---
title: Quantum Multigraphs Overview
url: https://www.emergentmind.com/topics/quantum-multigraphs
type: topic
---

# Quantum Multigraphs Overview

Quantum multigraphs are nonclassical generalizations of multigraphs that appear in several adjacent research programs, including operator-algebraic quantum graph theory, zero-error information theory, multipartite entanglement, quantum symmetry, quantum walks, and quantum statistical mechanics. In these settings, a multiedge need not be merely an integer multiplicity in an adjacency matrix: it may carry Kraus-space data, output labels, a noncommutative bimodule structure, or a controlled-phase operation. A recurring theme is that an ordinary single-edged graph is often recovered only after a collapse operation such as multiplication, partial trace, or explicit “edge counting” [2409.01951] [2604.06072].

## 1. Formal meanings of “quantum multigraph”

The literature contains several formal definitions of quantum multigraphs. In the confusability framework, a quantum multigraph on a pair \((M,N)\) of finite-dimensional algebras is a subspace \(V \subseteq B(H)\otimes N\) such that \(V\) is an \((M' \otimes 1)\)-bimodule and \((1 \otimes Z(N))V \subseteq V\). Here \(M\) plays the role of a quantum vertex set and \(N\) the role of a quantum label set, so multiplicity is built into the label algebra rather than represented only by an integer edge count [2604.06072].

A second operator-algebraic usage starts from a quantum graph specified by a triple \((V_G,\phi_G,A_G)\), where \(V_G\) is a finite quantum space, \(\phi_G\) is a faithful state, and \(A_G\) is a self-adjoint operator on the GNS Hilbert space \(L^2(C(G),\phi_G)\). In this setting, quantum multigraphs arise by allowing multiplicities or higher weights encoded in \(A_G\), and the associated operator system
\[
\mathcal{S}_G = \{ m(A \otimes X)m^* : X \in B(L^2(G)) \}
\]
captures the multiedged structure as well [2502.10521].

A third usage is algebraic-combinatorial rather than operator-algebraic. A quantum multigraph may be a formal real linear combination
\[
\alpha = \sum_{i=1}^n c_i F_i
\]
of finite undirected multigraphs, called nonnegative when
\[
t(\alpha,G) := \sum_i c_i\, t(F_i,G) \ge 0
\]
for all multigraphs \(G\). In that tradition, the object of study is not a single quantum system but a noncommutative analogue of graph-density inequalities and Positivstellensätze [1310.6903].

## 2. Operator-algebraic and categorical structure

A major unifying framework models finite quantum sets by finite-index inclusions of unital C\(^*\)-algebras
\[
A \subset B,
\]
with adjacency encoded by Schur idempotents \(\widehat{T} \in \mathrm{End}_{A-A}(B)\). The quantum Fourier transform
\[
\mathcal{F} : \mathrm{End}_{B-B}(B_1) \to \mathrm{End}_{A-A}(B)
\]
intertwines composition and the Schur product,
\[
\mathcal{F}(S_1 \circ S_2)=\mathcal{F}(S_1)\star \mathcal{F}(S_2), \qquad
S \star T := m \circ (S \otimes T) \circ m^\dagger.
\]
Within this formalism, classical multigraphs arise from integer-valued Schur idempotents, while quantum multiedges arise from higher-rank projections or sums thereof. The same language encompasses finite classical simple graphs, Cayley graphs of finite quantum groupoids, and all finite-dimensional quantum graphs [2409.01951].

The infinite-dimensional extension replaces bounded-operator bimodules by Hilbert–Schmidt quantum relations. For a von Neumann algebra \(M \subseteq \mathcal B(H)\), an HS quantum relation is a Hilbert–Schmidt norm-closed \(M'\)-bimodule in \(HS(H)\). Using an operator-valued weight \(\varphi^{-1}\), one obtains a self-dual Hilbert \(C^*\)-module that mediates a bijection between HS quantum relations and projections \(e \in M \bar\otimes M^{\mathrm{op}}\). When \(e\) is integrable for the slice map \(\operatorname{id}\otimes \varphi^{\mathrm{op}}\), there is an associated normal CP map \(A:M\to M\), interpreted as a quantum adjacency operator, with Kraus representation built from the HS quantum relation [2511.23121].

Concrete low-dimensional examples clarify the boundary between classical and genuinely quantum behavior. On the quantum space \(M_2\), all simple quantum graphs are classified, and all are quantum isomorphic to classical Cayley graphs of \(\mathbb Z_2 \times \mathbb Z_2\); the summaries of these examples state that quantum-specific multiple edges do not arise at that level [2109.13618].

## 3. Confusability, adjacency, and graph parameters

In channel theory, the quantum confusability multigraph refines the standard noncommutative confusability graph by retaining output information. For a quantum channel \(\Phi\) with linear Kraus space \(\mathcal K\), the confusability multigraph is
\[
\widetilde{S}_\Phi \cong \mathcal K^* \otimes \mathcal K,
\]
while the usual single-edged confusability graph is recovered by “counting” edges,
\[
S_\Phi = \mathcal K^* \mathcal K.
\]
This retains the full Kraus-space structure as a finer invariant than the operator system alone. The same work gives a necessary and sufficient condition: a quantum multigraph arises as a confusability multigraph of some quantum channel if and only if it is a symmetric decomposable quantum multi-relation [2604.06072].

