{"id":"6b1143de-ecc2-4a99-ba47-21c3a928c18f","arxiv_id":"2608.00340","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Introduces two extinction thresholds for Mukwembi's HIV graph automaton, proves bounds and the extremal characterization, and shows the thresholds can be arbitrarily far apart.","lead":"This paper studies a graph-based cellular automaton for HIV spread and defines two thresholds: the first replacement parameter that guarantees extinction, and the one beyond which extinction always happens. It proves general bounds, characterizes when the upper bound is reached, and shows the gap between the two thresholds can be arbitrarily large.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.11 is false as stated: for k=2, Table 2 claims extinction at R=3, but the specified initial state enters a 10-cycle, so hiv(F_2)≠3.","rationale":"The reader's ACCEPT rested on Theorem 2.11 and flagged Table 2 as the weakest point. Spot-checking that table reveals a concrete false entry, not merely a missing derivation. For k=2, the admissible state with one infected U-vertex and x,y,z healthy is latent at R=3 via a 10-cycle, contradicting the table's claimed t_ext=15 and the conclusion k+1∈E(F_k). Thus the stated theorem has a counterexample, and the central claim that the gap can be any prescribed integer is not established. Earlier bounds such as Theorem 2.10 appear sound, but the unbounded-gap construction is invalid as written. A corrected construction and an independently verified enumeration would be needed before acceptance.","tokens_in":17682,"tokens_out":42661,"duration_ms":346818,"concrete_test":"Run the automaton on F_2 with R=3 and initial state (x,y,z)=(0,0,0), u1=u2=0, u3=1. Verify that t6=t16 with configuration I={y}, D={u1,u2}, H={x,z,u3}, and that 0_G never appears. This reproduces the counterexample and settles that Table 2 and Theorem 2.11 are wrong for k=2.","verdict_should_be":"REJECT","load_bearing_attack":"In Theorem 2.11 take k=2. Then F_2 has U={u1,u2,u3}, Δ=5, and R=k+1=3. Table 2, block |U0|=k (first column), asserts that the admissible initial state with x,y,z healthy, u1,u2 healthy, and u3 infected has t_ext=15. Direct iteration of (1) from this state yields a 10-cycle beginning at t=6:\n\nt6: I={y}, D={u1,u2}, H={x,z,u3}\nt7: I={x,z,u3}, D={y}, H={u1,u2}\nt8: I={y,u1,u2}, D={x,z,u3}, H=∅\nt9: I={x}, D={y,u1,u2}, H={z,u3}\nt10: I={u3}, D={x}, H={y,z,u1,u2}\nt11: I={y}, D={u3}, H={x,z,u1,u2}\nt12: I={x,z,u1,u2}, D={y}, H={u3}\nt13: I={y,u3}, D={x,z,u1,u2}, H=∅\nt14: I={x}, D={y,u3}, H={z,u1,u2}\nt15: I={u1,u2}, D={x}, H={y,z,u3}\nt16 = t6.\n\nNo iterate is 0_G, so R=3 is latent for this admissible state. This directly contradicts Table 2 and the assertion k+1∈E(F_k). Consequently hiv(F_2)≠3 and Theorem 2.11 fails for k=2. The unboundedness claim might be salvageable through other graphs or for larger k, but the proof as written is not correct.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies Mukwembi's graph-based three-state cellular automaton for HIV dynamics. For a connected graph G it introduces two extinction parameters: hiv(G), the smallest R such that every admissible initial configuration reaches the all-healthy state, and HIV(G), the threshold beyond which every larger R guarantees extinction. The main results are: (i) 2 ≤ hiv(G) ≤ HIV(G) ≤ Δ(G)+1, with equality HIV(G)=Δ(G)+1 if and only if G is bipartite and Δ-regular (Theorem 2.10); (ii) the gap HIV(G)−hiv(G) is unbounded (Theorem 2.11, via the family F_k); (iii) the R-state transition digraph realizes cycles of every length (Theorem 3.3) and has a clean characterization of 2-cycles (Theorem 3.4); and (iv) closed formulas for cycles, complete graphs, and complete bipartite graphs, plus computational evidence for wheels. The proofs are largely first-principles combinatorial arguments. I also checked the circulated counterexample to Theorem 2.11 for k=2: it does not land. Direct iteration of the stated initial configuration with |U0|=2 and (x,y,z)=(H,H,H) reaches 0_G at t=15, exactly as Table 2 reports; the proposed 10-cycle mis-evaluates t14 because the dead vertex x has only two infected neighbors at t13, fewer than R=3, so it becomes healthy rather than infected.","tokens_in":18116,"tokens_out":37073,"duration_ms":294905,"significance":"If