This paper improves bounds on the rainbow saturation number of paths in edge-colored graphs, providing tighter upper bounds for certain path lengths and color counts.
Contribution
It presents improved upper bounds for the rainbow saturation number of paths, refining previous results for specific path lengths and color parameters.
Findings
01
Established new upper bounds for $sat_t(n,rak{R}(P_ ext{ell}))$
02
Showed bounds are tighter for $ ext{ell} extgreater= 5$ and $t extgreater= 2 ext{ell} - 5$
03
Enhanced understanding of rainbow saturation in colored graphs.
Abstract
For a fixed graph F and an integer t, the \dfn{rainbow saturation number} of F, denoted by sattβ(n,R(F)), is defined as the minimum number of edges in a t-edge-colored graph on n vertices which does not contain a \dfn{rainbow copy} of F, i.e., a copy of F all of whose edges receive a different color, but the addition of any missing edge in any color from [t] creates such a rainbow copy. Barrus, Ferrara, Vardenbussche and Wenger prove that sattβ(n,R(Pββ))β₯nβ1 for ββ₯4 and sattβ(n,R(Pββ))β€βββ1nβββ (2ββ1β) for tβ₯(2ββ1β), where Pββ is a path with β edges. In this short note, we improve the upper bounds and show that sattβ(n,R(Pββ))β€ββnβββ ((2ββ2β)+4) for ββ₯5 and tβ₯2ββ5.
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsLimits and Structures in Graph Theory Β· Graph theory and applications Β· Advanced Graph Theory Research
Full text
A note on rainbow saturation number of paths
Shujuan Cao 111
School of Mathematical Sciences, Tiangong University, Tianjin 300387, China;
[email protected]. ,βYuede Ma 222
School of Science, Xi an Technological University, Xi an, Shaanxi 710021, PR China;
[email protected]. βand βZhenyu Taoqiu333
Center for Combinatorics and LPMC,
Nankai University, Tianjin 300071, China;
[email protected] (the corresponding author).
Abstract
For a fixed graph F and an integer t, the rainbow saturation number of F, denoted by sattβ(n,R(F)), is defined as the minimum number of edges in a t-edge-colored graph on n vertices which does not contain a
rainbow copy of F, i.e., a copy of F all of whose edges receive a different color, but the addition of any missing edge in any color from [t] creates such a rainbow copy.
Barrus, Ferrara, Vardenbussche and Wenger prove that sattβ(n,R(Pββ))β₯nβ1 for ββ₯4 and sattβ(n,R(Pββ))β€βββ1nβββ (2ββ1β) for tβ₯(2ββ1β), where Pββ is a path with β edges. In this short note, we improve the upper bounds and show that sattβ(n,R(Pββ))β€ββnβββ ((2ββ2β)+4) for ββ₯5 and tβ₯2ββ5.
Throughout this note, all graphs are simple, undirected, and finite. Throughout we use
the terminology and notation of [8]. For a positive integer t, let [t] denote the set {1,β¦,t}. A t-edge-coloring of a graph G is a function f:E(G)β[t], and a graph equipped with such a coloring is a (t-edge-colored graph).
A graph G is called F-saturated if it is a maximal F-free graph. The classical saturation problem, first studied by ZykovΒ [9] and ErdΓΆs, Hajnal and MoonΒ [2], asks for the minimum number of edges in an F-saturated graph. For more results on saturation numbers, the reader should consult the excellent survey of Faudree, Faudree, and Schmitt [3]. A rainbow analog of this problem was recently introduced by Barrus, Ferrara, Vardenbussche and WengerΒ [1], where a t-edge-colored graph is defined to be rainbow F-saturated if it contains no rainbow copy of F, i.e., a copy of F all of whose edges receive a different color, but the addition of any missing edge in any color creates such a rainbow copy, denoted by (R(F),t)-saturated. This minimum size of a t-edge-colored rainbow F-saturated graph, denoted by sattβ(n,R(F)), is the rainbow saturation number of F, i.e.,
[TABLE]
In [1], the authors proved some results on sattβ(n,R(F)) of various families of graphs including complete graphs, trees and cycles. KorΓ‘ndi [7] prove a conjecture of Barrus et al. on the rainbow saturation number of complete graphs. For more results on this topic we refer to [4, 5, 6]. Especially, Barrus et al. [1] proved the following results on paths.
