---
title: Generalized Corona Product
url: https://www.emergentmind.com/topics/generalised-corona-product
type: topic
---

# Generalized Corona Product

Searching arXiv for recent and foundational papers on generalized corona products and related corona-type graph products.
The generalized corona product is a family of graph constructions that extends the classical corona by allowing nonuniform attachments to a base graph. In the unsigned setting, one standard form takes a graph \(G\) and a collection of graphs \(H_i\), attaching one \(H_i\) to each vertex of \(G\); an edge-based analogue attaches one \(H_i\) to each edge of \(G\). These constructions preserve a transparent block structure while substantially modifying graph invariants such as order, size, diameter, chromatic parameters, domination, independence, geodeticity, and spectra. The notion has also been extended to signed graphs, where the attachment edges inherit signs from vertex markings, enabling structural-balance and cospectrality analyses by means of signed coronals [2411.13139], [1712.04699], [2310.08057].

## 1. Core definitions and variants

For a simple connected graph \(G=(V(G),E(G))\) with \(V(G)=\{u_1,u_2,\dots,u_n\}\), and simple graphs \(H_i\) of orders \(t_i=|V(H_i)|\), the generalized corona is denoted
\[
G\;\tilde\circ\;\bigwedge_{i=1}^n H_i .
\]
It is obtained by taking one copy of \(G\), one copy of each \(H_i\), and then adding, for each \(i\), all edges joining \(u_i\) to every vertex of \(H_i\). Equivalently,
\[
V\bigl(G\;\tilde\circ\;\bigwedge_i H_i\bigr)
=V(G)\;\cup\;\bigsqcup_{i=1}^n V(H_i),
\]
and
\[
E\bigl(G\;\tilde\circ\;\bigwedge_i H_i\bigr)
=E(G)\;\cup\;\bigcup_{i=1}^n E(H_i)\;\cup\;
\bigl\{\{u_i,v\}:v\in V(H_i),\;i=1,\dots,n\bigr\}.
\]
When all \(H_i\cong H\), this reduces to the ordinary corona product \(G\circ H\) [2411.13139].

An edge-based generalization appears in the generalized edge corona product. If \(G\) is a simple graph with edge set \(E(G)=\{e_1,\dots,e_m\}\), where \(e_i=\{u_i,v_i\}\), and \(H_i\) are simple graphs, then
\[
G\diamond (H_1,\dots,H_m)
\]
has vertex set
\[
V\bigl(G\diamond(H_1,\dots,H_m)\bigr)
=
V(G)\;\cup\;\bigcup_{i=1}^m V(H_i),
\]
and edge set
\[
E\bigl(G\diamond(H_1,\dots,H_m)\bigr)
=
E(G)\;\cup\;\bigcup_{i=1}^m E(H_i)
\;\cup\;\biggl\{
\{u_i,w\},\,\{v_i,w\}\;\bigm|\;
1\le i\le m,\;w\in V(H_i)
\biggr\}.
\]
In words, for each edge \(e_i=\{u_i,v_i\}\) of \(G\), the graph \(H_i\) is attached by joining each of its vertices to both \(u_i\) and \(v_i\) [1712.04699].

A related uniform edge-corona construction uses a single graph \(H\), attaching a fresh copy \(H_e\cong H\) to each edge \(e\in E(G)\). In that notation,
\[
V\bigl(G\circledcirc H\bigr)
=
V(G)\;\biguplus\;\biguplus_{e\in E(G)}V_e,
\]
and
\[
E\bigl(G\circledcirc H\bigr)
=
E(G)\;\cup\;
\bigcup_{e=\{u,v\}\in E(G)}\bigl(E(H_e)\;\cup\;\{\{u,w\},\{v,w\} : w\in V_e\}\bigr).
\]
This is described as a natural “dual” of the classical vertex corona, because it attaches a copy of \(H\) to each edge of \(G\) rather than to each vertex [2002.12219].

