{"id":"d910a77a-8ec6-4c42-8a54-609f1021d654","arxiv_id":"1908.06628","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper claims a sufficient condition, lambda(β,c,α) > lambda_c, for strong survival of type 2 in the unequal-death-rate multitype contact process, but the proof's central comparison is not justified.","lead":"This paper gives a condition for one infection type to survive forever in a two-type contact process with unequal death rates. It is a step toward a long-standing conjecture, but the main proof appears to miss that the other type initially blocks the survivor's spread.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 4.1's 'unblocked' 2-arrows ignore the initially type-1 sites used in Theorem 2.2, so the comparison (4.5) and the main theorem are not established.","rationale":"The manuscript aims to prove a sufficient condition for strong survival of type 2 in the multitype contact process with unequal death rates. The mechanism is entirely comparison-based: the CPREE dominates the MCP in the type-2 set, and a standard supercritical contact process is embedded into the CPREE's usable 2-arrows. The load-bearing step is Proposition 4.1, whose proof claims that every contact-process arrow is also a traversable unblocked 2-arrow. The reader's objection is to the point: the proof's notion of 'blocked' records only whether a spontaneous type-1 birth from a 1-arrow below could have occupied the target before the 2-arrow. It does not record the fact that in the initial state of Theorem 2.2 every non-x site is already in state 1. Such a site cannot be changed to type 2 until its first type-1 death mark, since 1-to-2 is not a transition. Thus an unblocked 2-arrow into an initially type-1 site can be unusable. The proof tries to sidestep this by starting the background process in equilibrium and then saying that starting from all unblocked only helps; but the comparison needs a lower bound on the type-2 set, and all-unblocked is an upper bound. Starting from all-blocked is what the initial condition forces, and that process is not in the equilibrium required for Broman's lemma. The failure is not merely a matter of non-sharp constants: there is a positive-probability event in the proposed coupling in which the first standard arrow from an occupied site to an initially type-1 site occurs before that site's first type-1 death mark; then the standard contact process infects the site while the CPREE leaves it in state 1, violating {tilde xi = 1} subset {xi = 2} at that time. Since this inclusion is the bridge used in (4.5)-(4.6), Theorem 2.2 is not proved as written. The result may be true and repairable, for example by modifying the initial condition or by handling the transient blocking from initial type-1 sites, but the present proof does not establish it. I therefore recommend the same verdict as the reader: REJECT.","tokens_in":8928,"tokens_out":15372,"duration_ms":162753,"concrete_test":"Fix a finite path with sites 0 and 1. Let the CPREE/MCP start with site 0 in state 2 and site 1 in state 1, and let the standard contact process start with only site 0 infected. Let tau be the time of the first coupled point of the Broman process that gives an arrow from site 0 to site 1 in both processes. Condition on the positive-probability event that tau occurs before the first type-1 death mark on site 1, before the first death mark on site 0 or site 1, and before any 1-arrow on site 1's timeline. Then the 2-arrow at tau is 'unblocked' by the paper's definition, site 0 is still type 2, but site 1 is still initially type 1, so the CPREE state at site 1 remains 1. In the standard contact process, site 0 is still infected and site 1 is empty, so the arrow makes site 1 infected.","verdict_should_be":"REJECT","load_bearing_attack":"The proof of Proposition 4.1 defines a 2-arrow as blocked only when a 1-arrow lies below it on the target timeline with no type-1 death mark between. This definition does not register the initial state used in Theorem 2.2, where every site except x starts in state 1. A site in state 1 remains in state 1 until its first type-1 death mark, and the transition table allows only 0 to 2, so an 'unblocked' 2-arrow arriving before that first death mark cannot create type 2 at that site. The proof then starts the background processes in equilibrium and says that starting with all sites unblocked only helps the comparison. This is the wrong direction: the true usable-2-arrow process starts with every