On distance integral and distance Laplacian integral graphs S. Pirzada, Ummer Mushtaq, Leonardo de Lima Abstract Let G G be a connected graph on n n vertices and let D ⁡ ( G ) D(G) and D L ​ ( G ) D^{L}(G) be the distance and the distance Laplacian matrices associated with G G . A graph G G is said to be D D -integral (resp. D L D^{L} -integral) if all eigenvalues of D ⁡ ( G ) D(G) (resp. D L ​ ( G ) D^{L}(G) ) are integers. In this paper, we obtain various conditions under which the graphs a K m ¯ ∇ C n a\overline{K_{m}}\nabla C_{n} and K p , p ∇ C n K_{p,p}\nabla C_{n} are distance integral. We also obtain conditions on m m , n n under which the dumbbell graph 𝑫 ​ 𝑩 ​ ( W m , n ) \boldsymbol{DB}(W_{m,n}) is D L D^{L} -integral. 2020 MSC 05C50 1 Introduction We consider a connected simple graph G=(V,E) , where V represents the vertex set with cardinality n and E denotes the edge set. The adjacency matrix of graph G , denoted by A(G)=a_{ij} , is a square matrix with a_{ij}=1 if the vertex v_{i} is adjacent to v_{j} , and a_{ij}=0 otherwise. A graph is regular if each vertex has the same degree. If M represents the matrix associated with a graph G , then G is termed M -integral if all eigenvalues of M are integers. The M -spectrum of a graph G encompasses all eigenvalues of its matrix M along with their respective multiplicities. Let d_{G}(i,j) be the length of the shortest path between vertices v_{i} and v_{j}. The n\times n distance matrix of G , denoted by D(G) , has d_{G}(i,j) as its (i,j) -entry. Notice that D(G) is symmetric and so has real eigenvalues. Write Tr(G) for the transmission matrix of G which is the diagonal matrix of the row sums of D(G) . The distance Laplacian of G is given by D^{L}(G)=Tr(G)-D(G). The complement of G , denoted as \overline{G} , is the graph sharing the same vertex set as G , where two vertices are adjacent if and only if they are not adjacent in G . The union of two graphs G_{1} and G_{2} , denoted G_{1}\cup G_{2} , comprises the vertex set V(G_{1})\cup V(G_{2}) and the edge set E(G_{1})\cup E(G_{2}) . The join G_{1}\nabla G_{2} of two graphs G_{1} and G_{2} is the graph obtained from G_{1}\cup G_{2} by adding all possible edges from the vertices of G_{1} to the vertices of G_{2} . Conventionally, K_{n} denotes the complete graph, C_{n} represents the cycle and K_{p,q} denotes the complete bipartite graph on p+q vertices. Given m\geq 2 and n\geq 3, the generalized wheel graph W_{m,n} is the graph obtained by the join \overline{K_{m}}\nabla C_{n} . The dumbbell graph, denoted by DB(W_{m,n}) , is obtained from two copies of generalized wheel graph W_{m,n} by connecting m vertices in one copy with the corresponding vertices in the other copy [ 11 ] . For more definitions and notations, we refer to [ 15 ] . In spectral graph theory, a fundamental question is about identifying graphs where all the eigenvalues of a matrix associated with the graph are integers. This quest gained momentum in 1974 when Harary and Schwenk [ 7 ] introduced the concept of integral graphs, igniting widespread interest and prompting extensive research in this area. In 2002, Balaliska et al. [ 5 ] presented a comprehensive survey of findings concerning integral graphs. For further insights and results, we refer to [ 1 , 2 , 5 , 6 , 13 , 18 ] . Regarding the D -integral graph, the computational results of [ 16 ] point out that there are only 49 D -integral graphs up to 9 vertices. The rarity of D -integral graphs is a motivation to characterize graphs with this property, and some papers in the literature have addressed this subject. In [ 8 ] , Ilić characterized the distance spectra of