{"id":"70fa0da6-3602-4cd1-b89e-ca12577478d7","arxiv_id":"2508.18068","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Unitary Cayley graphs admit pretty good fractional revival and fractional revival exactly when the vertex count is 2 or twice a prime; quadratic unitary Cayley graphs admit pretty good fractional revival exactly for counts 2, 8, or twice a prime.","lead":"This paper gives complete lists of which two families of circulant graphs can transfer quantum states, or fragments of states, between pairs of vertices over time. Quantum network designers can read off from the number of vertices whether a graph supports perfect, pretty good, or fractional state transfer.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"All PGFR/FR classifications hinge on Theorem 2.6, an imported, unproved integer-relation criterion; a misstatement or misindexing there would collapse Theorems 3.8, 3.10, 6.13, and 7.3.","rationale":"The reader's weakest assumption correctly identifies the load-bearing role of Theorem 2.6: every PGFR and FR classification in the paper depends on an imported necessary-and-sufficient integer-relation criterion that is not proven or even fully stated in this manuscript. My reading of the application lemmas confirms that the arguments are essentially translations of eigenvalue data into the language of that criterion; there is no independent structural proof of the classifications. I did not find an internal contradiction that would force rejection, and the CRT labeling concern flagged by the reader appears to be a notational issue rather than a demonstrated error once the character-index convention from Theorem 2.12 is adopted. The periodic/integral assertion in Section 4 is also under-supported, but it is less load-bearing because the PGFR/FR results in Sections 6 and 7 do not depend on it. Therefore the honest verdict remains CONDITIONAL: the central claim is plausible and internally coherent, but it cannot be certified without a complete, correct proof of Theorem 2.6 and a check of the omitted lemma casework.","tokens_in":17304,"tokens_out":40078,"duration_ms":397864,"concrete_test":"Independently derive Theorem 2.6 from the proof in arXiv:2503.20367v2 and spot-check it on C_10 (i.e., G_10): compute the integer lattice {ell in Z^9 : sum_{r=1}^9 ell_r(lambda_r - lambda_0) = 0} using the true cycle eigenvalues 2cos(2pi r/10), and verify that no vector in this lattice has sum_{r odd} ell_r = +/-1. Also check that the eigenvalue values used in Lemma 6.3 for p=5 reproduce these true C_10 eigenvalues. If either check fails, the Section 6 classification needs rework.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The classifications in Sections 3 and 6, and hence the FR results in Sections 3 and 7, all reduce to Theorem 2.6, a necessary-and-sufficient criterion for PGFR on circulant graphs imported from the authors' own preprint [18] and stated without proof. The criterion is delicate: PGFR is claimed to hold exactly when no integer relation sum_{r=1}^{n-1} ell_r(lambda_r - lambda_0) = 0 has sum_{r odd} ell_r = +/-1. Every non-existence lemma (3.5, 3.6, 3.7, 6.4-6.12) and every existence lemma (3.4, 6.2, 6.3) is an application of this single tool. If the criterion is misstated, if the parity of r is tied to the eigenvector convention in Theorem 2.1 (where a_r has the same parity as r), or if the connection sets here do not satisfy a hidden hypothesis, then the classifications collapse. Several supporting lemmas are also dismissed with 'proof similar, omitted' (e.g., 3.7, 4.3, 4.5, 4.7, 4.9, 4.14, 6.12), so an independent reader cannot separate a wrong eigenvalue label from a wrong application. The periodic/integral step in Section 4 is a secondary support issue, since Sections 6-7 do not rely on it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies continuous-time quantum walks on unitary Cayley graphs X_n and quadratic unitary Cayley graphs G_n. It claims complete classifications for five phenomena: pretty good fractional revival and fractional revival on X_n, and periodicity, perfect state transfer, pretty good state transfer, pretty good fractional revival, and fractional revival on G_n. The main results are that X_n admits PGFR if and only if n is 2 or twice a prime, that X_n admits FR in exactly the same cases, that G_n admits PGFR if and only if n is in {2, 8, 2p} with p prime, and that G_n admits FR if and only if n is in {2, 4, 2p} with p ≡ 3 mod 4. The proofs reduce the problem to explicit eigenvalue formulas for the two graph families and to an integer-relation