diff --git a/notes/venue-retrieval-p5-20260909.md b/notes/venue-retrieval-p5-20260909.md new file mode 100644 index 00000000..a215757f --- /dev/null +++ b/notes/venue-retrieval-p5-20260909.md @@ -0,0 +1,293 @@ +# Venue retrieval — P5 / AP-2026-0311 after the JAP/AAP desk reject (issue #697) + +Date: 2026-09-09. Ticket: #697 (parent #650). Manuscript object: PR #688 draft +`docs/manuscripts/p5-exact-continuation-bounded-state/README.md` (read from local branch +`pr688`, **not merged**, not checked out as this branch) and the submitted proof PDF +AP-2026-0311, *"Homological Occupation Growth: Branching Tests and Cut-Network +Representations"* (Ying/Yu, submitted 2026-09-03, 17 pp., desk-rejected 2026-09-09 by +JAP/AAP with *"not in a suitable scope … insufficient methodological novelty from an +applied probability view"*). + +**Retrieval and notes only.** No STATUS/ROADMAP edits, no `references.bib` edits, no +gadget/rank computation, no Huawei. The AP proof PDF was read locally via `pdftotext` +and is **not** committed here. + +## 0. What the two objects are + +The submitted 17-page paper and the #688 draft are **one paper object**, but they are not +identical in coverage (this matters for the split question, §3): + +| Content | AP-2026-0311 (submitted) | P5 #688 draft | +|---|---|---| +| Cut-network representation + converse realization (torus, rank-one) | Thms 4.1, 4.4, 5.1, 5.2, 6.1 (submitted core) | Part I, Thms A–D | +| Depth-as-resource / no-bounded-radius / positive-vs-signed rank | not in submission | Thms B, C, D | +| Kalman finite-horizon no-go (`p_c = 1/2` vs `1/3`, identical responses) | not in submission | Part II, Thm E | +| P398 reflection quotient (w=4..8 instance) | not in submission | Part III, Thms F–H | +| 4×4 torus witness `(1,7/8,9/14,5/14,4/35,…)` + fork gap 1/98 | §§3, 6 (Thms 3.1, 3.3, 6.1) | Part I opening fact | + +Everything below treats them as one object; §3's split recommendation never implies two +overlapping submissions. + +## 1. Sources fetched, with marking + +| Source | URL / location | Marking | +|---|---|---| +| JAP about page (Cambridge, Applied Probability Trust) | cambridge.org/core/journals/journal-of-applied-probability | **PRIMARY_TEXT_READ** | +| JAP Vol 63 Issue 3 (Sept 2026) table of contents | cambridge.org/core/journals/journal-of-applied-probability/issue/E52E939B7D9921109FE29020814EB70D | **ABSTRACT_ONLY** (titles + one-line summaries; full texts not read) | +| AAP about page (Cambridge) | cambridge.org/core/journals/advances-in-applied-probability | **ABSTRACT_ONLY** (page fetched; contains no scope text — AAP scope is stated on the JAP page as the joint-submission companion) | +| AAP "Phase Transitions" collection preface (June 2025) | cambridge.org/core/journals/advances-in-applied-probability/collections/june-2025-phase-transitions | **PRIMARY_TEXT_READ** (preface only) | +| AAP "Branching Processes" collection (Aug 2025) article list | …/collections/august-2025-branching-processes | **ABSTRACT_ONLY** | +| CPC about page (Cambridge) | cambridge.org/core/journals/combinatorics-probability-and-computing | **PRIMARY_TEXT_READ** (scope quoted verbatim below) | +| EJP journal page (IMS) | imstat.org/journals-and-publications/electronic-journal-of-probability/ | **PRIMARY_TEXT_READ** (scope statement quoted below) | +| ALEA homepage + Vol XXIII TOC | alea.impa.br | **ABSTRACT_ONLY** (no scope statement on page) | +| EJC homepage + Vol 33 Issue 3 TOC | combinatorics.org | **ABSTRACT_ONLY** (issue TOC read; scope statement not on the page) | +| *Linear Algebra and its Applications* aims & scope (Elsevier) | sciencedirect.com/journal/linear-algebra-and-its-applications/about/aims-and-scope | **PRIMARY_TEXT_READ** | +| Random Structures & Algorithms (Wiley) | onlinelibrary.wiley.com/journal/10982418 | **NOT_FETCHED** (two attempts returned navigation chrome only; no scope text). Tripwire is satisfied by CPC + EJP. | +| J. Phys. A scope