In the proof of Theorem 1.1(iii), Barrus et al. use rainbow Kββ1β as construction components. Here we improve the upper bounds by changing the construction components, which yields Theorem 1.2.
Theorem 1.2**.**
For ββ₯5 and tβ₯2ββ5, sattβ(n,R(Pββ))β€ββnβββ ((2ββ2β)+4).
In Section 2, for all positive integers ββ₯5, we provide infinitely many t-edge-colored graphs with sattβ(n,R(Pββ)) attaining the upper bounds. In Section 3, the proof of the main result is presented.
2 Construction
In this section, we construct several rainbow Pββ-saturated graphs achieving the upper bounds in Theorem 1.2, for all positive integers ββ₯5.
Definition 2.1**.**
For any positive integer ββ₯5, let V(Kββ2β)={v0β,v1β,β¦,vββ3β} and c(viβvjβ)=i+j as the edge coloring rule of Kββ2β.
Let H be the graph obtained from Kββ2β by adding two vertices {vββ2β,x} and four edges vββ2βvββ3β,vββ2βvββ4β,xvββ3β,xvββ4β. Set c(vββ2βvββ3β)=c(xvββ4β)=2ββ5 amd c(vββ2βvββ4β)=c(xvββ3β)=2ββ6. Let Gβ=kH for all positive integers kβ₯2.
In the following we will study some properties of H and Gβ.
Lemma 2.2**.**
The graph H is t-proper-edge-colored, and H does not contain any rainbow copy of Pββ, where t=2ββ5 and ββ₯5.
Proof.
By the edge coloring rule of H, for any vβH, all edges incident to v have different colors. Thus H is t-proper-edge-coloring, and t=ββ2+ββ3=2ββ5.
Suppose H contains a rainbow copy of Pββ, denoted by Pββ²β. For β£Hβ£=β, we have x,vββ2ββPββ²β. Then there are at least two same colors in c(Pββ²β).
β‘
For a graph G and vβG, let Pvβ be a rainbow Hamilton path from v of G and c(Pvβ) be the color set of E(Pvβ). Let Gvβ be the set of all such Pvβ and c(Gvβ)=βc(Pvβ).
Lemma 2.3**.**
For any positive integer ββ₯5, let Hβ=Hβx and eβE(Gβ). Then Hβ+e contains a rainbow Pββ such that c(e) is any color from [t] and e is incident to viββHβ, where iβ/{ββ4,ββ3}.
Proof.
We only need to prove c(Hvββ)=β for any vβHββ{vββ4β,vββ3β}.
Suppose β is odd. Assume aβN+, then V(Hβ)={v0β,v1β,β¦,vββ4β,vββ3β,vββ2β}. By Definition 2.1, t=2ββ5 and vββ2β is only adjacent to vββ3β,Β vββ4β.
If v=v0β, for Hβ, then let Pv0β1β=v0β,v1β,v3β,β¦,v2a+1β,β¦,vββ4β,vββ2β,vββ3β,β¦,v2aβ,β¦,v4β,v2β and Pv0β2β={v0β,v1β,v2β,v3β,β¦,v2aβ,v2a+1β,β¦,vββ4β,vββ3β,vββ2β}. Then all elements in c(Pv0β1β) are even except for {1,2ββ5} and all elements in c(Pv0β2β) are odd. So we have c(Hv0βββ)β{1,2ββ5}. Change the positions of a few vertices in Pv0β2β, we can get Pv0β21β={v0β,v2β,v1β,v3β,β¦,vββ5β,vββ3β,vββ4β,vββ2β}. There is no {1,2ββ5} in c(Pv0β21β). Thus c(Hv0βββ)=β .