The literature therefore uses “generalized corona product” in more than one sense. In one line of work it denotes nonuniform vertex-attached coronae; in another, nonuniform edge-attached coronae. A plausible implication is that careful attention to notation is necessary, since the same adjective “generalized” refers to different attachment schemes in different papers.

## 2. Constructional viewpoint and immediate structural parameters

The generalized corona admits a direct constructive description: start with one copy of \(G\), place one copy of each attachment graph, and connect the designated base vertex \(u_i\) to every vertex of the corresponding fiber \(H_i\) [2411.13139]. The generalized edge corona proceeds analogously, but the attachment site is an edge \(e_i=\{u_i,v_i\}\), and every vertex of \(H_i\) is joined to both endpoints [1712.04699].

For the generalized corona \(G\;\tilde\circ\;\bigwedge_i H_i\), the basic counts are explicit. Its order is
\[
\bigl|V(G\;\tilde\circ\;\bigwedge_i H_i)\bigr|
=
|V(G)|+\sum_{i=1}^n |V(H_i)|
=
n+\sum_{i=1}^n t_i,
\]
and its size is
\[
\bigl|E(G\;\tilde\circ\;\bigwedge_i H_i)\bigr|
=
|E(G)| \;+\;\sum_{i=1}^n |E(H_i)|\;+\;\sum_{i=1}^n t_i.
\]
If both \(G\) and each \(H_i\) are connected, then the product is connected. Its diameter satisfies
\[
\mathrm{diam}\bigl(G\;\tilde\circ\;\bigwedge_i H_i\bigr)
=\mathrm{diam}(G)+2.
\]
For the ordinary corona \(G\circ H\), antipodal pairs in the fibers correspond to antipodes in the base graph: two vertices \(v_p^i\in H_i\) and \(v_q^j\in H_j\) are antipodal in \(G\circ H\) if and only if \(u_i\) and \(u_j\) are antipodal in \(G\) [2411.13139].

A simple example of the edge-attached construction is obtained by taking \(G=P_2\), the single edge \(\{u,v\}\), and \(H_1=K_2=\{w,x\}\). Then the resulting graph has vertices \(\{u,v,w,x\}\), with edges \(\{u,v\}\), \(\{w,x\}\), and the four attachment edges \(\{u,w\},\{u,x\},\{v,w\},\{v,x\}\). The result is the familiar diamond graph on four vertices [1712.04699].

These formulas show that generalized corona products alter metric structure in a tightly controlled way: attachment contributes one extra step at each end of a base-to-base geodesic in the vertex-attached case, while in the edge-attached case the local neighborhoods of edge endpoints expand by the orders of the attached graphs. This suggests why distance-based colorings and geodetic parameters are especially natural invariants for these products.

## 3. Distance coloring and related chromatic parameters

For the generalized edge corona, the paper “On Generalized Edge Corona Product of Graphs” derives exact and sharp formulas for several \(k\)-distance chromatic numbers. If \(\chi_{\le k}(X)\) denotes the minimum number of colors so that any two vertices at graph-distance \(\le k\) receive different colors, then the ordinary chromatic number satisfies
\[
\chi\bigl(G\diamond(H_1,\dots,H_m)\bigr)
=
\max\Bigl\{
\chi(G),\;\max_{1\le i\le m}\{\chi(H_i)+2\}
\Bigr\}.
\]
The lower bound comes from the induced copy of \(G\) and from the need to color each \(H_i\) with colors distinct from the two endpoint colors of the supporting edge \(e_i\). The upper bound is achieved by first properly coloring \(G\) and then coloring each \(H_i\) with the remaining \(t-2\) colors when \(t\ge \chi(H_i)+2\) [1712.04699].

