)=(k+n31)+c2+(d1+d2++dkc1c2). (3.14)

    It follows from (3.10) and (3.14) that

    e3,3(G)=2k+(kc1c2)3. (3.15)

    Combining (3.13) and (3.15), it yields

    e3,3(G)(2k3)+(k22n3+c22)=5k24n3+c22.

    Since n30, c20,

    e3,3(G)5k24. (3.16)

    If k is even,

    e3,3(G)5k24,

    the equality holds if and only if n3=c2=0, d1=d2==dkc1c2=3, that is, GC(n,k).

    If k is odd,

    e3,3(G)5k292,

    the equality holds if and only if n3+c2=1, d1=d2==dkc1c2=3, then n3=0,c2=1 or n3=1,c2=0, that is, GC(n,k).

    Proof of Theorem 1.2: Assume that under which k2 and n6k3 condition G has the minimum Sombor index in G(n,k). By Corollary 3.3 and Lemma 3.5, 2dG(v)3 for any vertex v in G. By Lemma 3.1,

    SO(G)=22n+(613102)(k1)+(52213)e3,3(G). (3.17)

    Thus, by Lemma 3.7,

    SO(G)22n+52(k22)+213(kk2+1),

    with equality if and only if GC(n,k).

    Recall that G(n,k) denotes the set of cacti of order n and with k cycles. In this paper, we establish a sharp upper bound for the Sombor index of a cactus in G(n,k) and characterize the corresponding extremal graphs. In addition, for the case when n6k3, we give a sharp lower bound for the Sombor index of a cactus in G(n,k) and characterize the corresponding extremal graphs as well. We believe that Theorem 1.2 is true for the case when 3kn6k4.

    Conjucture 4.1. Let k and n be two integers with n3k and k2. For any graph GG(n,k),

    SO(G)22n+52(kk22)+213(k2+1),

    with equality if and only if GC(n,k).

    The research of cactus graph is supported by the National Natural Science Foundation of China(11801487, 12061073).

    The authors declare no conflict of interest.