If v=v2aβ, for Hβ, then let Pv2aβ1β={v2aβ,v2aβ2β,β¦,v4β,v2β,v0β,v1β,v3β,β¦,vββ4β,vββ2β,vββ3β,β¦,v2a+2β} and Pv2aβ2β={v[β2a],v0β,v1β,v2β,v3β,β¦,v2aβ1β,v2a+1β,β¦,vββ4β,vββ3β,vββ2β}. Then all elements in c(Pv2aβ1β) are even except for {1,2ββ5} and all elements in c(Pv2aβ2β) are odd except for {2a,4a}. So we have c(Hv2aβββ)β{1,2ββ5,2a,4a}. Change the positions of a few vertices in Pv2aβ2β, we can get Pv2aβ3β={v2aβ,v2aβ1β,β¦,v3β,v1β,v2β,v0β,v2a+2β,v2a+1β,v2a+3β,v2a+4β,β¦,vββ5β,vββ3β,vββ4β,vββ2β}. There is no {1,2a,4a,2ββ5} in c(Pv2aβ3β). Thus c(Hv2aβββ)=β .
If v=v2a+1β, for Hβ, then let Pv2a+1β1β={v2a+1β,v2aβ1β,β¦,v3β,v1β,v0β,v2β,v4β,β¦,vββ3β,vββ2β,vββ4β,β¦,v2a+3β} and Pv2a+1β2β={v2a+1β,v1β,v0β,v2β,v3β,β¦,v2aβ,v2a+2β,β¦,vββ4β,vββ3β,vββ2β}. Then all elements in c(Pv2a+1β1β) are even except for {1,2ββ5} and all elements in c(Pv2a+1β2β) are odd except for {2,2a+2,4a+2}. So we have c(Hv2a+1βββ)β{1,2ββ5,2,2a+2,4a+2}. Change the positions of a few vertices in Pv2a+1β2β, we can get Pv2a+1β3β={v2a+1β,v2aβ,β¦,v3β,v1β,v2β,v0β,v2a+2β,v2a+3β,β¦,vββ3β,vββ4β,vββ2β} and Pv2a+1β21β={v2a+1β,v3β,v2β,v1β,v0β,v4β,β¦,vββ2β}. There is no {1,2,4a+2,2ββ5} in c(Pv2a+1β3β) and no 2a+2 in c(Pv2a+1β21β). Thus c(Hv2a+1βββ)=β .
If v=vββ5β, for Hβ, then let Pvββ5β1β={vββ5β,vββ7β,β¦,v4β,v2β,v0β,v1β,v3β,β¦,vββ4β,vββ2β,vββ3β} and Pvββ5β2β={vββ5β,v0β,v1β,v2β,v3β,β¦,vββ6β,vββ4β,vββ3β,vββ2β}. Then all elements in c(Pvββ5β1β) are even except for {1,2ββ5} and all elements in c(Pvββ5β2β) are odd except for {ββ5,2ββ10}. So we have c(Hvββ5βββ)β{1,2ββ5,ββ5,2ββ10}. Change the positions of a few vertices in Pvββ5β2β, we can get Pvββ5β3β={vββ5β,vββ6β,β¦,v3β,v1β,v2β,v0β,vββ3β,vββ4β,vββ2β}. There is no {1,2ββ5,ββ5,2ββ10} in c(Pvββ5β3β). Thus c(Hvββ5βββ)=β .
If v=vββ4β,Β ββ₯7, for Hβ, then let Pvββ4β1β={vββ4β,vββ6β,β¦,v3β,v1β,v0β,v2β,v4β,β¦,vββ3β,vββ2β} and Pvββ4β2β={vββ4β,v1β,v0β,v2β,v3β,β¦,vββ5β,vββ3β,vββ2β}. Then all elements in c(Pvββ4β1β) are even except for {1,2ββ5} and all elements in c(Pvββ4β2β) are odd except for {2,ββ3,2ββ8}. So we have c(Hvββ4βββ)β{1,2ββ5,2,ββ3,2ββ8}. Change the positions of a few vertices in Pvββ4β2β, we can get Pvββ4β3β={vββ4β,vββ2β,vββ3β,v0β,v2β,v1β,v3β,β¦,vββ5β} and Pvββ4β31β={vββ4β,vββ2β,vββ3β,v2β,v0β,v1β,v3β,β¦,vββ5β}. There is no {1,2,2ββ8} in c(Pvββ4β3β) and no ββ3 in c(Pvββ4β31β). Thus c(Hvββ4βββ)={2ββ5}.