A complementary development constructs large parametric families of quantum graphs that can still be treated discretely. In “Quantum Graph Theory by Example,” the examples are parametrised by triples of matrices \((A,B,C)\). The parametrisation decomposes the structure into classical and genuinely quantum parts: \(A\) and \(C\) are described by a classical weighted graph called the strange graph, while \(B\) supplies a purely quantum contribution with no classical analogue. On this basis, exact formulas or bounds are given for the number of connected components, the chromatic number, the independence number, and the clique number, yielding the first large parametric families of quantum graphs for which standard graph parameters can be computed analytically [2603.23651].

These two lines address different aspects of adjacency. The confusability construction asks how much channel-output information can be retained before collapsing to the usual operator system, whereas the \((A,B,C)\)-model isolates how much of a quantum graph’s combinatorics is still governed by a weighted classical graph and how much remains genuinely nonclassical. This suggests that multiplicity in quantum multigraphs is often inseparable from extra structural data, not merely from repeated incidence.

## 4. Quantum symmetry and automorphisms of multigraphs

For classical directed or undirected multigraphs, quantum symmetry has been developed as an extension of the Banica and Bichon quantum automorphism-group formalisms. If \((V,E)\) is a multigraph with adjacency matrix
\[
W^i_j = |\{ \tau \in E : s(\tau)=i,\ t(\tau)=j \}|,
\]
the theory in “Quantum symmetry in multigraphs (part I)” introduces three categories: \(\mathcal C^{Ban}_{(V,E)}\), \(\mathcal C^{Bic}_{(V,E)}\), and \(\mathcal C^{sym}_{(V,E)}\). For single-edged graphs these reduce to the already existing notions of quantum symmetry. For general multigraphs, universal objects always exist in the Banica-type and Bichon-type categories, while for the symmetric category existence is generally open. A central theorem states that any multigraph with at least two pairs of vertices with multiple edges among them possesses genuine quantum symmetry [2302.08726].

The continuation “part II” constructs an explicit non-Bichon type co-action on a multigraph: it preserves quantum symmetry of \((V,E)\) in the new sense but not always in Bichon’s sense. The construction is motivated by automorphisms of quantum graphs and works on a fibered vertex algebra rather than only through quantum permutations of the edge set. This makes precise that, for multigraphs, operator-level compatibility can be strictly broader than edge-permutation symmetry [2403.00481].

Quantum Mycielskians provide a further symmetry-sensitive construction. In that framework, a quantum graph is given by \((V_G,\phi_G,A_G)\), and quantum multigraphs arise by allowing multiplicities or higher weights in \(A_G\). The quantum Mycielski construction lifts quantum automorphisms from \(G\) to \(\mathfrak p_{r-1}(G)\); if \(G\) has no quantum twin vertices, then all quantum automorphisms of the quantum Mycielskian fixing the master vertex arise from those of \(G\). The same paper introduces a quantum distinguishing number and proves that, if \(G\) has no quantum twins, then
\[
D(\mathfrak p(G)) \le D(G)+1.
\]
For classical graphs with \(|V|\le 6\), the quantum and classical notions of twin vertices coincide [2502.10521].

## 5. Multigraph states, multihypergraph states, and entanglement structure

In multipartite state theory, a quantum multigraph state is an entangled \(N\)-qudit state built from edge-dependent controlled-phase operations. If \(\widehat G=(V,\widehat E)\) is a multigraph and \(m_e \in \mathbb Z_d\) denotes the multiplicity or weight of edge \(e\), the state is
\[
\left| \widehat{G} \right\rangle =
\left( \prod_{e \in \widehat{E}} \widehat{CZ}_e^{m_e} \right)
|+_d\rangle^{\otimes N}.
\]
The edges can be parameterized by tuples \(s=(s_{v_0},s_{v_1})\), so distinct multiedges between the same vertices may implement distinct controlled-phase gates. Standard graph states are recovered when all exponents are \(1\). The same paper introduces multihypergraph states and proves a one-to-one correspondence between \(d\)-dimensional multihypergraph states and generalized real equally weighted states when \(d\) is prime; for composite \(d\), multihypergraph states form a subset of the generalized real equally weighted states [2312.14399].

This formalism should be distinguished from quantum hypergraph states in the earlier stabilizer literature. There, a \(k\)-uniform hypergraph state is generated by \(k\)-body controlled-\(Z\) gates, with graph states as the \(k=2\) special case. The same source explicitly contrasts multigraphs and hypergraphs: a multigraph allows multiple edges between pairs of vertices but remains pairwise, whereas a hypergraph allows general \(k\)-body interactions. It also proves that \(G_k\) and \(G_{k'}\) are inequivalent under local Pauli operations for \(k \ne k'\), and that the set of all real equally weighted states coincides with the set of general hypergraph states,
\[
G_{\le n} = G_\pm.
\]
Certain states used in Grover’s algorithm require genuinely \(n\)-body controlled-\(Z\) operations and therefore cannot be generated using only graph states [1211.5554].