    [1] E. Viera-Martin, J. F. Gomez-Aguilar, J. E. Solis-Perez, J. A. Hernandez-Perez, R. F. Escobar-Jimenez, Artificial neural networks: a practical review of applications involving fractional calculus, Eur. Phys. J. Spec. Top., 231 (2022), 2059–2095. https://doi.org/10.1140/epjs/s11734-022-00455-3 doi: 10.1140/epjs/s11734-022-00455-3
    [2] J. T. Machado, V. Kiryakova, F. Mainardi, Recent history of fractional calculus, Commun. Nonlinear Sci. Numer. Simul., 16 (2011), 1140–1153. https://doi.org/10.1016/j.cnsns.2010.05.027 doi: 10.1016/j.cnsns.2010.05.027
    [3] S. P. Wen, Z. G. Zeng, T. W. Huang, Q. G. Meng, W. Yao, Lag synchronization of switched neural networks via neural activation function and applications in image encryption, IEEE Trans. Neural Netw. Learn. Syst., 26 (2015), 1493–1502. https://doi.org/10.1109/TNNLS.2014.2387355 doi: 10.1109/TNNLS.2014.2387355
    [4] N. Ghosh, A. Garg, B. K. Panigrahi, J. Kim, An evolving quantum fuzzy neural network for online state-of-health estimation of Li-ion cell, Appl. Soft Comput., 143 (2023), 110263. https://doi.org/10.1016/j.asoc.2023.110263 doi: 10.1016/j.asoc.2023.110263
    [5] Y. Zhang, X. P. Wang, E. G. Friedman, Memristor-based circuit design for multilayer neural networks, IEEE Trans. Circuits Syst. I Regul. Pap., 65 (2018), 677–686. https://doi.org/10.1109/TCSI.2017.2729787 doi: 10.1109/TCSI.2017.2729787
    [6] C. J. Wang, Z. L. Xu, An intelligent fault diagnosis model based on deep neural network for few-shot fault diagnosis, Neurocomputing, 456 (2021), 550–562. https://doi.org/10.1016/j.neucom.2020.11.070 doi: 10.1016/j.neucom.2020.11.070
    [7] G. R. Murthy, Toward optimal synthesis of discrete-time Hopfield neural network, IEEE Trans. Neur. Netw. Learn. Syst., 34 (2023), 9549–9554. https://doi.org/10.1109/TNNLS.2022.3156107 doi: 10.1109/TNNLS.2022.3156107
    [8] H. Zhang, Z. G. Zeng, Synchronization of recurrent neural networks with unbounded delays and time-varying coefficients via generalized differential inequalities, Neural Netw., 143 (2021), 161–170. https://doi.org/10.1016/j.neunet.2021.05.022 doi: 10.1016/j.neunet.2021.05.022
    [9] S. L. Chen, H. L. Li, Y. G. Kao, L. Zhang, C. Hu, Finite-time stabilization of fractional-order fuzzy quaternion-valued BAM neural networks via direct quaternion approach, J. Franklin Inst., 358 (2021), 7650–7673. https://doi.org/10.1016/j.jfranklin.2021.08.008 doi: 10.1016/j.jfranklin.2021.08.008
    [10] Z. J. Zhang, T. T. Yu, X. Zhang, Algebra criteria for global exponential stability of multiple time-varying delay Cohen-Grossberg neural networks, Appl. Math. Comput., 435 (2022), 127461. https://doi.org/10.1016/j.amc.2022.127461 doi: 10.1016/j.amc.2022.127461
    [11] H. Zhang, Y. H. Cheng, W. W. Zhang, H. M. Zhang, Time-dependent and Caputo derivative order-dependent quasi-uniform synchronization on fuzzy neural networks with proportional and distributed delays, Math. Comput. Simul., 203 (2023), 846–857. https://doi.org/10.1016/j.matcom.2022.07.019 doi: 10.1016/j.matcom.2022.07.019
    [12] F. Zhao, J. G. Jian, B. X. Wang, Finite-time synchronization of fractional-order delayed memristive fuzzy neural networks, Fuzzy Sets Syst., 467 (2023), 108578. https://doi.org/10.1016/j.fss.2023.108578 doi: 10.1016/j.fss.2023.108578
    [13] F. F. Du, J. G. Lu, Adaptive finite-time synchronization of fractional-order delayed fuzzy cellular neural networks, Fuzzy Sets Syst., 466 (2023), 108480. https://doi.org/10.1016/j.fss.2023.02.001 doi: 10.1016/j.fss.2023.02.001
    [14] H. L. Li, J. D. Cao, C. Hu, L. Zhang, H. J. Jiang, Adaptive control-based synchronization of discrete-time fractional-order fuzzy neural networks with time-varying delays, Neural Netw., 168 (2023), 59–73. https://doi.org/10.1016/j.neunet.2023.09.019 doi: 10.1016/j.neunet.2023.09.019
    [15] J. T. Fei, Z. Wang, Q. Pan, Self-constructing fuzzy neural fractional-order sliding mode control of active power filter, IEEE Trans. Neural Netw. Learn. Syst., 34 (2023), 10600–10611. https://doi.org/10.1109/TNNLS.2022.3169518 doi: 10.1109/TNNLS.2022.3169518