non-x site blocked by its initial type-1 infection, so starting all sites unblocked overcounts usable arrows and cannot be used to prove the inclusion {tilde xi_t = 1} subset {xi_t = 2}. Consequently the assertion that 'all arrows associated with Xtilde_t can be traversed by type 2 in the CPREE and never blocked from below by a pre-existing type 1 infection' is false: the pre-existing type-1 infection is exactly a blocker omitted from the definition of 'blocked'. Indeed, with positive probability a standard-contact-process arrow from an occupied site y to an initially type-1 site x occurs before x's first type-1 death mark; at that moment tilde xi_x = 1 while xi_x = 1, so the pathwise inclusion fails. The preservation of (4.5), and hence the chain of inequalities (4.6), is therefore not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a sufficient condition for strong survival of type 2 in the multitype contact process with unequal death rates, under parameters beta2 = c beta, delta2 = 1, beta1 = beta alpha, delta1 = alpha. The main result, Theorem 2.2, states that if the quantity lambda(beta,c,alpha) defined in (2.2) exceeds the contact process critical value lambda_c, then type 2 survives strongly, i.e. P^x(x in B_t infinitely often) > 0. The proof proceeds by comparing the multitype process to a CPREE-type process and then to a standard contact process through a coupling based on Broman's point-process domination lemma. The central step is Proposition 4.1, which asserts that every active path of the standard contact process is an active path for the CPREE through so-called unblocked 2-arrows.","tokens_in":9284,"tokens_out":6322,"duration_ms":67059,"significance":"If Theorem 2.2 were established, it would provide a nontrivial sufficient condition in the direction of the birth-to-death-ratio conjecture for the multitype contact process, and the use of Broman's coupling in this context would be an interesting contribution. The paper also gives useful consistency checks with the conjectured threshold and several remarks on extinction and translation-invariant initial states. However, the proof of the main theorem rests on a pathwise comparison whose crucial step is not justified: the definition of an unblocked 2-arrow ignores the initial type-1 occupation of the target site, which is exactly the initial condition used in Theorem 2.2. Because this gap is internal to the proof of Proposition 4.1 and is not repaired elsewhere, the main result is not established by the argument presented.","major_comments":[{"comment":"The definition of an unblocked 2-arrow ignores the initial state of the target site. A 2-arrow is called blocked only when a 1-arrow lies below it on the target timeline with no type-1 death mark between. In the application to Theorem 2.2, every site except x starts in state 1, and a type-1 site remains type 1 until its first type-1 death mark. Since the transition table only allows 0 -> 2, an unblocked 2-arrow arriving before that first death mark cannot create type 2. Therefore the assertion in the proof that 'all arrows associated with tilde X_t can be traversed by type 2 in the CPREE and never blocked from below by a pre-existing type 1 infection' is false: the initial type-1 infection is precisely a blocker omitted from the definition. The sentence 'starting with all sites unblocked only helps the comparison' is also in the wrong direction, because the true CPREE has fewer usable arrows than the coupled process with all sites unblocked, so an inclusion proved for the latter does not transfer to the former.","section":"Section 4, Proposition 4.1"},{"comment":"The inequality chain (4.5)-(4.6) depends entirely on the pathwise inclusion {tilde xi_t = 1} subset {xi_t = 2} for the CPREE starting with x in state 2 and all other sites in state 1. Since Proposition 4.1 does not establish this inclusion under that initial condition, the comparison P^x_cp(tilde xi_t(x)=1) <= P^x_cpree(xi_t(x)=2) <= P^x_mcp(eta_t(x)=2) is unsupported. The final assertion that 'whenever tilde xi_t(x)=1 ... eta_t(x)=2' and the consequent inequality (4.6) therefore do not follow. No alternative argument is supplied to repair this step.","section":"Section 4, proof of Theorem 2.2, (4.5)-(4.6)"}],"minor_comments":[{"comment":"In the sentence 'Since lambda <= lambda, Lemma 3.1 shows that there is a Poisson counting process...', the second lambda should be the function lambda(beta,c,alpha) from (2.2), not the birth rate of the standard