This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

On the first three minimum Mostar indices of tree-like phenylenes

Hechao Liu1, Lihua You1,, Hanlin Chen2, Zikai Tang3
1School of Mathematical Sciences, South China Normal University,
Guangzhou, 510631, P. R. China
[email protected], [email protected]
2
College of Computer Engineering and Applied Mathematics, Changsha University,
Changsha, Hunan 410022, P. R. China
[email protected]
3
School of Mathematics and Statistics, Hunan Normal University,
Changsha, Hunan 410081, P. R. China
[email protected]
Corresponding author
Abstract

Let G=(VG,EG)G=(V_{G},E_{G}) be a simple connected graph with its vertex set VGV_{G} and edge set EGE_{G}. The Mostar index Mo(G)Mo(G) was defined as Mo(G)=e=uvE(G)|nunv|Mo(G)=\sum\limits_{e=uv\in E(G)}|n_{u}-n_{v}|, where nun_{u} (resp., nvn_{v}) is the number of vertices whose distance to vertex uu (resp., vv) is smaller than the distance to vertex vv (resp., uu). In this study, we determine the first three minimum Mostar indices of tree-like phenylenes and characterize all the tree-like phenylenes attaining these values. At last, we give some numerical examples and discussion.

Keywords: Mostar index; Tree-like phenylene; Extremal value.

2020 Mathematics Subject Classification: 05C09, 05C92.

1 Introduction

Chemical indices are a class of numerical invariants that are closely related to the structure of a chemical graph. Chemical indices can be used to predicte the structural and physic-chemical properties of a chemical molecular. They are mainly used for the quantitative characterisation of chemical structures and thus contribute to the study of QSAR, QSTR and QSPR relationships of chemical structures. In recent years, chemical indices have been found a wide range of applications in chemical science, medical science, complex networks, toxicology, etc. The combination of quantum chemistry and chemical graph theory has also become a promising area in the study of QSAR and QSPR.

Our convention in this paper follows [4] for notations that we omit here. Phenylenes are a class of chemical compounds in which the carbon atoms form 6-membered cycles and 4-membered cycles. Each 4-membered cycle(=square) is adjacent to two disjoint 6-membered cycles(=hexagons), and no two hexagons are adjacent [25]. If there is a hexagon of phenylene adjacent to three squares, we call it the tree-type phenylene (the definition of tree-type phenylene is similar to tree-type hexagonal system, see [7]). Related structures include extended sp-carbon nets and heterocyclic analogs. These molecules have great theoretical and potential practical significance in finding new molecules with (super) conductive properties [28].

Nowadays, phenylene is still a hot topic in many experimental and theoretical studies. Some topological properties of phenylenes has been established such as (total π\pi-electron) energy [11], HOMO LUMO separation [13], cyclic conjugation [15], Kekulé structure count[12], Wiener index [10], PI index [6, 14], Detour index [21] and Kirchhoff index [23, 31].

Let 𝒫h\mathscr{P}_{h} be the set of tree-like phenylenes with hh hexagons and h1h-1 squares. And 𝒫=h=1𝒫h\mathscr{P}=\bigcup_{h=1}^{\infty}\mathscr{P}_{h}. Suppose P𝒫hP\in\mathscr{P}_{h}, and RR is a hexagon of PP. Then RR is called an ii-hexagon, if it has exactly ii (0i3)(0\leq i\leq 3) adjacent squares in PP. A 11-hexagon is called a terminal hexagon of PP. A 22-hexagon is called turn-hexagon of PP if its two 22-vertices (the vertices of degree 22) are adjacent in PP. A 33-hexagon is called a full-hexagon of PP. Let 𝒫h,i𝒫h\mathscr{P}_{h,i}\subseteq\mathscr{P}_{h} (0ih12)(0\leq i\leq\lfloor\frac{h-1}{2}\rfloor) be the set of phenylenes with ii full-hexagons. Each graph in 𝒫h,0\mathscr{P}_{h,0} (or denoted by 𝒞h\mathcal{C}_{h} ) is called a phenylene chain. A phenylene chain is called a linear phenylene chain (denoted by LhL_{h}) if it contains no turn-hexagons. Let 𝒞h,i𝒞h\mathcal{C}_{h,i}\subseteq\mathcal{C}_{h} be the set of phenylene chains with ii turn-hexagons.

A segment of a phenylene chain G𝒞hG\in\mathcal{C}_{h} is a maximal linear sub-chain. Denote by SS a non-terminal segment of a phenylene chain GG. We call SS a non-zigzag segment (resp., a zigzag segment) if it’s two neighboring segments lie on the same sides (resp., on different sides) of the line through centers of all hexagons and squares on SS.

Denote by CL(t1,t2,t3,,tk,tk+1)C_{L}(t_{1},t_{2},t_{3},\cdots,t_{k},t_{k+1}) the phenylene chain with hh hexagons and exactly k+1k+1 segments S1,S2,,S_{1},S_{2},\cdots, Sk+1S_{k+1} of lengths t1+1,t2+2,t3+2,,tk+2,tk+1+1t_{1}+1,t_{2}+2,t_{3}+2,\cdots,t_{k}+2,t_{k+1}+1, respectively, where S1S_{1} and Sk+1S_{k+1} are the terminal segments, all SiS_{i} (2ik)(2\leq i\leq k) are zagzig segments, 1t1tk+11\leq t_{1}\leq t_{k+1}, and i=1k+1ti+k=h\sum\limits_{i=1}^{k+1}t_{i}+k=h. Particularly, CL(j,n)𝒞h,1C_{L}(j,n)\in\mathcal{C}_{h,1} is the graph including two vertex-disjoint linear phenylene chains LjL_{j} and LnL_{n} as subgraphs, where jnj\leq n, and h=j+n+1h=j+n+1. CL(j,k,n)𝒞h,2C_{L}(j,k,n)\in\mathcal{C}_{h,2} is the graph including three vertex-disjoint linear phenylene chains LjL_{j}, LkL_{k} and LnL_{n} as subgraphs, where jnj\leq n, the second segment is a zigzag segment and h=j+k+n+2h=j+k+n+2.

Denote by PL(j,k,n)𝒫h,1P_{L}(j,k,n)\in\mathscr{P}_{h,1} the graph including three vertex-disjoint linear phenylene chains LjL_{j}, LkL_{k} and LnL_{n} as subgraphs, where jknj\leq k\leq n, and h=j+k+n+1h=j+k+n+1.

Došlić et al.[9] introduced Mostar index [9] of a graph GG, which is defined as

Mo(G)=e=uvE(G)|nunv|.Mo(G)=\sum_{e=uv\in E(G)}|n_{u}-n_{v}|.

Došlić et al. determined extremal values of Mostar index among trees and unicyclic graphs, then gave a cut method for computing the Mostar index of benzenoid systems.

Similarly, the edge Mostar index [20] is defined as

Moe(G)=e=uvE(G)|mumv|,Mo_{e}(G)=\sum_{e=uv\in E(G)}|m_{u}-m_{v}|,

where mum_{u} (resp.,mvm_{v}) is the number of edges whose distance to vertex uu (resp., vv) is smaller than the distance to vertex vv (resp., uu). We can refer to [1, 2, 3, 7, 8, 9, 16, 17, 18, 19, 20, 22, 26, 27, 29, 30] for more details about (edge) Mostar index.