criterion for PGFR on circulant graphs.","tokens_in":17550,"tokens_out":65017,"duration_ms":574312,"significance":"If the classifications are correct, the paper gives complete characterizations of several quantum-walk phenomena on two natural circulant families, and the result that FR and PGFR coincide for X_n is a clean and useful observation. The strategy of reducing PGFR to an integer-relation condition and FR to a rational-ratio condition is appropriate, and the spectral inputs from Klotz-Sander and Huang are made explicit. The claims are concrete and falsifiable in individual cases. However, the proof is not yet reliable: the central PGFR criterion is imported without proof, and Section 6 contains a concrete CRT-dependent eigenvalue error that invalidates a key non-existence lemma.","major_comments":[{"comment":"Theorem 2.6 is the sole criterion used to prove every PGFR existence and non-existence statement in Sections 3 and 6, yet it is imported without proof from the authors' own preprint [18] and is not restated with full hypotheses. The condition is delicate: it depends on the parity of the eigenvector index r as ordered in Theorem 2.1, and a misstatement or misapplication would invalidate Theorems 3.8, 3.10, 6.13, and 7.3. The authors should provide a complete proof of Theorem 2.6, or cite a published version, and verify explicitly that the connection sets considered satisfy its hypotheses.","section":"§2, Theorem 2.6"},{"comment":"The eigenvalue assignment λ_{3p} = λ_{(0,3)} = (1−p)/2 is incorrect. The CRT identification Z_{8p} ≅ Z_p × Z_8 sends r = 3p to (0, 3p mod 8), which is (0,1) for p ≡ 3 mod 8 and (0,5) for p ≡ 7 mod 8, never (0,3). For the concrete case p = 3, G_24 is the 24-cycle (all unit squares modulo 24 are equal to 1), and its eigenvalue at r = 9 is 2cos(3π/4) = −√2, not (1−p)/2 = −1. Therefore the integer relation displayed in Lemma 6.6 is not a relation among the true eigenvalues, and the exclusion of n = 8p in Theorem 6.13 is not established by the given proof. Lemma 6.7 is affected by the same CRT mislabeling.","section":"§6, Lemma 6.6"},{"comment":"Several lemmas that are essential to the final classifications are dismissed with \"proof similar\" and no details: Lemmas 3.7, 4.3, 4.5, 4.7, 4.9, 4.14, and 6.12. Given the CRT-indexing errors in Lemmas 6.4–6.7, these omissions are not merely stylistic: the reader cannot separate an incorrect eigenvalue label from an incorrect application of Theorem 2.6. Full proofs, or at least detailed derivations of the displayed eigenvalues, should be provided for every non-existence lemma used in Theorems 6.13 and 7.3.","section":"§§3, 4, 6"}],"minor_comments":[{"comment":"The abstract contains an incomplete sentence: \"we classify all X_n admitting pretty good fractional\" should read \"pretty good fractional revival.\"","section":"Abstract"},{"comment":"In the proof of Lemma 5.5, the displayed phase should be exp(−i(t_k λ_r + πr)), which is independent of the starting vertex a; as written with πar, the contradiction obtained at r = 3 and r = 2^{h−2} is not valid for general a.","section":"§5, Lemma 5.5"},{"comment":"Corollary 4.15.1 repeats Theorem 4.15 verbatim in the text; if the corollary is intended to assert periodicity rather than integrality, the distinction should be stated explicitly and the proof of the corollary should be supplied.","section":"§4, Corollary 4.15.1"},{"comment":"There are minor typographical errors: in Lemma 4.1 the index should be a = 2^{h−3} rather than 2h−3, and in Lemma 6.3 the phrase \"us write\" should be \"we write.\"","section":"§4, Lemma 4.1; §6, Lemma 6.3"}],"recommendation":"major_revision","confidential_remarks":"The paper depends on the authors' own unpublished preprint [18] for the central PGFR criterion, and the false eigenvalue in Lemma 6.6 is a substantive error in a load-bearing part of the proof. The final classifications may still be correct, but the submitted proof is not reliable in Section 6, and the omitted 'similar' proofs should be supplied or verified before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a standard classification paper done competently. It takes known state-transfer criteria and runs them on the unitary Cayley graphs X_n and the quadratic unitary Cayley graphs G_n. The results are new, the classifications look right, and I checked enough of the eigenvalue computations to believe the casework.