
The exponent problems of the three-colored digraphs containing n vertices, one n-cycle, one (n−3)-cycle, and one 3-cycle are considered. According to the coloring of the 3-cycle, we only consider the case where there are zero red arcs, one yellow arc, and two blue arcs in the 3-cycle. The primitive conditions of the cycle matrix are given, based on the discussion of different cases, the upper bound of the primitive exponents is found, and the extreme digraphs are characterized. The results are useful for the study of the primitive exponents of three-colored digraphs in general cases and the application of graph coloring problems.
Citation: Meijin Luo, Qiutao Qin. Exponents of a class of special three-colored primitive digraphs with n vertices in graph theory[J]. AIMS Mathematics, 2025, 10(4): 9415-9434. doi: 10.3934/math.2025435
[1] | Xiaoxue Hu, Jiangxu Kong . An improved upper bound for the dynamic list coloring of 1-planar graphs. AIMS Mathematics, 2022, 7(5): 7337-7348. doi: 10.3934/math.2022409 |
[2] | Jin Cai, Shuangliang Tian, Lizhen Peng . On star and acyclic coloring of generalized lexicographic product of graphs. AIMS Mathematics, 2022, 7(8): 14270-14281. doi: 10.3934/math.2022786 |
[3] | M. Haris Mateen, Muhammad Khalid Mahmmod, Dilshad Alghazzawi, Jia-Bao Liu . Structures of power digraphs over the congruence equation $ x^p\equiv y\; (\text{mod}\; m) $ and enumerations. AIMS Mathematics, 2021, 6(5): 4581-4596. doi: 10.3934/math.2021270 |
[4] | Zhihong Xie, Guoliang Hao, S. M. Sheikholeslami, Shuting Zeng . On the Italian reinforcement number of a digraph. AIMS Mathematics, 2021, 6(6): 6490-6505. doi: 10.3934/math.2021382 |
[5] | Nuttawoot Nupo, Sayan Panma . Certain structural properties for Cayley regularity graphs of semigroups and their theoretical applications. AIMS Mathematics, 2023, 8(7): 16228-16239. doi: 10.3934/math.2023830 |
[6] | Yanyi Li, Lily Chen . Injective edge coloring of generalized Petersen graphs. AIMS Mathematics, 2021, 6(8): 7929-7943. doi: 10.3934/math.2021460 |
[7] | Yindi Weng . Bounds and complexity results of rainbow vertex-disconnection colorings. AIMS Mathematics, 2025, 10(3): 5960-5970. doi: 10.3934/math.2025272 |
[8] | Gaixiang Cai, Fengru Xiao, Guidong Yu . The identification numbers of lollipop graphs. AIMS Mathematics, 2025, 10(4): 7813-7827. doi: 10.3934/math.2025358 |
[9] | Hongyu Chen, Li Zhang . A smaller upper bound for the list injective chromatic number of planar graphs. AIMS Mathematics, 2025, 10(1): 289-310. doi: 10.3934/math.2025014 |
[10] | Muhammad Kamran Jamil, Muhammad Imran, Aisha Javed, Roslan Hasni . On the first general Zagreb eccentricity index. AIMS Mathematics, 2021, 6(1): 532-542. doi: 10.3934/math.2021032 |
The exponent problems of the three-colored digraphs containing n vertices, one n-cycle, one (n−3)-cycle, and one 3-cycle are considered. According to the coloring of the 3-cycle, we only consider the case where there are zero red arcs, one yellow arc, and two blue arcs in the 3-cycle. The primitive conditions of the cycle matrix are given, based on the discussion of different cases, the upper bound of the primitive exponents is found, and the extreme digraphs are characterized. The results are useful for the study of the primitive exponents of three-colored digraphs in general cases and the application of graph coloring problems.
Graph theory originated from the famous problem of the seven bridges of Königsberg. In 1736, Euler used abstract analysis to prove that the seven bridges problem of Königsberg had no solution, which was the first graph theory problem and also marked the birth of graph theory as a discipline. With the development of graph theory, it has become a widely applied branch of mathematics. There are traces of graph theory and its applications in various fields such as mathematics, computer science, and biological science.
The graph coloring problem is one of the classic problems in the field of combinatorial optimization, which has penetrated into multiple disciplines such as physics, chemistry, biology, electronics, economics, management, and computer science. More and more people are interested in and attach importance to graph theory. Coloring the edges (arcs) or vertices of a graph can often be associated with a matrix, thereby effectively solving the problem. We conducted research on the complementarity spectrum of the directed graph and Gromov hyperbolicity of Johnson and Kneser graphs and obtained some mathematical conclusions [1,2]. Some research has been conducted on issues such as metal halide perovskites, perovskite light-emitting diodes, and predicting electronic band structures of quantum-confined nanostructures in physics, and some of these problems have been solved through the application of graph coloring [3,4,5]. We consider each atom in a molecule as a vertex in the graph and the chemical bonds between atoms as edges in the graph. By coloring the graph, we can solve the problem of molecular structure in biology and chemistry [6,7,8,9]. In fact, algorithms can be optimized through graph coloring problems. In [10], graph invariants were used to quantitatively characterize the topological properties of complex networks. In [11,12], hybrid heuristic coloring algorithms, intelligent optimization algorithms, and heuristic algorithms that are applicable to both large-scale and small-hard graphs were presented. The results obtained have guiding significance for the research of related problems.
