[P4 venue retrieval] wrapping/crossing literature, lattice sizes, venue fit (#696)

[LightChainr]

Retrieval-only notes per issue #696. Branch from main; P4 draft (pr688) read via git show, not merged.

Contents (notes/venue-retrieval-p4-20260909.md)

Sources fetched and read in full (PRIMARY_TEXT_READ):

• Langlands–Pouliot–Saint-Aubin, Conformal invariance in two-dimensional percolation, Bull. AMS 30, 1–61 (1994) — torus homology-class wrapping probabilities, 500×500 torus, four-angle rotation test (Table 3.3)
• Newman–Ziff, Phys. Rev. E 64, 016706 (2001) — R^(h,v,b,e,1) wrapping observables; 7.3×10⁹ samples at 128²; 1024² stretched-exponential
• Mertens–Ziff, Phys. Rev. E 94, 062152 (2016) — the matching function M_L(p) = R^x_L(p) − R̂^x_L(1−p); exact L=3–11, MC to L=128; corrections-to-scaling tower
• Pruessner–Moloney, J. Stat. Phys. 115, 839–853 (2004) — winding clusters, 300²–30000², published scalar-law non-identification
• Simmons–Kleban–Ziff, J. Phys. A 40, F771 (2007) — crossing densities, 512² bonds, 3.3×10¹¹ hulls

ABSTRACT_ONLY: Pinson, J. Stat. Phys. 75, 1167 (1994) (Springer/ADS bot-blocked — flagged as the one unresolved primary, needed before any channel-novelty claim); Akhunzhanov–Eserkepov–Tarasevich, J. Phys. A (2022). [LIT]: Cardy 1992; Watts 1996.

Key answers

• Q1: no published cos 4θ orientation contrast of a matching function found among the fetched texts; published odd objects are odd in p (Mertens–Ziff) or topological channel (NZ R^(1)); orientation only appears as LPS's rotation-irrelevance check (bounded at ~5×10⁻³, 10⁵ samples) and homology/aspect-ratio dependence. Caveat: Pinson full text unverified.
• Q2: MZ 2016 is the scalar-success-with-corrections analogue (two extra exponents 2−x=−3.42, 2−y=0.705; "agree only approximately"); P–M 2004 is the honest scalar-failure precedent ("unable to identify these functions"; "the main source of error is not statistical, but systematic").
• Q3: published MC wrapping FSS headlines are made at L ≈ 512–1024 (one at 3×10⁴); exact-enumeration branch at L ≤ 128. P4's 500M replicas are a normal statistical budget; L ≤ 425 is the objection; a single new geometry at L ≥ 512 is the realistic ask.
• Q4: PRE first (MZ/NZ home), JSTAT/JSP second (Pinson/P–M lineage), J. Phys. A if exact-algebra leaning (SKZ/Watts precedent), PRL no (tripwire checked: no fetched PRL analogue lacking new exponent/p_c), SciPost fallback only.
• Q5: single-lineage can pass clearly labelled (MZ precedent) but the second independent doubling curve will be asked for; combine it with the L ≥ 512 axis.

Grok assessment stress-tested: agree on all five points, with two refinements (the L ≥ 512 number; name and test the MZ-style correction-exponent alternative in the manuscript analysis).

No Monte Carlo, no p_c claim, no STATUS/ROADMAP edits, ticket/issue/PR #644/#688 untouched, do not merge.

[LightChainr]

Verified as notes-only for #696. Do not merge. Do not close #696 / #644 / #688. STATUS untouched; no Monte Carlo; P3 statistic not imported; Pinson 1994 correctly left ABSTRACT_ONLY rather than guessed.

Tripwire: wrapping/crossing papers named with lattice sizes and venues (LPS 1994 500²; NZ 2001 128²/1024²; MZ 2016 exact L=3–11 + MC to 128; P–M 2004 to 30000²; SKZ 2007 512²). PRL tripwire held: no analogue without a new exponent/p_c.

Spot-check: LPS BAMS 1994, Newman–Ziff PRE 64 016706, Mertens–Ziff PRE 94 062152, Simmons–Kleban–Ziff J. Phys. A 40 F771 are real. Pruessner–Moloney J. Stat. Phys. 115 (2004) is a real winding-cluster paper (cond-mat/0310361). I did not re-fetch every PDF.

vs the prior Grok venue note (the point of #696)

Prior Grok	This note	Call
PRE/JSTAT after writing; not PRL	Agree, independently from MZ→PRE and P–M/Pinson→JSP	agree
Most physics-new of the four if the signal survives larger N	Agree; “larger” here means L ≥ 512, not 10⁴	agree, with a number I did not have
N~10² is the referee objection; 500M replicas do not buy systematics	Agree and sharpened: 500M is a normal statistical budget (NZ 7.3e9, P–M ~1e9); the deficit is linear size (425 vs 512–1024). 10⁴ is for tail probabilities, not amplitude/shape	agree; I overstated the size demand
Stop at low-rank mixing; no LCFT module	Agree; MZ absorb residuals into correction exponents, P–M refuse to identify C(n)	agree
Do not use the N=580 ladder as physics	Agree (already a P4 limitation)	agree

Two refinements that should enter the manuscript, not just this note:

1. Name and test the MZ correction-exponent tower (2−x=−3.42, 2−y=0.705) as the conservative alternative to Jordan mixing. The draft’s D′ vs S′ split (−0.009σ vs +2.70σ) is the discriminator; it currently lives in P50 REPORT.md.
2. Channel novelty: matching function (odd in p) and NZ R^(1) (odd in topological channel) are published; a cos 4θ orientation contrast of the matching function was not found in the texts read. Soften until Pinson 1994 is PRIMARY_TEXT_READ.

#695 (P3) still running.