Mostar indices can be used to measure the peripherality of chemical graphs, so in addition to chemical applications, Mostar indices have a wide range of applications in complex networks. It can be used to describe structural properties of the network. Mostar indices can also be used to extend quantum estimates and expand reactivity based on electronic descriptors. In this study, our aim is to solve the extremal problem of tree-like phenylenes with respect to Mostar indices. Using the methods of [7], we determine the first three minimum values of the Mostar index of tree-like phenylenes with a fixed number of hexagons and characterize all the tree-like phenylenes attaining these values.

2 Preliminaries

An orthogonal cut is a line segment that starts from the middle of a peripheral edge of a phenylene, goes orthogonal to this edge and ends at the first next peripheral edge that it intersects. Let P𝒫hP\in\mathscr{P}_{h}. Denote by 𝒪Puv\mathcal{O}_{P}^{uv} the set of edges that parallel with uvuv in PP, and |𝒪Puv|=oPuv|\mathcal{O}_{P}^{uv}|=o_{P}^{uv}. Let 𝒪P\mathcal{O}_{P} be the set of all disjoint parallel classes in PP. Note that 𝒪Puv\mathcal{O}_{P}^{uv} is an edge cut, denote by GPuG_{P}^{u} (resp.,GPv)(resp.,G_{P}^{v}) the connected components of P𝒪PuvP-\mathcal{O}_{P}^{uv} contain uu (resp.,v)(resp.,v). Denote by rPur_{P}^{u} (resp.,rPv)(resp.,r_{P}^{v}) the number of hexagons in GPuG_{P}^{u} (resp.,GPv)(resp.,G_{P}^{v}). Note that, for any P𝒫hP\in\mathscr{P}_{h}, |VP|=6h|V_{P}|=6h and |EP|=8h2|E_{P}|=8h-2.

Bearing in mind that PL(j,k,n)𝒫h,1P_{L}(j,k,n)\in\mathscr{P}_{h,1} is the graph including three vertex-disjoint linear phenyene chains LjL_{j}, LkL_{k} and LnL_{n} as subgraphs, where jknj\leq k\leq n, and h=j+k+n+1h=j+k+n+1. By the definition of Mo(G)Mo(G), we can calculate the value of Mo(PL(j,k,n))Mo(P_{L}(j,k,n)).

Lemma 2.1

Given a phenylene G=PL(j,k,n)𝒫h,1G=P_{L}(j,k,n)\in\mathscr{P}_{h,1} with three branches Lj,Lk,LnL_{j},L_{k},L_{n} (1jkn)(1\leq j\leq k\leq n) and hh (h=j+k+n+1)(h=j+k+n+1) hexagons, then

(1)(1) If nh2n\leq\lfloor\frac{h}{2}\rfloor, then Mo(G)=24(2kj+3nj+4kn+k+2n)Mo(G)=24(2kj+3nj+4kn+k+2n).

(2)(2) If nh2+1n\geq\lfloor\frac{h}{2}\rfloor+1, then Mo(G)=6(4j+3j2+8k+3k2+4n+3n2+14kj+6jn+10kn+1)Mo(G)=6(4j+3j^{2}+8k+3k^{2}+4n+3n^{2}+14kj+6jn+10kn+1) for even hh; Mo(G)=6(4j+3j2+8k+3k2+4n+3n2+14kj+6jn+10kn)Mo(G)=6(4j+3j^{2}+8k+3k^{2}+4n+3n^{2}+14kj+6jn+10kn) for odd hh.

Proof. By using the cut method to PL(j,k,n)P_{L}(j,k,n), we have

Mo(G)=\displaystyle Mo(G)= 6{2(j+1)(nk)+2(k+1)(nj)+2(n+1)(kj)\displaystyle 6\{2(j+1)(n-k)+2(k+1)(n-j)+2(n+1)(k-j)
+4i=1j(h+12i)+4i=1k(h+12i)+4i=1n|h+12i|\displaystyle+4\sum_{i=1}^{j}(h+1-2i)+4\sum_{i=1}^{k}(h+1-2i)+4\sum_{i=1}^{n}|h+1-2i|
+2i=1j(h2i)+2i=1k(h2i)+2i=1n|h2i|}.\displaystyle+2\sum_{i=1}^{j}(h-2i)+2\sum_{i=1}^{k}(h-2i)+2\sum_{i=1}^{n}|h-2i|\}.

(1)(1) If nh2n\leq\lfloor\frac{h}{2}\rfloor, then

Mo(G)=\displaystyle Mo(G)= 6{2(j+1)(nk)+2(k+1)(nj)+2(n+1)(kj)\displaystyle 6\{2(j+1)(n-k)+2(k+1)(n-j)+2(n+1)(k-j)
+4i=1j(h+12i)+4i=1k(h+12i)+4i=1n(h+12i)\displaystyle+4\sum_{i=1}^{j}(h+1-2i)+4\sum_{i=1}^{k}(h+1-2i)+4\sum_{i=1}^{n}(h+1-2i)
+2i=1j(h2i)+2i=1k(h2i)+2i=1n(h2i)}\displaystyle+2\sum_{i=1}^{j}(h-2i)+2\sum_{i=1}^{k}(h-2i)+2\sum_{i=1}^{n}(h-2i)\}
=\displaystyle= 24(2kj+3nj+4kn+k+2n).\displaystyle 24(2kj+3nj+4kn+k+2n).

(2)(2) If nh2+1n\geq\lfloor\frac{h}{2}\rfloor+1, and hh is even, then

Mo(G)=\displaystyle Mo(G)= 6{2(j+1)(nk)+2(k+1)(nj)+2(n+1)(kj)\displaystyle 6\{2(j+1)(n-k)+2(k+1)(n-j)+2(n+1)(k-j)
+4i=1j(h+12i)+4i=1k(h+12i)+4i=1h2(h+12i)4i=h2+1n(h+12i)\displaystyle+4\sum_{i=1}^{j}(h+1-2i)+4\sum_{i=1}^{k}(h+1-2i)+4\sum_{i=1}^{\frac{h}{2}}(h+1-2i)-4\sum_{i=\frac{h}{2}+1}^{n}(h+1-2i)
+2i=1j(h2i)+2i=1k(h2i)+2i=1h2(h2i)2i=h2+1n(h2i)}\displaystyle+2\sum_{i=1}^{j}(h-2i)+2\sum_{i=1}^{k}(h-2i)+2\sum_{i=1}^{\frac{h}{2}}(h-2i)-2\sum_{i=\frac{h}{2}+1}^{n}(h-2i)\}
=\displaystyle= 6(4j+3j2+8k+3k2+4n+3n2+14kj+6jn+10kn+1).\displaystyle 6(4j+3j^{2}+8k+3k^{2}+4n+3n^{2}+14kj+6jn+10kn+1).

If nh2+1n\geq\lfloor\frac{h}{2}\rfloor+1, and hh is odd, similarly, we have Mo(G)=6(4j+3j2+8k+3k2+4n+3n2+14kj+6jn+10kn)Mo(G)=6(4j+3j^{2}+8k+3k^{2}+4n+3n^{2}+14kj+6jn+10kn).

This completes the proof. \blacksquare

3 The minimal tree-like phenylenes

Let G𝒞hG\in\mathcal{C}_{h} and R1R_{1}, RhR_{h} are two terminal hexagons of GG. Denote by xi,1,xi,2,,xi,6x_{i,1},x_{i,2},\cdots,x_{i,6} the six clockwise successive vertices in RiR_{i} for i=1,hi=1,h, where dG(xi,j)=2d_{G}(x_{i,j})=2 for j=1,2,3,4j=1,2,3,4. Let eij=xi,jxi,j+1e_{ij}=x_{i,j}x_{i,j+1} for i=1,hi=1,h and j=1,2,,6j=1,2,\cdots,6 (let xi,7:=xi,1x_{i,7}:=x_{i,1}).