In this article, we establish corresponding relationships between nonnegative matrices and their adjoint directed graphs, or between directed graphs and their adjacent matrices. By coloring the edges (arcs) of the directed graph, we discuss and find the upper bound of the exponent in the primitive case, providing reference for the application of graph coloring problems. The following basic concepts are given.
A three-colored digraph D is a graph with a direction and whose arc contains three colors (usually red, yellow, and blue). If D has a sequence of vertices v1v2⋯vtvt+1, then it is a walk from v1 to vt+1 of length t; if v1v2⋯vtvt+1 are different, then it is a path from v1 to vt+1 of length t[13]. The arcs of the same walk or path can be the same color or different colors. D is a strongly connected graph if any two vertices can be reached by a walk. ν's decomposition vector is expressed as (u(ν),v(ν),w(ν)) or (u(ν),v(ν),w(ν))T, then ν is said to be an (u(ν),v(ν),w(ν))-walk, where u(ν), v(ν), and w(ν) represent respectively the numbers of red arcs, the numbers of yellow arcs and the numbers of blue arcs in ν.
For any (vi,vj), there exists an (h1,h2,h3)-walk from vi to vj, where h1, h2 and h3 are nonnegative integers and
h1+h2+h3>0, |
then the minimum value of h1+h2+h3 is the primitive exponent of D, denoted as exp(D).
Let
C={γ1,γ2,⋯,γt−1,γt} (i=1,2,⋯,t) |
be the set of cycles, M1 be the cycle matrix of D. Thus
M1=[u1u2⋯ut−1utv1v2⋯vt−1vtw1w2⋯wt−1wt] |
for some nonnegative integers ui,vi,wi (i=1,2,⋯,t), where the ith column of M1 represents the decomposition vector of the γi-cycle. The content of M1, usually denoted as content(M1), is defined to be 0 if r(M1)<3; otherwise, content(M1) is defined to be the greatest common factor of all 3×3 minors of M1, where r(M1) is the rank of M1.
Lemma 1. [14] If D is a three-colored primitive digraph, then D is strongly connected and
content(M1)=1. |
Combinatorial mathematics is an important branch of mathematics that is used widely in many fields such as computer networks, economics, operations research, information coding, etc. Nonnegative matrix theory and graph theory are two important research contents in combinatorial mathematics. In solving the exponential problems of nonnegative matrices, the equivalent relation is often established with their associated digraphs so as to solve the problems of nonnegative matrices with associated digraphs. At present, the research on single nonnegative matrices and nonnegative matrix pairs is basically perfect, and it is an inevitable trend to extend the research to nonnegative matrix clusters. Analogous to the correspondence between a single matrix and its associated digraphs. Three-colored digraphs are associated digraphs of nonnegative matrix clusters, and there is a one-to-one correspondence between them. According to the simple and intuitionistic characteristics of the graph, we can use three-colored digraphs to solve the problems related to the primitive exponent of nonnegative matrix clusters.
Up to now, some achievements have been made on the primitive exponents of nonnegative matrix pairs (or two-colored digraphs) and nonnegative matrix clusters (or three-colored digraphs). There are some basic concepts of graph theory that are needed to carry out the research in this paper [13,14]. Gao and Shao are the earliest experts in the study of primitive exponents of two-colored digraphs in China, and the studying method of primitive exponents of two-colored digraph has a good reference effect for the research of three-colored digraphs [15]. The range of primitive exponent is found by taking a class of two-colored digraph (or nonnegative matrix pairs) as an example, respectively, and the coloring of the arcs that reach the upper and lower bounds of the exponent are given [15,16,17,18,19]. By coloring all arcs, the exponent set of the corresponding digraph is found [20,21]. With the in-depth study of the primitive exponents of two-colored digraphs (or nonnegative matrix pairs), most of the problems of primitive exponents have been solved. The research on the extension of two-colored digraphs to three-colored digraphs is a relatively new field and has achieved few achievements. Some representative classes of three-colored digraphs are selected respectively for research; some upper bounds of exponents and the extremal three-colored digraphs are found on the premise of primitives [22,23,24,25].
For n≥5, we study a class of three-colored digraphs whose uncolored digraph is shown in Figure 1.