\n\nWhat's actually new: for X_n, they classify PGFR and FR — both happen exactly for n = 2 or n = 2p, so FR iff PGFR. For G_n, they classify periodicity (n = 2, 4, p^s, 2p^s with p ≡ 3 mod 4), PST (n = 2, 4), PGST (n = 2, 4, 8), PGFR (n = 2, 8, 2p), and FR (n = 2, 4, 2p with p ≡ 3 mod 4). The FR classifications are the most interesting, and the paper points out that there are infinite families where PGFR occurs but FR does not. That is a genuinely useful piece of landscape.\n\nSoft spots, in proportion. The big one is Theorem 2.6, imported without proof from the authors' own preprint. It is the necessary-and-sufficient criterion for PGFR on circulant graphs, and every non-existence lemma in Sections 3 and 6 is an application of it. If that theorem were misstated or inapplicable, the classifications for both families would collapse. That is a real dependency, but it is not a circularity or a free parameter — the theorem is stated cleanly and the applications are direct. Still, a referee would be right to ask for a proof or a published reference before accepting the paper. The second issue is the number of 'proof similar, omitted' lemmas (3.7, 4.3, 4.5, 4.7, 4.9, 4.14, 6.12). In a classification paper this is acceptable when the similarity is obvious; here it is mostly obvious, but there are enough of them that an independent reader cannot fully verify the casework without redoing it. The stress-test note worried about sign ambiguities in the G_2p eigenvalue lists; I did not find a concrete error, but the character labeling is delicate and deserves a careful look.\n\nWho this is for: algebraic graph theorists working on continuous-time quantum walks. It is a within-subfield advance, not a reorganization of anything, but it is solid. I would send it to peer review and expect it to be published after the authors either prove Theorem 2.6 or cite a published version, and after the omitted proofs are filled in. I would cite it if I were working on these families myself.","headline":"A clean, correct-in-outline classification of state transfer on two circulant families; the main caveat is how much weight rests on an unproved prior theorem and on omitted 'similar' proofs.","tokens_in":18127,"tokens_out":9616,"would_cite":true,"duration_ms":84726,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11A07","15A16","05C50","81P45"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper fully classifies fractional revival and pretty good fractional revival on unitary Cayley graphs and on quadratic unitary Cayley graphs, and also classifies periodicity, perfect state transfer, and pretty good state transfer on…","keywords":["unitary Cayley graph","quadratic unitary Cayley graph","periodicity","perfect state transfer","pretty good state transfer","fractional revival","pretty good fractional revival","circulant graph"],"falsifier":"For a predicted negative case, take $n=12$ on the unitary Cayley graph side. The eigenvalues $\\lambda_1=0$, $\\lambda_2=2$, $\\lambda_8=-2$ satisfy the forbidden integer relation of Theorem 2.6 with $\\sum_{r\\text{ odd}}\\ell_r=1$, so the theorem predicts that no sequence of times $t_k$ can make $H(t_k)e_0$ approach $\\alpha e_0+\\beta e_6$. A direct numerical search of the transition matrix over large times that finds such an approach would refute the classification; the same check can be repeated for $G_{16}$, which is predicted to have no pretty good fractional revival.","tokens_in":17045,"feed_emoji":"⚛️","tokens_out":16793,"duration_ms":147905,"temperature":0.7,"pith_summary":"The paper asks exactly which graphs in two families of circulant graphs — unitary Cayley graphs $X_n$ and quadratic unitary Cayley graphs $G_n$ — support fractional revival and its approximation, pretty good fractional revival, during a continuous-time quantum walk. For $X_n$ it proves a dichotomy: fractional revival occurs precisely when $n=2$ or $n=2p$ for a prime $p$, and pretty good fractional revival occurs for the same set, so the two notions coincide throughout the family. For $G_n$ they diverge: pretty good fractional revival occurs exactly for $n \\in \\{2,8,2p\\}$ with $p$ prime, while fractional revival occurs exactly for $n \\in \\{2,4,2p\\}$ with $p$ an odd prime congruent to $3 \\pmod{4}$. The paper also classifies periodicity ($n \\in \\{2,4,p^s,2p^s\\}$ with $p \\equiv 3 \\pmod{4}$), perfect state transfer ($n=2,4$), and pretty good state transfer ($n=2,4,8$) on $G_n$. These exact lists settle, for two large families of qubit networks, which can transmit a state exactly, which only approximately, and which can entangle two vertices through fractional revival.","feed_headline":"Unitary Cayley graphs revive only for n=2 or n=2p","feed_subtitle":"The quadratic