For trees, the 2-distance chromatic number is determined by the maximum degree and the orders of the attachment graphs. If \(T\) is a tree of maximum degree \(\Delta\), and \(e_{i_1},\dots,e_{i_\Delta}\) are the edges incident to a fixed vertex \(v\), with \(\lvert V(H_{i_k})\rvert=n_{i_k}\), then
\[
\chi_{\le2}\!\bigl(T\diamond(H_1,\dots,H_m)\bigr)
=
\Delta+1\;+\;\sum_{k=1}^\Delta n_{i_k}.
\]
In the uniform case \(H_i\cong H\) with \(\lvert V(H)\rvert=n\),
\[
\chi_{\le2}(T\diamond H)
=
(\,n+1\,)\,\Delta \;+\;1.
\]

For complete graphs \(K_n\), writing \(t=\max_i\lvert V(H_i)\rvert\) and \(\nu(K_n)\) for the matching number, the paper gives
\[
\chi_{\le2}\bigl(K_n\diamond(H_1,\dots)\bigr)
\le
\begin{cases}
n(t+1)-t, & n\text{ even},\\
n(t+1),   & n\text{ odd}.
\end{cases}
\]
In the uniform case \(H_i\cong H\) with \(\lvert V(H)\rvert=n_2\),
\[
\chi_{\le2}(K_n\diamond H)
=\begin{cases}
n(n_2+1)-n_2,&n\text{ even},\\
n(n_2+1),   &n\text{ odd}.
\end{cases}
\]

For complete bipartite graphs, the 3-distance chromatic number is exact:
\[
\chi_{\le3}\bigl(K_{m,n}\diamond(H_1,\dots,H_{mn})\bigr)
=
m+n \;+\;\sum_{i=1}^{mn}\lvert V(H_i)\rvert,
\]
and in the uniform case \(H_i\cong H\) of order \(n_2\),
\[
\chi_{\le3}(K_{m,n}\diamond H)
=m+n+mn\,n_2.
\]
The proof strategy combines the increase in maximum degree under the edge-corona operation, the inequality \(\chi_{\le2}(X)\ge \Delta(X)+1\), and partitioning-of-edge-sets arguments for complete graphs [1712.04699].

These results clarify a frequent misconception: generalized corona-type products do not merely shift the chromatic number by a fixed additive constant. In the edge-corona case, ordinary coloring depends on \(\chi(H_i)+2\), but distance coloring can scale with the maximum degree of the base graph and the total orders of multiple attachments around a high-degree vertex.

## 4. Domination, independence, and strong geodeticity

In the generalized edge corona, domination and independence admit exact evaluations when the base graph is connected. A set \(D\subseteq V(X)\) is dominating if every \(v\notin D\) has a neighbor in \(D\), and the domination number \(\gamma(X)\) is the size of a smallest dominating set. If \(G\) is connected, then
\[
\gamma\bigl(G\diamond(H_1,\dots,H_m)\bigr)
=
\beta(G),
\]
where \(\beta(G)\) is the vertex-cover number of \(G\). The proof uses the observation that any dominating set in the corona must dominate each attached edge-plus-fiber subgraph, and that placing dominating vertices inside the \(H_i\) pieces is never advantageous compared with moving them to endpoints in \(G\) [1712.04699].

For the same product, the independence number satisfies
\[
\alpha\bigl(G\diamond(H_1,\dots,H_m)\bigr)
=
\sum_{i=1}^m \alpha\bigl(H_i\bigr),
\]
again for connected \(G\). This formula isolates the independent behavior entirely in the fiber graphs, rather than in the base graph [1712.04699].