Obviously, D is strongly connected. It is known that the coloring of the arcs in the 3-cycle can be the same or different. That is to say, the three arcs can be one color, two colors, or three colors in the 3-cycle. If the three arcs in the 3-cycle can contain only one color, and content(M1) must be a multiple of 3 at this point, so
content(M1)≠1 |
and D is non-primitive. If the coloration of all three arcs in the 3-cycle is different, that is, the path is decomposed into (1,1,1)T, which has been discussed [22]. If the three arcs in the 3-cycle have two different patterns, then the walks can be decomposed into six possibilities, specifically (0,1,2)T or (0,2,1)T or (1,0,2)T or (1,2,0)T or (2,0,1)T or (2,1,0)T.
Since the calculation methods of the six coloring cases of the three arcs are similar when two colors are included in the 3-cycle, in this paper, we only research when there are zero red arcs, one yellow arc, and two blue arcs in the 3-cycle, that is, the decomposition vector of the 3-cycle is (0,1,2)T on the basis of [22]. In this case, we can assume D's cycle matrix
M2=[ac0bd1n−a−bn−3−c−d2], | (1) |
for some nonnegative integers a–d.
From Figure 1, the n-cycle and 3-cycle in D have two common arcs, vn→v1 and v1→v2. The coloring of arcs vn→v1 and v1→v2 can be the same or different in a 3-cycle. The n-cycle and (n−3)-cycle in D only have five non-common arcs vn→v1→v2→v3→v4 and vn→v4. Thus
{c−1≤a≤c+4,d−1≤b≤d+4,n−4−c−d≤n−a−b≤n−c−d+1,c+d−1≤a+b≤c+d+4. | (2) |
This section gives the primitive conditions that D needs to satisfy. Combined with Figure 1, for the sake of discussion, we can assume
b=d+k, |
where k is an integer and
−1≤k≤4. |
Theorem 1. If D is primitive, then
3ad−an+3a−3bc+cn=±1. |
Proof. From formula (1), we have
|M2|=3ad−an+3a−3bc+cn. |
According to Lemma 1, we know if
content(M2)=1, |
then
|M2|=±1. |
Therefore,
3ad−an+3a−3bc+cn=±1. |
Theorem 2. If a=c−1, then D is primitive,
|M2|=3c−3d+n−3−3ck=±1, |
n is not a multiple of 3, and
b=d+k (k=0,1,2,3,4). |
Proof. Because
a=c−1, |
from formulas (1) and (2), we can see
b=d+k (k=0,1,2,3,4), |M2|=3c−3d+n−3−3ck. |
According to Lemma 1, we know if
content(M2)=1, |
then
3c−3d+n−3−3ck=±1. |
Since n is not a multiple of 3, we can discuss it in the following five cases:
(1) If k=0, then
|M2|=3c−3d+n−3. |
So D is primitive and
c=3d−n+3±13. |
(2) If k=1, then
|M2|=−3d+n−3. |
So D is primitive and
d=n−3∓13. |
(3) If k=2, then
|M2|=−3c−3d+n−3. |
So D is primitive and
c=−3d+n−3∓13. |
(4) If k=3, then
|M2|=−6c−3d+n−3. |
So D is primitive and
c=−3d+n−3∓16. |
(5) If k=4, then
|M2|=−9c−3d+n−3. |
So D is primitive and
c=−3d+n−3∓19. |
Theorem 3. If
a=c, |
then
|M2|=3c−3ck≠±1 |
and D is non-primitive.
Theorem 4. If
a=c+1, |
then D is primitive,
|M2|=3c+3d−n+3−3ck=±1, |
n is not a multiple of 3 and
b=d+k (k=−1,0,1,2,3). |
Theorem 5. If
a=c+2, |
then D is primitive,
|M2|=3c+6d−2n+6−3ck=±1, |
n is not a multiple of 3, c is not a multiple of 2 and
b=d+k (k=0,2). |
Theorem 6. If
a=c+3, |
then
|M2|=3c+9d−3n+9−3ck≠±1 |
and D is non-primitive.
Theorem 7. If
a=c+4, |
then D is primitive,
|M2|=3c+12d−4n+12−3ck=±1, |
n is not a multiple of 3, c is not a multiple of 2, and b=d.
Because of the complexity of the calculation, we will use "maple" software to calculate all primitive exponents of three-colored digraphs D, compare them, and find the maximal primitive exponent for the case
a=c−1, b=d+1. |
Since the calculation method is similar, we shall only find out the upper bound on the exponents when
a=c−1, b=d+1. |
Theorem 8. If
a=c−1, b=d+1, |
and D is primitive, then
exp(D)≤8n3−20n2−31n−39. |
Proof. For every pair of vertices (vi,vj) in D, we say that pvivj represents the shortest path from vi to vj, and denote
u(pvivj)=x, v(pvivj)=y |
and
w(pvivj)=z. |
We can always find the same walk to go from vi, along pvivj to vj, and the numbers of revolutions around the n-cycle, the (n−3)-cycle, and the 3-cycle are p1 times, p2 times, and p3 times, respectively.