side is settled: exact revival only for n=2,4,2p with p=3 mod 4; near-revival for n=2,8,2p.","key_machinery":"The machinery has four parts. The spectral decomposition of a circulant graph $\\mathrm{Cay}(\\mathbb{Z}_n,S)$ expresses the transition matrix $H(t)=\\exp(-itA)$ as a sum over Fourier eigenprojectors indexed by the eigenvalues $\\lambda_r$. For $X_n$ these eigenvalues are given by $\\lambda_r=\\mu(c(r,n))\\,\\varphi(n)/\\varphi(c(r,n))$ with $c(r,n)=n/\\gcd(r,n)$; for $G_n$ they are products of quadratic Gaussian sums attached to the prime-power factors of $n$, obtained by identifying $\\mathbb{Z}_n$ with a product of the $\\mathbb{Z}_{p_j^{k_j}}$ through the Chinese remainder theorem. Two structural criteria carry the argument. Theorem 2.5 decides fractional revival from the rationality of ratios of eigenvalue differences. Theorem 2.6 decides pretty good fractional revival from the absence of an integer relation $\\sum_{r=1}^{n-1}\\ell_r(\\lambda_r-\\lambda_0)=0$ whose odd-index coefficients sum to $\\pm1$. The classifications reduce to checking these two criteria case by case through the prime factorization of $n$.","core_discovery":"The central discovery is a set of exact classifications. A unitary Cayley graph $X_n$ — the circulant graph on $\\mathbb{Z}_n$ whose edges connect numbers with a unit difference — admits fractional revival exactly when $n=2$ or $n=2p$ for a prime $p$, and admits pretty good fractional revival for exactly the same values of $n$; the two phenomena therefore coincide for the whole family. The quadratic unitary Cayley graph $G_n$, whose edges connect numbers whose difference is a quadratic residue or the negative of one modulo $n$, splits the two notions: it admits pretty good fractional revival exactly for $n \\in \\{2,8,2p\\}$ with $p$ prime, and fractional revival exactly for $n \\in \\{2,4,2p\\}$ with $p$ an odd prime satisfying $p \\equiv 3 \\pmod{4}$. Along the way the paper proves that $G_n$ is periodic exactly for $n \\in \\{2,4,p^s,2p^s\\}$ with $p \\equiv 3 \\pmod{4}$, that it has perfect state transfer only for $n=2,4$, and that it has pretty good state transfer only for $n=2,4,8$. In particular there are infinitely many $G_n$ that approximate revival but never achieve it, and infinitely many that achieve fractional revival without perfect state transfer.","pith_inferences":["The pattern suggests a testable conjecture for $k$-th power unitary Cayley graphs: admissible $n$ will again be controlled by one odd prime factor and by whether $-1$ lies in the subgroup of $k$-th powers modulo that prime.","Since the pretty good fractional revival criterion is imported as a cited theorem rather than proved in this paper, an independent proof of that criterion would place the two PGFR classifications on a fully self-contained footing.","The contrast between the unitary family, where fractional revival and pretty good fractional revival coincide, and the quadratic family, where they diverge, indicates that quadratic-residue structure rather than integrality alone is what opens the gap between approximate and exact revival."],"forward_implications":["For unitary Cayley graphs, the paper fixes the exact list: $X_n$ has fractional revival, and equivalently pretty good fractional revival, only for $n=2$ or $n=2p$ with $p$ prime.","The quadratic family gives infinitely many concrete networks, $G_{2p}$ with $p \\equiv 3 \\pmod{4}$, where two vertices can be entangled by fractional revival even though no perfect state transfer occurs.","It also gives infinitely many networks, $G_{2p}$ with $p \\equiv 1 \\pmod{4}$, that achieve pretty good fractional revival but neither exact fractional revival nor pretty good state transfer; the approximability hierarchy is therefore strict and populated.","Periodicity, perfect state transfer, and pretty good state transfer on $G_n$ are pinned down for every $n$, so the five phenomena are now closed questions for both families."],"supporting_citations":[{"why":"Supplies Theorem 2.6, the necessary-and-sufficient integer-relation criterion for pretty good fractional revival on circulant graphs; the FR and PGFR classifications for both families rest on it.","marker":"[18]"},{"why":"Supplies Theorem 2.5, the criterion for fractional revival on circulant graphs used to prove existence of FR for the admissible vertex counts.","marker":"[27]"},{"why":"Provides the quadratic Gaussian-sum eigenvalue formulas (Theorems 2.11 and 2.12) used throughout the quadratic unitary Cayley graph sections.","marker":"[17]"},{"why":"Supplies the eigenvalue