A different parameter family arises in recent work on strong geodeticity for corona-type products. For the generalized corona \(G\;\tilde\circ\;\bigwedge_i H_i\), let \(Sg'(H_i)\) denote the strong 2-geodetic number of \(H_i\), and \(\eta'_{Sg}(H_i)\) a corresponding basis. Then
\[
\eta_{Sg}\bigl(
G\;\tilde\circ\;\bigwedge_{i=1}^n H_i
\bigr)
=
\bigsqcup_{i=1}^n \eta'_{Sg}(H_i),
\qquad
Sg\bigl(
G\;\tilde\circ\;\bigwedge_{i=1}^n H_i
\bigr)
=
\sum_{i=1}^n Sg'(H_i).
\]
In the ordinary corona case, if each \(H_i\cong H\) and \(Sg'(H)=s\), then
\[
Sg(G\circ H)=n\cdot s.
\]
Special cases include
\[
Sg(G\circ \overline{K_m})=n\cdot m
\]
when \(H\) is edgeless on \(m\) vertices, and
\[
Sg(G\circ K_m)=n\cdot m
\]
when \(H\) is complete [2411.13139].

Very similar formulas hold for the generalized edge-corona and neighborhood-corona, in each case expressing the strong geodetic number of the product as the sum of the strong 2-geodetic numbers of the attached graphs [2411.13139]. This suggests a unifying principle across corona-type products: geodetic coverage is controlled by the fibers rather than by the base, once the attachment mechanism forces all long geodesics to pass through the base interface.

## 5. Spectral theory and coronal methods

A substantial branch of the theory concerns spectral factorization. For generalized corona products of signed graphs, the construction begins with a signed graph
\[
G_s=(V_G,E_G,\sigma_G,p),\qquad |V_G|=n,
\]
and signed graphs
\[
H_s^i=(V_{H_i},E_{H_i},\sigma_{H_i},p_i),\qquad |V_{H_i}|=k_i.
\]
The generalized corona product
\[
G_s\;\circ\;\{H_s^1,\dots,H_s^n\}
\]
is formed by attaching every vertex of \(H_s^i\) to the \(i\)th vertex of \(G_s\), with the sign of a new edge \(\{v_i,u\}\) declared to be
\[
\sigma\bigl(\{v_i,u\}\bigr)=p(v_i)\,p_i(u).
\]
Under a natural ordering of the vertices, the adjacency matrix has block form
\[
A\bigl(G_s\circ\{H_s^i\}\bigr)
=\begin{pmatrix}
A(G_s)&B\\
B^T&\displaystyle\bigoplus_{i=1}^n A(H_s^i)
\end{pmatrix},
\]
where the coupling block \(B\) is built from the marking vectors [2310.08057].

The key tool is the signed coronal
\[
X_{H_s}(X) = p_H^T\,(X\,I - A(H_s))^{-1}\,p_H,
\]
together with Laplacian and signless-Laplacian analogues
\[
X^{(L)}_{H_s}(a)
=
p_H^T\!\bigl((a-1)I - L(H_s)\bigr)^{-1}p_H,
\qquad
X^{(Q)}_{H_s}(b)
=
p_H^T\!\bigl((b-1)I - Q(H_s)\bigr)^{-1}p_H.
\]
Using Schur complements, the adjacency-characteristic polynomial factors as
\[
f_{\,G_s\circ\{H_s^i\}(X)
=
\prod_{i=1}^n f_{H_s^i}(X) \;\times\;
g\bigl(X_{H_s^1}(X),\dots,X_{H_s^n}(X);\,G_s\bigr),
\]
where
\[
g\bigl(x_1,\dots,x_n;\,G_s\bigr)
=
\det\Bigl(\operatorname{diag}(1-x_1,\dots,1-x_n)-A(G_s)\Bigr).
\]
If all \(H_s^i=H_s\), then
\[
f_{G_s\circ H_s}(X)
=
f_{H_s}(X)^{\,n}\;\times\;f_{G_s}\!\bigl(X - X_{H_s}(X)\bigr).
\]
Parallel factorizations hold for Laplacian and signless-Laplacian polynomials [2310.08057].