If
a=c−1, b=d+1, |
from Theorem 2, we have
−3d+n−3=±1, |
that is
d=n−3∓13. |
Combined with Figure 1, the n-cycle has one more red arc than the (n−3)-cycle, so we can see that vn→v4 is a red arc. The following situations are discussed from
−3d+n−3=1 |
and
−3d+n−3=−1. |
(1) According to formula 1, if
−3d+n−3=1, |
then
d=n−43, |
M3=[c−1c0n−13n−4312n+43−c2n−53−c2] | (3) |
and
M−13=[c−1−2cc−(c−2)2c−2−(c−1)−n+73−c2n−53+2c−n+43−c]. | (4) |
From formula (3), we know that the elements in the cycle matrix M3 are nonnegative, so
1≤c≤2n−53. |
The positivity or negativity of each element in the matrix M−13 depends on the value of c. Therefore, we can determine the symbol of each element in M−13 that does not include −c+2. The value of −c+2 is directly related to the size of c, which is discussed in the following three cases.
(ⅰ) c=1.
According to formulas (3) and (4), we have
M4=[010n−13n−4312n+132n−832] |
and
M−14=[0−21100−n−432n+13−n−13]. |
At this point, we know
−c+2=1>0. |
Taking
p1=2n+13+2y−z, p2=1−x |
and
p3=2n2−n−19+n−43x−2n+13y+n−13z, |
we see that
[xyz]+p1[0n−132n+13]+p2[1n−432n−83]+p3[012]=[12n2−n−19+n−43+2n2−n−194n2+4n+19+2n−83+4n2−2n−29]. |
Noting that
0≤x≤1, 0≤y≤n−13and0≤z≤2n+13. |
Obviously, we can obtain
p1≥0,p2≥0andp3≥0. |
This gives
exp(D)≤1+2n2−n−19+n−43+2n2−n−19+4n2+4n+19+2n−83+4n2−2n−29=4n2+3n−103. |
(ⅱ) c=2.
According to formulas (3) and (4), we have
M5=[120n−13n−4312n−232n−1132] |
and
M−15=[1−4202−1−n+132n+73−n−23]. |
At this point, we know
−c+2=0. |
Taking
p1=4n−13−x+4y−2z, p2=2n−23−2y+z |
and
p3=2n2+5n−79+n−13x−2n+73y+n+23z, |
we see that
[xyz]+p1[1n−132n−23]+p2[2n−432n−113]+p3[012] |
=[4n−13+4n−434n2−5n+19+2n2−10n+89+2n2+5n−798n2−10n+29+4n2−26n+229+4n2+10n−149]. |
Noting that
0≤x≤2, 0≤y≤n−13,and0≤z≤2n−23. |
If x=2, then
0≤y≤n−43, 0≤z≤2n−113, |
and pvivj must contain the arc vn→v4, and the initial and terminal vertices of pvivj must be on the (n−3)-cycle. If x=1, then
0≤y≤n−13, 0≤z≤2n−23. |
If
y=n−13, |
then x≥0, z≥0. If
z=2n−23, |
then x≥0, y≥0. Obviously, we can obtain
p1≥0,p2≥0,andp3≥0. |
This gives
exp(D)≤4n−13+4n−43+4n2−5n+19+2n2−10n+89+2n2+5n−79+8n2−10n+29+4n2−26n+229+4n2+10n−149=8n2−4n−13. |
(ⅲ) 3≤c≤2n−53.