Taken together, these results separate two distinct generalizations of graph states. Multigraph states enlarge the class of pairwise constructions by allowing multiple, parameterized edges between the same vertices; hypergraph states enlarge it by raising the interaction order from \(2\) to arbitrary \(k\).

## 6. Quantum walks, search, and multipartite exclusivity

Quantum-walk algorithms on multigraphs have been developed in both search and circuit-synthesis settings. For arbitrary bipartite multigraphs, adapted Szegedy quantum walks and adapted staggered quantum walks yield a search algorithm with quadratic speedup over classical Markov-chain search. The query complexity is \(\tilde O(\sqrt{HT})\), where \(HT\) is the hitting time of the underlying reversible Markov chain, and the same framework extends via line graphs to 2-tessellable graphs [2504.12586].

In coined discrete-time quantum walks, the shift operator can be read as a unitary form of the adjacency matrix, but the transformation generally changes the original graph into a directed multigraph by splitting single edges or arcs into multiple arcs. The shift operator is written as
\[
S = \sum_{i=0}^{m-1}\sum_{j=0}^{m-1} |c_i\rangle\langle c_j| \otimes \mathcal B^{\intercal}_{ij},
\]
and the generalized coin operator is
\[
\mathcal C = \sum_{k=0}^{n-1} C_k \otimes |v_k\rangle\langle v_k|.
\]
The paper states that any quantum circuit acting on a bipartite system is suitable to be the shift operator in a coined quantum walk, and any such circuit is therefore associated to a multigraph in this representation [2304.01582].

A different multigraph role appears in quantum non-locality. For an \(N\)-partite Bell inequality, an edge-coloured multigraph with \(N\) colour classes encodes which party is responsible for each exclusivity relation. The relevant invariant is the multigraph Lovász number
\[
\theta(\textsf G,w)=\max \sum_{i\in V} w_i \langle \psi | \Pi_i | \psi \rangle,
\]
where each projector factorizes across parties according to the coloured factors. It satisfies
\[
\theta(\textsf G,w) \le \vartheta(G,w),
\]
with \(G\) the simple exclusivity graph obtained by merging colours, and it yields tighter bounds than the ordinary Lovász number for inequalities such as \(I_3\), \(I_{3322}\), and the pentagon examples discussed in the paper [1407.5340].

## 7. Reconstruction, statistical mechanics, gravity, and positivity

Quantum multigraphs also appear as objects to be reconstructed or quantized in their own right. In graph reconstruction via operator algebras, the boundary crossed product
\[
A_X := C(\Lambda_X)\rtimes \Gamma_X
\]
associated to the universal covering of a multigraph determines only topological information: for graphs with first Betti number \(g_X>1\),
\[
K_0(A_X)=\mathbb Z^{g_X}\oplus \mathbb Z/(g_X-1).
\]
By contrast, after equipping the algebra with a quantum statistical mechanical time evolution based on the Busemann function, the resulting QSM system reconstructs finite multigraphs with minimal degree at least three up to graph isomorphism [1209.5783].

In causal quantum gravity, effective multigraph ensembles arise by collapsing each spatial slice at distance \(n\) in a causal triangulation to a single vertex and retaining the edges between levels \(n\) and \(n+1\). For these rooted multigraphs, the spectral and Hausdorff dimensions are related by
\[
d_s = \frac{2d_H}{2+\rho},
\]
where \(\rho\) is the exponent governing anomalous resistance to infinity. The equality \(d_s=d_H\) holds if and only if \(\rho=0\). The same framework introduces scale-dependent spectral dimension and exhibits flows from \(2\) to \(1\) in a Galton–Watson-based model and from \(4\) to \(2\) in a model motivated by four-dimensional CDT [1202.6322].

A further step is the direct quantization of finite multigraphs as microscopic degrees of freedom. In “Dynamical Quantum Multigraphs,” each unordered edge \(\{i,j\}\) carries its own finite-dimensional Hilbert space, and the total Hilbert space for labeled multigraphs is
\[
\mathcal H_{MG}=\bigotimes_{j>i=1}^N \mathcal H_{ij}.
\]
For labeled quantum simple graphs, the free theory is exactly the Erdős–Rényi–Gilbert \(G(N,p)\) model and has analytic free energy, hence no thermodynamic phase transition. For unlabeled quantum graphs, both the free and the ferromagnetic Ising models exhibit proper thermodynamic phase transitions, characterized by divergence in the specific heat and by an order parameter given as the fraction of vertices in the largest connected component [2509.08296].

The graph-density tradition adds an algebraic-positivity perspective. There, Positivstellensätze for quantum multigraphs characterize nonnegativity through approximation by preorders and sums of squares, but also show limitations of exact certification. In particular, Robinson-polynomial-type examples yield nonnegative quantum multigraphs that are not sums of squares, so approximation terms such as \(\varepsilon\) are generally unavoidable [1310.6903].

Source: https://www.emergentmind.com/topics/quantum-multigraphs