1. What this unit proves 2
2. What it uses 3
3. Strategy 4
4. Steps 4
5. Where the difficulty lies 5
6. Notation 5
7. What remains unexplained 5
Code dependencies 5
Lean source 5
Additional source locations 6
Key statements 12
Part IV — Infinite volume.
Extracts an infinite-volume limit from fixed-volume correlation functions and establishes its Euclidean axioms.
Status: edited. Condensed from the unit's detailed notes, which were checked against the linked Lean in two rounds and edited; the local limitations still apply. Status meanings.
Overview · All units · Detailed notes · ← Previous: III.16 · Next: V.1 →
Results · Inputs · Strategy · Proof steps · Notation · What remains unexplained · Code dependencies · Lean source · Key statements
Reading scope. For the main theorem, the Wilson family is supplied by III.14. III.13 describes an alternative family; it is not substituted into that route here.
Part: IV, infinite volume (the related units).
Lean: YM/Extraction, YM/Cells, YM/Slice,
YM/Euclid, YM/Dash,
YM/Next/EXTRACTION, YM/Lane;
D01–D06. Size: 7,444 lines in 1,049
line-counted declarations in the source allocation.
Source inventory and conventions
Let \(C\) be a Wilson family for a compact matrix gauge group. (H1) is its specification (II.1–II.2, III.13; Family.Makes). (H2) comprises fixed-volume centered budgets, ordered-tail power budgets, reflection positivity and chessboard inequalities, including root subadditivity (CenterPort). Extractions select strictly increasing spacing and volume indices; ordered tails mean \(k\) large, then \(j\) large depending on \(k\). For every fixed \(p\) and label \(v\), at every sufficiently fine scale \(m\), (H2) gives a constant bounding all centered roots and the interior condition for all sites and large \(j\), with \(N_j \geqslant \max (m, \text{ref} (d))\). For every degree \(d\), \(R > 0\) and \(\varepsilon > 0\), it also gives \(m_0 \geqslant \text{ref} (d)\) and \(A \geqslant 0\) such that, for every \(m \geqslant m_0\) and \(\eta > 0\), eventually along ordered tails with \(N \geqslant m\), all labels \(\|v\| \leqslant R\) and sites satisfy the centered interior condition and root bound \(Aa_m^{- d - \varepsilon} + \eta\). Positivity and chessboard laws hold at every cutoff and volume.
Proposition IV.1.1 (raw limits). Under (H1)–(H2) along any extraction \(f\), there exist a refinement \(((N_j), (p_k))\) and multilinear tables \(S, P_p\) such that, for every arity \(n\) and label list \(v\):
(a) For every \(p\), \(P_p (v ; \cdot)\) is continuous on \(\text{Gen}^{(p)}_n\). For every compact \(K\) therein and \(\varepsilon > 0\), eventually in \(j\), \(|\mathcal{C}_{N_j, p} (v ; r) - P_p (v ; a_{N_j} r) | < \varepsilon\) for all lifted lattice tuples \(r\) with \(a_{N_j} r \in K\).
(b) \(S (v ; \cdot)\) is continuous on \(\text{Gen}_n\), and \(P_{p_k} \to S\) locally uniformly there.
(c) For every \(t \in \mathcal{D}_n\), \(t\) is eventually admissible at \(p_k\); at every admissible \(p\), \(V_p = \lim_j \mathcal{C}_{N_j, p} (v ; t)\) exists, and \(V_{p_k} \to \int tS (v ; \cdot)\).
(d) \(tS (v ; \cdot)\) is integrable for every \(t \in \mathcal{D}_n\).
Status: theorem. Lean: Extraction.crops.
Proposition IV.1.2 (label derivatives). Suppose \(S\) satisfies IV.1.1(c)–(d), with (H1)–(H2) along \(f\). Assume the relay properties of §2 for every degree, projections \(\pi_d (N, p)\) on \(\text{Label}_d\) converging in operator norm along ordered tails to projections \(\pi_d\), and \(\dim \text{Null}_d (S) \leqslant \dim \ker \pi_d\). Then \(\ker \pi_d = \text{Null}_d (S)\), and there exist linear maps \(\kappa_{d, e} : \text{Label}_d \to \text{Label}_e\) for every \(d \leqslant e\), and \(\tau_{d, a} : \text{Label}_d \to \text{Label}_{d + 1}\) for every axis, such that for all arities, slots, labels and \(t \in \mathcal{D}_n\),
Status: conditional theorem. Lean: Work.D04.answer, lifts_model.