At this point, we know
−c+2<0. |
According to formulas (3) and (4), taking
p1=2cn−2c+33−(c−1)x+2cy−cz, p2=2cn−2c−2n+23+(c−2)x−(2c−2)y+(c−1)z |
and
p3=2n2+6cn−6c−7n+59+n+3c−73x−2n+6c−53y+n+3c−43z, |
we see that
[xyz]+p1[c−1n−132n+43−c]+p2[cn−432n−53−c]+p3[012] |
=[(c−1)(2cn−2c+3)3+2c2n−2c2−2cn+2c3(n−1)(2cn−2c+3)9+(n−4)(2cn−2c−2n+2)9+2n2+6cn−6c−7n+59(2n−3c+4)(2cn−2c+3)9+(2n−3c−5)(2cn−2c−2n+2)9+4n2+12cn−12c−14n+109]. |
Noting that
0≤x≤c, 0≤y≤n−13,and0≤z≤2n−3c+43. |
If x=c, then
0≤y≤n−43, 0≤z≤2n−3c−53, |
and pvivj must contain the arc vn→v4, and the initial and terminal vertices of pvivj must be on the (n−3)-cycle. If x=c−1, then
0≤y≤n−13 |
and
0≤z≤2n−3c+43. |
If
y=n−13, |
then x≥0, z≥0. If
z=2n−3c+43, |
then x≥0, y≥0. Obviously, we can obtain p1≥0, p2≥0, and p3≥0. This gives
exp(D)≤(c−1)(2cn−2c+3)3+2c2n−2c2−2cn+2c3+(n−1)(2cn−2c+3)9+(n−4)(2cn−2c−2n+2)9+2n2+6cn−6c−7n+59+(2n−3c+4)(2cn−2c+3)9+(2n−3c−5)(2cn−2c−2n+2)9+4n2+12cn−12c−14n+109=2cn2−2cn+3n3+(n−3)(2cn−2c−2n+2)3+2n2+6cn−6c−7n+53. |
Let f1(c) be a function of c and
f1(c)=2cn2−2cn+3n3+(n−3)(2cn−2c−2n+2)3+2n2+6cn−6c−7n+53, |
then
f′1(c)=4n2−4n3>0, |
and f1(c) is an increasing function. Because
3≤c≤2n−53, |
then
exp(D)≤f1(2n−53)=8n3−28n2+32n−39. |
(2) According to formula (1) and Theorem 2, if
−3d+n−3=−1, |
then
d=n−23, |
M6=[c−1c0n+13n−2312n+23−c2n−73−c2] | (5) |
and
M−16=[−(c+1)2c−cc−(2c−2)c−1n+13+c−2n+73−2cn−23+c]. |
From formula (5), we know that the elements in the cycle matrix M6 are nonnegative, so
1≤c≤2n−73. |
The positivity or negativity of each element in the matrix M−16 depends on the value of c. So, based on the range of values for c and n≥5, we can judge the sign of each element in M−16.
Taking
p1=2cn+2c3+(c+1)x−2cy+cz, p2=2cn+2c−2n−23−cx+(2c−2)y−(c−1)z |
and
p3=2n2+6cn+6c−5n−79−(n+13+c)x+(2n−73+2c)y−(n−23+c)z, |
we see that
[xyz]+p1[c−1n+132n+23−c]+p2[cn−232n−73−c]+p3[012] |
=[(2cn+2c)(c−1)3+2c2n+2c2−2cn−2c32cn2+4cn+2c9+(n−2)(2cn+2c−2n−2)9+2n2+6cn+6c−5n−79(2cn+2c)(2n−3c+2)9+(2n−3c−7)(2cn+2c−2n−2)9+4n2+12cn+12c−10n−149]. |
Noting that
0≤x≤c, 0≤y≤n+13,and0≤z≤2n−3c+23. |
If x=c, then
0≤y≤n−23 |
and
0≤z≤2n−3c−73, |
and pvivj must contain the arc vn→v4, the initial and terminal vertices of pvivj must be on the (n−3)-cycle. If
x=c−1, |
then
0≤y≤n+13 |
and
0≤z≤2n−3c+23. |
If
y=n+13, |
then x≥0, z≥0. If
z=2n−3c+23, |
then x≥0, y≥0. Obviously, we can obtain
p1≥0,p2≥0,andp3≥0. |
This gives
exp(D)≤(2cn+2c)(c−1)3+2c2n+2c2−2cn−2c3+2cn2+4cn+2c9+(n−2)(2cn+2c−2n−2)9+2n2+6cn+6c−5n−79+(2cn+2c)(2n−3c+2)9+(2n−3c−7)(2cn+2c−2n−2)9+4n2+12cn+12c−10n−149=2cn2+2cn3+(n−3)(2cn+2c−2n−2)3+2n2+6cn+6c−5n−73. |
Let f2(c) be a function of c and
f2(c)=2cn2+2cn3+(n−3)(2cn+2c−2n−2)3+2n2+6cn+6c−5n−73, |
then
f′2(c)=4n2+4n3>0, |
and f(c) is an increasing function. Because
1≤c≤2n−73, |
then
exp(D)≤f2(2n−73)=8n3−20n2−31n−39. |
In summary, the upper bounds of all primitive exponents of
a=c−1, b=d+1 |
can be obtained. Thus
exp(D)≤{4n2+3n−103,(|M2|=1,d=n−43,c=1),8n2−4n−13,(|M2|=1,d=n−43,c=2),8n3−28n2+32n−39,(|M2|=1,d=n−43,3≤c≤2n−53),8n3−20n2−31n−39,(|M2|=−1,d=n−23). |
Comparing all the values of exp(D), we obtain
exp(D)≤8n3−20n2−31n−39. |
We will give the coloring of the arcs of D such that the equality holds, that is, the extremal digraphs with the maximal primitive exponent.
Theorem 9. If
a=c−1, b=d+1, |
and D is primitive, then
exp(D)=8n3−20n2−31n−39 |
if and only if the n+13 yellow arcs are consecutive on the n-cycle.