If v=vββ3β,Β ββ₯7, for Hβ, then let Pvββ3β1β={vββ3β,vββ5β,β¦,v4β,v2β,v0β,v1β,v3β,β¦,vββ4β,vββ2β} and Pvββ3β2β={vββ3β,v0β,v1β,v2β,β¦,vββ5β,vββ4β,vββ2β}. Then all elements in c(Pvββ3β1β) are even except for 1 and all elements in c(Pvββ3β2β) are odd except for {ββ3,2ββ6}. So we have c(Hvββ3βββ)β{1,ββ3,2ββ6}. Change the positions of a few vertices in Pvββ3β2β, we can get Pvββ3β21β={vββ3β,v0β,v2β,v1β,v3β,β¦,vββ2β} and Pvββ3β22β={vββ3β,v2β,v0β,v1β,v3β,v4β,β¦,vββ2β}. There is no 1 in c(Pvββ3β21β) and no ββ3 in c(Pvββ3β22β). Thus c(Hvββ3βββ)={2ββ6}.
If v=vββ2β, for Hβ, then let Pvββ2β1β={vββ2β,vββ3β,vββ5β,β¦,v4β,v2β,v0β,v1β,v3β,β¦,vββ4β} and Pvββ2β2β={vββ2β,vββ3β,v0β,v1β,v2β,β¦,vββ5β,vββ4β}. Then all elements in c(Pvββ2β1β) are even except for {1,2ββ5} and all elements in c(Pvββ2β2β) are odd except for ββ3. So we have c(Hvββ2βββ)β{1,2ββ5,ββ3}. Change the positions of a few vertices in Pvββ2β2β, we can get Pvββ2β21β={vββ2β,vββ3β,v0β,v2β,v1β,v3β,β¦,vββ4β} and Pvββ2β22β={vββ2β,vββ4β,vββ3β,v2β,v0β,v1β,v3β,β¦,vββ5β}. There is no 1 in c(Pvββ2β21β) and no {ββ3,2ββ5} in c(Pvββ2β22β). Thus c(Hvββ2βββ)=β .
The proof is similar when β is even.
β‘
Corollary 2.4**.**
The graph H is an (R(Pββ),t)-saturated graph for ββ₯7.
Proof.
By Lemma 2.3, for any eβE(HΛ), then H+e contains a rainbow Pββ such that c(e) is any color from [t] and e is incident to viββHβ, where iβ/{ββ4,ββ3}. By Definition 2.1, c(vββ2βvββ3β)=2ββ5=c(xvββ4β) and c(vββ2βvββ4β)=2ββ6=c(xvββ3β). Then we have c(Hxβ)=c(Hvββ2ββ)=β .
If ββ₯7, then H+e contains a rainbow Pββ also holds for e which is incident to vββ4β,vββ3β by Definition 2.1.
Together with Lemma 2.2, H is (R(Pββ),t)-saturated for ββ₯7.
β‘
Proposition 2.5**.**
For β=5, Gβ in Definition 2.1 is (R(P5β),5)-saturated and β£E(Gβ)β£=57nβ.
Proof.
Similar as the proof in Lemma 2.3, c(Hv0ββ)=β ,Β c(Hv3ββ)=c(Hxβ)=β .
If v=v1β, then there are only Pv1β1β={v1β,v0β,v2β,v3β} and Pv1β2β={v1β,v3β,v2β,v0β} in Hβ. Then c(Hv1βββ)={2,5}.
If v=v2β, then there are only Pv2β1β={v2β,v0β,v1β,v3β} and Pv2β2β={v2β,v3β,v1β,v0β} in Hβ. Then c(Hv2βββ)={1,4}.