contact process.","section":"Section 4, Proposition 4.1"},{"comment":"The statement that lambda is increasing in c and alpha is made without a proof; a short derivation or a reference to a supplementary calculation would improve readability.","section":"Section 2.2"},{"comment":"The line 'Solving lambda = 2/d >= lambda_c' is ambiguous: it should read 'Solving lambda(beta,c,alpha) = 2/d', since lambda is elsewhere a function.","section":"Section 4, after (4.6)"}],"recommendation":"reject","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Joe,\n\nThe main thing to know: the central comparison in Theorem 2.2 does not go through as written. Proposition 4.1's definition of an 'unblocked' 2-arrow only looks at 1-arrows below it on the timeline; it ignores the initial occupancy of the target site. In Theorem 2.2, every site except x starts in state 1. A 2-arrow arriving before that site's first type-1 death mark cannot create type 2, because the site isn't empty. The proof says these arrows 'can be traversed by type 2 ... and never blocked from below by a pre-existing type 1 infection,' but the initial state is exactly such an infection. So the pathwise inclusion {tilde ξ_t = 1} ⊂ {ξ_t = 2} can fail with positive probability, and the chain (4.6) is unsupported.\n\nWhat the paper does well: it is a genuine extension of Broman's point-process coupling to a CPREE-type model, and the sufficient condition (2.2) is a concrete, new functional of the parameters. The consistency check with the BDR conjecture (c > 1 required) is nice and correctly identifies that the result doesn't contradict the conjecture. The writing is unusually candid about the limitations: the author explicitly notes the condition is far from the conjectured threshold and requires extreme parameter values. That part is honest.\n\nThe flaw is not a minor technicality; it is load-bearing. The comparison (4.5) is the only mechanism transferring strong survival from the contact process to the MCP. If it fails, Theorem 2.2 has no proof. I would not be surprised if the argument could be repaired—one would need to either alter the definition of unblocked to account for the initial type-1 state, or start the comparison after all initial type-1 infections have had a chance to die, which would lose the finite-time coupling—but that repair is not in the manuscript.\n\nThere is also a secondary weakness: even if the gap were fixed, the sufficient condition is far from the BDR threshold, so the payoff is partial. That is not a flaw in itself, but it tempers the significance. This is a paper for specialists in interacting particle systems; they should read it knowing the main result is not established.\n\nIf this came to me as a referee, I would recommend rejection as written, with the suggestion that the authors fix Proposition 4.1. The paper deserves a serious referee, not a desk reject, because the error is subtle and the technique is potentially useful.","headline":"A clear but flawed extension of Broman's coupling: the proof of the main theorem ignores initial type-1 infections, so the comparison (4.5) and strong survival conclusion are not established.","tokens_in":9806,"tokens_out":6130,"would_cite":false,"duration_ms":58716,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K35"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves an explicit rate condition under which the stronger particle type survives strongly in the multitype contact process.","keywords":["multitype contact process","stochastic domination","strong survival","birth-to-death ratio","contact process in a randomly evolving environment","point process coupling","graphical construction","critical value"],"falsifier":"Take the graphical construction with site x initially type 2 and all other sites type 1. For each site, locate the first 2-arrow on its timeline that is unblocked in the sense of Proposition 4.1 and ask whether a type-1 death mark occurs on that timeline before the arrow. If such an arrow arrives before the first type-1 death, the target site is still type 1 at arrival, so type 2 cannot traverse it; exhibiting one such arrow on a positive-probability event would contradict the assertion that all unblocked arrows are traversable and would break the domination chain in (4.5).","tokens_in":8711,"feed_emoji":"🦠","tokens_out":7486,"duration_ms":73673,"temperature":0.7,"pith_summary":"The paper addresses a long-standing question in the multitype contact process: when two competing