    [16] T. Nitta, Orthogonality of decision boundaries in complex-valued neural networks, Neural Comput., 16 (2004), 73–97. https://doi.org/10.1162/08997660460734001 doi: 10.1162/08997660460734001
    [17] I. Cha, S. A. Kassam, Channel equalization using adaptive complex radial basis function networks, IEEE J. Sel. Areas Commun., 13 (1995), 122–131. https://doi.org/10.1109/49.363139 doi: 10.1109/49.363139
    [18] H. L. Li, C. Hu, J. D. Cao, H. J. Jiang, A. Alsaedi, Quasi-projective and complete synchronization of fractional-order complex-valued neural networks with time delays, Neural Netw., 118 (2019), 102–109. https://doi.org/10.1016/j.neunet.2019.06.008 doi: 10.1016/j.neunet.2019.06.008
    [19] X. L. Zhang, H. L. Li, Y. G. Yu, L. Zhang, H. J. Jiang, Quasi-projective and complete synchronization of discrete-time fractional-order delayed neural networks, Neural Netw., 164 (2023), 497–507. https://doi.org/10.1016/j.neunet.2023.05.005 doi: 10.1016/j.neunet.2023.05.005
    [20] S. Yang, H. J. Jiang, C. Hu, J. Yu, Synchronization for fractional-order reaction-diffusion competitive neural networks with leakage and discrete delays, Neurocomputing, 436 (2021), 47–57. https://doi.org/10.1016/j.neucom.2021.01.009 doi: 10.1016/j.neucom.2021.01.009
    [21] Z. Y. Yang, J. Zhang, J. H. Hu, J. Mei, New results on finite-time stability for fractional-order neural networks with proportional delay, Neurocomputing, 442 (2021), 327–336. https://doi.org/10.1016/j.neucom.2021.02.082 doi: 10.1016/j.neucom.2021.02.082
    [22] C. Chen, L. X. Li, H. P. Peng, Y. X. Yang, Fixed-time synchronization of inertial memristor-based neural networks with discrete delay, Neural Netw., 109 (2019), 81–89. https://doi.org/10.1016/j.neunet.2018.10.011 doi: 10.1016/j.neunet.2018.10.011
    [23] M. Hui, N. Yao, H. H. C. Iu, R. Yao, L. Bai, Adaptive synchronization of fractional-order complex-valued neural networks with time-varying delays, IEEE Access, 10 (2022), 45677–45688. https://doi.org/10.1109/ACCESS.2022.3170091 doi: 10.1109/ACCESS.2022.3170091
    [24] G. L. Chen, D. S. Li, L. Shi, O. van Gaans, S. V. Lunel, Stability results for stochastic delayed recurrent neural networks with discrete and distributed delays, J. Differ. Equ., 264 (2018), 3864–3898. https://doi.org/10.1016/j.jde.2017.11.032 doi: 10.1016/j.jde.2017.11.032
    [25] L. Z. Si, M. Xiao, G. P. Jiang, Z. S. Cheng, Q. K. Song, J. D. Cao, Dynamics of fractional-order neural networks with discrete and distributed delays, IEEE Access, 8 (2019), 46071–46080. https://doi.org/10.1109/ACCESS.2019.2946790 doi: 10.1109/ACCESS.2019.2946790
    [26] Z. Y. Yang, J. Zhang, J. H. Hu, J. Mei, Some new Gronwall-type integral inequalities and their applications to finite-time stability of fractional-order neural networks with hybrid delays, Neural Process. Lett., 55 (2023), 11233–11258. https://doi.org/10.1007/s11063-023-11373-3 doi: 10.1007/s11063-023-11373-3
    [27] W. Q. Gong, J. L. Liang, C. J. Zhang, Multistability of complex-valued neural networks with distributed delays, Neural Comput. Appl., 28 (2017), 1–14. https://doi.org/10.1007/s00521-016-2305-9 doi: 10.1007/s00521-016-2305-9
    [28] Y. J. Gu, Y. G. Yu, H. Wang, Synchronization for fractional-order time-delayed memristor-based neural networks with parameter uncertainty, J. Franklin Inst., 353 (2016), 3657–3684. https://doi.org/10.1016/j.jfranklin.2016.06.029 doi: 10.1016/j.jfranklin.2016.06.029
    [29] Y. L. Huang, S. H. Qiu, S. Y. Ren, Z. W. Zheng, Fixed-time synchronization of coupled Cohen-Grossberg neural networks with and without parameter uncertainties, Neurocomputing, 315 (2018), 157–168. https://doi.org/10.1016/j.neucom.2018.07.013 doi: 10.1016/j.neucom.2018.07.013
    [30] Z. Chen, Complete synchronization for impulsive Cohen-Grossberg neural networks with delay under noise perturbation, Chaos Solitons Fract., 42 (2009), 1664–1669. https://doi.org/10.1016/j.chaos.2009.03.063 doi: 10.1016/j.chaos.2009.03.063