Proposition IV.1.3 (smooth limit). If \(S\) satisfies IV.1.1 along an extraction where IV.1.2 applies, there exists \(T\), smooth on every \(\text{Conf}_n\), equal to \(S\) on \(\text{Gen}_n\), satisfying IV.1.1(a)–(d) with the same extraction and \(P_p\), and the basic identities below. The maps descend to \(\text{rise}_{d, e} : F_d \to F_e\) and \(\partial_{d, a} : F_d \to F_{d + 1}\). For every arity, label list, slot and \(x \in \text{Conf}_n\),
The smooth ordered limit on \(\text{Conf}_n\) along a given extraction is unique. Status: theorem. Lean: Org.Arrives.serves.
Proposition IV.1.4 (Euclidean hierarchy). Assume (H1), IV.1.1(a)–(c) for \(T\) along \(f\), and smoothness on every \(\text{Conf}_n\). Then, for every arity, degree list and labels, on \(\text{Conf}_n\):
(E1) Simultaneous permutations preserve \(T\); the empty correlation is \(1\); inserting \(1_d\) deletes that slot.
(E2) \(T (\rho (s) v ; sx) = T (v ; x)\) for every \(s \in H_4\), and \(T (\bar{v} ; x) = \overline{T (v ; x)}\).
(E3) \(T (v ; x + b) = T (v ; x)\) for every \(b \in \mathbb{R}^4\).
(E4) For every axis \(a\), finite collection of words \(w_i\) with distinct points within each word and \(x^a > 0\), and \(c \in \mathbb{C}^m\), \(\sum_{i, j} \bar{c}_i c_j T (\Theta_a w_i w_j)\) is real and nonnegative.
(E5) The same holds for smeared words \((v^{(i)}, t_i)\), \(t_i \in \mathcal{D}_{n_i}\) supported where all \(x_k^a > 0\), with entries \(\iint \overline{t_i (y)} t_j (z) T (\Theta_a (v^{(i)}, y) (v^{(j)}, z))\); these integrals converge absolutely.
With (E7), this gives (E1)–(E7): (E6) is smoothness; (E7) requires, for every arity, degree list, labels and \(j \geqslant 0\), some \(A \in \mathbb{N}, C \geqslant 0\) with \(\|D^j T (v ; x)\| \leqslant Cr^{- A}\) for all \(0 < r \leqslant 1\) and all \(r\)-separated configurations. The constants are independent of \(r, x\); \(D^j\) has its real Fréchet operator norm. Status: theorem, conditional on (E7) for assembly. Lean: Order.assembles.
Proposition IV.1.5 (short-distance bound). Assume (H1)–(H2) along \(f\), smoothness of \(T\) for every arity and labels, and IV.1.1(a)–(b). For every arity, degree list \((d_i)\) and labels \(v\), some \(C \geqslant 0\), independent of \(r, x\), satisfies \(|T (v ; x) | \leqslant Cr^{- 4 n (1 + \max_i d_i)}\) for all \(0 < r \leqslant 1\) and \(r\)-separated \(x\). The empty maximum is \(0\). With IV.1.3, B10 (V.1) supplies (E7). Status: theorem. Lean: Work.D05.answer.
Proposition IV.1.6 (symmetries of derivatives). For smooth, permutation-symmetric \(T\) satisfying (E2), let \((\text{rise}, \partial)\) satisfy IV.1.3's identities. For every \(d \leqslant e\), axis and \(s \in H_4\), these maps commute with conjugation, \(\text{rise}\) commutes with \(\rho (s)\), and
Status: theorem. Lean: Work.D06.compat.
II.1–II.2, III.13: Wilson family specification (H1).
III.11: centered budgets (H2), supplied by MergeProof.Accent.gardena.
III.10–III.11: relay approximations to differences and inclusions, null-label detection and the separate dimension comparison (MergeProof.Scotia, savings).
III.16: positive-kernel trace Lipschitz estimates. I.1, III.2, V.1: gentle scales and subpolynomial column products. V.1: B10.
Osterwalder–Seiler positivity; chessboard estimates; Schatten Hölder, three-lines and Cauchy estimates; Arzelà–Ascoli, dominated convergence and Fourier regularity.
The relay hypotheses require bounded projections at each fixed volume and along ordered tails; degree-\((d + 1)\) approximants to degree-\(d\) forward differences and degree-\(e\) approximants to degree-\(d\) labels for \(d \leqslant e\), with subpolynomial cutoff coordinates; and a null-label detector. For each approximation error there must be one rate \(y (N, p) \to 0\): for every integer \(q \geqslant 0\) and lower scale \(m\), some \(k \geqslant m\) gives, eventually along ordered tails with \(N \geqslant k\), interior errors with roots \(\leqslant y (N, p)\) in every translated bond cell of side \(a_k\), at relative lattice coordinates satisfying \(M \leqslant z^a\) and \(z^a + M + 1 \leqslant a_k / a_N\), where \(M = \lfloor (a_k / a_N) / (8 (q + 1)) \rfloor\). Projection-kernel labels with subpolynomial growth must be chessboard-close to bounded measurable observables whose separated spectator correlations vanish exactly, eventually along ordered tails. Subpolynomial means \(\|B_{d, N, p}^{- 1} w (N, p)\| \leqslant a_N^{- \varepsilon}\) eventually for every \(\varepsilon > 0\), where \(B\) is the terminal transfer to reference coordinates. The dimension comparison is an additional hypothesis (RelayBin).