particle types have different death rates, does the type with the larger birth-to-death ratio drive the other to extinction? It proves a sufficient condition for the stronger type to survive, expressed through an explicit rate $\\lambda(\\beta,c,\\alpha)$ computed from the birth and death parameters. If this rate exceeds the critical value $\\lambda_c$ of the ordinary contact process, then from a single occupied site surrounded by the weaker type, the stronger type returns to that site infinitely often with positive probability. The argument couples the two-type process to a standard contact process through a randomly evolving environment, so that every infection path of the standard process is a type-2 path in the original system. The condition is consistent with the long-standing birth-to-death-ratio conjecture and gives a quantitative handle on the unequal-death-rate case.","feed_headline":"Survival threshold found for the multitype contact process","feed_subtitle":"The proof couples the two-type process to a single infection, matching the long-standing birth-to-death ratio conjecture.","key_machinery":"The argument turns on the unblocked 2-arrow counting process in the graphical construction. A 2-arrow from a neighbor into a site is called blocked if a 1-arrow below it on the same timeline has no type-1 death mark between them; blocked arrows are discarded. For each neighbor, the counting process of unblocked 2-arrows is a two-rate point process that is blocked with equilibrium probability $p=(1+2d\\beta)^{-1}$, has arrival rate $0$ while blocked and $c\\beta$ while unblocked, and flips between these states at rates $\\alpha$ and $2d\\beta\\alpha$. A point-process domination lemma [2] then guarantees a Poisson process of rate $\\lambda(\\beta,c,\\alpha)$ whose points can be embedded among the unblocked arrows. Those Poisson points are identified with the births of a standard contact process, yielding the comparison chain: multitype process $\\geq$ randomly-evolving-environment process $\\geq$ standard contact process.","core_discovery":"The central claim is Theorem 2.2: for the nearest-neighbor multitype contact process on $\\mathbb{Z}^d$ with rates $\\beta_2=c\\beta$, $\\delta_2=1$, $\\beta_1=\\beta\\alpha$, $\\delta_1=\\alpha$, and $\\beta>\\lambda_c$, define $$\\$\\lambda$(\\$\\beta$,c,\\$\\alpha$)=\\frac12\\left(c\\$\\beta$+\\$\\alpha$+2d\\$\\beta$\\$\\alpha$-\\sqrt{(c\\$\\beta$-\\$\\alpha$-2d\\$\\beta$\\$\\alpha$)^2+8d\\$\\alpha$ c\\$beta^{2}$}\\right).$$ If $\\lambda(\\beta,c,\\alpha)>\\lambda_c$, then type 2 survives strongly: starting with one site in state 2 and every other site in state 1, the starting site is in state 2 infinitely often with positive probability. The proof builds a chain of stochastic dominations from the multitype process to a contact process in a randomly evolving environment and then to a standard contact process with birth rate $\\lambda$, so that every active path of the standard process is an active path of type 2 in the original process. This gives the first sufficient condition of its kind for strong survival of the dominant type when the two types die at different rates.","pith_inferences":["The threshold is likely conservative: the proof ignores blocked 2-arrows entirely, so tracking when those arrows become usable after type-1 deaths may push the survival region toward smaller $c$, closer to the conjectured $c>1$ boundary.","Since the threshold depends on dimension through $\\lambda_c$, identical nominal rates could put the process in the survival regime in low dimensions and below it in high dimensions; a simulation sweep over dimension would test this mechanism directly.","The comparison treats type 1 as a spontaneously regenerating environment rather than as a true competitor; refining the randomly-evolving-environment step to preserve the actual type-1 dynamics might produce a sharper condition than (2.2)."],"forward_implications":["For parameters where the computed $\\lambda(\\beta,c,\\alpha)$ exceeds $\\lambda_c$, type 2 returns to its starting site infinitely often with positive probability even though all other sites begin in state 1.","Because $\\lambda$ increases in both $c$ and $\\alpha$, survival is guaranteed for large $c$ whenever $\\alpha>\\lambda_c$, and for large $\\alpha$ whenever $c\\beta>(1+2d\\beta)\\lambda_c$.","The condition forces $c>1$, so the theorem never contradicts the conjecture that the type with the larger birth-to-death ratio wins.","The