integral circulant graphs and demonstrated their D -integrality. In 2015, Pokorný et al. [ 16 ] delineated D -integral graphs within various classes, including complete split graphs, multiple complete split-like graphs, extended complete split-like graphs, and multiple extended complete split-like graphs, also proving the non- D -integrality of nontrivial trees. In the same year, Yang and Wang [ 19 ] established a necessary and sufficient condition for the D -integrality of complete r -partite graphs K_{p_{1},p_{2},\dots,p_{r}}\cong K_{a_{1}p_{1},a_{2}p_{2},\dots,a_{s}p_{s}} and generated numerous new classes of D -integral graphs for s=1,2,3,4 . In 2016, Zhao et al. [ 21 ] provided necessary and sufficient conditions for determining the D^{Q} -integrality of complete r -partite graph such as K_{{p}_{1},p_{2},\dots,p_{r}}\cong K_{a_{1}p_{1},a_{2}p_{2},\dots,a_{s}p_{s}} and introduced new classes of D^{Q} -integral graphs for s=1,2,3 . Subsequently, in 2017, da Silva et al. [ 17 ] investigated the concept of D^{L} -integrality and integrality of the distance signless Laplacian eigenvalues in various types of graphs, including complete split graphs, multiple complete split-like graphs, extended complete split-like graphs, and multiple extended complete split-like graphs, drawing on Pokorný’s work [ 16 ] . Additional insights and advancements in this field can be found in [ 9 , 14 ] . In 2023, Lu et al. [ 13 ] obtained all D -integral generalized wheel graphs aK_{m}\nabla C_{n} . The rest of this paper is organized as follows. In Section 2, we completely characterize all D -integral graphs of the forms a\overline{K_{m}}\nabla C_{n} and K_{p,p}\nabla C_{n}, and prove that there are no D -integral dumbbell graphs. In Section 3, we derive all D^{L} -integral dumbbell graphs \boldsymbol{DB}(W_{m,n}) . 2 D -integral of a\overline{K_{m}}\nabla C_{n} and K_{p,p}\nabla C_{n} In this section, we shall provide the distance spectrum of the generalized graph a\overline{K_{m}}\nabla C_{n} and K_{p,p}\nabla C_{n} . We denote a\overline{K_{m}}\nabla C_{n} by EGW(a,m,n) and completely determine all D -integral EGW(a,m,n) graphs. The following observations will be used in the sequel. Lemma 2.1 . [ 4 ] The adjacency spectrum of the cycle C_{n} is \displaystyle\Bigg\{2cos\left(\frac{2k\pi}{n}\right):k=1,2,\dots,n-1\Bigg\}. Lemma 2.2 . [ 4 ] Let M be a square matrix of order n that can be written in the blocks as M=\begin{pmatrix}M_{11}&M_{12}&\dots&M_{1k}\\ M_{21}&M_{22}&\dots&M_{2k}\\ \dots&\dots&\dots&\dots\\ \dots&\dots&\dots&\dots\\ M_{k1}&M_{k2}&\dots&M_{kk}\end{pmatrix} where M_{ij},1\leq i,j\leq k, is the n_{i}\times m_{j} matrix such that its lines have constant row sum equal to c_{ij} . Let \overline{M}=[c_{ij}]_{k\times k} . Then the eigenvalues of \overline{M} are also eigenvalues of M . Lemma 2.3 . [ 20 ] Let M be defined as in Lemma 2.2 such that M_{ij}=s_{ij}J_{n_{i},n_{j}} for i\neq j , and M_{ii}=s_{ii}J_{n_{i},n_{i}}+p_{i}I_{n_{i}} . Then the quotient matrix of M is B=b_{ij} with b_{ij}=s_{ij}n_{j} if i\neq j , and b_{ii}=s_{ii}n_{i}+p_{i} . Moreover, \displaystyle\sigma(M)=\sigma(B)\cup\Big\{p_{1}^{n_{1}-1},p_{2}^{n_{2}-1},\dots,p_{k}^{n_{k}-1}\Big\}, where \sigma(A) represents the spectrum of matrix A . The following lemma gives the distance spectrum of \overline{K_{m}}\nabla C_{n} . Lemma 2.4 . The distance spectrum of \overline{K_{m}}\nabla C_{n} is \displaystyle\Bigg\{-2^{m-1},(m+n-3)\pm\sqrt{(m+n-3)^{2}-(3mn-8m-4n+8)}, \displaystyle-2-2cos\left(\frac{2k\pi}{n}\right):k=1,2,\dots,n-1\Bigg\}. Proof. Consider W_{m,n}=\overline{K_{m}}\nabla C_{n},m\geq 1,n\geq 3 , the join of the empty graph