- The paper initially studies fractional revival (FR) on *oriented* graphs, establishing necessary and sufficient conditions for proper FR between strongly cospectral vertices and framing new spectral characteristics for FR.
- Author's confirm that no proper fractional revival (FR) exists on oriented Cayley graphs over finite abelian groups, emphasizing the skew-symmetric adjacency.
- The work complete characteristics of proper FR on oriented semi-Cayley graphs over abelian groups, detailing practical cases and providing constructive examples of graphs with explicit transfer probabilities.
- questions
- What are the practical implications of the bidirectional nature of FR in oriented graphs?
- How does the skew-symmetry of the adjacency matrix $A_\Gamma$ impact the feasibility of FR in oriented graphs?
- In what ways does the character table of a group influence the feasibility of FR in oriented Cayley and semi-Cayley graphs?
- For which types of groups beyond abelian groups, might the authors' results be extendable to study FR?
- Find recent papers about quantum transport phenomena on directed graphs.
Overview and contribution
This paper by Jiang, Liu, and Wang initiates the study of fractional revival (FR) on oriented graphs, a setting previously unexamined for this quantum transport phenomenon. Working with the real skew-symmetric adjacency matrix AΓ as Hamiltonian — so that the transition matrix is U(t)=exp(AΓt), which is real unitary — the authors establish three main results: (i) necessary and sufficient spectral conditions for proper FR between strongly cospectral vertices in arbitrary oriented graphs; (ii) a nonexistence theorem showing that oriented Cayley graphs over finite abelian groups admit no proper FR at all; and (iii) a full characterization of proper FR on oriented semi-Cayley graphs OSC(G,R,L,S) over abelian groups. The paper also extends the theory of strong cospectrality to the oriented setting, where eigenvalues are purely imaginary or zero and eigenvectors are generally complex.
Spectral characterization of FR between strongly cospectral vertices
The paper adapts strong cospectrality to oriented graphs: vertices u and v are strongly cospectral if for each eigenvalue λr there exists a phase factor eiπqr(u,v) with Eλreu=eiπqr(u,v)Eλrev, with phases constant across eigenspaces of equal eigenvalue. Strongly cospectral vertices are shown to be cospectral.
The central structural result is that FR in the oriented setting is inherently bidirectional: if (α,β)-FR occurs from u to U(t)=exp(AΓt)0 at time U(t)=exp(AΓt)1 between cospectral vertices, then U(t)=exp(AΓt)2-FR occurs from U(t)=exp(AΓt)3 to U(t)=exp(AΓt)4 at the same time. This follows from unitarity of the real transition matrix together with equality of diagonal entries. The proof forces the entire U(t)=exp(AΓt)5-column of U(t)=exp(AΓt)6 to be supported only on U(t)=exp(AΓt)7 — a rigidity that later drives both nonexistence results.
The main characterization states that U(t)=exp(AΓt)8-FR with U(t)=exp(AΓt)9 occurs between strongly cospectral OSC(G,R,L,S)0 at time OSC(G,R,L,S)1 iff for every eigenvalue OSC(G,R,L,S)2 in the support of OSC(G,R,L,S)3:
OSC(G,R,L,S)4
An immediate consequence is that all relevant phase factors must equal OSC(G,R,L,S)5, partitioning the support into sets OSC(G,R,L,S)6 and OSC(G,R,L,S)7. Because OSC(G,R,L,S)8 is skew-symmetric, eigenvalues come in conjugate pairs OSC(G,R,L,S)9, and the authors prove that u0 iff u1. This symmetry collapses the periodicity condition to a single congruence class: fixing u2, FR occurs exactly when u3 for all u4.
A notable feature of the oriented case is that FR can occur between more than one pair simultaneously — unlike undirected graphs, where Kay's uniqueness theorem restricts PST to a single partner per vertex. The paper does not pursue multi-pair FR systematically here, leaving it as an open question.
Nonexistence of proper FR on oriented abelian Cayley graphs
For u5 with u6 abelian of order u7 and u8, the character table u9 diagonalizes v0, and all vertices are cospectral: v1 depends only on v2. This translation invariance is fatal for FR. Suppose v3-FR occurs from v4 to v5. By translation invariance, FR also occurs from v6 to v7, while the bidirectionality theorem gives v8. If v9, then column λr0 has squared entries λr1, contradicting orthogonality of columns of λr2. If instead λr3, comparing entries yields λr4, forcing λr5.
This negative result is unconditional: no choice of connection set λr6 in any finite abelian group produces proper FR. The obstruction is purely structural — vertex-transitivity combined with skew-symmetric adjacency makes the two-sided FR pattern incompatible with unitarity. It contrasts sharply with the undirected case, where Wang et al. exhibited genuine families of abelian Cayley graphs admitting FR.