For c(v2βx)=4,Β c(v1βx)=5, we have c(Hv1ββ)={2} and c(Hv2ββ)={1}.
For H1,H2βGβ, let e1β=v11βv12ββE(Gβ) and c(e1β)=2. We can get a rainbow P5β={v21β,v31β,v11β,v12β,v22β} in Gβ+e1β.
Let e2β=v21βv22ββE(Gβ) and c(e2β)=1, we can get a rainbow P5β={v11β,v31β,v21β,v22β,v12β} in Gβ+e2β.
Let e3β=v11βv22ββE(Gβ) and c(e3β)β{1,2}, we can get a rainbow P5β={v21β,v31β,v11β,v22β,v12β} in Gβ+e3β.
Therefore, Gβ is (R(P5β),5)-saturated and β£E(Gβ)β£=57nβ.
β‘
Proposition 2.6**.**
For β=6, Gβ in Definition 2.1 is (R(P6β),7)-saturated and β£E(Gβ)β£=35nβ.
Proof.
Similar as in the proof of Lemma 2.3, c(Hv0ββ)=β ,Β c(Hv1ββ)=β ,Β c(Hv4ββ)=c(Hxβ)=β .
If v=v2β, then there are only Pv2β1β={v2β,v0β,v1β,v3β,v4β}, Pv2β2β={v2β,v4β,v3β,v0β,v1β} and Pv2β3β={v2β,v4β,v3β,v1β,v0β} in Hβ. Then c(Hv2βββ)={1,7}.
If v=v3β, then there are only Pv3β1β={v3β,v1β,v0β,v2β,v4β}, Pv3β2β={v3β,v4β,v2β,v1β,v0β} and Pv3β3β={v3β,v4β,v2β,v0β,v1β} in Hβ. Then c(Hv3βββ)={1,6}.
For c(v2βx)=7,Β c(v3βx)=6, we have c(Hv2ββ)={1}=c(Hv3ββ).
For H1,H2βGβ, let e1β=v21βv21ββE(GβΛ) and c(e1β)=1. We can get a rainbow P6β={v01β,v31β,v41β,v21β,v22β,v02β} in Gβ+e1β.
Let e2β=v31βv32ββE(GβΛ) and c(e2β)=1, we can get a rainbow P6β={v11β,v21β,v41β,v31β,v32β,v12β} in Gβ+e2β.
Let e3β=v21βv32ββE(GβΛ) and c(e3β)=1, we can get a rainbow P6β={v01β,v31β,v41β,v21β,v32β,v12β} in Gβ+e3β.
Therefore, Gβ is (R(P6β),7)-saturated and β£E(Gβ)β£=35nβ.
β‘
From Corollary 2.4, Propositions 2.5 and 2.6, we get the following theorem.
Theorem 2.7**.**
The graph Gβ is an (R(Pββ),t)-saturated graph and β£E(Gβ)β£=ββnβββ ((2ββ2β)+4), where t=2ββ5 and ββ₯5.
Let G be an (R(Pββ),t)-saturated graph with n vertices. We present the proof by considering the following cases.
Case 1.β=5.
If t=5 and nβ‘0Β (modΒ 5), then all components of G are H in Definition 2.1, where β£Hβ£=5, i.e., G=Gβ, which yields β£E(G)β£=57nβ.
If t=6 and nβ‘1Β (modΒ 5), then all components of G are H except for two rainbow triangles, which yields β£E(G)β£=57nββ512β.
If t=5 and nβ‘2Β (modΒ 5), then all components of G are H except for an edge e that c(e)=3, which yields β£E(G)β£=57nββ59β.
If t=5 and nβ‘3Β (modΒ 5), then all components of G are H except for a rainbow triangle, which yields β£E(G)β£=57nββ56β.
If t=5 and nβ‘4Β (modΒ 5), then all components of G are H except for a rainbow K4β, which yields β£E(G)β£=57nβ+52β.
Thus sattβ(n,R(P5β))β€β57nββ.
Case 2.β=6.
If t=7 and nβ‘0Β (modΒ 6), then all components of G are H in Definition 2.1, where β£Hβ£=6, i.e., G=Gβ, which yields β£E(G)β£=35nβ.