the results hold, the paper provides a genuine refinement of extinction thresholds for Mukwembi's automaton and shows that hiv and HIV are not redundant invariants. The extremal characterization in Theorem 2.10 is elegant, and the unbounded-gap construction in Theorem 2.11 is a strong structural result. The phase-space/digraph viewpoint in Section 3 is a useful framework for further work. Strengths of the manuscript include its first-principles proofs, the absence of fitted parameters, the explicit constructions, and the reproducibility of the numerical part through the provided code. I found no circularity or fitted quantities. The main presentation weakness is that the proof of Theorem 2.11 relies on a large finite-case table that is asserted without derivation; however, I spot-checked several entries and found them correct, and the alleged stress-test counterexample is invalid. With a small clarification of how Table 2 is obtained, the central claims should be accepted.","major_comments":[],"minor_comments":[{"comment":"The proof that k+1 ∈ E(F_k) is an unproved finite-case computation. The table is plausible and I verified several entries, including the k=2, |U0|=2, (x,y,z)=(H,H,H) case, but a short derivation of the block recurrences, or an explicit statement that the linked code verifies all 8(2k) cases, would remove the only non-transparent step in a headline theorem.","section":"Theorem 2.11, Table 2"},{"comment":"The sub-case tables near the end of the proof are also asserted without derivation. A sentence saying that each row follows by direct substitution into (1), and perhaps a short example, would help the reader check the exhaustiveness of the cases.","section":"Theorem 4.3"},{"comment":"There are several typos: 'is is latent' in the Introduction, 'stablish' in Section 1, 'conformed' for 'composed' in the Conclusions, and 'States sequence' in the Figure 1 caption. These should be corrected.","section":"General presentation"},{"comment":"The n=3 case refers to the proof of Theorem 2.11 but does not specify which choice of k and R is intended. Please state explicitly, e.g., k=2 and R=3 in F_2.","section":"Theorem 3.3"},{"comment":"The layout of Table 2 is dense; the z/y/x rows repeat across blocks and the column headers are easy to misread. A supplementary machine-readable table, or splitting the table into separate blocks per |U0|, would improve verifiability.","section":"Table 2 layout"},{"comment":"The extinction-time formula max_v dist(v,I0)+2 is stated after assuming D_t ∩ I_{t+1} = ∅ for all t. That assumption should be repeated in the sentence to avoid ambiguity.","section":"Paragraph after Proposition 2.5"}],"recommendation":"minor_revision","confidential_remarks":"The paper is mathematically sound in its main lines, and the stress-test counterexample is incorrect. My only substantive request is to make the finite-case verification in Theorem 2.11 more transparent; this is a local fix and does not affect the validity of the central claims. I recommend minor revision rather than acceptance as-is solely because of that presentation gap."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: the paper's headline claim about the unbounded gap is not just unproven, it's false as stated. I checked the k=2 case of Theorem 2.11. For F_2, take the admissible initial state with x,y,z healthy, u1,u2 healthy, and u3 infected, with R=3 (which is k+1). Table 2 lists extinction at t=15 for this state, but the transition rules (1) actually put the system into a 10-cycle starting at t=6 (I_{t+10}=I_t, etc.). So R=3 is latent, hiv(F_2)≠3, and the proof of Theorem 2.11 collapses. The unbounded-gap claim may still be true via another construction, but this paper doesn't show it.\n\nNow for what's actually good. The paper does introduce a useful distinction between hiv(G) and HIV(G), and the general bounds 2≤hiv≤HIV≤Δ+1 (Theorem 2.8) and the extremal characterization (Theorem 2.10) look solid. The proof for bipartite regular graphs is clean, and the converse argument for non-regular or non-bipartite graphs seems convincing. The R-state transition digraph section is genuinely interesting: cycles of arbitrary length exist, and the characterization of 2-cycles via a partition with cross-degree at least R is neat. The formulas for cycles and complete graphs are plausible, and the wheel data is honestly labeled as computational.