Characterization for oriented semi-Cayley graphs
The semi-Cayley construction λr7 has vertex set λr8 with intra-side arcs governed by λr9 and eiπqr(u,v)0 and cross-side arcs from side 0 to side 1 governed by eiπqr(u,v)1. Over an abelian eiπqr(u,v)2, the character table block-diagonalizes the adjacency matrix into eiπqr(u,v)3 blocks
eiπqr(u,v)4
yielding explicit eigenvalues eiπqr(u,v)5 and Kronecker-product eigenprojectors.
Three results structure the analysis:
- No FR within a side: proper FR from eiπqr(u,v)6 to eiπqr(u,v)7 is impossible, via the same three-column unitarity contradiction as in the Cayley case.
- Block form of the transition matrix: if FR occurs from eiπqr(u,v)8 to eiπqr(u,v)9, then necessarily
Eλreu=eiπqr(u,v)Eλrev0
for a permutation matrix Eλreu=eiπqr(u,v)Eλrev1 determined by Eλreu=eiπqr(u,v)Eλrev2 and an orthogonal matrix Eλreu=eiπqr(u,v)Eλrev3.
- Characterization: writing Eλreu=eiπqr(u,v)Eλrev4, FR from Eλreu=eiπqr(u,v)Eλrev5 to Eλreu=eiπqr(u,v)Eλrev6 requires Eλreu=eiπqr(u,v)Eλrev7 plus explicit exponential constraints on Eλreu=eiπqr(u,v)Eλrev8 involving Eλreu=eiπqr(u,v)Eλrev9 and (α,β)0. In terms of strong cospectrality, the conditions become: (a) no character annihilates (α,β)1; (b) (α,β)2 for all (α,β)3; (c) a phase-coherence condition tying equal eigenvalues to consistent values of (α,β)4. Given strong cospectrality, FR reduces to a partition (α,β)5 according to the sign of (α,β)6, with the periodicity conditions (α,β)7 and (α,β)8, and (α,β)9, u0 read off from u1. Note that condition (b), u2 for all u3, is derived rather than assumed, and it constrains the admissible pairs u4 considerably.
Two concrete constructions confirm feasibility. For u5 with u6, the graph admits u7-FR from u8 to u9 at time U(t)=exp(AΓt)00; here the transfer probability U(t)=exp(AΓt)01 approaches 1 as U(t)=exp(AΓt)02 grows, though never reaching it for finite U(t)=exp(AΓt)03. A second example with nonempty U(t)=exp(AΓt)04, U(t)=exp(AΓt)05 over U(t)=exp(AΓt)06 achieves U(t)=exp(AΓt)07-FR at time U(t)=exp(AΓt)08, i.e., transfer probability U(t)=exp(AΓt)09.
These examples show that the index-two-like doubling of the vertex set is what enables FR: the cross-side coupling through U(t)=exp(AΓt)10 breaks the translation-invariance obstruction that kills the Cayley case. The characterization is complete but conditional — it presumes FR endpoints lie on opposite sides and are strongly cospectral; FR between non-cospectral vertices is explicitly left open.
Limitations and open questions
The paper's results are confined to abelian underlying groups, where character-table diagonalization is available; the nonabelian case is not addressed. The general characterization (Theorem FRNS) applies only to strongly cospectral endpoint pairs, and the authors do not determine whether proper FR can occur without strong cospectrality in oriented graphs. Within the semi-Cayley analysis, the necessity direction of the FR characterization relies on the Fourier inversion argument applied entrywise, which presupposes the FR endpoints have the specific form U(t)=exp(AΓt)11; whether asymmetric choices of U(t)=exp(AΓt)12 beyond those satisfying U(t)=exp(AΓt)13 could still support FR is excluded only under strong cospectrality. Three problems are stated explicitly: existence of FR on oriented Cayley graphs over nonabelian groups; characterization of FR between non-cospectral vertices; and construction of oriented graphs admitting FR between every pair of vertices, the fractional analogue of universal perfect state transfer.
Conclusion
The paper supplies the first systematic treatment of fractional revival on oriented graphs. Its principal findings are a phase-based spectral criterion for FR between strongly cospectral vertices, a definitive impossibility result for oriented abelian Cayley graphs, and a workable, checkable characterization for oriented semi-Cayley graphs over abelian groups, supported by explicit cyclic constructions attaining transfer probabilities up to U(t)=exp(AΓt)14. The separation between the Cayley and semi-Cayley cases isolates vertex-transitivity as the decisive obstruction, and the stated open problems indicate where the theory must next be extended.