If t=7 and nβ‘1Β (modΒ 6), then all components of G are H except for a rainbow triangle and a rainbow K4β, which yields β£E(G)β£=35nββ38β.
If t=7 and nβ‘2Β (modΒ 6), then all components of G are H except for an edge e that c(e)=4, then β£E(G)β£=35nββ37β.
If t=7 and nβ‘3Β (modΒ 6), then all components of G are H except for a rainbow triangle, which yields β£E(G)β£=35nββ2.
If t=7 and nβ‘4Β (modΒ 6), then all components of G are H except for a rainbow K4β, which yields β£E(G)β£=35nββ32β.
If t=7 and nβ‘5Β (modΒ 6), then all components of G are H except for two rainbow K4β and a rainbow triangle, which yields β£E(G)β£=35nββ310β.
Thus sattβ(n,R(P6β))β€β35nββ.
Case 3.ββ₯7.
If t=2ββ5 and nβ‘0Β (modΒ β), then all components of G are H in Definition 2.1, where β£Hβ£=β, i.e., G=Gβ, which yields β£E(G)β£=βnββ ((2ββ2β)+4).
If t=2ββ5 and nβ‘1Β (modΒ β), then all components of G are H except for a vertex v, which yields β£E(G)β£=βnβ1ββ ((2ββ2β)+4).
If t=2ββ5 and nβ‘2Β (modΒ β), then all components of G are H except for an edge e, where c(e)β[t], which yields β£E(G)β£=βnβ2ββ ((2ββ2β)+4)+1.
If t=2ββ5 and nβ‘3Β (modΒ β), then all components of G are H except for a rainbow triangle, which yields β£E(G)β£=βnβ3ββ ((2ββ2β)+4)+3.
If t=2ββ5 and nβ‘4Β (modΒ β), then all components of G are H except for a rainbow K4β, which yields β£E(G)β£=βnβ4ββ ((2ββ2β)+4)+6.
β¦
If t=2ββ5 and nβ‘aΒ (modΒ β), then all components of G are H except for a rainbow Kaβ, which yields β£E(G)β£=βnβaββ ((2ββ2β)+4)+(2aβ). It is easy to check eaββ€ββnβββ ((2ββ2β)+4), since 1β€aβ€ββ1 and then 1β€aβ€(ββ3)2+8β.
From all the above cases, we have for ββ₯5 and tβ₯2ββ5, sattβ(n,R(Pββ))β€ββnβββ ((2ββ2β)+4). The proof is thus complete. β‘
Acknowledgements. Shujuan Cao is partially supported by the National Natural Science Foundation of China (No. 11801412 ).
Bibliography9
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] M. D. Barrus, M. Ferrara, J. Vandenbussche and P. S. Wenger, Colored saturation parameters for rainbow subgraphs, J. Graph Theory , 86 (2017), 375β386.
2[2] P. ErdΓΆs, A. Hajnal and J.W. Moon, A problem in graph theory, Amer. Math. Monthly , 71 (1964), 1107β1110.
3[3] J. Faudree, R. Faudree and J. Schmitt, A survey of minimum saturated graphs, Electron. J. Combin. 18 (2011), #DS 19.
4[4] M. Ferrara, D. Johnston, S. Loeb, F. Pfender, A. Schulte, H.C. Smith, E. Sullivan, M. Tait, C. Tompkins, On edge-colored saturation problems, ar Xiv:1712.00163.
5[5] A. GirΓ£o (2019). Extremal and Structural Problems of Graphs (Doctoral thesis). https://doi.org/10.17863/CAM.32787
6[6] A. GirΓ£o, D. Lewis and K. Popielarz, Rainbow saturation of graphs, ar Xiv:1710.08025.
7[7] D. KorΓ‘ndi, Rainbow saturation and graph capacities, SIAM J. Discrete Math. 32 (2)(2018), 1261β1264.
8[8] D.B. West, Introduction to Graph Theory, Prentice Hall, Inc., Upper Saddle River, NJ, 2001.