\n\nSoft spots beyond the false theorem: the case tables in Theorem 4.3 are too terse; I couldn't verify them by eye, and they're load-bearing for the complete bipartite result. The conclusion also mentions graphs with extinction sets comprising several intervals, but that's just computational observation. The paper's own admission that the table in Theorem 2.11 is an enumeration with no derivation is a red flag, and the stress-test confirms the table is wrong.\n\nWho's this for? Readers interested in graph-theoretic models of epidemics and cellular automaton phase spaces. The valid results deserve a serious referee, and the false theorem needs to be caught. I wouldn't cite the unbounded-gap claim, but the rest might be usable after a rewrite. Send it to review, but expect a major revision.","headline":"The paper's unbounded-gap theorem is false as stated, but the bounds and digraph results justify a careful referee.","tokens_in":18617,"tokens_out":9886,"would_cite":false,"duration_ms":77521,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37B15","05C75","92D30"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that two natural extinction thresholds for a graph-based HIV infection automaton always satisfy 2 ≤ hiv(G) ≤ HIV(G) ≤ Δ(G)+1, and that the gap HIV(G)−hiv(G) can be arbitrarily large.","keywords":["HIV graph automaton","extinction threshold","latency set","cellular automaton","replacement parameter","bipartite regular graphs","state-transition digraph","non-monotone extinction"],"falsifier":"Simulate the deterministic rules on F_3 (vertices u1,...,u5, x, y, z; edges xy, yz, and x-u_i, y-u_i) with R=4, starting from each of the 2^8 admissible configurations, and check that every trajectory reaches the all-healthy state within the extinction times stated in Table 2; a single failure would disprove the unbounded-gap theorem.","tokens_in":17620,"feed_emoji":"🦠","tokens_out":6374,"duration_ms":51840,"temperature":0.7,"pith_summary":"The paper studies a synchronous cellular automaton on graphs proposed as a model of HIV dynamics in lymph-node networks: healthy cells become infected when they have an infected neighbor, infected cells die, and dead cells are replaced by infected or healthy cells depending on how many infected neighbors they have, measured against a replacement parameter R. It introduces two parameters for a graph G: hiv(G), the smallest R that guarantees extinction from every starting configuration, and HIV(G), the threshold beyond which every larger R guarantees extinction. The main results are that these parameters always lie between 2 and Δ(G)+1, that the upper bound is attained exactly for bipartite regular graphs, and that the difference between the two parameters is unbounded — one can build graphs where the gap is any prescribed integer. The paper also computes the parameters for cycles, complete graphs, and complete bipartite graphs, and studies the phase-space digraph of the automaton, showing it can contain cycles of any length. The upshot is that extinction in this model is controlled by graph structure, not just by the replacement parameter, and it is not a monotone property in R.","feed_headline":"The gap between two extinction thresholds can be any integer","feed_subtitle":"A graph-theoretic proof pins down both thresholds and shows their gap is arbitrarily large","key_machinery":"The two extinction parameters hiv(G) and HIV(G) — the first replacement value that forces extinction for every admissible initial configuration, and the smallest threshold after which every larger replacement value does so. The main mechanism is the four-stage classification of vertices (never-infected, infected, dead, cured): once R ≥ Δ(G)+1, a vertex can only move forward through these stages, never back. The characterization of the extremal case is carried by a partition/expansion argument showing that when G is not bipartite-regular, R=Δ(G) forces a cured vertex to appear and then propagate to the whole graph. The unbounded-gap construction F_k — a star-like graph with a long tail of deg","core_discovery":"The central claim is a complete structural