randomly-evolving-environment process used in the proof also survives strongly whenever $\\lambda>\\lambda_c$.","When $c>\\alpha>1$, the weaker type dies out under translation-invariant initial states while type 2 survives, matching the conjecture in that parameter range."],"supporting_citations":[{"why":"Supplies the two-rate point-process domination lemma used to embed a Poisson process among unblocked 2-arrows.","marker":"[2]"},{"why":"Proves the equal-death-rate multitype contact process result and the survival notion this paper generalizes.","marker":"[7]"},{"why":"Gives the randomly-evolving-environment model and the comparison construction that Proposition 4.1 adapts.","marker":"[8]"},{"why":"Establishes monotonicity and attractiveness of multitype contact processes used in the domination chain.","marker":"[9]"},{"why":"Confirms attractiveness for the multitype contact process parameter range with unequal death rates.","marker":"[1]"},{"why":"Provides the standard contact process survival and critical-value facts used in Theorem 2.2.","marker":"[5]"},{"why":"Introduces the graphical construction that underlies all couplings and active-path arguments.","marker":"[3]"}],"fun_headline_variants":["Sufficient condition for dominant type survival","Coupling proves survival of stronger type","Dominant type survives under this condition","Stochastic comparison yields survival criterion","Unequal death rates: when dominant type wins"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that every arrow it counts as available to the stronger type can really be used, even though the starting configuration fills all surrounding sites with the weaker type, and a weaker-type site cannot switch to the stronger type until it first dies.","fun_headline_variants_meta":{"raw":{"variants":["Sufficient condition for dominant type survival","Coupling proves survival of stronger type","Dominant type survives under this condition","Stochastic comparison yields survival criterion","Unequal death rates: when dominant type wins"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00021,"raw_usage":{"total_tokens":1354,"prompt_tokens":831,"completion_tokens":523,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":447,"completion_tokens_details":{"reasoning_tokens":461}},"tokens_in":447,"tokens_out":523,"duration_ms":5694,"temperature":1.0,"reasoning_tokens":461,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:41:16.358111+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the graphical construction with site x initially type 2 and all other sites type 1. For each site, locate the first 2-arrow on its timeline that is unblocked in the sense of Proposition 4.1 and ask whether a type-1 death mark occurs on that timeline before the arrow. If such an arrow arrives before the first type-1 death, the target site is still type 1 at arrival, so type 2 cannot traverse it; exhibiting one such arrow on a positive-probability event would contradict the assertion that all unblocked arrows are traversable and would break the domination chain in (4.5).","supporting_citations":[{"cited_title":"I.: Stochastic domination for a hidden markov chain with applic a- tions to the contact process in a randomly evolving environm ent, Ann","cited_arxiv_id":null,"evidence_quote":"Supplies the two-rate point-process domination lemma used to embed a Poisson process among unblocked 2-arrows."},{"cited_title":"3, 467–506","cited_arxiv_id":null,"evidence_quote":"Proves the equal-death-rate multitype contact process result and the survival notion this paper generalizes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the randomly-evolving-environment model and the comparison construction that Proposition 4.1 adapts."},{"cited_title":"Attractive n-type contact processes","cited_arxiv_id":"1006.5723","evidence_quote":"Establishes monotonicity and attractiveness of multitype contact processes used in the domination chain."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Confirms attractiveness for the multitype contact process parameter range with unequal death rates."},{"cited_title":"MR- 1717346","cited_arxiv_id":null,"evidence_quote":"Provides the standard contact process survival and critical-value facts used in Theorem 2.2."},{"cited_title":"E.: Additive set-valued markov processes and graphical method s, Ann","cited_arxiv_id":null,"evidence_quote":"Introduces the graphical construction that underlies all couplings and active-path arguments."}],"review_version":1}