Chessboard bounds control separated correlations. Transfer kernels turn movement of an insertion into movement between gaps; complex interpolation gives equicontinuity. Two compactness arguments select spacing and volume subsequences. Relay approximations give distributional derivatives as raised-label kernels, hence smoothness. Lattice symmetries and positivity pass to the limit; scale estimates give polynomial bounds.
Under (H1), reflection positivity and repeated Cauchy–Schwarz give chessboard estimates; translation log-convexity transfers centered bounds to interior positions. Around.interior_from.
Under (H1)–(H2), translated cells separate insertions and bound correlations on distinct-point compact sets. Around.size_done.
Under the same hypotheses, slab traces and chessboard column moments give equicontinuity on generic compact sets through interpolation and Cauchy's estimate. Difference.completed.
Two Arzelà–Ascoli extractions give IV.1.1(a)–(b); domination gives (c)–(d). first_surface, Order.ordinary.
Under IV.1.2's hypotheses, summation by parts identifies weak derivatives; small chessboard errors vanish. Normalized projected approximants and the dimension count give fixed derivative labels. Work.D03.answer, Dash.brace.
Iterating these identities gives rapid localized Fourier decay and the smooth representative. Dash.awake.
Under IV.1.4's hypotheses, density passes symmetries; limiting Gram matrices pass positivity. Under IV.1.5's hypotheses, two scale choices give its bound. The derivative plan then promotes it to (E7). Order.realPositive, Work.D05.answer, Work.B10.capped_tags.
Uncentered roots and transfer-kernel equicontinuity require the main estimates. Fixed derivative labels require normalized remainders and the separate rank comparison. Equicontinuity holds on generic configurations; smoothness extends to all distinct-point configurations. The label maps are determined only modulo null labels, while their quotient maps are intrinsic. Continuous rotation symmetry comes later.
\(\text{Label}_d\) is the finite-dimensional complex label space; \(F_d = \text{Label}_d / \text{Null}_d\), where null labels have zero smeared kernels in every arity and slot against all spectator labels and tests. \(\text{Conf}_n\) means distinct points; \(\text{Gen}_n\) means separation in every coordinate, modulo \(L_p\) for \(\text{Gen}^{(p)}_n\). \(\mathcal{D}_n = C_c^{\infty} (\text{Conf}_n)\); admissibility means support inside \((- L_p / 2, L_p / 2)^{4 n}\). \(a_N \downarrow 0\) and \(L_p \uparrow \infty\) are spacings and periods. \(\mathcal{C}_{N, p} (v ; r) = \langle \prod_i I_{N, p} (v_i, r_i) \rangle\); smearing sums against \(t (\hat{r})\) with weight \(a_N^{4 n}\), where \(\hat{r}\) is the physical site position. Lifted tuples are reduced modulo \(L_p / a_N\) for evaluation. \(v_{i \to w}\) replaces slot \(i\) by \(w\). Separation uses the supremum norm. \(\rho_d\) acts on labels and their quotients. \(H_4\) consists of signed coordinate permutations, with matrices \((s_{ab})\); \(s_a\) reverses axis \(a\); \(\Theta_a\) reverses words, reflects points and replaces \(v\) by \(\rho (s_a) \bar{v}\).
The upstream relay, rank and full fixed-volume kernel arguments remain unexplained, as does the budget's relation to III.12. Extraction is subsequential in both indices; III.13 supplies the full-spacing upgrade (UVExtension.Reports.useFull). Independence from the volume subsequence is not asserted.
Detailed notes: the full account of this unit, with proof sketches for every step, is on the detailed notes page.
Code dependencies (from the compiled Lean). These lists follow references between declarations allocated to units; they can include shared definitions, record fields and alternative constructions. They are not a mathematical reading order. Only unit pairs with at least 25 references are shown; see the counting rules.
References to: I.1, II.1, III.16, III.1, III.10.
Referenced by: VIII.1, VII.4, VI.2, VI.1, VII.2, V.1, III.11, V.2, V.3, III.13, III.6, VII.3, III.10, III.12, VII.6, VII.1, II.1, VII.5, III.8, III.3, III.2.
Source allocation: 7,444 attributed lines; 1,049 line-counted declarations; 1,464 assigned declaration names. Of the assigned names, 1,385 are in the compiled closure of the seventeen headline results. These quantities have different scopes.