with m vertices and 2-regular cycle graph with n vertices. The distance matrix of W_{m,n} can be written as \displaystyle D(W_{m,n})=\begin{bmatrix}A&B\\ C&D\end{bmatrix}, where A=2(J-I)_{m\times m} , B=J_{m\times n} , C=J_{n\times m} , D=2(J-I)_{n\times n}-A(C_{n}) , and J_{m,n} is all ones matrix, A(C_{n}) is the adjacency matrix of the cycle C_{n} . Assume that V(\overline{K_{m}})=\{v_{1},v_{2},\dots,v_{m}\} and V(C_{n})=\{u_{1},u_{2},\dots,u_{n}\} be the vertex set of the graphs \overline{K_{m}} and C_{n} , respectively. Then the vertex set of W_{m,n} is V(W_{m,n})=V(\overline{K_{m}})\cup V(C_{n}) . Since \overline{K_{m}},m\geq 2 is an empty graph of order m , in D(W_{m,n}) , the shortest distance between the vertices v_{i}\in\overline{K_{m}},~i=1,2,\dots,m is two. Let e_{m} be all ones vector of length m , say e_{m}=(1,1,\dots,1)^{T} . In D(W_{m,n}) , the block matrix A has eigenvector [e_{m}\hskip 14.22636pt0^{T}_{n}]_{(m+n)\times 1} corresponding to the eigenvalue 2m-2 , while the remaining eigenvectors corresponding to the eigenvalue -2 with multiplicity m-1 are orthogonal to [e_{m}\hskip 14.22636pt0^{T}_{n}]_{(m+n)\times 1} . Also, A(C_{n}) has all ones vector e_{n}=(1,1,\dots,1)^{T} as an eigenvector corresponding to the eigenvalue 2 , the remaining eigenvectors are orthogonal to e_{n} . In D(W_{m,n}) , the block matrix D has an eigenvector [0^{T}_{m}\hskip 14.22636pte_{n}]_{(m+n)\times 1} corresponding to the eigenvalue 2n-4 while the remaining eigenvectors corresponding to the eigenvalues -2-2cos\left(\frac{2k\pi}{n}\right) , k=1,2,\dots,n-1 , are orthogonal to [0^{T}_{m}\hskip 14.22636pte_{n}]_{(m+n)\times 1} . The remaining two distance eigenvalues are \{(m+n-3)\pm\sqrt{(m+n-3)^{2}-(3mn-8m-4n+8)}\} , which are the zeros of the characteristic polynomial of the quotient matrix \begin{bmatrix}2m-2&n\\ m&2n-4\end{bmatrix} . This completes the proof. ∎ In the following lemma, we find the distance spectrum of the join of a copies of the empty graph \overline{K_{m}} and the cycle C_{n} . Lemma 2.5 . The distance spectrum of the extended EGW(a,m,n) consists of the eigenvalues \displaystyle\Bigg\{-2^{(am-1)},(am+n-3)\pm\sqrt{(am+n-3)^{2}-(3amn-8am-4n+8)}, \displaystyle-2-2cos\left(\frac{2k\pi}{n}\right);k=1,2,\dots,n-1\Bigg\}. Proof. Consider the graph a\overline{K_{m}}\nabla C_{n},~a\geq 2,m\geq 1,n\geq 3 , the join of a copies of the empty graph \overline{K_{m}} with m vertices and 2-regular cycle graph with n vertices. The distance matrix of a\overline{K_{m}}\nabla C_{n} can be written as D(a\overline{K_{m}}\nabla C_{n})=\begin{pmatrix}\begin{array}[]{cccc|c}X_{11}&X_{12}&\dots&X_{1a}&J_{a\times n}\\ X_{21}&X_{22}&\dots&X_{2a}&J_{a\times n}\\ \dots&\dots&\dots&\dots&\dots\\ \dots&\dots&\dots&\dots&\dots\\ X_{a1}&X_{a2}&\dots&X_{aa}&J_{a\times n}\\ \hline\cr J_{n\times a}&J_{n\times a}&\dots&J_{n\times a}&2(J-I)_{n\times n}-A(C_{n}),\\ \end{array}\\ \end{pmatrix} where \displaystyle X_{ij}=\begin{cases}2(J-I)_{a\times a},\qquad\mbox{if}~i=j\\ 2J_{a\times a},\qquad\mbox{if}~i\neq j,\end{cases} J is the all ones matrix and A(C_{n}) is the adjacency matrix of cycle C_{n} . Let V_{p}(\overline{K_{m}})=\{u^{p}_{1},u^{p}_{2},\dots,u^{p}_{m}\} and V(C_{n})=\{v_{1},v_{2},\dots,v_{n}\} be the vertex sets of the graphs p^{th} copy of \overline{K_{m}} , where 1\leq p\leq a , and C_{n} . The vertex set of a\overline{K_{m}}\nabla C_{n} is V(a\overline{K_{m}}\nabla C_{n})=\cup_{p=1}^{a}V_{p}(\overline{K_{m}})\cup V(C_{n}) . Let v^{1}_{m}=(1,1,\dots,1)^{T} be all ones vector of length, say m . Then in the matrix D(a\overline{K_{m}}\nabla C_{n}) , v^{1}_{m} is an eigenvector of the block X_{11} corresponding