Proof. Obviously, the corresponding cycle matrix is formula (5). We can see that if the equality in Theorem 8 holds, then
|M6|=−1. |
This completes the proof.
Sufficiency. Suppose (h1,h2,h3) is a 3-tuple of nonnegative integers, and there is an (h1,h2,h3)-walk from vi to vj for every pair of vertices (vi,vj).
Taking vi and vj to be the same vertex on the n-cycle, then there exist nonnegative integers μ1–μ3 satisfying
[h1h2h3]=M6[μ1μ2μ3]. |
We can suppose that the starting vertex of n+13 continuous yellow arcs on the n-cycle is vi and the ending vertex is vj. Combined with Figure 1, there is only one path from vi to vj, and the vector of the walk is decomposed into (0,n+13,0). Thus
M6μ=[h1h2−n+13h3] |
has a nonnegative integer solution that satisfies the conditions. Therefore,
μ=M−16[h1h2−n+13h3]=M−16[h1h2h3]−M−16[0n+130]=[μ1μ2μ3]−M−16[0n+130]=[μ1μ2μ3]−[−(c+1)2c−cc−(2c−2)c−1n+13+c−2n+73−2cn−23+c][0n+130]=[μ1μ2μ3]−[2cn+2c3−2cn+2c−2n−23−2n2+6cn+6c−5n−79]≥0. |
Then
μ1≥2cn+2c3. |
Similarly, we can suppose that the ending vertex of n+13 continuous yellow arcs on the n-cycle is vi and the starting vertex is vj. Combined with Figure 1, there is only one path from vi to vj, and the vector of the walk is decomposed into (c−1,0,2n+23−c). Thus
M6μ=[h1−(c−1)h2h3−(2n+23−c)] |
has a nonnegative integer solution that satisfies the conditions. Therefore,
μ=M−16[h1−(c−1)h2h3−(2n+23−c)]=M−16[h1h2h3]−M−16[c−102n+23−c]=[μ1μ2μ3]−M−16[c−102n+23−c]=[μ1μ2μ3]−[−c−12c−cc−2c+2c−1n+13+c−2n+73−2cn−23+c][c−102n+23−c]=[μ1μ2μ3]−[−(c2−1)−(2cn−3c2+2c)32cn+2c−2n−232n2+6cn+6c−5n−79]≥0. |
Then
μ2≥2cn+2c−2n−23, μ3≥2n2+6cn+6c−5n−79. |
Thus
h1+h2+h3=[111]M6[μ1μ2μ3]≥[nn−33][2cn+2c32cn+2c−2n−232n2+6cn+6c−5n−79]=2cn2+2cn3+(n−3)(2cn+2c−2n−2)3+2n2+6cn+6c−5n−73. |
Let f3(c) be a function of c and
f3(c)=2cn2+2cn3+(n−3)(2cn+2c−2n−2)3+2n2+6cn+6c−5n−73, |
then
f′3(c)=4n2+4n3>0, |
and f3(c) is an increasing function. Because
1≤c≤2n−73, |
then
exp(D)≤f3(2n−73)=8n3−20n2−31n−39. |
Include proofs in the form
exp(D)≥8n3−20n2−31n−39, |
such that the equation is satisfied.
Necessity. We can use proof by contradiction. We just have to show
exp(D)<8n3−20n2−31n−39 |
if there are no continuous yellow arcs of n+13 length.
In this case, there are at most n−23 long continuous yellow paths on the n-cycle. For every pair of vertices (vi,vj) in D, we say that pvivj represents the shortest path from vi to vj, and denote
u(pvivj)=x,v(pvivj)=y,andw(pvivj)=z. |
We can always find the same walk to go from vi, along pvivj to vj, and the numbers of revolutions around the n-cycle, the (n−3)-cycle, and the 3-cycle are p1 times, p2 times, and p3 times, respectively.