The folders and result chains below are source inventories. Their line counts cover each whole group, including any share allocated to another unit; percentages use attributed lines plus one per assigned name. They need not sum to the unit total.
work package YM/Extraction/, with YM/Slice/, YM/Cells/ — 4,642 lines.
work package YM/Euclid/ — 1,079 lines.
work package YM/Dash/ — 530 lines.
work package YM/Next/EXTRACTION/ — 249 lines.
work package YM/Lane/ — 101 lines.
work package YM/Further/DENSE_MIXED_KERNELS/ — 12 lines.
Further pinned locations for the definitions and comparisons discussed in this unit.
In the selected statement order. K- identifiers name this generated list; proposition and lemma numbers in the outline are separate. Each sentence is an automatically drafted summary of a Lean statement; follow its link for the exact hypotheses and conclusion. The outline status does not certify these summaries. Main result and key result describe importance, not correctness. Full names are in the statement table.
K-IV.1.1 Interior (numbered result) — A structure packaging a measurable local assignment on a set of sites with a uniform bound independent of the continuum limit.
K-IV.1.2 enumerate — The copy map for a cell sends the base chamber into the chamber of that cell.
K-IV.1.3 existence (numbered result) — For a fold-closed predicate satisfied by some assignment, some constant assignment also satisfies it.
K-IV.1.4 Hierarchy (key result) — A bundle of Euclidean-covariance-type properties (axioms) of a Schwinger-function table, required before establishing continuous rotation invariance.
K-IV.1.5 awake (key result) — Given a live, suitably regular family of distributions with enough derivatives, there is a smooth table matching it on the regular part.
K-IV.1.6 pathway (numbered result) — If a function along a sequence of translates has bounded root-test at both endpoints and all are interior, then it is bounded at all intermediate offsets.
K-IV.1.7 fold_both (key result) — Cauchy–Schwarz-type fold estimate: the square of the root power of a mixed assignment is bounded by the product of the two homogeneous ones.
K-IV.1.8 tiles (main result) — Chessboard estimate: the root power of a product of cell-localized functions is bounded by the product of single-function root tests.
K-IV.1.9 physical (numbered result) — For large period and small mesh, evaluating a shifted test function at lattice cells after a one-slot jump equals evaluating at the translated point, with no wrapping.
K-IV.1.10 answer (key result) — Forward differences in a single slot are controlled in the ordered extraction.
K-IV.1.11 step_positive (key result) — For nonnegative β, the temporal heat-kernel transfer step of the Wilson lattice is a nonnegative (positive-type) kernel.
K-IV.1.12 enough (key result) — For a reflection-symmetric brick tiling, the p-th root-power of the norm of the column moment is bounded by the common root-test bound A of the function and its reflection.
K-IV.1.13 reflected_positive (key result) — The Wilson lattice measure is reflection positive across a seam for nonnegative β.
K-IV.1.14 laws (key result) — For nonnegative coupling, the cell reflection-positivity seam property holds for all sizes.
K-IV.1.15 seamless (key result) — For a family satisfying the makes condition, the torus tiling at every side and coupling scale has the seam property.
K-IV.1.16 assembles (main result) — An ordered extraction with point limits, smooth and capped, yields a full Euclidean-positive hierarchy of limiting correlation functions.
K-IV.1.17 answer (key result) — The D05 requirement holds: the extraction order's evaluation, with tie handling and cofinal sequences, is established for the construction.
K-IV.1.18 answer (key result) — Transfer from match rents to the all-shot monitor holds, discharging a required hypothesis for the extraction.
K-IV.1.19 uniqueLift (key result) — Compatible finite-degree linear maps lift uniquely to a linear map on the union (colimit) space.
K-IV.1.20 powering (key result) — Shifting a length l of time between two bricks changes the moment by at most the product of column bounds times l/r.
K-IV.1.21 organized (numbered result) — For well-separated time points, cyclic gap arithmetic reorganizes the product of column moments with a bound in terms of the K_i.
K-IV.1.22 primitive (key result) — Lattice correlation functions have a Lipschitz bound for a single-point shift in one axis, with constant given by a product of root-test bounds divided by the separation scale r.
K-IV.1.23 completed (key result) — Centered bounds on lattice data give Lipschitz equicontinuity of the transfer from the exact cell estimates.
K-IV.1.24 crops (key result) — From centered bounds on the projected data one can extract a subsequence with raw continuum limits.
K-IV.1.25 compat (key result) — For a smooth symmetric table with real spin and a differential plan with derivatives and inclusions, the plan is compatible with orientation.
K-IV.1.26 Valid (numbered result) — A validity predicate bundling conditions on a common aligned series of coefficients for detecting sets at each weight.