    [31] S. C. Jia, C. Hu, J. Yu, H. J. Jiang, Asymptotical and adaptive synchronization of Cohen-Grossberg neural networks with heterogeneous proportional delays, Neurocomputing, 275 (2018), 1449–1455. https://doi.org/10.1016/j.neucom.2017.09.076 doi: 10.1016/j.neucom.2017.09.076
    [32] L. Wang, H. L. Li, L. Zhang, C. Hu, H. J. Jiang, Quasi-synchronization of fractional-order complex-value neural networks with discontinuous activations, Neurocomputing, 560 (2023), 126856. https://doi.org/10.1016/j.neucom.2023.126856 doi: 10.1016/j.neucom.2023.126856
    [33] X. Wang, J. D. Cao, B. Yang, F. B. Chen, Fast fixed-time synchronization control analysis for a class of coupled delayed Cohen-Grossberg neural networks, J. Franklin Inst., 359 (2022), 1612–1639. https://doi.org/10.1016/j.jfranklin.2022.01.026 doi: 10.1016/j.jfranklin.2022.01.026
    [34] P. Liu, M. X. Kong, Z. G. Zeng, Projective synchronization analysis of fractional-order neural networks with mixed time delays, IEEE Trans. Cybernet., 52 (2022), 6798–6808. https://doi.org/10.1109/TCYB.2020.3027755 doi: 10.1109/TCYB.2020.3027755
    [35] I. Stamova, Global Mittag-Leffler stability and synchronization of impulsive fractional-order neural networks with time-varying delays, Nonlinear Dyn., 77 (2014), 1251–1260. https://doi.org/10.1007/s11071-014-1375-4 doi: 10.1007/s11071-014-1375-4
    [36] H. Zhang, Y. H. Cheng, H. M. Zhang, W. W. Zhang, J. D. Cao, Hybrid control design for Mittag-Leffler projective synchronization on FOQVNNs with multiple mixed delays and impulsive effects, Math. Comput. Simul., 197 (2022), 341–357. https://doi.org/10.1016/j.matcom.2022.02.022 doi: 10.1016/j.matcom.2022.02.022
    [37] J. Y. Chen, C. D. Li, X. J. Yang, Global Mittag-Leffler projective synchronization of nonidentical fractional-order neural networks with delay via sliding mode control, Neurocomputing, 313 (2018), 324–332. https://doi.org/10.1016/j.neucom.2018.06.029 doi: 10.1016/j.neucom.2018.06.029
    [38] I. Podlubny, Fractional differential equations, Academic Press, 1998.
    [39] T. Yang, L. B. Yang, C. W. Wu, L. O. Chua, Fuzzy cellular neural networks: theory, In: 1996 Fourth IEEE International Workshop on Cellular Neural Networks and their Applications Proceedings (CNNA-96), 1996,181–186. https://doi.org/10.1109/CNNA.1996.566545
    [40] S. Yang, J. Yu, C. Hu, H. J. Jiang, Quasi-projection synchronization of fractional-order complex-valued recurrent neural networks, Neural Netw., 104 (2018), 104–113. https://doi.org/10.1016/j.neunet.2018.04.007 doi: 10.1016/j.neunet.2018.04.007
    [41] Z. Y. Wu, G. R. Chen, X. C. Fu, Synchronization of a network coupled with complex-variable chaotic systems, Chaos, 22 (2012), 023127. https://doi.org/10.1063/1.4717525 doi: 10.1063/1.4717525
    [42] J. Yu, C. Hu, H. J. Jiang, Corrogendum to "projective synchronization for fractional neural networks", Neural Netw., 67 (2015), 152–154. https://doi.org/10.1016/j.neunet.2015.02.007 doi: 10.1016/j.neunet.2015.02.007
    [43] S. Liu, R. Yang, X. F. Zhou, W. Jiang, X. Y. Li, X. W. Zhao, Stability analysis of fractional delayed equations and its applications on consensus of multi-agent systems, Commun. Nonlinear Sci. Numer. Simul., 73 (2019), 351–362. https://doi.org/10.1016/j.cnsns.2019.02.019 doi: 10.1016/j.cnsns.2019.02.019
  • This article has been cited by:

    1. Taha Koleilat, Hojat Asgariandehkordi, Hassan Rivaz, Yiming Xiao, 2024, Chapter 60, 978-3-031-72389-6, 643, 10.1007/978-3-031-72390-2_60
    2. Lina Du, Chunjie Ma, Junhao Chen, Huimin Zheng, Xiushan Nie, Zan Gao, Survey on deep learning-based weakly supervised salient object detection, 2025, 09574174, 127497, 10.1016/j.eswa.2025.127497
    3. Xia Li, Xinran Liu, Lin Qi, Junyu Dong, Weakly supervised camouflaged object detection based on the SAM model and mask guidance, 2025, 02628856, 105571, 10.1016/j.imavis.2025.105571
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(909) PDF downloads(51) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog