## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Roger B. Eggleton

Mathematics Department, Illinois State University Normal, IL 61790, U.S.A.

**For Severino V. Gervacio, on his 65th birthday.
**

Abstract: This paper provides an introduction to graphs which are linked pairs, linked chains, or linked cycles. These structures allow for a generalized notion of graph factorization, permitting regular decomposition of many graphs into isomorphic links.

1. An unconventional “factorization” Consider the graph H in Fig. 1. This is an identity graph, that is, its only automorphism is the identity map on its vertices. Nevertheless it has a more subtle structural regularity which is of interest. For any subset V of the vertex set of H, let H[V] be the subgraph of H induced by V. In particular, for A = {1, 2, 3, 4, 5,6,7} and B = {1, 2, 3, 4,8,9,10} let G0 = H[ A] and G1 = H[B] . These induced subgraphs both have order 7 and size 6. Indeed, each is isomorphic to the smallest identity tree T (a 3-legged spider with legs of length 1, 2 and 3), so H contains two isomorphic induced subgraphs. In ! two subgraphs comprise ! H. Thus H “factorizes” into fact G0 " G1 = H , so these ! two copies of the tree T. However G0 " G1 = H[1, 2, 3, 4] # 2 P2 , so G0 and G1 are not edge-disjoint. This “factorization” differs from a conventional factorization in two ways: it is less demanding insofar as its “factors” may share edges, but it is more demanding insofar as its “factors” are not simply subgraphs, they are ! by their vertex sets. ! ! subgraphs which are induced 3 7 4 5 6 H[A]

! !

9

8 H

2

1

10

H[ B ]

Figure 1. An unconventional “factorization” of H. 2. Linked pairs Let us generalize from the example in Fig. 1. Choose any finite simple graph G of order g ≥ 2, and an induced proper subgraph K of order k, where 1 " k < g . Label the vertices of K with the positive integers [1..k], and extend this labeling so that the positive integers [1..g] label all the vertices of G. Let K* be an

!

. so " (G0 ) = G1 in Fig.(2 g " k )]. H = G " G#. There are eight possible isomorphisms from K* to K. ! ! graph H is the linked pair with initial link G The new 0 = G . ! ! ! g ! ! G ! c b d K* = K b a e f a d e σ: 9 σ(a) = d σ(b) = e σ(d) = b σ(e) = a 6 5 2 1 G0 3 4 7 8 4 3 2 G1 1 10 Figure 2.GRAPHS WHICH ARE LINKED STRUCTURES 2 ! induced subgraph of G isomorphic to K. The subgraph is the kernel of the linked pair. " (7) = 10 . and " ( j ) = g + i if j is the ith vertex of G not in K*. . k ] " [( g + 1).. The resulting linked pair H = G0 " G1 is precisely the graph H in Fig.. K. " (2) = 4. as specified in Fig. Then " is the bijection " (1) = 3.. labeled as in Fig.. in [1]. so K = H[ A " B] = H[1. ! " and link isomorphism . is labeled by the positive integers [1. These notions were introduced and developed in ! ! purpose of collaboration with Peter Adams and James MacDougall.7] so that K is labeled with [1. " (3) = 2.(2 g % k )] . K* and σ. graph isomorphism from G to ! The resulting graph. the 3-legged spider with legs of length 1. 2. 2 and 3. The procedure is the linked pair construction ! with ingredients G. ! " (4) = 1.. k ] .. One possible relabeling of G with [1. K = G " G 0 1 ! ! ! K* is the prekernel ... a copy of G labeled by [1.4] yields G0 in Fig. let G be T. g ] # [1. and σ is the ! shift. g ] and B = [1. there is always at least one available choice for K*. " (6) = 9. k ] " [( g + 1). Choose one of them to be σ. k ] $ [( g + 1). 2. where ! G0 = G = H[ A] and G1 ! = G" = H[B].. and let K* = K. 2.(2 g # k )]. Let K be the induced subgraph 2 P2 shown. Extend σ to a bijection " : [1. by requiring " (i ) = # (i ) for every vertex i of K*. since K* could be K itself. Then " induces a G". copy link G1 = G". ! ! Note that if A = [1. 1. The ! ! formal introduction to the main constructions the present paper is to give a less in that paper. Let " : K * # K be an isomorphism. " (5) = 8. For example.(2 g # k )] . 2. Linked pair construction for H.. then H != G0 " G1 ..

e . along with the new linked pair H = G0 " G1 . " (7) = 10 . " ( e) = e . " ( d ) = d . For the first. " (5) = 2. A third linked pair. Relabeling G like G0 in Fig. K* be the same as in Fig. let G and K be the same as in Fig. Then G1 = " (G0 ) . 3.GRAPHS WHICH ARE LINKED STRUCTURES 3 To see more of the richness of the linked pair construction. 2. but for σ take the isomorphism " ( a) = b . " (c ) = b . 2 results in " being " (1) = 8. " ( e) = e . 1. let us examine two relatives of the example constructed in Fig. K. " (3) = 3. ! ! 6 ! 5 2 1 9 8 1 2 G1 3 4 10 G0 6 3 7 ! 4 ! ! 4 10 7 3 5 8 9 2 H 1 Figure 3. For a further example. let the ingredients G. " (4) = 4. 2 results in " being " (1) = 2. A linked pair related to Fig. " (6) = 9. 2 and “deconstructed” in Fig. 4. ! ! ! ! 6 ! ! H 5 4 7 3 3 2 4 7 5 6 1 G0 10 2 1 8 G1 9 4 3 ! ! 9 10 2 1 8 Figure 4. are as shown in Fig. " (3) = 9. " (5) = 8. but now choose K * = G[ b . " (7) = 3 . " (4) = 4. . " (6) = 10. " (2) = 1. and the new linked pair H = G0 " G1 . 2. Relabeling G like G0 in Fig. f ] and let σ be the isomorphism " ( b) = a . The corresponding G1 = " (G0 ) is shown in Fig. c . " (2) = 1. " ( b) = a . " ( f ) = d . 2.

This larger graph is a linked chain. let us iterate the linked pair construction shown in Fig. 56}. " ) is explicitly described by the equations H[ A " B " C " D] = H . " (10) = 13 . " (2) = 4. Thus. take the last copy link as ! link and act on it with the newly extended link isomorphism. ! A " B = B " C = C " D = k. In our example the last ! chain of isomorphisms is actually a chain of three equalities. Linked chains Iterating the linked pair construction produces a larger graph. The regular structure of H = H 4 (G.. B. 34. 35. iteration and make as many links as we ! ! ! ! 6 5 2 1 12 11 1 2 ! 3 G0 4 7 9 8 4 3 15 4 G2 3 13 3 4 G3 14 1 2 16 G1 2 1 10 Figure 5. The union of the first four links is the 4-linked chain H = G0 " G1 " G2 " G3 . The equations ! imply that A. C . As an example. For instance. 5. " (3) = 2. the first four links are shown necessary to stop there: we could continue ! in Fig. " (13) = 16 . " (4) = 1. D are subsets of the vertices of H. because we chose K* = K in the ingredients of the construction. 47. Then the copy link is G1 = " (G0 ) . H[ A] " H[B] " H[C ] " H[D] " G. 25. 2. " (6) = 9. " (5) = 8. extend " by adding an image vertex for each vertex not in the current domain of " . extend " by " (11) = 14. such that A = B = C = D = g. For G = G0 take the tree T with vertex set [1. " (7) = 10 . and so on. producing G3 = " (G2 ) . For each iteration.7] and edge set {12. ! " (12) = 15. pro! ducing! the copy link G2 = " (G1 ) .GRAPHS WHICH ARE LINKED STRUCTURES 4 3. ! . " (9) = 12. Four consecutive links of a linked chain. ! where G is the initial link and K is the kernel of the linked pair construction. !new initial ! for the first iteration " is extended by adding " (8) = 11. H[ A # B] " ! H[B # C ] " H[C # D] " K . comprising any desired number of isomorphic induced subgraphs which are linked together in a regular way. and for " take " (1) = 3. For the second iteration. But it is not ! like.

The behaviour of the link isomorphism " is central to the structure of a linked chain. For example. A single iteration of a! linked pair construction simply ! ! involves extending its link isomorphism by adjoining to each open trajectory one new directed edge ! ! ! to a new terminal vertex..13] and D = [1. B = [1.. k ] " [( g + 1).... ! Then the " -digraph is the directed! graph D(" ) with vertex set [1. g ] . on the other hand.(2 g # k )] ... and every vertex has positive total degree. 2. With ! ! ! ! the general labeling conventions used in Section 2.16] for our example. let Dn (" ) denote the extended digraph after n further iterations of the linked pair construction.(2 g " k )].4] " [14. and a directed cycle is a closed G which have indegree 0 in D(" ) : each is the initial vertex of an open trajectory ! linked pair construction has at least one open trajectory...4] " [11. Since " is one-to-one and has ! range [1.. every vertex in D(" ) has outdegree at most 1 and ! indegree at most 1. There are g " k > 0 vertices of ! trajectory. all µ vertices of a closed trajectory of order µ belong to K * "K # G = G0 . also A " B = B " C = C " D = [1.. Therefore every ! ! component of D(" ) has size at least 1. ! ! ! .g] and K is labeled with [1. !1 vertices of G = G0 and one Note that an open trajectory of order µ comprises µ " ! terminal vertex which is in G1 but not G0 . C = [1. ! ! ! 1 3 1 3 4 5 6 7 2 8 9 10 5 6 7 4 8 9 10 2 11 12 13 14 15 16 D0 (" ) D2 (" ) Figure 6... 2 and 5.k]. The components of D(! " ) are the trajectories of " : a directed path is an open ! trajectory. it is possible to have no closed trajectory. from the current terminal vertex Multiple iterations ! ! simply repeat this procedure: writing D0 (" ) = D(" ) for the " -digraph of the linked pair. 6 shows D0 (" ) and D2 (" ) for the linked pair construction in Fig.GRAPHS WHICH ARE LINKED STRUCTURES 5 We have A = [1. and is! either a ! directed path or a directed ! cycle.(2 g " k )] and with a directed edge i " # (i ) for each i " [1.7] .4] " [8.. where G is labeled with [1. so g = 7 and k = 4. the linked pair H = G0 " G1 is ! labeled by [1.10].4]. On the other so every ! ! for a linked pair construction hand.. so we clarify the behaviour of " with a visual representation. representing respectively the first two links ! ! 5. The " -digraphs for Figs. and all four links shown in Fig. Fig.

The iterated " -digraph D2 (" ) reveals the essential structure of H 4 (G. ! ! the linear ! ordering to the trajectories of " corresponding to Assign a linear ordering of their least members. The " -digraph D(" ) has another important use: the trajectories of " show us how to create an intrinsic labeling of the vertices of the relevant linked pair.. the " -digraph D(" ) in Fig. The new intrinsic labeling of any vertex v has the form (i.10 . j) with 0 " j " µ # 1.8 ! ! ! ! ! . 7). where the primary index i indicates the trajectory which contains v. A simple method would be to take the arrangement of 16 vertices used for D2 (" ) in Fig.k]. Another method is to arrange the vertices of each trajectory in D2 (" ) in a ! ! circle to reflect the rotational symmetry of the closed trajectory (Fig. insofar as they more clearly show the structural regularity of H.. 2. has its trajectories in the linear order < 6. j). 6. " ) = G0 # G1 # G2 # G3 comprising the links in Fig. 2. and insert the edges of the four links shown in ! ! Fig. ! Begin with the initial link G labeled by [1. ! Note that . " ) . 6. " ) . so it is desirable to apply D2 (! " ) to construct a drawing of H 4 (G. " ) with links in Fig.g] and the kernel K labeled by [1. such that any associated linked chain has a natural extension of this labeling. For instance.. ! and the secondary index j indicates the place of v in the sequence of all vertices ! comprising its trajectory. 5.9 < 7. ! (directed path). 5. 4) <!5. sequentially its vertices have intrinsic labels (i. 5? There is no “best” drawing of H. If the ith trajectory has order µ ..GRAPHS WHICH ARE LINKED STRUCTURES 6 ! How should we draw the 4-linked chain H = H 4 (G. ! 12 10 11 15 H[ D] 1 14 H[A] ! 7 H 4 3 13 H[C] 8 9 2 16 5 6 H[ B ] Figure 7. corresponding to the linked pair in Fig. omit all the directed edges.. indicates an open trajectory and (. and j " Zm if the trajectory is closed ( m = µ is the modulus (1. but surely some drawings are better than others. The 4-linked chain H 4 (G.. with the convention that the! first vertex of a directed cycle is its least member in the prior labeling. 3. where j " Z if the trajectory is open.) indicates a closed trajectory (directed cycle).

4 " (1. 3) mod 4. and the n-linked chain is ! H n (G.0). 2: 1 " (1. ! 2) (4. depending on whether the ith trajectory is a directed path. " is a graph isomorphism. j ) = (i . 1) (3. 3 " (1. " ) = U0 #r< n Gr = U0 #r< n " r (G) . where the second index increments by integer addition or by addition ! modulo m. 7 " (4. if the initial link is G0 = G. 3 and 4 to compare their structures with that For Fig. 2). 0) (2. in Figs. in the second line all secondary indices are simply integers. 7. 5 " (2. 3) (4. 3) (1. 3.10 " (4. 2 " (1. 1) (1.1). Given the intrinsic labeling of any linked pair. 3) (3. the linear ordering of the trajectories in D(" ) is ! ! ! . In the first line all secondary indices are residues modulo 4 (the modulus of the ! closed trajectory is 4). " ) in Fig. ! 9 " (3.1) . or a directed cycle of order µ = m. The key is the simplicity of the action of " on the intrinsic labels of the vertices: " (i . 8 shows ! the intrinsic labeling of the 4-linked chain in Fig. ! ! (3. 2) H[τ2(V)] (2. 1) (4. j + r) for every integer r ≥ 0. j + 1) . The intrinsic labeling of the 4-linked chain H 4 (G. we can now neatly specify ! every linked chain which it generates. 8 " (2. 2. ! 3) H[τ3(V)] H[V] (2. then the rth r copy link is Gr = " (G) . Fig. 0) (4. If V is the vertex set of the initial ! link G0 = G. 2) (2. Of course. 6 determine the following intrinsic vertex labeling for the linked pair in Fig. ! ! where " r (i . For instance. Now that we have seen how the " -digraph D(" ) reveals the structure of a linked pair and its related linked chains.0). 1) Figure 8. 0) H (1.GRAPHS WHICH ARE LINKED STRUCTURES 7 of the trajectory).1). To illustrate.1). 6 " (3. j ) = (i . 0) H[τ(V)] ] (3. the trajectories of D(" ) in Fig.0). and its action on the edges is induced by its action on! the vertices. we can briefly examine the linked pairs ! of the linked pair in Fig. 2) (1. 7.0). then " r (V ) is the vertex set of Gr = " r (G) . So.

4. 4 are ! 5 " (1. the linear ordering of the trajectories in D(" ) is ! 5. j # 1) for! and to form the infinite digraph D"# ($ ) by modifying D(" ) so that each open trajectory extends backward endlessly. 3 The interested reader is invited to use this data to create drawings of the two ! related to the one in Fig. For Fig. 4 " (3. 7 " (2. 6 " (5. so the intrinsic labels for the vertices of the linked pair in Fig. $ ) " H1 " n (G.1). ! ! ! ! . $ ) = Ur%0 G"r = Ur%0 $ "r (G) .0). 7 " (6. $ ) is isomorphic to an induced proper subgraph of itself. 8.0).1. ! Again. " ) . For a given linked pair with " -digraph D(" ) . " n ( H ±# (G.1). 6 " (4. the union of all forward and backward links.0).1).0).1) .GRAPHS WHICH ARE LINKED STRUCTURES 8 (1. ! H +" (G .9 < 7. !" (2. so H"# (G. 2 " (1. for every n ≥ 1. $ ) H"# (G r& n ! ! above infinite ! linked chains is However. so for every n ∈ Z. It is also natural to specify " #1 (i . 2. 3. # ) . 4-linked chains ! 4. more important than either of the the free linked chain. 2). # ) . # )) = Ur% n # r (G) = H +" (G. 5 "! (4. 3).8 < 6. 4 " (3.0) mod 1. but it is also natural to consider linked chains with infinitely many links. ! The mapping " is a nontrivial automorphism of H ±" (G. ! ! H"# (G. 9 " (2. 10 " (6. 2 " (1. $ ) % $ " n ( H"# (G.0). ! so the intrinsic labels for the vertices of the linked pair in Fig. # ) $ # n ( H +" (G. j ) .10 . 3 " (2. Indeed. 10 " (4. 8 " (4. Indeed. ! !. 2). It is consistent with our earlier notation ! to write G"r = (# "1 )r (G) = # "r (G) for each integer r ≥ 0. ! ! The mapping " is a contractive isomorphism on H +" (G. 9 " (5. Free linked chains We have discussed n-linked chains for any positive integer n. for every n ≥ 1. 3 are 1 " (1. H ±" (G.1). # ) = Ur$0 Gr = Ur$0 # r (G) . so H +" (G. # ) . the mapping " #1 is a contractive isomorphism on H"# (G. 2) < (3) < (4) < 5. 8 " (1. ) = (i .10 . " )) = H ±# (G. The union of all the corresponding links is the forward linked chain.0) mod 1.0). # ) is isomorphic to an induced proper subgraph of itself.1) mod 2. j! each intrinsic label (i . The union of all these links ! ! is the backward linked chain. we write D+" (# ) to denote the extended digraph in which each open trajectory has been extended forward endlessly. 1 " (1.8 < 7. # ) = Ur $ Z Gr = Ur $ Z # r (G) . $ )) = U $ "r (G) = H"# (G.9 < ( 4) < 6. # ) & H n &1 (G. $ ) . ! H +" (G.0).1) .0) mod 1.

. # )/( m1 . 2) ! ! … … ! ! … (2. 2.. ! ! ! ! 1 " (1. If there are t trajectories. a) = (i . 3) ! (3. It is natural to call ! m the modulus of the collapsed trajectory. m2 . # ) we choose a positive integer m " µ and make the identification (i . produced by collapsing each open trajectory of " into a closed trajectory. Linked cycles A linked cycle is ! a special kind of homomorphic image of a free linked chain. Also # ". Fig.. "1) 6 " (3. ! ! ! ! ! .. "1) 7 " (4. # ) = H ±$ (G. The free " -digraph for the linked pair in Fig. j ) = (i . 3) ! ! ! ! ±" ! ! ! ! Figure 9. mt ) . in the ! !are adjacent whenever they are equivalent to two resulting graph two vertices vertices which are adjacent in the free linked chain. 9 shows the free " -digraph D±" (# ) for the linked pair introduced in Fig. is an automorphism. How is this collapsing of open trajectories achieved? If the ith trajectory in ! D(" ) is an open trajectory of order µ . 3) ! ! ! ! (4.1) 12 " (3. b) whenever a " b (mod m) . Free linked chains are particularly important because of their role in this construction. We close this section by noting that the infinite linked chains described here correspond to the infinite extensions D+" (# ). 2) 16 " (4.0) 4 " (1. defined by # "(i . to collapse the corresponding trajectory of the free linked chain H ±" (G.. ! … … … ! D (# ) 5. D$" (# ) and D±" (# ) of the " -digraph D(" ) of the underlying linked pair construction. Formally.1) 2 " (1. Collapse all open trajectories place for closed in this way.1) 11 " (2.0) 10 " (4. 2. . matching the terminology already in ! ! trajectories. j + 1) mod mi for 1 " i " t . the resulting graph may be denoted by ! ! H "(G. "1) 5 " (2. this replaces Z by Zm as the secondary index set for vertices in the ith trajectory.0) 9 " (3.. and ( m1 .0) 8 " (2. 2) 14 " (2. . m2 . so that the resulting graph is a cyclic sequence of isomorphic induced subgraphs (links) in which consecutive pairs of links all have isomorphic intersections. In the next section we use free linked chains to construct infinite families of new graphs with nontrivial automorphism groups.1) 13 " (4. 3) ! ! 3 " (1. 2) 15 " (3.GRAPHS WHICH ARE LINKED STRUCTURES 9 Thus free linked chains are highly symmetric: they have infinite automorphism groups. mt ) is the list of moduli.

For ( m1 . Any its modulus for H "(G. a4 .. However.. # ) to qualify as a linked cycle constructed from G and " we require it to be appropriately composed of induced copies of G. m"} # max{µ . 2 yields an n–linked cycle with n = 4 " lcm{ a2 . ! ! We a class of linked cycles constructed from G and " . ! ! ! . Improper adjacencies can only ! if at least one of them is a collapsed trajectory... there can be no improper adjacency if m|m" and m" # µ"... and Z2 as the second index set of the other four trajectories. 3 yields an n–linked ! cycle with n = 2 " lcm{ a ..6 a2 . 4 yields a 4–linked cycle. mt ) =! to the free linked chain generated by the linked pair in Fig... m2 .. 2. Again. mt ) = (4... Let ! ! can now specify ! m0 be the least common multiple of the moduli of the closed trajectories ( m0 = 1 ! ! ! if there for product over an ! are no closed trajectories.. Indeed. mt ) satisfies all the sufficient conditions for H "(G. mt ) = (4. 7 ! but with Z4 as the second index set of all four trajectories. . a3 . 4 yields an n–linked cycle free linked chain generated by the linked with n = 6 " lcm{ a1 . (3) H "(G. .. it suffices for its modulus to satisfy m " 2µ # 2 . 4 a2 .. 3 but with Z1 as the second index set ! ! of the second and third trajectories. a4 }. m2 .1. let µ0 be the maximum of the moduli of the open trajectories. # ) is the union of finitely many ! (4) the intersections Gr" # Gr"+1 are isomorphic for ! all r " 0 . µ"} . the list of moduli !illustrate. 2) . in general there are subtle further constraints on the choice of moduli for vertices of G do !collapsing the open trajectories of H ±" (G. so that nonadjacent ! not become adjacent as a result of the collapsing operation. for any positive integers a4 . ! ! To avoid improper adjacencies within a collapsed trajectory of order µ . for! ! ( m1 .. by the usual convention ! ! empty set). # ) . a3 . 4) applied to the free linked chain generated by ! example. and let ! c be a positive integer such that cm0 " 2µ0 # 2 . If occur between two trajectories both trajectories are collapsed trajectories. . # ) to be mi = cm0 ni . specifically: (1) G is the subgraph induced by the vertex set V = {(i . of orders µ and µ" . a4 }. the 1 2 t 4 5 6 2-linked cycle is precisely the graph in Fig. choose !integer. a2 .. .6 a 4 ) to collapse the ! ! ! pair in Fig. where ni is any positive such choice of ( m1 .1. when ( m1 . 2 a5 . a . 2. 4) . it suffices for their ! trajectory of moduli m and m" to satisfy gcd{m. j ) : 0 " j < mi . If one is a closed order µ and modulus m ! = µ . 4. a }. Indeed. "! (2) G = G and !is isomorphic to each of the subgraphs Gr" such that G0 Gr"+1 = # "(Gr" ) for all r " 0 . m (2.. m .GRAPHS WHICH ARE LINKED STRUCTURES 10 In order for H "(G. . and the other is a collapsed trajectory of order µ" and modulus m". 1 " i " t } . the 4-linked cycle is precisely the graph in Fig. a6 .1. a 5 . m2 . note that the lists of moduli constructed above are in general not ! the only possibilities for collapsing a free linked chain into a linked cycle. m2 ... ! the linked pair in Fig.1. 2 a6 ) applied !2 . m ) = (2.1. . 4 a 4 ) applied to the free linked chain ! generated by the ! ! linked pair in Fig. Again. If the ith trajectory is open. mt ) = (4. ! To any choice of positive integers a2 . # ) to ! be a linked cycle constructed from G and " . m 2 . ! !Hence.1. 4. 4 a3 . mt ) = (6 a1 . ! of the subgraphs Gr".. when ( m . 2 a4 . the list of moduli ( m1 . . 4. applying ( m1 .

. # ) = H ±$ (G. F is a linked cycle with links isomorphic ! to a proper induced subgraph G. mt ) .1. Symmetric graphs as linked cycles ! showed how it We began this discussion with a particular graph (Fig.) ! 6. The resulting infinite graph is a constrained linked chain. and see that the presence of symmetry in a graph typically allows us to “factorize” it as a linked cycle. we can adapt the earlier class of linked cycle solutions to include constrained linked chains by allowing ai to be a ! nonnegative integer rather than requiring it to be a strictly positive integer. and its link isomorphism is # " = $ . the fourth is derived.. while leaving the other open trajectories unchanged. This is an example of the “analytic” viewpoint. such that v " # ( v) is a directed edge for each vertex v of F. 10. every ! finite graph with a derived automorphism can be “cyclically factorized” into isomorphic links (proper induced ! subgraphs) in such a way that consecutive !pairs intersect isomorphically. This terminology can be extended to " itself: we say " is primal if all ! its trajectories are primal. 2.. or the range of possible final products. in ! every other case at least one trajectory has! order greater than 1. If " is the identity map. Let " be any automorphism of a given finite graph F.. Since " is a bijection. Beginning with the pieces in Fig. Its complement is the “synthetic” viewpoint. consistent with the constraints already noted. In this instance " has two trajectories of order 1 and two of order 2. In this final section we return to the analytic viewpoint. by choosing moduli for a proper subset of the open trajectories. 10 has an automorphism " of order 2. The components of D(" ) are the trajectories of " . Thus. The first three trajectories are primal. and the corresponding # " is an automorphism. its free linked chain can be partially collapsed. and is derived if at least one of its trajectories is derived. A trajectory of " ! !trajectories ! a subgraph K or K in F is primal if its vertex set induces . To illustrate. ! Let F be a finite graph with a derived automorphism " . observe that if a linked pair has more than one open trajectory. assigning mi = 0 to be its modulus is notationally consistent and leaves the trajectory uncollapsed. then linked cycles and constrained linked chains. 2) . and we are interested in the final product. whence the intrinsic labeling of vertices in Fig. which fit together to form our first graph. Take a map " ! " with free digraph comprising three closed trajectories identical with D ( # ) ± " ! those of " and one infinite trajectory. then finite linked chains. m2 . all other 1 2 are derived. The " -digraph is the directed graph D(" ) on the vertices of F. mt ) and " = $ # = $ /(m1 . 1) and can be “factorized” into two isomorphic subgraphs with nonempty intersection. then free linked chains. so that " = $ # = $ /(1. the ! graph F in Fig.. We can always find !link ! a proper induced subgraph G and a isomorphism " such that ! ! F = H "(G. # )/( m1 . so each trajectory is closed ! ! (a directed cycle).. most of our discussion has been from the synthetic viewpoint: we have seen how to build linked pairs. since its vertices induce a subgraph 2 K1 in F. then all trajectories have order 1. 2. where something is taken apart so that its pieces can be examined. Then " itself has closed trajectories identical with those of D±" (# ) . (If the ith trajectory is open. where separate pieces are assembled into a whole. In other words... every vertex has outdegree 1 and indegree 1. and an open trajectory of ! ! ! ! ! ! ! ! .GRAPHS WHICH ARE LINKED STRUCTURES 11 To close this section. m2 .

0). and the !modified graph would no longer be a linked cycle. This fascinating graph has many delightful features.1) " (G) Figure 10. so the order of the first trajectory must satisfy 2 " µ " 5 . Both trajectories have order 5. (2. so the vertices of the open trajectories of ! ! induced subgraph G = P[V ] is the initial link of a free linked chain H ±" (G. ! F (1. 10). The intrinsic labeling of P in Fig. (2. ! As a final example. (2.0) (4. the ! bound imposed by the modulus m = 5 applied to the trajectories of D±" (# ) to give ! " # =$. Graph Theory and Combinatorics. (4. the Petersen graph1.0). λ) is adjacent ! to (i.0)} . Of course " is the ±" ±$ ! ! restriction of " to G. Gervacio Conference on ! !24-25.GRAPHS WHICH ARE LINKED STRUCTURES 12 ! order µ " 2 . Then the order of the ith trajectory must satisfy µ > " . consider P. Thus F is a 2-linked cycle with initial link G = P5 (Fig. and both induce the subgraph C5. # ) . define the longest edge ! length λ of the ith trajectory to be the largest integer λ such that (i. 2)} . The infinite trajectory of D±" (# ) requires a modulus m = 2. (2. # )/(1. 2009. and F != H "(G. (3. (1.1). the shift " is the restriction of " to K*.0). The set comprising the nonterminal !" is V = {(1. so we have µ = 2 .0). ! ! ! Given that the ith trajectory is open. The smallest possible solutions minimize the order of the initial link.0) ! ! ! (2.0). # ) = H (G. so both are derived. 2) . But ! note that adding an edge to F between (4. comprising the vertices of the closed trajectories and the nonterminal vertex of the open trajectory of " . (3.1) G (3. The set V = {(1.0) and (4.0). 2. and that of the second trajectory must satisfy 3 " µ " 5 .1). A symmetric graph shown as a linked cycle.1) would make " primal. (3.0) (4. # ) such ! ! ! ! 1 The Petersen graph was used as the logo for the Severino V.1)] = 2 K2 . so take µ = 2 for the first trajectory and µ = 3 for the second. De La Salle University. The order of each must satisfy µ " 5 . 0) in P.0) (3. (3.1). 5) . Take a map " with free " -digraph D±" (# ) comprising two infinite trajectories. (2. ! ! ! and ! the prekernel is K * = " #1 ( K ) = K .1.0). is the vertex set of the initial link G = F[V ] of the free ! linked chain H (G.0). also m " µ is necessary. 11 corresponds to the trajectories of an automorphism " of order 5. The kernel of the underlying linked pair is ! ! K = G " # (G) = F[V " # (V )] = F[(1. Then " itself has two open trajectories. April ! ! . so that " = $ # = $ /(5.

2)} . (1.2) α(G) (1. Here G = P5 generates the Petersen graph P as a 5-linked cycle (Fig.0). The Petersen graph as a linked cycle. 11 is for the pair of orders (2. Peter Adams. The pair (5. Graphs which are linked structures. 2)] = K1 $ !2 the prekernel is K * = " #1 ( K ) = P[(1. 10. The pair (3. # ) = H ±$ (G. 4) also has G = F . K . to appear. the kernel and ! smallest solution.2) α3(G) (1.3) (2. The pair (2. (2.4) (2. ! ! (2. (2. (2.1) α4(G) G (2. having two derived trajectories. and the shift is " = " K*.0).0) (1.1). (2.0). and each gives ! rise to a “factorization” of P as a 5-linked cycle. This introduction is the first paper to appear on graphs as linked structures. so ! 2).1). prekernel contain the corresponding subgraphs of the ! (1. 3).0) (1.4) (2.! ! in this case the initial link is G = P[V ] = F . the interested reader is invited to calculate the “factorization” of P resulting from an automorphism of order 2. Reference [1] Roger Eggleton. 11). The kernel of the underlying linked pair is K = G " # (G) = P[V " # (V )] = P[(1.0). Note that there are 12 solutions for the orders of the " trajectories.1) (2.1)] . where F happens to be a 5-cycle with a ! pendant vertex. and " is the restriction of " to G. .1). but on a different vertex set V. (2. (2. # )/(5. 5) . We anticipate the forthcoming publication of a more formal account [1].GRAPHS WHICH ARE LINKED STRUCTURES 13 ! that P = H "(G. and in this case G is a ! in common. the graph in Fig. As another application of this analytic approach. 5) is the largest solution. moreover. 3) has V = {(1.3) P α2(G) Figure 11. (1. The case in Fig.1). In all 12 solutions the initial link pair of 5-cycles with a single edge ! G contains the initial link P5 of the smallest solution. and James MacDougall.

- Problem Solveingchp 3
- Query Languages for Graph Databases
- Combinatorics Notes
- graph
- Networkx Reference
- 3.IJAEST Vol No 5 Issue No 2 Content Based 3D Model Classification Using Geometric Graph Representation 116 122
- konsul1agama
- Chapter 5 - Graphs
- ijcopi12
- Mehvish I. Poshni, Imran F. Khan and Jonathan L. Gross- Genus distributions of graphs under edge-amalgamations
- 5.Combinatorics
- 10.1.1.64
- Probabilistic Network Theory
- ECE242.Fall11
- Peter Hamburger, Gyorgy Petruska and Attila Sali- Saturated Chain Partitions in Ranked Partially Ordered Sets, and Non-Monotone Symmetric 11-Venn Diagrams
- 09 Clustering
- Module 3
- Theory of Computer Science
- Python Examples
- C10-1121
- icc07channel
- Protecting the Sensitive Information on Osn
- Choi-A Decomposition Approach
- Chap 7
- AGT Take Home Exam - Gaston Bengolea
- 1 in indaia ans to malaysia wsing
- MC0063(B)-Unit10-fi
- Advanced Data Structures UVT
- IJETTCS-2014-04-20-099
- 27287326-Algorithms-and-Data-Structures

- Edwin Abbott - Flatland - A Romance of Many Dimensions.pdf
- sample_letter.pdf
- Inversions of Permutations in Symmetric, Alternatingv and Dihedral Groups .pdf
- info42-7.pdf
- ReferenceLetters.pdf
- On realizing all simple graphs with a given.pdf
- Real Analysis Lecture Notes.pdf
- atiyah.pdf
- BridgetoAbstractMathematicsMathematicalProofandStructures.pdf
- atiyah.pdf
- ec1
- Complex_Variables_with_Applications_2004_-_Wunsch.pdf
- Discrete Financial Mathematics
- solutionsCA.pdf

Sign up to vote on this title

UsefulNot usefulRead Free for 30 Days

Cancel anytime.

Close Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Loading