Taking
p1=2cn−c3+(c+1)x−2cy+cz, p2=2cn+2c−2n−23−cx+(2c−2)y−(c−1)z |
and
p3=2n2+6cn+6c−5n−79−(n+13+c)x+(2n−73+2c)y−(n−23+c)z, |
we see that
[xyz]+p1[c−1n+132n+23−c]+p2[cn−232n−73−c]+p3[012] |
=[(2cn−c)(c−1)3+2c2n+2c2−2cn−2c32cn2+cn−c9+(n−2)(2cn+2c−2n−2)9+2n2+6cn+6c−5n−79(2n−3c+2)(2cn−c)9+(2n−3c−7)(2cn+2c−2n−2)9+4n2+12cn+12c−10n−149]. |
Noting that
0≤x≤c, 0≤y≤n+13,and0≤z≤2n−3c+23. |
If x=c, then
0≤y≤n−23 |
and
0≤z≤2n−3c−73, |
and pvivj must contain the arc vn→v4, the initial and terminal vertices of pvivj must be on the (n−3)-cycle. If x=c−1, then
0≤y≤n+13 |
and
0≤z≤2n−3c+23. |
If
y=n+13, |
then x≥1, or z≥1. If
y=n−23, |
then
0≤x≤c |
and
0≤z≤2n−3c−73 |
or
0≤x≤c−1 |
and
0≤z≤2n−3c+23. |
So we can obtain p1≥0, p2≥0, and p3≥0. This gives
exp(D)≤(2cn−c)(c−1)3+2c2n+2c2−2cn−2c3+2cn2+cn−c9+(n−2)(2cn+2c−2n−2)9+2n2+6cn+6c−5n−79+(2n−3c+2)(2cn−c)9+(2n−3c−7)(2cn+2c−2n−2)9+4n2+12cn+12c−10n−149=2cn2−cn3+(n−3)(2cn+2c−2n−2)3+2n2+6cn+6c−5n−73. |
Let f4(c) be a function of c and
f4(c)=2cn2−cn3+(n−3)(2cn+2c−2n−2)3+2n2+6cn+6c−5n−73, |
then
f′4(c)=4n2+n3>0, |
and f4(c) is an increasing function. Because
1≤c≤2n−73, |
then
exp(D)≤f4(2n−73)=8n3−26n2−10n−39<8n3−20n2−31n−39. |
On the basis of previous references, we study the primitive problems of a class of three-colored digraphs shown in Figure 1 when there are zero red arcs, one yellow arc, and two blue arcs in the 3-cycle. We discussed the coloring of Figure 1 in different situations, found the conditions that the primitive digraph needs to satisfy, and obtained Theorems 1–7. In the primitive state, the maximum primitive exponent of the three-colored digraphs was found when
a=c−1, b=d+1, |
and the conclusion of Theorem 8 was obtained. Finally, the extreme digraph that reaches the upper bound of the exponent was characterized, and Theorem 9 was given. The research methods and conclusions of this article provide reference for using digraph coloring to solve some problems in physics, computer science, biology, and other fields.
Meijin Luo: the conceptualization, proof techniques, checking the calculations, and writing the manuscript; Qiutao Qin: the collection and organization of literature, research analysis, and manuscript. All authors have read and approved the final published version of the manuscript.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
This work was supported by the High-level Talent Start-up Project of Hechi University (No. 2022GCC008), National Natural Science Foundation of China (No. 31702049), Project of Young and Mid-aged College Teachers of Guangxi in 2024 (No. 2024KY0632) and Project of Hechi College (No. 2022YLXK001).
The authors declare that they have no conflicts of interest.
[1] |
D. Bravo, F. Cubría, M. Fiori, V. Trevisan, Characterization of digraphs with three complementarity eigenvalues, J. Algebraic Comb., 57 (2023), 1173–1193. https://doi.org/10.1007/s10801-023-01218-6 doi: 10.1007/s10801-023-01218-6
![]() |
[2] |
J. Méndez, R. Reyes, J. M. Rodríguez, J. M. Sigarreta, Gromov hyperbolicity of Johnson and Kneser graphs, Aequationes Math., 98 (2024), 661–686. https://doi.org/10.1007/s00010-024-01076-y doi: 10.1007/s00010-024-01076-y
![]() |
[3] |
T. H. Han, K. Y. Jang, Y. T. Dong, R. H. Friend, E. H. Sargent, T. W. Lee, A roadmap for the commercialization of perovskite light emitters, Nat. Rev. Mater., 7 (2022), 757–777. https://doi.org/10.1038/s41578-022-00459-4 doi: 10.1038/s41578-022-00459-4
![]() |
[4] |
J. Guo, Y. H. Fu, W. J. Zheng, M. Y. Xie, Y. C. Huang, Z. Y. Miao, et al., Entropy-driven strongly confined low-toxicity pure-red perovskite quantum dots for spectrally stable light-emitting diodes, Nano Lett., 24 (2024), 417–423. https://doi.org/10.1021/acs.nanolett.3c04214 doi: 10.1021/acs.nanolett.3c04214
![]() |
[5] |
Z. F. Wang, S. Z. Ye, H. Wang, Q. J. Huang, J. He, S. Chang, Graph representation-based machine learning framework for predicting electronic band structures of quantum-confined nanostructures, Sci. China Mater., 65 (2022), 3157–3170. https://doi.org/10.1007/s40843-022-2103-9 doi: 10.1007/s40843-022-2103-9
![]() |
[6] |