For ordinary iterated corona graphs generated from a seed graph \(S\), the adjacency matrix of the \(k\)th iterate \(C_k\) has an explicit block structure, and the adjacency, Laplacian, and signless Laplacian spectra can be built in closed form from the spectra of the previous iterate and of the seed when the seed is regular [1507.05335]. In particular, if \(S\) is \(r\)-regular, the adjacency spectrum of \(C_k\) consists of two branches \(\lambda^\pm_{i,k}\) defined recursively from the eigenvalues \(\mu_i\) of \(S\), together with the old eigenvalues \(\mu_i\) for \(i=1,\dots,n-1\), each with multiplicity \(n(n+1)^{k-1}\). Corresponding branch formulas are also given for Laplacian and signless-Laplacian eigenvalues [1507.05335].

The edge-corona network model on complete graphs goes further by giving explicit spectra for the normalized Laplacian of the iterated family
\[
\G_q(g)
:=
\underbrace{\bigl(\cdots\bigl(\K_{q+2}\circledcirc \K_q\bigr)\circledcirc \K_q\cdots\bigr)}_{g\text{ times }\,\circledcirc\,\K_q},
\]
with recursive spectral relation
\[
\Lambda_{g+1}
=
\Bigl\{\tfrac{\lambda_i(g)+q}{q+1}:i=1,\dots,N_q(g)\Bigr\}
\;\cup\;
\Bigl\{\tfrac{q-1}{q+1}\Bigr\}^{\,M_q(g)-N_q(g)}
\;\cup\;
\Bigl\{-\tfrac1{q+1}\Bigr\}^{\,(q-1)M_q(g)+N_q(g)},
\]
and corresponding normalized-Laplacian formula
\[
\Sigma_{g+1}
=
\Bigl\{\tfrac{\sigma_i(g)}{q+1}\Bigr\}^{N_q(g)}
\cup
\Bigl\{\tfrac{2}{q+1}\Bigr\}^{\,M_q(g)-N_q(g)}
\cup
\Bigl\{\tfrac{q+2}{q+1}\Bigr\}^{\,(q-1)M_q(g)+N_q(g)}.
\]
This explicit solvability supports exact formulas for mixing time, mean hitting time, and the number of spanning trees [2002.12219].

## 6. Iterative models, network structure, and signed extensions

Iterated corona constructions have been used as deterministic network models. For a fixed seed graph \(S\), the iterated corona graphs are defined by
\[
C_0=S,\qquad C_1=S\circ S,\qquad C_k=C_{k-1}\circ S,\quad k\ge 2.
\]
When \(S\) is regular, the cumulative degree distribution of \(C_k\) is
\[
P_c(k)=(n+1)^{(r+1-k)/n},
\]
which decays exponentially in \(k\). When the seed is a clique \(K_n\), the cumulative betweenness distribution satisfies
\[
P_c(b)\propto b^{-1},
\]
that is, a power law with exponent near \(2\) [1507.05335].

For the edge-corona simplicial-network model based on repeated attachment of \(K_q\) to the edges of \(K_{q+2}\), the numbers of edges and vertices are
\[
M_q(g)
=\Bigl(\tfrac{(q+1)(q+2)}2\Bigr)^{\,g+1},
\qquad
N_q(g)
=\frac{2}{q+3}\Bigl(\tfrac{(q+1)(q+2)}2\Bigr)^{\,g+1}
+\frac{2(q+2)}{q+3}.
\]
A vertex created at iteration \(g_v\) has degree
\[
d_v(g)=(q+1)^{\,g-g_v+1},
\]
and the resulting network is scale-free with exponent
\[
\gamma
=2+\frac{\ln(q+2)}{\ln(q+1)}-\frac{\ln2}{\ln(q+1)},
\qquad 2<\gamma<3.
\]
Its diameter satisfies
\[
\mathrm{diam}\bigl(\G_q(g)\bigr)
=\begin{cases}
g+1,&q=1,\\
2g+1,&q\ge 2,
\end{cases}
\qquad
\mathrm{diam}\sim O(\ln N_q(g)),
\]
and its clustering obeys
\[
C_v=\frac{q+1}{d},
\qquad
C\bigl(\G_q(g)\bigr)\xrightarrow[g\to\infty]{}
\frac{q^2+4q+3}{q^2+4q+5}.
\]
The model also yields closed-form clique counts:
\[
N_k(g)
=\frac{\bigl(\tfrac{(q+1)(q+2)}2\bigr)^{\,g+1}-1}
{\tfrac{(q+1)(q+2)}2-1}
\binom{q+2}{k},
\qquad 3\le k\le q+2,
\]
with \(N_{k>q+2}(g)=0\) and \(N_2(g)=M_q(g)\) [2002.12219].