formula for unitary Cayley graphs used to compute the eigenvalue differences in Section 3.","marker":"[19]"},{"why":"Supplies the necessary condition for pretty good state transfer and a useful criterion for periodic Cayley graphs that narrow the candidates for $G_n$.","marker":"[20]"},{"why":"Supplies the sufficient condition (Theorem 5.3) used to rule out pretty good state transfer when n has an odd prime factor.","marker":"[21]"},{"why":"Supplies the fractional revival classification for cycles (Theorem 2.4), used to exclude the eight-vertex cycle from the fractional revival list.","marker":"[10]"},{"why":"Supplies the necessary condition for perfect state transfer on integral abelian Cayley graphs (Theorem 2.2), used to exclude the $2p^s$ cases.","marker":"[25]"}],"fun_headline_variants":["Unitary Cayley graphs: fractional revival only for n=2 or 2p","Quadratic Cayley: pretty good fractional revival for n=2,8,2p","Fractional and pretty good fractional revival coincide on X_n","Perfect state transfer on quadratic Cayley graphs only for n=2,4","Revival classification complete for unitary and quadratic Cayley graphs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole classification stands on the borrowed criterion that a circulant graph has pretty good fractional revival exactly when no integer combination of its eigenvalue gaps has the forbidden parity sum; if that criterion is false, all the listed vertex counts would have to be rechecked.","fun_headline_variants_meta":{"raw":{"variants":["Unitary Cayley graphs: fractional revival only for n=2 or 2p","Quadratic Cayley: pretty good fractional revival for n=2,8,2p","Fractional and pretty good fractional revival coincide on X_n","Perfect state transfer on quadratic Cayley graphs only for n=2,4","Revival classification complete for unitary and quadratic Cayley graphs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002112,"raw_usage":{"total_tokens":8278,"prompt_tokens":1091,"completion_tokens":7187,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":707,"completion_tokens_details":{"reasoning_tokens":7089}},"tokens_in":707,"tokens_out":7187,"duration_ms":51466,"temperature":1.0,"reasoning_tokens":7089,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:02:18.380641+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a predicted negative case, take $n=12$ on the unitary Cayley graph side. The eigenvalues $\\lambda_1=0$, $\\lambda_2=2$, $\\lambda_8=-2$ satisfy the forbidden integer relation of Theorem 2.6 with $\\sum_{r\\text{ odd}}\\ell_r=1$, so the theorem predicts that no sequence of times $t_k$ can make $H(t_k)e_0$ approach $\\alpha e_0+\\beta e_6$. A direct numerical search of the transition matrix over large times that finds such an approach would refute the classification; the same check can be repeated for $G_{16}$, which is predicted to have no pretty good fractional revival.","supporting_citations":[{"cited_title":"Pretty good fractional revival on abelian Cayley graphs","cited_arxiv_id":"2503.20367","evidence_quote":"Supplies Theorem 2.6, the necessary-and-sufficient integer-relation criterion for pretty good fractional revival on circulant graphs; the FR and PGFR classifications for both families rest on it."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Theorem 2.5, the criterion for fractional revival on circulant graphs used to prove existence of FR for the admissible vertex counts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the quadratic Gaussian-sum eigenvalue formulas (Theorems 2.11 and 2.12) used throughout the quadratic unitary Cayley graph sections."},{"cited_title":"Klotz and T","cited_arxiv_id":null,"evidence_quote":"Supplies the eigenvalue formula for unitary Cayley graphs used to compute the eigenvalue differences in Section 3."},{"cited_title":"Pal and B","cited_arxiv_id":null,"evidence_quote":"Supplies the necessary condition for pretty good state transfer and a useful criterion for periodic Cayley graphs that narrow the candidates for $G_n$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the sufficient condition (Theorem 5.3) used to rule out pretty good state transfer when n has an odd prime factor."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the fractional revival classification for cycles (Theorem 2.4), used to exclude the eight-vertex cycle from the fractional revival list."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the necessary condition for perfect state transfer on integral abelian Cayley graphs (Theorem 2.2), used to exclude the $2p^s$ cases."}],"review_version":2}