to the eigenvalue 2m-2 , while the remaining eigenvalues corresponding to -2 with algebraic multiplicity m-1 are orthogonal to v^{1}_{m} . Under these circumstances, -2 is an eigenvalue of D(a\overline{K_{m}}\nabla C_{n}) of multiplicity m-1 and the corresponding eigenvectors are orthogonal to [v^{1}_{m}\hskip 8.53581pt,0^{T}_{n},\dots,0^{T}_{m},0^{T}_{n}]_{(am+n)\times 1} . Similarly, for each diagonal block matrix X_{ii},~1\leq i\leq a , -2 is an eigenvalue of D(a\overline{K_{m}}\nabla C_{n}) of multiplicity m-1 and the corresponding eigenvectors are orthogonal to \displaystyle[0^{T}_{m},0^{T}_{m},\dots,v^{i}_{m}\hskip 8.53581pt\dots,0^{T}_{m},0^{T}_{n}]_{(am+n)\times 1}. Also, A(C_{n}) and 2(J-I)_{n\times n} have all ones vector e_{n}=(1,1,\dots,1)^{T} as an eigenvector corresponding to the eigenvalue 2 and 2n-2 , respectively. Then, the block 2(J-I)_{n\times n}-A(C_{n}) has an eigenvector e_{n} corresponding to the eigenvalue 2n-4. Therefore, -2-2cos(\frac{2k\pi}{n}),k=1,2,\dots,n-1 are the eigenvalues of D(a\overline{K_{m}}\nabla C_{n}) and the corresponding vectors are orthogonal to [0^{T}_{m},0^{T}_{m},\dots,0^{T}_{m},e_{n}]_{(am+n)\times 1} . In this way, we get -2-2cos(\frac{2k\pi}{n}),k=1,2,\dots,n-1 and -2 with multiplicity am-a as eigenvalues of D(a\overline{K_{m}}\nabla C_{n}) . The remaining eigenvalues are the eigenvalues of the quotient matrix \begin{pmatrix}\begin{array}[]{cccc|c}2m-2&2m&\dots&2m&n\\ \dots&\dots&\dots&\dots&\dots\\ 2m&2m&\dots&2m-2&n\\ \hline\cr n&n&n&n&2n-4\end{array}\\ \end{pmatrix}_{(a+1)\times(a+1)}. By Lemma 2.3 , clearly -2 is an eigenvalue of the above matrix with multiplicity a-1 and the remaining two eigenvalues are the zeros of the characteristic polynomial of the following matrix \displaystyle\begin{bmatrix}2am-2&n\\ am&2n-4\end{bmatrix}. By combining all the above, the complete spectrum of D(a\overline{K_{m}}\nabla C_{n}) is \displaystyle\Bigg\{-2^{(am-1)},(am+n-3)\pm\sqrt{(am+n-3)^{2}-(3amn-8am-4n+8)}, \displaystyle-2-2cos\left(\frac{2k\pi}{n}\right):k=1,2,\dots,n-1\Bigg\}. ∎ Now, we have the following result. Theorem 2.6 . The graph \overline{K_{m}}\nabla C_{n} with m+n vertices is D -integral if and only if one of the following holds: (i) n=3,~m=1 , (ii) n=3,~m=4 , (iii) n=4,~m=2 , (iv) n=6,~m=4 , (v) n=6,~m=12 . Proof. By Lemma 2.4 , the distance spectrum of \overline{K_{m}}\nabla C_{n} consists of the eigenvalue -2 with multiplicity m-1 and -2-2cos(\frac{2k\pi}{n}),k=1,2,\dots,n-1 and the remaining two eigenvalues are (m+n-3)\pm\sqrt{(m+n-3)^{2}-(3mn-8m-4n+8)}. Clearly, (m+n-3)\pm\sqrt{(m+n-3)^{2}-(3mn-8m-4n+8)} are integers if and only if (m+n-3)^{2}-(3mn-8m-4n+8) is a perfect square. Also, it is apparent that 2cos(\frac{2k\pi}{n}) is an integer for any 1\leq k\leq n-1 if and only if n\in\{3,4,6\} . Therefore, \overline{K_{m}}\nabla C_{n} is D -integral if and only if (m+n-3)^{2}-(m-4n+8) is a perfect square and n\in\{3,4,6\} . To establish sufficiency involves straightforward calculations, indicating that \overline{K_{m}}\nabla C_{n} becomes D -integral when condition (i), (ii), (iii) or (iv) holds. Now, delve into the necessity aspect. Let t=(m+n-3)^{2}-(3mn-8m-4n+8)=c^{2} . We consider the following three cases. Case 1. Let n=3 . Then, we have t=m^{2}-m+4=c^{2} , that is, m^{2}-m+4-c^{2}=0 . This gives, m=\frac{1\pm\sqrt{4c^{2}-15}}{2} . Since m is an integer which is possible only if 4c^{2}-15 is an odd perfect square number. Let 4c^{2}-15=p^{2} . This implies that 4c^{2}-p^{2}=15 , further implies that (2c-p)(2c+p)=15 , implies that c=4 or c=2 . Substituting back values of c , we get m=4 or m=1 . Case 2. Let n=4 . We have t=(m+1)^{2}-4m+8=c^{2} , that is, 8=c^{2}-(m-1)^{2} . This gives, 8=(c+1-m)(c-1+m) . Now, 8=8\times 1=2\times 4 . For the case (c+1-m)(c-1+m)=8\times 1 , we have 2c=9 , which is not possible. For the case (c+1-m)(c-1+m)=2\times 4 , we have c=3 , which gives m=2 . Case 3. Let n=6 . So, we have t=(m+3)^{2}-10m+16=c^{2} , that is, 21=c^{2}-(m-2)^{2} . We have 21=3\times 7=1\times 21 . Now, For the case (c-m+2)(c+m-2)=1 , we have c=11 , which gives m=12 . For the case (c-m+2)(c+m-2)=3\times 7 , we have c=5 , which gives m=4 . ∎ Theorem 2.7 . Let a\geq 1 and n\geq 3 . Then the extended graph EGW(a,m,n) on am+n vertices is D -integral if and only if the ordered triple (a,m,n)\in S, where \displaystyle S=\{(1,4,3),(4,1,3),(2,2,3),(1,1,3),(1,2,4),(2,1,4),(1,4,6),(2,2,6),(4,1,6), \displaystyle(1,12,6),(2,6,6),(3,4,6),(4,3,6),(6,2,6),(12,1,6)\} Proof. By Lemmas 2.4 and 2.5 , the distance spectrum of a\overline{K_{m}}\nabla C_{n} consists of the eigenvalues -2 with multiplicity am-1 and -2-2cos(\frac{2k\pi}{n}),k=1,2,\dots,n-1 , and the remaining two eigenvalues are (am+n-3)\pm\sqrt{(am+n-3)^{2}-(3amn-8am-4n+8)}. It is clear that (m+n-3)\pm\sqrt{(am+n-3)^{2}-(3amn-8am-4n+8)} are integers if and only if (am+n-3)^{2}-(3amn-8am-4n+8) is a perfect square. Moreover, it is obvious that 2cos(\frac{2k\pi}{n}) is an integer for any 1\leq k\leq n-1 if and only if n\in\{3,4,6\} . Therefore, the extended graph EGW(a,m,n) is D -integral if and only if (am+n-3)^{2}-(3amn-8am-4n+8) is a perfect square and n\in\{3,4,6\} . Let t=(am+n-3)^{2}-(3amn-8am-4n+8)=c^{2} . We consider the following three cases. Case 1. Let n=3 . Then, we have t=(am)^{2}-am+4=c^{2} , that is, (am)^{2}-am+4-c^{2}=0 . This gives, am=\frac{1\pm\sqrt{4c^{2}-15}}{2} . Since am is an integer which is possible only if 4c^{2}-15 is an odd perfect square. We suppose that 4c^{2}-15=p^{2} , for some positive integer p . Then 4c^{2}-p^{2}=15 , that is, (2c-p)(2c+p)=15 . Since 15=1\times 15=5\times 3 , this gives c=4 or c=2 . By substituting the value of c in am=\frac{1\pm\sqrt{4c^{2}-15}}{2} , we get am=4 or am=1 . Since 4=4\times 1=2\times 2=1\times 4 , therefore the possible values of a and m are a=1,~m=4 or a=4,~m=1 or a=2,~m=2 . Similarly, am=1 gives the possible values as a=1,~m=1 . Thus, for n=3 , the possible extended distance integral graphs EGW(a,m,n) are EGW(1,4,3),~EGW(4,1,3),~EGW(2,2,3),~EGW(1,1,3) . Case 2. Let n=4 . We have t=(am+1)^{2}-4am+8=c^{2} , that is, (am-1)^{2}+8=c^{2} . This gives c^{2}-(am-1)^{2}=8 . Now by factorizing, we get (c-am+1)(c+am-1)=2\times 4=1\times 8 . Then the possible values of c is 3. By back substituting the value of c=3 , we get am=2 . Since 2=1\times 2=1\times 2 , therefore, the possible values of a and m corresponding to n=4 are a=1,~m=2 or a=1,~m=1 . Therefore, for n=4 , the possible extended distance integral graphs are EGW(1,2,4) and EGW(2,1,4) . Case 3. Let n=6 . In this case, we have t=(am+3)^{2}-10am+16=c^{2} , that is, (am-2)^{2}+21=c^{2} . This gives (c-am+2)(c+am-2)=1\times 21=3\times 7 . Here the possible values c are 5 and 11 . By back substituting the values of c , we get am=4 and am=12 . Since 12=1\times 12=2\times 6=3\times 4=4\times 3=6\times 2=12\times 1 and 4=1\times 4=2\times 2=4\times 1 , therefore, the possible values of a and m corresponding to n=6 are a=1,~m=4 or a=4,~m=1 or a=2,~m2 or a=1,~m=12 or a=2,~m=6 or a=3,~m=4 or a=4,~m=3 or a=6,~m=2 or a=12,~m=1 . Thus, for n=6 , the possible extended distance integral graphs EGW(a,m,n) are EGW(1,4,6),EGW(2,2,6),EGW(4,1,6),EGW(1,12,6),EGW(2,6,6),\\ EGW(3,4,6) , EGW(4,3,6) , EGW(6,2,6) , EGW(12,1,6) . ∎ The following observation gives the distance