Y. Tokuhara, T. Akutsu, J. M. Schwartz, J. C. Nacher, A practically efficient algorithm for identifying critical control proteins in directed probabilistic biological networks, npj Syst. Biol. Appl., 10 (2024), 87. https://doi.org/10.1038/s41540-024-00411-y doi: 10.1038/s41540-024-00411-y
![]() |
[7] |
A. Rahimi, A. Mirzazadeh, S. Tavakolpour, Genetics and genomics of SARS-CoV-2: a review of the literature with the special focus on genetic diversity and SARS-CoV-2 genome detection, Genomics, 113 (2021), 1221–1232. https://doi.org/10.1016/j.ygeno.2020.09.059 doi: 10.1016/j.ygeno.2020.09.059
![]() |
[8] |
Y. L. Wang, Y. F. Sun, Z. L. Shen, C. Wang, J. Qian, Q. Y. Mao, et al., Fully human single-domain antibody targeting a highly conserved cryptic epitope on the Nipah virus G protein, Nat. Commun., 15 (2024), 6892. https://doi.org/10.1038/s41467-024-51066-6 doi: 10.1038/s41467-024-51066-6
![]() |
[9] |
Y. M. Si, D. H. Zhou, J. Zhao, Y. F. Peng, P. Z. Chen, J. L. Fan, et al., Radiation chemistry of a novel zinc-oxo cluster crosslinking strategy for EUV patterning, Sci. China Mater., 67 (2024), 1588–1593. https://doi.org/10.1007/s40843-023-2827-8 doi: 10.1007/s40843-023-2827-8
![]() |
[10] | Y. J. Wang, Research on graph invariants as well as their applications in complex networks, North University of China, 2021. |
[11] | H. P. Sun, A study of heuristic algorithms for graph coloring problem, Guangzhou University, 2023. |
[12] |
J. H. Song, X. F. Wang, S. M. Hu, J. W. Jia, D. Yan, Algorithmic research overview on graph coloring problems, Comput. Eng. Appl., 60 (2024), 66–77. https://doi.org/10.3778/j.issn.1002-8331.2403-0434 doi: 10.3778/j.issn.1002-8331.2403-0434
![]() |
[13] | R. A. Brualdi, H. J. Ryser, Combinatorial matrix theory, Cambridge University Press, 1991. https://doi.org/10.1017/cbo9781107325708 |
[14] |
B. L. Shader, S. Suwilo, Exponents of nonnegative matrix pairs, Linear Algebra Appl., 363 (2003), 275–293. https://doi.org/10.1016/s0024-3795(01)00566-3 doi: 10.1016/s0024-3795(01)00566-3
![]() |
[15] |
Y. B. Gao, Y. L. Shao, Exponents of two-colored digraphs with two cycles, Linear Algebra Appl., 407 (2005), 263–276. https://doi.org/10.1016/j.laa.2005.05.004 doi: 10.1016/j.laa.2005.05.004
![]() |
[16] | X. Li, Primitive exponents of a class of special two-colored digraphs, J. Jiangsu Normal Univ., 30 (2012), 6–8. |
[17] | J. Q. Lin, Exponents of a class of two-colored digraphs with directed and double directed cycles alternated, J. Taiyuan Normal Univ., 11 (2012), 16–19. |
[18] |
Mulyono, H. Sumardi, S. Suwilo, The scrambling index of primitive two-colored two cycles whose lengths differ by 1, Far East J. Math. Sci., 96 (2015), 113–132. https://doi.org/10.17654/FJMSJan2015_113_132 doi: 10.17654/FJMSJan2015_113_132
![]() |
[19] | M. J. Luo, X. Li, S. Tao, Upper bound of primitive exponent of a class of nonnegative matrix pairs, Proceedings of the 4th International Conference on Information Technology and Management Innovation, 2015, 34–37. https://doi.org/10.2991/icitmi-15.2015.7 |
[20] | P. Yang, Y. B. Gao, Primitive exponents of a class of two-colored digraphs with two cycles, J. North Univ. China, 30 (2009), 405–409. |
[21] | Y. Ma, S. X. Chen, The base set of primitive non-powerful signed simple graphs, Math. Theory Appl., 34 (2014), 61–65. |
[22] | M. J. Luo, X. D. Li, Z. Y. Hou, X. Li, Exponents of a class of special primitive three-colored digraphs with three cycles, Ars Comb., 148 (2020), 33–46. |
[23] | Q. Yan, Y. L. Shao, Exponents of a class of three-colored digraphs, J. North Univ. China, 30 (2009), 512–517. |
[24] |
H. Q. Liu, Primitive exponents of a class of three-colored digraphs with two m-cycles, Shanxi Agric. Univ., 30 (2010), 380–384. https://doi.org/10.13842/j.cnki.issn1671-8151.2010.04.007 doi: 10.13842/j.cnki.issn1671-8151.2010.04.007
![]() |
[25] |
H. L. Zhou, Q. Y. Song, Exponents of a class of three-colored digraphs, J. Jilin Normal Univ., 5 (2011), 82–89. https://doi.org/10.16862/j.cnki.issn1674-3873.2011.02.022 doi: 10.16862/j.cnki.issn1674-3873.2011.02.022
![]() |