Suppose that P1,P2𝒫P_{1},P_{2}\in\mathscr{P} and ui,viu_{i},v_{i} be two adjacent 22-vertices (vertices with degree 2) in PiP_{i} for i=1,2i=1,2. Let P=P1(u1,v1)P2(u2,v2)P=P_{1}(u_{1},v_{1})\Box P_{2}(u_{2},v_{2}) be the phenylene obtained from P1,P2P_{1},P_{2} by connecting u1u_{1} with u2u_{2}, and v1v_{1} with v2v_{2}, respectively.

Refer to caption
Figure 1: The tree-like phenylenes P1P_{1}, P2P_{2} and P3P_{3} of Lemma 3.1.

For uvE(G)uv\in E(G), we denote nGun_{G}^{u} (or nun_{u}) the number of vertices in GG lying closer to vertex uu than to vertex vv. Some symbols involved in the proof are described in Section 2.

Lemma 3.1

Let P𝒫nP\in\mathscr{P}_{n} and Pi=P(s,t)Lk(xk,i,xk,i+1)P_{i}=P(s,t)\Box L_{k}(x_{k,i},x_{k,i+1}) for i=1,2,3i=1,2,3, see Figure 1. And k2k\geq 2, nk1n\geq k-1, then

(1)(1) Mo(P2)Mo(P1)Mo(P_{2})\leq Mo(P_{1}), with equality iff |rPtrPs|k1|r_{P}^{t}-r_{P}^{s}|\geq k-1 and min{rPs,rPt}=0\min\{r_{P}^{s},r_{P}^{t}\}=0.

(2)(2) Mo(P2)<Mo(P3)Mo(P_{2})<Mo(P_{3}).

Proof. Let 𝒪Pst\mathcal{O}_{P}^{st} denote the set of edges that are parallel to edge stst in PP. Assume that rPsrPtr_{P}^{s}\leq r_{P}^{t}. Denote Ed={ek1}{ek3}{ek4}{ek6}𝒪Pst𝒪Lkek2E_{d}=\{e_{k1}\}\cup\{e_{k3}\}\cup\{e_{k4}\}\cup\{e_{k6}\}\cup\mathcal{O}_{P}^{st}\cup\mathcal{O}_{L_{k}}^{e_{k2}}. Denote ϕi(Ed)=uvEd|nPiunPiv|\phi_{i}(E_{d})=\sum\limits_{uv\in E_{d}}|n_{P_{i}}^{u}-n_{P_{i}}^{v}| (i=1,2,3i=1,2,3), we have Mo(Pi)Mo(P2)=ϕi(Ed)ϕ2(Ed)Mo(P_{i})-Mo(P_{2})=\phi_{i}(E_{d})-\phi_{2}(E_{d}) (i=1,3i=1,3).

Note that nk1n\geq k-1, then

ϕ1(Ed)=6{2(nk+1)+2kn+(oPst+2)|(rPtrPs)(k1)|}\phi_{1}(E_{d})=6\{2(n-k+1)+2kn+(o_{P}^{st}+2)|(r_{P}^{t}-r_{P}^{s})-(k-1)|\};

ϕ2(Ed)=6{4(nk+1)+(oPst+2k)(rPtrPs)}\phi_{2}(E_{d})=6\{4(n-k+1)+(o_{P}^{st}+2k)(r_{P}^{t}-r_{P}^{s})\};

ϕ3(Ed)=6{2(nk+1)+2kn+(oPst+2)[(rPtrPs)+(k1)]}\phi_{3}(E_{d})=6\{2(n-k+1)+2kn+(o_{P}^{st}+2)[(r_{P}^{t}-r_{P}^{s})+(k-1)]\}.

Bearing in mind that k2k\geq 2 and n=12oPst+rPt+rPsn=\frac{1}{2}o_{P}^{st}+r_{P}^{t}+r_{P}^{s}.

(1)(1) If rPtrPsk1r_{P}^{t}-r_{P}^{s}\geq k-1, then ϕ1(Ed)ϕ2(Ed)=6(k1)(2noPst2(rPtrPs))=24(k1)rPs0\phi_{1}(E_{d})-\phi_{2}(E_{d})=6(k-1)(2n-o_{P}^{st}-2(r_{P}^{t}-r_{P}^{s}))=24(k-1)r_{P}^{s}\geq 0, with equality iff rPs=0r_{P}^{s}=0.

If rPtrPs<k1r_{P}^{t}-r_{P}^{s}<k-1, then ϕ1(Ed)ϕ2(Ed)=12(oPst+2)(k1(rPtrPs))>0\phi_{1}(E_{d})-\phi_{2}(E_{d})=12(o_{P}^{st}+2)(k-1-(r_{P}^{t}-r_{P}^{s}))>0.

(2)(2) ϕ3(Ed)ϕ2(Ed)=6(k1)(2n+4+oPst2(rPtrPs))=12(k1)(oPst+2rPs+2)>0\phi_{3}(E_{d})-\phi_{2}(E_{d})=6(k-1)(2n+4+o_{P}^{st}-2(r_{P}^{t}-r_{P}^{s}))=12(k-1)(o_{P}^{st}+2r_{P}^{s}+2)>0.

This completes the proof. \blacksquare

By Lemma 3.1, we have

Corollary 3.2

Let P𝒫nP\in\mathscr{P}_{n} with s,ts,t being two adjacent 22-vertices of its turn hexagon. Denote by CkC_{k} a phenylene chain with kk (1)(\geq 1) hexagons, and u,vu,v are two adjacent 22-vertices of its one terminal hexagon. Then Mo(P(s,t)Lk(xk,2,xk,3))Mo(P(s,t)Ck(u,v))Mo(P(s,t)\Box L_{k}(x_{k,2},x_{k,3}))\leq Mo(P(s,t)\Box C_{k}(u,v)).

Let P𝒫n+1P\in\mathscr{P}_{n+1}, where nmax{j,k}n\geq max\{j,k\}, j,k1j,k\geq 1. t1,t,s,s1t_{1},t,s,s_{1} are vertices in Figure 2, and rPsrPtr_{P}^{s}\leq r_{P}^{t}, h=j+k+n+1h=j+k+n+1. Then we have

Refer to caption
Figure 2: The tree-type phenylenes P1P_{1}, P2P_{2} of Lemma 3.3.
Lemma 3.3

Let P1={P(t,t1)Lk(yk,2,yk,3)}(s1,s)Lj(xj,2,xj,3)P_{1}=\{P(t,t_{1})\Box L_{k}(y_{k,2},y_{k,3})\}(s_{1},s)\Box L_{j}(x_{j,2},x_{j,3}) and P2=P(s,t)Lk+jP_{2}=P(s,t)\Box L_{k+j} (xk+j,2,xk+j,3)(x_{k+j,2},x_{k+j,3}), see Figure 2, then Mo(P2)<Mo(P1)Mo(P_{2})<Mo(P_{1}).

Proof. Denote Ed,i=(EPiEP)ER𝒪PRstE_{d,i}=(E_{P_{i}}-E_{P})\cup E_{R}\cup\mathcal{O}_{P-R}^{st} (i=1,2i=1,2). Let ϕi(Ed,i)=uvEd,i|nPiunPiv|\phi_{i}(E_{d,i})=\sum\limits_{uv\in E_{d,i}}|n_{P_{i}}^{u}-n_{P_{i}}^{v}| (i=1,2i=1,2), then Mo(P1)Mo(P2)=ϕ1(Ed,1)ϕ2(Ed,2)Mo(P_{1})-Mo(P_{2})=\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2}).

Since nmax{j,k}n\geq\max\{j,k\}, then

ϕ1(Ed,1)=\displaystyle\phi_{1}(E_{d,1})= 6{2(j+1)(nk)+2(k+1)(nj)+oPst|rPt+k(rPs+j)|\displaystyle 6\{2(j+1)(n-k)+2(k+1)(n-j)+o_{P}^{st}|r_{P}^{t}+k-(r_{P}^{s}+j)|
+4i=1j(h+12i)+4i=1k(h+12i)+2i=1j(h2i)+2i=1k(h2i)}.\displaystyle+4\sum_{i=1}^{j}(h+1-2i)+4\sum_{i=1}^{k}(h+1-2i)+2\sum_{i=1}^{j}(h-2i)+2\sum_{i=1}^{k}(h-2i)\}.
ϕ2(Ed,2)=\displaystyle\phi_{2}(E_{d,2})= 6{(oPst+2(j+k))(rPtrPs)+4i=1j+k+1|h+12i|+2i=1j+k|h2i|}.\displaystyle 6\{(o_{P}^{st}+2(j+k))(r_{P}^{t}-r_{P}^{s})+4\sum_{i=1}^{j+k+1}|h+1-2i|+2\sum_{i=1}^{j+k}|h-2i|\}.

Note that n=12oPst+rPt+rPs1n=\frac{1}{2}o_{P}^{st}+r_{P}^{t}+r_{P}^{s}-1, j,k1j,k\geq 1 and oPst2o_{P}^{st}\geq 2.

Case 1. rPt+krPs+jr_{P}^{t}+k\geq r_{P}^{s}+j.

Subcase 1.1. j+k+1h2j+k+1\leq\lfloor\frac{h}{2}\rfloor.

ϕ1(Ed,1)ϕ2(Ed,2)=12(oPstk+2(k+j)rPs+4kj)>0\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2})=12(o_{P}^{st}k+2(k+j)r_{P}^{s}+4kj)>0.

Subcase 1.2. j+k+1h2+1j+k+1\geq\lfloor\frac{h}{2}\rfloor+1.

If hh is even, then j+kn+1j+k\geq n+1, and

ϕ1(Ed,1)ϕ2(Ed,2)=\displaystyle\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2})= 6{2oPstk+4(k+j)rPs3j23k23n2+2kj+6jn+6kn4k\displaystyle 6\{2o_{P}^{st}k+4(k+j)r_{P}^{s}-3j^{2}-3k^{2}-3n^{2}+2kj+6jn+6kn-4k
4j+4n1}\displaystyle-4j+4n-1\}
=\displaystyle= 6{3k(nk)+3(j+1)(nj)+2k(oPst+rPs2)+2j(k+2rPs1)\displaystyle 6\{3k(n-k)+3(j+1)(n-j)+2k(o_{P}^{st}+r_{P}^{s}-2)+2j(k+2r_{P}^{s}-1)
+3n(j+kn1)+4(n1)+2krPs+j+3}>0.\displaystyle+3n(j+k-n-1)+4(n-1)+2kr_{P}^{s}+j+3\}>0.

If hh is odd, then j+knj+k\geq n, and

ϕ1(Ed,1)ϕ2(Ed,2)=\displaystyle\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2})= 6{2oPstk+4(k+j)rPs3j23k23n2+2kj+6jn+6kn4k\displaystyle 6\{2o_{P}^{st}k+4(k+j)r_{P}^{s}-3j^{2}-3k^{2}-3n^{2}+2kj+6jn+6kn-4k
4j+4n}\displaystyle-4j+4n\}
=\displaystyle= 6{3k(nk)+3(j+1)(nj)+2k(oPst+rPs2)+2j(k+2rPs1)\displaystyle 6\{3k(n-k)+3(j+1)(n-j)+2k(o_{P}^{st}+r_{P}^{s}-2)+2j(k+2r_{P}^{s}-1)
+3n(j+kn)+n+2krPs+j}>0.\displaystyle+3n(j+k-n)+n+2kr_{P}^{s}+j\}>0.

Case 2. rPt+k<rPs+jr_{P}^{t}+k<r_{P}^{s}+j.

Note that njk+12n\geq j\geq k+1\geq 2 and oPst2o_{P}^{st}\geq 2.

Subcase 2.1. j+k+1h2j+k+1\leq\lfloor\frac{h}{2}\rfloor.

ϕ1(Ed,1)ϕ2(Ed,2)=12{oPst(rPs+jrPt)+2(k+j)rPs+4kj}>0\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2})=12\{o_{P}^{st}(r_{P}^{s}+j-r_{P}^{t})+2(k+j)r_{P}^{s}+4kj\}>0.

Subcase 2.2. j+k+1h2+1j+k+1\geq\lfloor\frac{h}{2}\rfloor+1.

If hh is even, then j+kn+1j+k\geq n+1, and

ϕ1(Ed,1)ϕ2(Ed,2)=\displaystyle\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2})= 6{2oPst(rPs+jrPt)+4(k+j)rPs3j23k23n2+2kj+6jn\displaystyle 6\{2o_{P}^{st}(r_{P}^{s}+j-r_{P}^{t})+4(k+j)r_{P}^{s}-3j^{2}-3k^{2}-3n^{2}+2kj+6jn
+6kn4k4j+4n1}\displaystyle+6kn-4k-4j+4n-1\}
=\displaystyle= 6{3k(nk)+3(j+1)(nj)+3n(j+kn1)+2oPst(rPs+j\displaystyle 6\{3k(n-k)+3(j+1)(n-j)+3n(j+k-n-1)+2o_{P}^{st}(r_{P}^{s}+j
rPtk1)+2k(j2)+2(oPst2)+(nj)+4(k+j)rPs\displaystyle-r_{P}^{t}-k-1)+2k(j-2)+2(o_{P}^{st}-2)+(n-j)+4(k+j)r_{P}^{s}
+2oPstk+3n+3}>0.\displaystyle+2o_{P}^{st}k+3n+3\}>0.

If hh is odd, then j+knj+k\geq n, and

ϕ1(Ed,1)ϕ2(Ed,2)=\displaystyle\phi_{1}(E_{d,1})-\phi_{2}(E_{d,2})= 6{2oPst(rPs+jrPt)+4(k+j)rPs3j23k23n2+2kj+6jn\displaystyle 6\{2o_{P}^{st}(r_{P}^{s}+j-r_{P}^{t})+4(k+j)r_{P}^{s}-3j^{2}-3k^{2}-3n^{2}+2kj+6jn
+6kn4k4j+4n}\displaystyle+6kn-4k-4j+4n\}
=\displaystyle= 6{3k(nk)+3(j+1)(nj)+3n(j+kn)+2oPst(rPs+j\displaystyle 6\{3k(n-k)+3(j+1)(n-j)+3n(j+k-n)+2o_{P}^{st}(r_{P}^{s}+j
rPtk1)+2k(j2)+2(oPst2)+(nj)+4(k+j)rPs\displaystyle-r_{P}^{t}-k-1)+2k(j-2)+2(o_{P}^{st}-2)+(n-j)+4(k+j)r_{P}^{s}
+2oPstk+4}>0.\displaystyle+2o_{P}^{st}k+4\}>0.

Thus, Mo(P2)<Mo(P1)Mo(P_{2})<Mo(P_{1}). This completes the proof. \blacksquare

By Lemma 3.1, we can directly obtain the smallest Mostar index among 𝒞h\mathcal{C}_{h}.

Lemma 3.4

[5] Let G𝒞hG\in\mathcal{C}_{h}, then Mo(G)Mo(Lh)=72h2h224h2Mo(G)\geq Mo(L_{h})=72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor, with equality iff GLhG\cong L_{h}.

By Corollary 3.2, Lemma 3.3 and Lemma 3.4, we can obtain the smallest Mostar index among 𝒫h\mathscr{P}_{h}. The proof of Theorem 3.5 follows from the same arguments as the proof of Theorem 1.3 of [7], thus we omit the proof.

Theorem 3.5

Let G𝒫hG\in\mathscr{P}_{h}, then Mo(G)Mo(Lh)=72h2h224h2Mo(G)\geq Mo(L_{h})=72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor, with equality iff GLhG\cong L_{h}.

4 The second minimal tree-like phenylenes

Bearing in mind that CL(j,n)𝒞h,1C_{L}(j,n)\in\mathcal{C}_{h,1} is the graph including two vertex-disjoint linear phenylene chains LjL_{j} and LnL_{n} as subgraphs, where jnj\leq n, and h=j+n+1h=j+n+1. CL(j,k,n)𝒞h,2C_{L}(j,k,n)\in\mathcal{C}_{h,2} is the graph including three vertex-disjoint linear phenylene chains LjL_{j}, LkL_{k} and LnL_{n} as subgraphs, where jnj\leq n, the second segment is a zigzag segment and h=j+k+n+2h=j+k+n+2. In the following, we give some useful results for our proofs of main theorem.

Using the cut method to CL(j,hj1)C_{L}(j,h-j-1) and LhL_{h}, and comparing the change of Mostar index among CL(j,hj1)C_{L}(j,h-j-1) and LhL_{h}, we also have

Lemma 4.1

[5] Let G=CL(j,hj1)G=C_{L}(j,h-j-1), 1jh121\leq j\leq\lfloor\frac{h-1}{2}\rfloor, be the phenylene chain with hh hexagons, then Mo(CL(1,h2)<Mo(CL(2,h3)<Mo(CL(3,h4)<<Mo(CL(h12,h12)Mo(C_{L}(1,h-2)<Mo(C_{L}(2,h-3)<Mo(C_{L}(3,h-4)<\cdots<Mo(C_{L}(\lfloor\frac{h-1}{2}\rfloor,\lceil\frac{h-1}{2}\rceil).

Refer to caption
Figure 3: The phenylenes CL(1,h2)C_{L}(1,h-2), CL(1,h4,1)C_{L}(1,h-4,1) of Lemma 4.2.

By Lemma 3.1, Lemma 4.1 and Lemma 3.4, we can also obtain the second minimum Mostar index among phenylene chains 𝒞h\mathcal{C}_{h}.

Lemma 4.2

[5] If G𝒞hG\in\mathcal{C}_{h} and G≇LhG\not\cong L_{h}, then Mo(G)Mo(CL(1,h2))=Mo(CL(1,h4,1))=72h2h224h2+24(h1)Mo(G)\geq Mo(C_{L}(1,h-2))=Mo(C_{L}(1,h-4,1))=72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor+24(h-1) with equality iff GCL(1,h2)G\cong C_{L}(1,h-2) or GCL(1,h4,1)G\cong C_{L}(1,h-4,1), see Figure 3.

Lemma 4.3

Given a phenylene PL(j,k,n)𝒫h,1P_{L}(j,k,n)\in\mathscr{P}_{h,1} with three branches Lj,Lk,LnL_{j},L_{k},L_{n} (1jkn)(1\leq j\leq k\leq n) and hh (h=j+k+n+1)(h=j+k+n+1) hexagons. Then Mo(PL(j,k,n))>Mo(CL(1,j+k+n1))Mo(P_{L}(j,k,n))>Mo(C_{L}(1,j+k+n-1)) for n2n\geq 2, and Mo(PL(1,1,1))<Mo(CL(1,2))Mo(P_{L}(1,1,1))<Mo(C_{L}(1,2)) for n=1n=1.

Proof. Note that Mo(CL(1,j+k+n1)=Mo(CL(1,h2)=72h2h224h2+24(h1)Mo(C_{L}(1,j+k+n-1)=Mo(C_{L}(1,h-2)=72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor+24(h-1). Then, by Lemma 2.1, we have

Case 1. nh2n\leq\lfloor\frac{h}{2}\rfloor

Subcase 1.1. If hh is even, then nj+k+1n\leq j+k+1, and we have

Mo(PL(j,k,n))Mo(CL(1,j+k+n1))=6{3j23k23n2+2kj+6jn+10kn4k8j1}=6{3j(kj)+3k(nk)+3(n+1)(j+k+1n)+k(nj)+3(k+j)(n4)+5k+j4}=6{3j(kj)+3k(nk)+3(n+1)(j+k+1n)+k(nj)+3(k+j)(n3)+2k2j4}>0.\begin{split}&\quad Mo(P_{L}(j,k,n))-Mo(C_{L}(1,j+k+n-1))\\ &=6\{-3j^{2}-3k^{2}-3n^{2}+2kj+6jn+10kn-4k-8j-1\}\\ &=6\{3j(k-j)+3k(n-k)+3(n+1)(j+k+1-n)+k(n-j)+3(k+j)(n-4)+5k\\ &\quad+j-4\}\\ &=6\{3j(k-j)+3k(n-k)+3(n+1)(j+k+1-n)+k(n-j)+3(k+j)(n-3)+2k\\ &\quad-2j-4\}>0.\end{split}

whenever n4n\geq 4, or (j,k,n){(1,1,3),(2,2,3),(1,3,3),(3,3,3)}(j,k,n)\in\{(1,1,3),(2,2,3),(1,3,3),(3,3,3)\}, or (j,k,n)=(1,2,2)(j,k,n)=(1,2,2), whereas Mo(PL(1,1,1))Mo(CL(1,2))=24<0Mo(P_{L}(1,1,1))-Mo(C_{L}(1,2))=-24<0.

Subcase 1.2. If hh is odd, then nj+kn\leq j+k, and we have

Mo(PL(j,k,n))Mo(CL(1,j+k+n1))=6{3j23k23n2+2kj+6jn+10kn4k8j}=6{3j(kj)+3k(nk)+3(n+1)(j+kn)+k(nj)+3(k+j)(n4)+3n+5k+j}=6{3j(kj)+3k(nk)+3(n+1)(j+kn)+k(nj)+3(k+j)(n3)+2(kj)+3n}>0.\begin{split}&\quad Mo(P_{L}(j,k,n))-Mo(C_{L}(1,j+k+n-1))\\ &=6\{-3j^{2}-3k^{2}-3n^{2}+2kj+6jn+10kn-4k-8j\}\\ &=6\{3j(k-j)+3k(n-k)+3(n+1)(j+k-n)+k(n-j)+3(k+j)(n-4)+3n\\ &\quad+5k+j\}\\ &=6\{3j(k-j)+3k(n-k)+3(n+1)(j+k-n)+k(n-j)+3(k+j)(n-3)\\ &\quad+2(k-j)+3n\}>0.\end{split}

whenever n3n\geq 3, or (j,k,n){(2,2,2),(1,1,2)}(j,k,n)\in\{(2,2,2),(1,1,2)\}.

Case 2. nh2+1n\geq\lfloor\frac{h}{2}\rfloor+1

Mo(PL(j,k,n))Mo(CL(1,j+k+n1))=6{4(j+n)(k1)+4kj}>0Mo(P_{L}(j,k,n))-Mo(C_{L}(1,j+k+n-1))=6\{4(j+n)(k-1)+4kj\}>0.

The proof is completed \blacksquare

By Lemma 3.3, Theorem 3.5, Lemma 4.2 and Lemma 4.3, we obtain the second minimum Mostar index of tree-like phenylenes 𝒫h\mathscr{P}_{h}.

Theorem 4.4

Let G𝒫hG\in\mathscr{P}_{h} (h4)(h\geq 4), and GLhG\ncong L_{h}, then

(1)(1) If h5h\geq 5, Mo(G)Mo(CL(1,h2))=Mo(CL(1,h4,1))=72h2h224h2+24(h1)Mo(G)\geq Mo(C_{L}(1,h-2))=Mo(C_{L}(1,h-4,1))=72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor+24(h-1) with equality iff GCL(1,h2)G\cong C_{L}(1,h-2) or GCL(1,h4,1)G\cong C_{L}(1,h-4,1).

(2)(2) If h=4h=4, Mo(G)Mo(PL(1,1,1))=288Mo(G)\geq Mo(P_{L}(1,1,1))=288, with equality iff GPL(1,1,1)G\cong P_{L}(1,1,1).

5 The third minimal tree-like phenylenes

Bearing in mind CL(t1,t2,t3,,tk,tk+1)C_{L}(t_{1},t_{2},t_{3},\cdots,t_{k},t_{k+1}) is the phenylene chian with hh hexagons and exactly k+1k+1 segments S1,S2,,Sk+1S_{1},S_{2},\ldots,S_{k+1} of lengths t1+1,t1+2,t3+2,,tk+2,tk+1+1t_{1}+1,t_{1}+2,t_{3}+2,\ldots,t_{k}+2,t_{k+1}+1, respectively, where S1S_{1} and Sk+1S_{k+1} are the terminal segments, all SiS_{i} (2ik)(2\leq i\leq k) are zagzig segments, 1t1tk+11\leq t_{1}\leq t_{k+1}, and i=1k+1ti+k=h\sum\limits_{i=1}^{k+1}t_{i}+k=h. In the following, we give the following Lemma 5.1 and Lemma 5.2, which are important for our proofs of main theorem 5.3. At first, we give the third minimum Mostar index among phenylene chains.

Refer to caption
Figure 4: Seven phenylene chains of Lemma 5.1.

Let CL(j,k,n)𝒞h,2C_{L}(j,k,n)\in\mathcal{C}_{h,2} with second segment is a zigzag segment. By Lemma 3.1, Lemma 4.1, we also have

Lemma 5.1

[5] If G𝒞hG\in\mathcal{C}_{h} and G≇{Lh,CL(1,h2),CL(1,h4,1)}G\not\cong\{L_{h},C_{L}(1,h-2),C_{L}(1,h-4,1)\}, then Mo(G)72h2h224h2+48(h2)Mo(G)\geq 72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor+48(h-2) with equality iff G{CL(2,h3),CL(1,0,h3),CL(1,h5,2),CL(2,h6,2),CL(1,0,h5,1),CL(1,0,h6,2),CL(1,0,h6,0,1)}G\in\{C_{L}(2,h-3),C_{L}(1,0,h-3),C_{L}(1,h-5,2),C_{L}(2,h-6,2),C_{L}(1,0,h-5,1),C_{L}(1,0,h-6,2),C_{L}(1,0,h-6,0,1)\}, see Figure 4.

Lemma 5.2

Given a phenylene G=PL(j,k,n)𝒫h,1G=P_{L}(j,k,n)\in\mathscr{P}_{h,1} with three branches Lj,Lk,LnL_{j},L_{k},L_{n} (1jkn)(1\leq j\leq k\leq n) and hh (h=j+k+n+1)(h=j+k+n+1) hexagons. Then

(1)(1) If G{PL(1,1,h3),PL(1,2,2),PL(2,2,2)}G\notin\{P_{L}(1,1,h-3),P_{L}(1,2,2),P_{L}(2,2,2)\}, then Mo(G)>Mo(CL(2,h3))Mo(G)>Mo(C_{L}(2,h-3)).

(2)(2) If G{PL(1,1,h3),PL(2,2,2)}G\in\{P_{L}(1,1,h-3),P_{L}(2,2,2)\}, then Mo(G)<Mo(CL(2,h3))Mo(G)<Mo(C_{L}(2,h-3)).

(3)(3) If GPL(1,2,2)G\cong P_{L}(1,2,2), then Mo(G)=Mo(CL(2,3))Mo(G)=Mo(C_{L}(2,3)).

Proof. Note that Mo(CL(2,j+k+n2)=Mo(CL(2,h3)=72h2h224h2+48(h2)Mo(C_{L}(2,j+k+n-2)=Mo(C_{L}(2,h-3)=72\lfloor\frac{h}{2}\rfloor\lceil\frac{h}{2}\rceil-24\lfloor\frac{h}{2}\rfloor+48(h-2). Then by Lemma 2.1, we have

Case 1. nh2n\leq\lfloor\frac{h}{2}\rfloor

Subcase 1.1. If hh is even, then nj+k+1n\leq j+k+1, and we have

Mo(PL(j,k,n))Mo(CL(2,j+k+n2))=6{3j23k23n2+2kj+6jn+10kn8k12j4n+7}=6{3j(kj)+3k(nk)+(3n+7)(j+k+1n)+k(nj)+3(k+j)(n6)+3kj}=6{3j(kj)+3k(nk)+(3n+7)(j+k+1n)+k(nj)+3(k+j)(n5)4j}=6{3j(kj)+3k(nk)+(3n+7)(j+k+1n)+k(nj)+3(k+j)(n4)7j3k}>0.\begin{split}&\quad Mo(P_{L}(j,k,n))-Mo(C_{L}(2,j+k+n-2))\\ &=6\{-3j^{2}-3k^{2}-3n^{2}+2kj+6jn+10kn-8k-12j-4n+7\}\\ &=6\{3j(k-j)+3k(n-k)+(3n+7)(j+k+1-n)+k(n-j)+3(k+j)(n-6)\\ &\quad+3k-j\}\\ &=6\{3j(k-j)+3k(n-k)+(3n+7)(j+k+1-n)+k(n-j)+3(k+j)(n-5)-4j\}\\ &=6\{3j(k-j)+3k(n-k)+(3n+7)(j+k+1-n)+k(n-j)+3(k+j)(n-4)-7j\\ &\quad-3k\}>0.\end{split}

whenever n6n\geq 6, or (j,k,n){(5,5,5),(3,5,5),(1,5,5),(4,4,5),(2,4,5),(3,3,5),(1,3,5),(j,k,n)\in\{(5,5,5),(3,5,5),(1,5,5),(4,4,5),(2,4,5),(3,3,5),(1,3,5), (2,2,5)}(2,2,5)\}, or (j,k,n){(3,4,4),(1,4,4),(2,3,4),(1,2,4)}(j,k,n)\in\{(3,4,4),(1,4,4),(2,3,4),(1,2,4)\} , or (j,k,n){(3,3,3),(1,3,3),(j,k,n)\in\{(3,3,3),(1,3,3), (2,2,3)}(2,2,3)\}, whereas Mo(PL(1,2,2))Mo(CL(2,3))=0Mo(P_{L}(1,2,2))-Mo(C_{L}(2,3))=0, Mo(PL(1,1,3))Mo(CL(2,3))=24<0Mo(P_{L}(1,1,3))-Mo(C_{L}(2,3))=-24<0, Mo(PL(1,1,1))Mo(CL(1,2))=48<0Mo(P_{L}(1,1,1))-Mo(C_{L}(1,2))=-48<0.

Subcase 1.2. If hh is odd, then nj+kn\leq j+k, and we have

Mo(PL(j,k,n))Mo(CL(2,j+k+n2))=6{3j23k23n2+2kj+6jn+10kn8k12j4n+8}=6{3j(kj)+3k(nk)+(3n+4)(j+kn)+k(nj)+3(k+j)(n5)+3kj+8}=6{3j(kj)+3k(nk)+(3n+4)(j+kn)+k(nj)+3(k+j)(n4)4j+8}=6{3j(kj)+3k(nk)+(3n+4)(j+kn)+k(nj)+3(k+j)(n3)7j3k+8}>0.\begin{split}&\quad Mo(P_{L}(j,k,n))-Mo(C_{L}(2,j+k+n-2))\\ &=6\{-3j^{2}-3k^{2}-3n^{2}+2kj+6jn+10kn-8k-12j-4n+8\}\\ &=6\{3j(k-j)+3k(n-k)+(3n+4)(j+k-n)+k(n-j)+3(k+j)(n-5)+3k-j\\ &\quad+8\}\\ &=6\{3j(k-j)+3k(n-k)+(3n+4)(j+k-n)+k(n-j)+3(k+j)(n-4)-4j+8\}\\ &=6\{3j(k-j)+3k(n-k)+(3n+4)(j+k-n)+k(n-j)+3(k+j)(n-3)-7j-3k\\ &\quad+8\}>0.\end{split}

whenever n5n\geq 5, or (j,k,n){(4,4,4),(2,4,4),(3,3,4),(1,3,4),(2,2,4)}(j,k,n)\in\{(4,4,4),(2,4,4),(3,3,4),(1,3,4),(2,2,4)\}, or (j,k,n){(2,3,3),(1,2,3)}(j,k,n)\in\{(2,3,3),(1,2,3)\}, whereas Mo(PL(2,2,2))Mo(CL(2,4))=24<0Mo(P_{L}(2,2,2))-Mo(C_{L}(2,4))=-24<0, Mo(PL(1,1,2))Mo(CL(2,2))=24<0Mo(P_{L}(1,1,2))-Mo(C_{L}(2,2))=-24<0.

Case 2. nh2+1n\geq\lfloor\frac{h}{2}\rfloor+1

Mo(PL(j,k,n))Mo(CL(2,j+k+n2))=6{4(j+n)(k2)+4k(j1)+8}>0,Mo(P_{L}(j,k,n))-Mo(C_{L}(2,j+k+n-2))=6\{4(j+n)(k-2)+4k(j-1)+8\}>0,

whenever k2k\geq 2, whereas (j,k,n)=(1,1,n)(j,k,n)=(1,1,n) (n4)(n\geq 4) with Mo(PL(1,1,n))Mo(CL(2,n))=24(1n)<0Mo(P_{L}(1,1,n))-Mo(C_{L}(2,n))=24(1-n)<0.

Thus, we have (1)(1) If G{PL(1,1,h3),PL(1,2,2),PL(2,2,2)}G\notin\{P_{L}(1,1,h-3),P_{L}(1,2,2),P_{L}(2,2,2)\}, then Mo(G)>Mo(CL(2,h3))Mo(G)>Mo(C_{L}(2,h-3)); (2)(2) If G{PL(1,1,h3),PL(2,2,2)}G\in\{P_{L}(1,1,h-3),P_{L}(2,2,2)\}, then Mo(G)<Mo(CL(2,h3))Mo(G)<Mo(C_{L}(2,h-3)); (3)(3) If GPL(1,2,2)G\cong P_{L}(1,2,2), then Mo(G)=Mo(CL(2,3))Mo(G)=Mo(C_{L}(2,3)).

The proof is completed \blacksquare

Comparing the Mostar indices of PL(1,2,2)P_{L}(1,2,2) with PL(1,1,3)P_{L}(1,1,3), PL(2,2,2)P_{L}(2,2,2) with PL(1,1,4)P_{L}(1,1,4). By Lemma 2.1, we have Mo(PL(1,1,3))Mo(PL(1,2,2))=24(3032)=48<0Mo(P_{L}(1,1,3))-Mo(P_{L}(1,2,2))=24(30-32)=-48<0, and Mo(PL(1,1,4))Mo(PL(2,2,2))=24(4042)=48<0Mo(P_{L}(1,1,4))-Mo(P_{L}(2,2,2))=24(40-42)=-48<0. By Lemma 3.3, Theorem 3.5, Theorem 4.4, Lemma 5.1 and Lemma 5.2, we obtain the third minimum Mostar index of tree-like phenylenes .

Theorem 5.3

Let G𝒫hG\in\mathscr{P}_{h}, and G{Lh,CL(1,h2),CL(1,h4,1)}G\ncong\{L_{h},C_{L}(1,h-2),C_{L}(1,h-4,1)\}. Then Mo(G)Mo(PL(1,1,h3))Mo(G)\geq Mo(P_{L}(1,1,h-3)), with equality iff GPL(1,1,h3)G\cong P_{L}(1,1,h-3).

6 More about (edge) Mostar indices

In this section, we investegate the correlation between boiling points (BP) of benzenoid hydrocarbons and edge Mostar indices. The 21 benzenoid hydrocarbons were shown in Figure 5. The experimental values of boiling points of benzenoid hydrocarbons of Table 1 were taken from [24]. The experimental values of Mostar indices of 2121 benzenoid hydrocarbons of Table 1 were taken from [8]. With the data of Figure 1, scatter plots between BP and edge Mostar indices were shown in Figures 6. We obtain that the correlation coefficient (RR) between boiling points and edge Mostar indices is about 0.9647, and

BP=0.9092×Moe(G)+252.6.BP=0.9092\times Mo_{e}(G)+252.6.

From [8], we know that the correction coefficient (R) between boiling ponits of benzenoid hydrocarbons and Wiener index is 0.9642, Mostar index is 0.9573, the first status connectivity index is 0.9677, the second status connectivity index is 0.9165, the first eccentric connectivity index is 0.9315, the second eccentric connectivity index is 0.8263. We compare the correction coefficient of edge Mostar index with other distance-based indices, we find the edge Mostar index is also a good predictor. The boiling points and edge Mostar indices are highly correlated since the correction coefficient (R) between boiling ponits of benzenoid hydrocarbons and edge Mostar index is 0.9647. It is worth noting that the regression model for the boilding point and edge Mostar index only applies to benzenoid hydrocarbons. We do not know whether it applies to phenylenes, which needs further study.

Refer to caption
Figure 5: 21 benzenoid hydrocarbons.
Table 1: Experimental values of BP and Moe(G)Mo_{e}(G), Mo(G)Mo(G) of 2121 benzenoid hydrocarbons
No. BP(oC)BP(^{o}C) Mo(G)Mo(G) Moe(G)Mo_{e}(G) No. BP(oC)BP(^{o}C) Mo(G)Mo(G) Moe(G)Mo_{e}(G)
11 218218 3232 4040 1212 542542 224224 300300
22 338338 8888 110110 1313 535535 248248 310310
33 340340 6464 8080 1414 536536 232232 290290
44 431431 160160 200200 1515 531531 264264 330330
55 425425 160160 200200 1616 519519 256256 320320
66 429429 144144 180180 1717 590590 252252 342342
77 440440 128128 160160 1818 592592 302302 390390
88 496496 198198 258258 1919 596596 300300 388388
99 493493 184184 240240 2020 594594 300300 388388
1010 497497 172172 222222 2121 595595 322322 412412
1111 547547 236236 316316
Refer to caption
Figure 6: Scatter plot between BP of benzenoid hydrocarbons and Moe(G)Mo_{e}(G).

7 Concluding Remarks

Among the research of quantum chemistry, computational chemistry and mathematical chemistry, the research of chemical indices is currently one of the more popular areas, as these chemical indices have proven to have a wide range of applications in QSAR, QSPR relationships for new drug discovery, molecular design, hazard estimation of compounds, numerical coding of chemical structures, database search, prediction of bioactivity, prediction of physicochemical properties of molecular materials. In this study, we determine the first three minimum values of the Mostar index of tree-like phenylenes with a fixed number of hexagons and characterize all the tree-like phenylenes attaining these values. Quite unexpectedly, the minimum and second minimum tree-like phenylenes are all in the phenylene chains, but the third minimum tree-like phenylenes are not in the phenylene chains. The results could be of some interest to researchers working in chemical applications of graph theory.

References

  • [1] M. Arockiaraj, J. Clement, N. Tratnik. Mostar indices of carbon nanostructures and circumscribed donut benzenoid systems, Int. J. Quantum. Chem. 119 (2019) #e26043.
  • [2] M. Arockiaraj, J. Clement, N. Tratnik, S. Mushtaq, K. Balasubramanian. Weighted Mostar indices as measures of molecular peripheral shapes with applications to graphene, graphyne and graphdiyne nanoribbons, SAR QSAR Environ. Res. 31 (2020) 187–208.
  • [3] K. Balasubramanian, Topological peripheral shapes and distance-based characterization of fullerenes C20-C720 : existence of isoperipheral fullerenes, Polycyclic Aromat. Compd. doi: 10.1080/10406638.2020.1802303.
  • [4] F. R. K. Chung, Spectral Graph theory, American Mathematical Society, Providence, RI, USA, 1997.
  • [5] H. Chen, H. Liu, Q. Xiao, J. Zhang, Extremal phenylene chains with respect to the Mostar index, Discrete Math., Algor. Applicat. doi:10.1142/S1793830921500750.
  • [6] H. Deng, S. Chen, J. Zhang, The PI index of phenylenes, J. Math. Chem. 41 (2007) 63–69.
  • [7] K. Deng, S. Li, Extremal catacondensed benzenoids with respect to the Mostar index, J. Math. Chem. 58 (2020) 1437–1465.
  • [8] K. Deng, S. Li, On the extremal values for the Mostar index of trees with given degree sequence, Appl. Math. Comput. 390 (2021) #125598.
  • [9] T. Došlić, I. Marthinjak, R. Škrekovski, S. Tipurić Spužević, I. Zubac, Mostar index, J. Math. Chem. 56 (2018) 2995–3013.
  • [10] B. Furtula, I. Gutman, T. Zeljko, A. Vesel, I. Pesek, Wiener-type topological indices of phenylenes, Indian. J. Chem. A. 41 (2002) 1767–1772.
  • [11] I. Gutman, B. Furtula, The total π\pi-electron energy saga, Croat. Chem. Acta. 90 (2017) 359–368.
  • [12] I. Gutman, B. Furtula, A Kekulé structure basis for phenylenes, J. Mol. Struct.: THEOCHEM, 770 (2006) 67–71.
  • [13] I. Gutman, P. Petković, P. V. Khadikar, Bounds for the total π\pi-electron energy of phenylenes, Rev. Roum. Chim. 41 (1996) 637–643.
  • [14] I. Gutman, A. R. Ashrafi, On the PI index of phenylenes and their hexagonal squeezes, MATCH Commun. Math. Comput. Chem. 60 (2008) 135–142.
  • [15] I. Gutman, Ž. Tomović, Cyclic conjugation in terminally bent and branched phenylenes, Indian. J. Chem. A. 40 (2001) 678–681.
  • [16] F. Gao, K. Xu, T. Došlić, On the difference of Mostar index and irregularity of graphs, Bull. Malays. Math. Sci. Soc. 44 (2021) 905–926.
  • [17] S. Huang, S. Li, M. Zhang, On the extremal Mostar indices of hexagonal chains, MATCH Commun. Math. Comput. Chem. 84 (2020) 249–271.
  • [18] F. Hayat, B. Zhou, On Mostar index of trees with parameters, Filomat, 33 (2019) 6453–6458.
  • [19] F. Hayat, B. Zhou, On cacti with large Mostar index, Filomat, 33 (2019) 4865–4873.
  • [20] M. Imran, S. Akhter, Z. Iqbal, Edge Mostar index of chemical structures and nanostructures using graph operations, Int. J. Quantum Chem. 120 (2020) #e26259.
  • [21] H. Liu, X. Fang, Extremal phenylene chains with respect to detour indices, J. Appl. Math. Comput. 67 (2021) 301–316.
  • [22] H. Liu, L. Song, Q. Xiao, Z. Tang, On edge Mostar index of graphs, Iranian J. Math. Chem. 11 (2020) 95–106.
  • [23] J. B. Liu, Q. Zheng, Z. Q. Cai, S. Hayat, On the Laplacians and normalized Laplacians for graph transformation with respect to the dicyclobutadieno derivative of [n] Phenylenes, Polycyclic Aromat. Compd. doi:10.1080/10406638.2020.1781209.
  • [24] Milano Chemometrics & QSAR Research Group, Molecular Descriptors: the free online resource, Milano Chemometrics and QSAR Research Group, http://www .moleculardescriptors.eu/dataset/dataset.htm. Accessed Dec 2017.
  • [25] L. Pavlovic, I. Gutman, Wiener numbers of phenylenes: an exact result, J. Chem. Inf. Comput. Sci 37 (1997) 355–358.
  • [26] A. Tepeh, Extremal bicyclic graphs with respect to Mostar index, Appl. Math. Comput. 355 (2019) 319–324.
  • [27] N. Tratnik, Computing the Mostar index in networks with applications to molecular graphs, Iranian J. Math. Chem. 12 (2021) 1–18.
  • [28] K. P. C. Vollhardt, The phenylenes, Pure Appl. Chem. 65 (1993) 153–156.
  • [29] Q. Xiao, M. Zeng, Z. Tang, H. Deng, H. Hua, Hexagonal chains with the first three minimal Mostar indices, MATCH Commun. Math. Comput. Chem. 85 (2021) 47–61.
  • [30] Q. Xiao, M. Zeng, Z. Tang, H. Hua, H. Deng, The hexagonal chains with the first three maximal Mostar indices, Discrete Appl. Math. 288 (2021) 180-191.
  • [31] Z. Zhu, J. B. Liu, The normalized Laplacian, degree-Kirchhoff index and the spanning tree numbers of generalized phenylenes, Discrete Appl. Math. 254 (2019) 256-267.