Signed generalizations introduce a different layer of structure. In the corona product of signed graphs, the spoke-edge signs are determined by vertex markings \(\mu_1,\mu_2\), and one obtains closed-form counts of positive and negative edges, triads of types \(T_0,\dots,T_3\), and spectral formulas for adjacency and signed Laplacian matrices [1908.10018]. Iterating the signed corona
\[
G^{(m)}=G^{(m-1)}\circ G
\]
produces a deterministic growing signed-network model with
\[
|V(G^{(m)})| = n\cdot (n+1)^m,
\qquad
|E(G^{(m)})| = [|E(G)| + n]\cdot (n+1)^m - n.
\]
The algebraic conflict is tracked via iterates of a quadratic spectral map, and in a specific uniform-\(d^{-}\) regime the minimum Laplacian eigenvalue can remain equal to that of the seed for all iterations [1908.10018].

## 7. Structural balance, cospectrality, and scope of the concept

For generalized corona products of signed graphs, structural balance is not determined solely by the balance of the constituents. One result states that if any constituent \(H_s^i\) or \(G_s\) is unbalanced then so is the product. Conversely, even when all \(G_s,H_s^i\) are balanced, the added coronal edges can create unbalanced 3-triads if there is an edge in some \(H_s^i\) of one of three types: a positive edge joining two opposite-marked vertices, a negative edge joining two negatively marked vertices, or a negative edge joining two positively marked vertices [2310.08057].

The same coronal machinery yields cospectrality criteria. If \(G_s\) is fixed and two collections \(\{H_s^i\}\) and \(\{\tilde H_s^i\}\) satisfy
\[
f_{H_s^i}(X)=f_{\tilde H_s^i}(X)
\quad\text{and}\quad
X_{H_s^i}(X)=X_{\tilde H_s^i}(X)
\quad\text{for all }i,
\]
then the corresponding generalized corona products are adjacency-cospectral. Analogous statements hold for Laplacian and signless-Laplacian cospectrality when the corresponding signed coronals agree. If the attachments are all identical, then cospectral base graphs produce cospectral corona products as well [2310.08057].

A broader perspective emerges from later work on digraphs, where two corona-type constructions are introduced: the vertex-corona and the arc-corona. The arc-corona includes a symmetric variant that attaches copies along undirected edges of the underlying graph, with adjacency blocks written using incidence matrices. This suggests that the passage from vertex-based to edge-based attachment is not confined to undirected graphs but fits into a wider coronal framework spanning adjacency, Laplacian, and signless-Laplacian spectra [2509.14481].

Taken together, these results show that “generalised corona product” is best understood as a family of attachment operators rather than a single graph product. The shared feature is a one-copy base graph together with multiple attached fibers, possibly heterogeneous, attached either at vertices or at edges. What varies across the literature is the attachment site, the role of markings or signs, and the target invariant: distance coloring in generalized edge coronae [1712.04699], strong geodeticity in corona-type products [2411.13139], recursive and closed-form spectral theory for iterated corona graphs [1507.05335], analytically solvable edge-corona network models [2002.12219], and structural balance and cospectrality in signed generalizations [2310.08057], [1908.10018].

Source: https://www.emergentmind.com/topics/generalised-corona-product