page (IOP) | iopscience.iop.org/journal/1751-8121/page/Scope | **NOT_FETCHED** (404) | +| Callan–Smiley, *Noncrossing partitions under rotation and reflection*, arXiv:math/0510447 (2005) | arxiv.org/abs/math/0510447 | **ABSTRACT_ONLY** (abstract read; confirms reflection-invariant NC partition count via the Kreweras involution; Thm 1 text not read in this pass) | +| D'Angeli–Donno, *The lumpability property for a family of Markov chains on poset block structures*, Adv. Appl. Math. 51(3):367–391 (2013), arXiv:1304.4180 | arxiv.org/abs/1304.4180 | **ABSTRACT_ONLY** | +| AP-2026-0311 proof PDF (17 pp.) | local, `/Users/lc/Downloads/AP-2026-0311_Proof_hi (1).pdf`, read-only | **PRIMARY_TEXT_READ** (full text extracted; not committed) | +| P5 README on `pr688` | `git show pr688:docs/manuscripts/p5-exact-continuation-bounded-state/README.md` | **PRIMARY_TEXT_READ** | +| Kemeny–Snell *Finite Markov Chains*; Wonham (Kalman decomposition); Schur/Wigner–Eckart; Ding 2016 (Miami PhD); Yannakakis / Fiorini et al. (slack-matrix nonnegative rank); Balle–Panangaden–Precup LICS 2015; Larsen–Skou 1991; Deng–Feng 2017 | — | **[LIT]** — cited from the AP PDF's own reference list and the #688 draft; primaries not fetched in this pass | + +Tripwire check: JAP scope fetched; CPC and EJP scope fetched. All three quoted below. + +## 2. Q1 — Was the desk reject predictable from the journals' own scope? + +**Fetched scope text (JAP, verbatim, Cambridge Core):** + +> "With a publication record spanning more than five decades, the *Journal of Applied +> Probability* is the oldest journal devoted to the publication of research in the field +> of applied probability. It is an international journal published by the Applied +> Probability Trust, and it serves as a companion publication to the *Advances in +> Applied Probability*. Its wide audience includes leading researchers across the entire +> spectrum of applied probability, including biosciences applications, operations +> research, telecommunications, computer science, engineering, epidemiology, financial +> mathematics, the physical and social sciences, and any field where stochastic modeling +> is used. +> A submission to Applied Probability represents a submission that may, at the +> Editor-in-Chief's discretion, appear in either the *Journal of Applied Probability* or +> the *Advances in Applied Probability*. Typically, shorter papers appear in the +> *Journal*, with longer contributions appearing in the *Advances*." + +Note the scope statement is *application-domain* broad, not *method* broad. Nothing in it +excludes reliability, graphs on tori, or finite Markov chains — and indeed the AP PDF +itself cites two JAP signature papers (Boland, JAP **38** (2001) 597–603; +Navarro–Samaniego–Balakrishnan, JAP **47** (2010) 235–253) — so "reliability is foreign +here" would be the wrong reading of the reject. + +**The journal's actual diet.** From the fetched JAP 63(3) (Sept 2026) TOC — 20 articles — +the percolation / branching / Markovian-network items are (all **ABSTRACT_ONLY**): + +1. M. Marivain, *"Brochette first-passage percolation"* — a new FPC model; time constant + and shape theorem. +2. D. Bertacchi, F. Zucca, *"Strong survival and extinction for branching random walks + via new order for generating functions"* — classification theorem for multitype + branching random walks. +3. B. Lodewijks, *"A star is born: Explosive Crump–Mode–Jagers branching processes"* — + explosion criterion, applied to preferential attachment. +4. D. Goldsztajn, K. Avrachenkov, *"Asymptotically optimal policies for weakly coupled + Markov decision processes"* — fluid limits, asymptotic optimality. +5. Ü. Işlak, B. Yeşiloğlu, *"Analysis of clustering and degree index in random graphs + and complex networks"* — limit theorems for network indices in ER/BA/WS models. + +The AAP side shows the same shape: the fetched June-2025 *Phase Transitions* collection +preface frames the journal's interest as "bulk properties … undergo an abrupt change as a +parameter of the system moves across a critical value" in "probabilistic models of +statistical physics … percolation and spin systems … stochastic models of networks, +branching and replication, epidemics, algorithms". + +**Comparison verdict.** Every sampled analogue is an *asymptotic* paper: a shape theorem, +a survival/extinction classification, an explosion criterion, a limit theorem, a CLT. +AP-2026-0311 contains no asymptotic statement of any kind — its theorems are exact finite +representation results (cut along an essential cycle ↔ two-terminal connectivity; every +cofacial plane network realizes on a torus; k+1 predictive classes on 8k vertices) plus +finite witnesses. It also *concedes in its own §8.1* that the testing instrument is +prior: "The abstract ability of copying tests to refine linear trace information is +therefore prior theory." So the desk-reject letter's two clauses — scope, and +"insufficient methodological novelty from an applied probability view" — are both +consistent with what the journals' own pages and TOCs show. **The reject was predictable +as an audience call, and it is an audience call, not a finding that any construction is +false.** + +Per the tripwire: among the fetched items above, **none is a construction/gadget-based +paper like AP-2026-0311** — the closest (Marivain) introduces a new model but immediately +proves asymptotics. Therefore this note does **not** recommend resubmitting the same text +to JAP/AAP, in any form. (Allowed answer: **no**.) + +## 3. Q3 — Split vs keep-one (taken before Q4 so the venue rows have a target) + +Independent of Grok, and keeping the one-paper-object rule: + +- **Part I content (= essentially the submitted AP paper + Thms B/C/D).** This is a + self-contained representation theorem with exact witnesses. It is the submission-ready + unit. It should go to **one** venue as one paper (see §4; best ticket-list fit: EJC). +- **Part II (Kalman no-go).** Theorem E is genuinely disjoint from Part I: different + object (declared `(G,B,C)` tasks), different method (realization theory), different + literature (control, identifiability). As a standalone counterexample paper it does + not overlap the Part I submission — no shared text, no shared theorems. A separate + submission of Part II alone is therefore **not** a violation of the one-paper rule, + but it is only worth doing if the honesty ladder (§4c of the #688 draft) survives + review as a framing, since Theorem E alone is a short paper. +- **Part III (P398 quotient).** Not ready: the quotient identification is proved at five + widths (`w = 4..8`, blocks `10, 26, 76, 232, 750`) and is stated as an **instance**, + not a theorem; the generic-rates lemma is unwritten (#688 draft §8.1). No venue + recommendation for Part III in its current state. It should **not** be split off now, + and should not ride along as a third of a submission either — if Part I goes out + alone, Part III stays in the repo until the lemma exists. + +Net: **keep-one for what was submitted (Part I, possibly enriched by Thms B/C/D); one +optional separate short paper for Part II; nothing for Part III yet.** This never puts +the same theorems in front of two editors. + +## 4. Q4 — Venue fit, with fetched scopes + +- ***Combinatorics, Probability & Computing.*** Scope (fetched, verbatim): "devoted to + the three areas of combinatorics, probability theory and theoretical computer science. + Topics covered include classical and algebraic graph theory, extremal set theory, + matroid theory, probabilistic methods and random combinatorial structures; + combinatorial probability and limit theorems for random combinatorial structures; the + theory of algorithms …". Plausible on paper, but the center of gravity is random + structures and asymptotics. No analogue of an exact finite representation theorem with + witnesses was fetched from CPC. **Second-best ticket-list fit for Part I** (the + graph-theory clause is real), but it would be a framing stretch. +- ***Random Structures & Algorithms.*** **NOT_FETCHED** (see §1) — no recommendation + beyond Grok's own, which this retrieval can neither confirm nor contradict. Flagged + for a future pass. +- ***Electronic Journal of Probability / ALEA.*** Scope (fetched, EJP, verbatim): + "publishes full-length research articles in probability theory. Short papers, those + less than 12 pages, should be submitted first to its sister journal, the Electronic + Communications in Probability (ECP)." The board (fetched) is strongly + interacting-particles/percolation weighted (Toninelli ed.; Arguin, Manolescu, Camia, + Teixeira, …). The fetched ALEA Vol XXIII TOC is likewise limit-theorem-heavy. Fit: + **Part II**, reframed as an exact identifiability/counterexample result for finite + Markov chains — and only via ECP if kept under 12 pages. No fetched EJP/ALEA analogue + of either part; EJP for Part I would be a poor fit (no asymptotics). +- ***Electronic Journal of Combinatorics.*** No scope statement on the fetched page + (ABSTRACT_ONLY), but the current issue (33(3), 2026, fetched TOC) contains exact + finite combinatorics of exactly AP-2026-0311's flavor: vertex-connectivity and + reliability-adjacent graph papers, enumeration with small witnesses, Cayley-graph + random-walk papers. **Best ticket-list fit for Part I** — the cut-network theorems and + the seven-vertex witness are finite combinatorics of networks, and the paper's + reliability citations ([6] Brown et al., *Networks* 76 (2020)) show where the + neighboring literature lives. Risk to state honestly: EJC editors will ask for the + combinatorial payoff to be front-and-center, i.e. the "homological" wrapper becomes + context, not headline. +- ***Linear Algebra and its Applications*** (and *Systems & Control Letters*, not + fetched). LAA scope (fetched, verbatim): "publishes articles that contribute new + information or new insights to matrix theory and finite dimensional linear algebra … + and … significant applications of matrix theory or linear algebra to other branches of + mathematics and to other sciences provided they contain ideas and/or statements that + are interesting from linear algebra point of view." Theorem E is literally a statement + about controllable/observable ranks of `(G,B,C)` triples — **LAA is a credible Part II + home**, arguably better than EJP because the audience already owns Kalman machinery. + SCL (**NOT_FETCHED**) would be the control-native alternative. +- ***J. Phys. A / JSTAT.*** **NOT_FETCHED** (404 on the IOP scope page). No + recommendation recorded. Note the draft's own boundary — P398 is not percolation — + makes these venues a framing trap for Part III anyway. +- ***JAP/AAP as a resubmit.*** **No.** Per §2: no fetched analogue as + construction-based as this paper, so the tripwire forbids recommending it. + +## 5. Q2 — Classical ingredients: used or reproved? + +| Ingredient (primary) | Status in P5/AP | Verdict | +|---|---|---| +| Kalman controllable/observable rank, finite-horizon realization (Wonham; standard control texts) **[LIT]** | Part II *uses* it: exact task order **defined** as the Kalman CO dimension; Phase H factorization is "the Kalman decomposition applied to a C₂-equivariant realization — cite Wonham" (#688 draft §5) | **Uses** | +| Exact lumpability of Markov chains (Kemeny–Snell **[LIT]**) | AP PDF Thm 3.3 *applies* the Kemeny–Snell criterion ("Direct exact backward refinement … splits four survival classes … Hence the survival partition fails the Kemeny–Snell criterion"); P5 Part III *uses* coarsest-lumping refinement as a tool | **Uses** | +| Schur / Wigner–Eckart vanishing of odd matrix elements **[LIT]** | P5 Part III Thm G: the vanishing is the classical statement; the paper's claim is **language + pointwise verification** on a declared grid, "claimed as a remark citing Schur, not a theorem" (#688 draft §5) | **Uses, with a new pointwise-formulation remark** | +| Callan–Smiley reflection-fixed noncrossing counts (arXiv:math/0510447 — **ABSTRACT_ONLY**, abstract confirms the reflection-invariance count via the Kreweras involution); Ding 2016 **[LIT]**; Burnside | P5 Part III: counts `[Catalan(w)+C(w,⌊w/2⌋)]/2 = 10,26,76,232,750` are **cited, not claimed** — "The paper cites; it claims nothing here" (#688 draft §5). Draft boundary §7.1 is explicit | **Uses** | +| D'Angeli–Donno 2013 (AAM 51(3):367–391; arXiv:1304.4180 — **ABSTRACT_ONLY**) | Fetched abstract confirms the load-bearing reading: "every lumping can be obtained from the action of a suitable subgroup … **if and only if** the poset … is totally ordered" — i.e. lumpings not arising from group actions exist, so "coarsest lumping = Aut-orbit" needs hypotheses P398 does not satisfy. P5 cites Prop. 13 as the *negative* boundary on its own w=4..8 instance | **Uses (as the reason the instance is not a theorem)** | +| Nonnegative vs ordinary rank of slack matrices (Yannakakis; Fiorini et al., regular n-gon **[LIT]**) | P5 Part I Thm D **reproduces the standard regular-n-gon separation** as a self-contained family (`rank ≤ 3`, nonnegative rank `n`) inside the response-matrix story; the draft itself says "Background the paper must cite but not claim" for the neighboring reduction | **Reproves a known instance as background** — must stay cited-not-claimed in prose | +| Probabilistic testing / copying tests (Larsen–Skou; van Breugel et al.; Deng–Feng **[LIT]**) | AP PDF §8.1 states the fork "is the same algebraic pattern used in probabilistic testing" and "the abstract ability of copying tests to refine linear trace information is therefore prior theory" | **Uses** | + +**Published cousin of the no-go "bounded task rank does not determine a threshold"?** +Within what was fetched: **none found**. The closest fetched adjacency is D'Angeli–Donno +(lumping structure ≠ group structure) and the weighted-automaton canonical-form line +(Balle–Panangaden–Precup **[LIT]**, cited by the AP PDF itself as prior for +predictive/trace representations). No fetched source states "identical finite-horizon +responses, different closing-gap thresholds (`1/2` vs `1/3`)". This negative is bounded +to this pass's searches — hidden-Markov identifiability and interacting-particle-systems +literature were **not** swept and remain the open place a cousin could live. The #688 +draft's own framing (Theorem E as a two-point separation with the honesty ladder) is the +correct hedge: even if a cousin exists, the torus/occupation connection keeps Part I +distinct, and Part II would still be an *explicit, verified* counterexample. + +## 6. Q5 — Title/framing + +The submitted title, *"Homological Occupation Growth: Branching Tests and Cut-Network +Representations"*, leads with a coinage ("homological occupation growth") that no +probability or combinatorics editor can map onto a known theorem class. What the paper +actually proves: (a) complete survival laws (lifetime signatures) do not determine +branching continuations, with gap 1/98; (b) the missing state is a cut-network +two-terminal connectivity state, update-closed; (c) every cofacial plane two-terminal +network realizes on a torus; (d) k+1 predictive classes on 8k vertices inside one +survival class. Suggested titles, theorem-naming, no marketing: + +1. **"Complete survival laws do not determine branching continuations: cut-network + states and a converse realization theorem for rank-one occupation growth on the + torus."** +2. **"Lifetime signatures versus copying tests: an exact cut-network representation and + unbounded branching-predictive classes on 8k vertices."** +3. **"Two-terminal network states for torus occupation growth: representation, + realization, and the limits of reliability summaries."** + +Each names either the separation (survival law vs branching) or the representation +theorem, in vocabulary an editor already parses. For the full P5 object (if/when Part II +is bundled), keep "no-go" language out of the title until the critical-sector-completeness +hedge is stated inside the abstract, not the introduction. + +## 7. Stress-testing the prior Grok assessment + +Point by point, with the fetched evidence: + +1. **"The JAP desk reject is a correct audience call, not a finding that the + constructions are false."** — **Agree**, and the fetch sharpens it: JAP's scope + statement is application-broad but its sampled TOC (§2) is uniformly + asymptotic-theorem-shaped; AP-2026-0311 has no asymptotic statement. The reject + letter's own second clause matches the paper's §8.1 concession that the testing + instrument is prior. +2. **"Methods are classical; packaging is new."** — **Agree with a refinement.** The + *instrument* (copying tests, lumpability criterion, Kalman ranks) is classical and the + paper says so. But Grok's phrasing undersells Thms 4.1/5.2: the cut-network + representation + converse realization is a genuinely new two-sided bridge between + rank-one homological continuation and planar two-terminal reliability. It is new + *structural* methodology — it simply is not *applied-probability* methodology, which + is what the EiC's clause was measuring. The right venue is one whose audience owns + structure, not limits (§4: EJC first). +3. **"Split: cut-network rank + no bounded radius → CPC or RSA; Kalman no-go → EJP or + ALEA; P398 quotient waits."** — **Partially agree; refine two ways.** (i) The + submitted paper is *already* the Part-I unit; splitting "cut-network rank" from "no + bounded radius" would thin an already-complete object — enrich, don't shatter. (ii) + CPC/RSA: CPC's fetched scope lists graph theory, so it is *arguable*, but its + probabilistic clauses are all random-structures/asymptotic; RSA could not be fetched, + so no RSA recommendation is recorded here. EJC (fetched TOC, §4) is the better + ticket-list target for Part I. (iii) EJP/ALEA for Part II: plausible, but the + fetched EJP scope routes <12-page papers to ECP, and LAA's fetched scope makes LAA at + least as good a fit because Theorem E is a Kalman-rank statement. (iv) P398 waits: + **agree exactly** — five widths are an instance, the generic-rates lemma is unwritten, + and D'Angeli–Donno (fetched) is precisely why it cannot be written as a general + theorem. +4. **"The submitted title hid the theorem."** — **Agree**; §6 proposes replacements that + name the separation and the representation theorem. +5. **"Not PTRF / Ann. Probab. / CMP."** — **Agree**, trivially: no asymptotics anywhere + in the object; nothing fetched suggests otherwise. Not re-litigated. + +## 8. Claim boundaries kept (restated for the record) + +- P5 and AP-2026-0311 are one paper object; §3's split never double-submits theorems. +- The w=4..8 lumping–orbit equality is an **instance**, not a theorem (D'Angeli–Donno + fetched and quoted as the reason). +- No wrapping identities (`D=r-1`, `F=(1+M)/2`) appear here or are recommended anywhere. +- No JAP/AAP appeal; no JAP/AAP resubmission of the same text (no fetched analogue; + §2's analogue list is explicitly not construction-based). +- The AP proof PDF is not committed; no `references.bib` entries were added. +- `docs/STATUS.md` and `docs/ROADMAP.md` untouched; #644, #650, #688 untouched. + +## 9. Not fetched / open for a follow-up pass + +- **RSA scope** (Wiley page returned no text) — needed before any RSA call. +- **SCL, J. Phys. A, JSTAT scope** — needed before a Part II control-physics call. +- Primaries for Kemeny–Snell, Wonham, Ding 2016, Yannakakis / Fiorini et al., + Larsen–Skou, Deng–Feng, Balle et al. — marked **[LIT]** above; a venue-prep pass + should fetch and transcribe them into `references.bib` per the #601 packet convention. +- Hidden-Markov / identifiability literature sweep for a published cousin of Theorem E.