characterization of when the naive upper bound is tight. For every connected graph with at least one edge, 2 ≤ hiv(G) ≤ HIV(G) ≤ Δ(G)+1; the equality HIV(G)=Δ(G)+1 holds precisely when G is bipartite and regular, in which case both parameters equal Δ(G)+1. Conversely, if G is not both bipartite and regular, then R=Δ(G) already guarantees extinction from every admissible initial state, pulling HIV(G) below the bound. A second main claim is that the two parameters are genuinely independent: for every integer k≥2 there is a connected graph F_k with HIV(F_k)−hiv(F_k)=k, so the gap can be made arbitrarily large. Along the way the paper proves that extinc","pith_inferences":["The two-threshold distinction is likely portable: any synchronous cellular automaton with a tunable reinfection parameter and a dead/replacement stage should exhibit analogous first-threshold vs eventual-threshold divergence whenever reinfection is needed for persistence.","If the paper's conjecture on wheel graphs is correct, it would give an infinite family where hiv(G)=3 or 4 but HIV(G)=n−1, making the gap essentially as large as possible while keeping the graph sparse; that would sharpen the unbounded-gap theorem from existence to a natural family.","The length-2 cycle characterization is equivalent to a bipartite-sparseness condition (a 'spanning R-dense cut'); asking for its algorithmic detection, or for analogues with longer periods, could connect to computational complexity questions the paper leaves open.","The isolated small values in E(W_n) (e.g., R=3 guarantees extinction for even wheels while intermediate R allow latency) suggest that low replacement values can accidentally extinguish infection by failing to reinfect enough dead cells, a non-monotonicity that may have epidemiological analogues."],"forward_implications":["If the graph is bipartite and regular, both extinction parameters coincide at Δ(G)+1; no finer threshold distinction exists for such graphs.","If the graph is not both bipartite and regular, extinction is already guaranteed at R=Δ(G), so the eventual threshold drops by at least one.","For any chosen integer gap k, some connected graph realizes HIV−hiv = k; the two thresholds are not coupled.","On cycles, complete graphs, and complete bipartite graphs, both parameters admit closed formulas: 3/2 for cycles (even/odd), ⌈(n+1)/2⌉ for K_n, and max{m+1, ⌊n/2⌋+1} for K_{m,n}.","The state-transition digraph can have cycles of every positive length, and a length-2 cycle exists exactly when V splits into two parts each sending at least R edges to the other."],"fun_headline_variants":["HIV extinction thresholds can be arbitrarily far apart","Bipartite regular graphs hit the upper bound in HIV model","HIV extinction gap takes every integer ≥2","HIV extinction gap proven unbounded","Graph structure determines HIV extinction thresholds"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The unbounded-gap construction relies on a finite case table (Table 2) that lists extinction times for every admissible initial configuration of the graph F_k at R=k+1; if any entry is wrong or the case split is not exhaustive, the proof that hiv(F_k)=k+1 collapses.","fun_headline_variants_meta":{"raw":{"variants":["HIV extinction thresholds can be arbitrarily far apart","Bipartite regular graphs hit the upper bound in HIV model","HIV extinction gap takes every integer ≥2","HIV extinction gap proven unbounded","Graph structure determines HIV extinction thresholds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000738,"raw_usage":{"total_tokens":3159,"prompt_tokens":794,"completion_tokens":2365,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":2299}},"tokens_in":538,"tokens_out":2365,"duration_ms":17895,"temperature":1.0,"reasoning_tokens":2299,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T00:41:45.207051+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate the deterministic rules on F_3 (vertices u1,...,u5, x, y, z; edges xy, yz, and x-u_i, y-u_i) with R=4, starting from each of the 2^8 admissible configurations, and check that every trajectory reaches the all-healthy state within the extinction times stated in Table 2; a single failure would disprove the unbounded-gap theorem.","supporting_citations":[],"review_version":1}