spectrum of the graph K_{p,p}\nabla C_{n} . Theorem 2.8 . The distance spectrum of K_{p,p}\nabla C_{n} is \displaystyle\Bigg\{-2^{2p-2},-2-2cos\left(\frac{2k\pi}{n}\right);k=1,2,\dots,n-1, \displaystyle\frac{2n+3p-6\pm\sqrt{(2n+3p-6)^{2}-16pn+48p+16n-32}}{2}\Bigg\} Fact 1. An ordered triplet of positive integers (a,b,c) is called Pythagorean triplet if a^{2}+b^{2}=c^{2} . We note that (3,4,5) is the only Pythagorean triplet whose one component is 4 . The following result gives a necessary and sufficient condition for the graph K_{p,p}\nabla C_{n} to be D -integral. Theorem 2.9 . The graph K_{p,p}\nabla C_{n} is D -integral if and only if (i) n=3,\hskip 2.84526ptp=1 and (ii) n=6,\hskip 2.84526ptp=4 . Proof. From Theorem 2.8 , the distance spectrum of the graph K_{p,p}\nabla C_{n} includes eigenvalues of -2 of multiplicity 2p-2 , as well as -2-2cos(\frac{2k\pi}{n}) , where k ranges from 1 to n-1 . Additionally, the remaining two eigenvalues are \displaystyle\frac{2n+3p-6\pm\sqrt{(2n+3p-6)^{2}-16pn+48p+16n-32}}{2}. It is evident that 2n+3p-6\pm\sqrt{(2n+3p-6)^{2}-16pn+48p+16n-32} will yield integer values if and only if (2n+3p-6)^{2}-16pn+48p+16n-32 is a perfect square. Furthermore, it is clear that 2\cos(\frac{2k\pi}{n}) results in the integer values for any 1\leq k\leq n-1 if and only if n belongs to the set \{3,4,6\} . Consequently, the graph K_{p,p}\nabla C_{n} is D -integral if and only if (2n+3p-6)^{2}-16pn+48p+16n-32 is a perfect square and n belongs to the set \{3,4,6\} . Let t=(2n+3p-6)^{2}-16pn+48p+16n-32=c^{2} . Now, we discuss the following three cases of n . Case 1. Let n=3 . Then, we have t=(3p)^{2}+4^{2}=c^{2} . Since each term on both sides is a perfect square and the left-hand side is the sum of two perfect square numbers with one number 4^{2} , so according to Fact 1, 3p should be equal to 3 , implying that p equals to 1 . Therefore, the only possible distance integral graph in this case is K_{1,1}\nabla C_{3} . Case 2. Let n=4 . In this case, we have t=(3p+2)^{2}-16p+32=c^{2} , that is, 9p^{2}-4p+36=c^{2} . This gives p=\frac{4\pm\sqrt{4c^{2}-1040}}{18} . We first check the values of c , for which 4c^{2}-1040 is a perfect square number. Assume that 4c^{2}-1040=q^{2} . Then (2c-q)(2c+q)=1040 . The different possibilities to write 1040 as a product of two factors are \displaystyle 1040=1\times 1040=2\times 520=4\times 260=8\times 130=10\times 104 \displaystyle=20\times 52=40\times 26=16\times 65=80\times 13=5\times 208. We eliminate the possibilities 1040=1\times 1040=2\times 520=8\times 130=40\times 26=16\times 65=80\times 13=10\times 104=5\times 208 , because none of these pairs has the sum a multiple of 4 . So we only consider two cases, namely (2c-q)(2c+q)=1040=4\times 260 and (2c-q)(2c+q)=1040=20\times 52 . For the case (2c-q)(2c+q)=4\times 260 , we get 4c=264 , implying c=66 and for the case (2c-q)(2c+q)=20\times 52 , we have c=18 . By substituting these values of c in p=\frac{4\pm\sqrt{4c^{2}-1040}}{18} , we get p=\frac{20}{18} and p=\frac{132}{18} , which is absurd. Thus, for n=4 , there is no value of p for which K_{p,p}\nabla C_{n} is distance integral. Case 3. Let n=6 . Here, we have t=(3p+6)^{2}-48p+64=c^{2} , that is, 9p^{2}-12p+100=c^{2} . This gives p=\frac{12\pm\sqrt{4c^{2}-3456}}{18} . Continuing with the same approach as in Case 2, we find the only integer value of p is 4 . Therefore, the only possible D -integral graph in this case is K_{4,4}\nabla C_{6} . This completes the proof. ∎ Next, we show that there is no dumbbell graph that is D-integral. Lemma 2.10 . [ 10 ] The distance spectrum of dumbbell graph \boldsymbol{DB}(W_{m,n}) consists of the set of eigenvalues \displaystyle\Bigg\{-4^{m-1},0^{m-1},\frac{1}{2}[-(m+n+4)\pm\sqrt{(m+n+4)^{2}-16m}], \displaystyle\frac{1}{2}[(5m+5n-8)\pm\sqrt{25m^{2}+25n^{2}-14mn}],-2-2cos\Bigg(\frac{2k\pi}{n}\Bigg);k=2,\dots,n\Bigg\}, where each of the eigenvalues -2-2cos(\frac{2k\pi}{n});k=2,3,\dots,n is of multiplicity 2 . Theorem 2.11 . There are no D -integral dumbbell graphs. Proof. By Lemma 2.10 , the D -spectrum of the dumbbell graph \boldsymbol{DB}(W_{m,n}) is given by \displaystyle\Bigg\{-4^{m-1},0^{m-1},\frac{1}{2}[-(m+n+4)\pm\sqrt{(m+n+4)^{2}-16m}], \displaystyle\frac{1}{2}[(5m+5n-8)\pm\sqrt{25m^{2}+25n^{2}-14mn}],-2-2cos(\frac{2k\pi}{n});k=2,\dots,n\Bigg\}, where each of the eigenvalues -2-2cos(\frac{2k\pi}{n});k=2,3,\dots,n is of multiplicity 2 . Since -(m+n+4) and (m+n+4)^{2}-16m are of the same parity and also (5m+5n-8) and 25m^{2}+25n^{2}-14mn are of the same parity, therefore it suffices to find the values of m and n for which both \frac{1}{2}[-(m+n+4)\pm\sqrt{(m+n+4)^{2}-16m} and \frac{1}{2}[(5m+5n-8)\pm\sqrt{25m^{2}+25n^{2}-14mn}] are perfect square numbers. As cos\frac{2k\pi}{n} is an integer only if n\in\{3,4,6\} , therefore, DB(W_{m.n}) is D -integral if and only if \sqrt{(m+n+4)^{2}-16m} and \sqrt{25m^{2}+25n^{2}-14mn} are integers, where n\in\{3,4,6\} . Assume that t=(m+n+4)^{2}-16m=c^{2} . We consider the following three cases. Case 1 . Let n=3 . We have t=(m+7)^{2}-16m=c^{2} , that is, 48=c^{2}-(m-1)^{2} which is equivalent to 48=(c-(m-1))(c+(m+1)) . As 48=1\times 48=2\times 24=3\times 16=6\times 8=4\times 12 , the sum of factors of 48 should be the multiple of 2 . Therefore, we consider only (c-(m-1))(c+(m-1))=4\times 12 , (c-(m-1))(c+(m-1))=2\times 24 and (c-(m-1))(c+(m-1))=6\times 8 . This gives c=8 , c=13 and c=7 . Substituting back the values of c , we get m=2 , m=5 and m=12 as the only integer values of m . Now, \sqrt{25m^{2}+25n^{2}-14mn} is not a perfect square number when m=2,5,12 and n=3 . Therefore, \boldsymbol{DB}(W_{m,n}) is not D -integral in this case. Case 2 . Let n=4 . Then t=(m+8)^{2}-16m=c^{2} , that is, 64=(c-m)(c+m) . So the possible values of c are 8,10 and 17 . Corresponding to these values, the possible values of m are 15 and 6 . But for m=6,15 and n=4 , we note that \sqrt{25m^{2}+25n^{2}-14mn} is not a perfect square number. Thus, \boldsymbol{DB}(W_{m,n}) is not D -integral in this case. Case 3 . For n=6 , we see that t=(m+2)^{2}+96=c^{2} . This gives, (c-(m+2))(c+(m+2))=96 . Proceeding similarly as in above cases, we observe that there exists no positive integral value of m , for which both \sqrt{(m+n+4)^{2}-16m} and \sqrt{25m^{2}+25n^{2}-14mn} are integers. Hence no distance integral dumbbell graphs exist in this case.∎ 3 D^{L} -integral dumbbell graphs In this section, we find all the possible distance Laplacian integral dumbbell graphs. We prove that there are only 8 scattered graphs in this family. The following result gives the distance Laplacian spectrum of a dumbbell graph. Lemma 3.1 . [ 9 ] The distance Laplacian spectrum of dumbbell graph \boldsymbol{DB}(W_{m,n}) consists of the eigenvalues \displaystyle\Bigg\{0,3m+3n,(5m+3n)^{m-1},(5m+3n-4)^{m-1},\Big\{(3m+5n-2)^{n-1}+2cos\left(\frac{2k\pi}{n}\right)\Big\}^{2}; \displaystyle k=1,2,\dots,n-1,\frac{1}{2}\Big[(9m+9n-4)\pm\sqrt{(3m-3n-4)^{2}+4mn}\Big]\Bigg\} Now, we show that there exist only 9 distance Laplacian integral dumbbell graphs. Theorem 3.2 . The only D^{L} -integral dumbbell graphs are \boldsymbol{DB}(W_{4,3}) , \boldsymbol{DB}(W_{5,3}) , \boldsymbol{DB}(W_{5,4}),\boldsymbol{DB}(W_{6,4}) , \boldsymbol{DB}(W_{14,4}) , \boldsymbol{DB}(W_{7,6}) , \boldsymbol{DB}(W_{8,6}) , \boldsymbol{DB}(W_{12,6}) and \boldsymbol{DB}(W_{19,6}) . Proof. The D^{L} -spectrum of the dumbbell graph \boldsymbol{DB}(W_{m,n}) is given in Lemma 3.1 . Since (9m+9n-4) and (3m-3n-4)^{2} are of the same parity, therefore, \frac{1}{2}\Big[(9m+9n-4)\pm\sqrt{(3m-3n-4)^{2}+4mn}\Big] is an integer if and only if (3m-3n-4)^{2}+4mn is a perfect square. Also, cos\frac{2k\pi}{n} is an integer only if n\in\{3,4,6\} . Therefore, DB(W_{m.n}) is D^{L} -integral if and only if \sqrt{(3m-3n-4)^{2}+4mn} is an integer where n\in\{3,4,6\} . Assume that t=(3m-3n-4)^{2}+4mn=c^{2} . We consider the following three cases. Case 1 . Let n=3 . In this case, we have t=(3m-11)^{2}+48=c^{2} , that is, 48=c^{2}-(3m-11)^{2} , which is equivalent to 48=(c-(3m-11))(c+(3m-11)) . Since 48=1\times 48=2\times 24=3\times 16=6\times 8=4\times 12 , so the sum of the factors of 48 should be the multiple of 2 . Therefore, we consider only (c-(3m-11))(c+(3m-11))=4\times 12 , (c-(3m-11))(c+(3m-11))=2\times 24 and (c-(3m-11))(c+(3m-11))=6\times 8 . This gives c=8 , c=13 and c=7 . Substituting back the values of c , we get m=4 and m=5 to be the only integer values of m . Thus, the only distance Laplacian integral dumbbell graphs when n=3 , are DB(W_{4,3}) and DB(W_{5,3}) . Case 2 . Let n=4 . Here, we have t=(3m-16)^{2}+16m=c^{2} , that is, 9m^{2}-80m+256-c^{2}=0 . This gives, m=\frac{80\pm\sqrt{36c^{2}-2816}}{18} . Since m is an integer, so 36c^{2}-2816 should be perfect square number. Let 36c^{2}-2816=p^{2} , that is, (6c-p)(6c+p)=2816 . Now the different possibilities for the factors (6c-p) and (6c+p) are \displaystyle 2816=4\times 704=2\times 1408=1\times 2816=256\times 11=128\times 22 \displaystyle 2816=64\times 44=32\times 88=16\times 176=8\times 352. The cases (6c-p)(6c+p)=1\times 2816 , (6c-p)(6c+p)=256\times 46 , (6c-p)(6c+p)=128\times 22 and (6c-p)(6c+p)=2\times 1408 are not possible, because the sum of these factors is not a multiple of 6 . For the case (6c-p)(6c+p)=8\times 52 , c=30 and the only possible positive integral value of m corresponding to c=30 is m=14 . Similarly, for the cases (6c-p)(6c+p)=16\times 176,(6c-p)(6c+p)=32\times 8,~(6c-p)(6c+p)=64\times 44 and (6c-p)(6c+p)=4\times 704 , the only possible integer values of m are 5 and 6 . Therefore, the only distance Laplacian integral dumbbell graphs when n=4 are DB(W_{5,4}),DB(W_{6,4}) and DB(W_{14,4}) . Case 3 . Taking n=6 , we see that t=(3m-22)^{2}+24m=c^{2} , that is, (3m-18)^{2}+160=c^{2} . This gives c^{2}-(3m-18)^{2}=160 , which implies that (c-(3m-18){2})(c-(3m-18){2})=160 . Now the different possibilities to write 160 as product of two factors are \displaystyle 160=1\times 160=5\times 32=10\times 16=20\times 8=40\times 4=80\times 2. The first two cases 160=1\times 160=5\times 32 and 1605=5\times 32 are not possible, as the sum of these factors is not a multiple of 2 . For the case 160=10\times 16 , c=13 and the only possible positive integral value of m corresponding to c=13 is m=7 . For the case (c-(3m-18){2})(c-(3m-18){2})=20\times 8 , we have c=14 . Using back the value of c gives m=8 . Similarly for the remaining two cases, the values of m are 12 and 19 . Thus, the only D^{L} -integral dumbbell graphs when n=6 are DB(W_{7,6}) , DB(W_{8,6}) , DB(W_{12,6}) and DB(W_{19,6}) . ∎ Acknowledgements. The research of S. Pirzada is supported by NBHM-DAE research project No. NBHM/02011/20/2022. The research of Leonardo de Lima is partially supported by CNPq grant 305988/2025-5. Data availability Data sharing does not apply to this article as no data sets were generated or analyzed during the current study. Conflict of interest. The authors declare that they have no conflict of interest.