Home·The theorem·The proof in one reading·Architecture·Units·Manuscript·Glossary·FAQ·Work items·Data·Method·Provenance·Changes
← II.1 Field labels and their lattice representativesPart II·Detailed notesIII.1 The block-spin step and the symmetric gauge fixing →

II.2. Marker algebra, polymer combinatorics and Gaussian estimates

Contents

1. What this unit proves 1

2. What it uses 4

3. Strategy 4

4. Steps 5

5. Where the difficulty lies 5

6. Notation 5

7. What remains unexplained 5

Code dependencies 6

Lean source 6

Additional source locations 6

Key statements 14

Part II — Data of the construction.

Develops the algebra and estimates used to multiply and sum polymer activities.

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: II.1 · Next: III.1 →

Results · Inputs · Strategy · Proof steps · Notation · What remains unexplained · Code dependencies · Lean source · Key statements

Reading scope. For activity definitions see III.1, for field labels and Wilson-family data II.1, and for reflection-positive cells IV.1. The source allocation below does not define these reading routes.

Source inventory and conventions

Part: II, Data of the construction. Lean: YM/Marker/, YM/Ac/, YM/Ovens/Coffee.lean, YM/Further/SHELL_BIRTH_49/Fusing/Coal.lean, YM/Extra/MULTISOURCE_TRAJECTORIES/, YM/Push/GRID_APPROXIMATION/Nil.lean, YM/MergeProof/, YM/Next/RANK/, YM/Vac/Treasury.lean. Size: 3,691 lines in 793 line-counted declarations in the source allocation.

1. What this unit proves

Proposition II.2.1 (marker algebra). For finite \(V\) and a downward-closed family \(R \subseteq 2^V\) containing \(\emptyset\) and singletons, \(\mathcal{A}_R =\mathbb{C}^R\) is a commutative unital algebra: \(\delta_x \delta_y = \delta_{x \cup y}\) for disjoint \(x, y\) with \(x \cup y \in R\), and zero otherwise; \(1 = \delta_{\emptyset}\). Constant coefficients multiply. If \(a (\emptyset) = 0\), then \(a^k (x) = 0\) for \(|x| < k\), hence \(a^k = 0\) for \(k > |V|\). Status: theorem. Lean: ring structure, Poly.stops.

Proposition II.2.2 (norm). For positive \(w_v\), set \(w^x = \prod_{v \in x} w_v\), \(\|a\|_w = \sum_{x \in R} w^x |a (x) |\) and \(\text{Exp} (a) = \sum_{k = 0}^{|V|} a^k / k!\). Then \(\|1\|_w = 1\), \(\|c \delta_x \|_w = w^x |c|\), \(\|ab\|_w \leqslant \|a\|_w \|b\|_w\), and \(\| \text{Exp} (a)\|_w \leqslant e^{\|a\|_w}\). A countable family with \(\|f_i \|_w \leqslant b_i\) and \(\sum_i b_i < \infty\) has a coefficientwise sum of norm at most \(\sum_i b_i\). Status: theorem. Lean: multiplication, budgeted.

Proposition II.2.3 (connected polynomial). For zero-constant \(a, b\), \(\text{Exp} (a + b) = \text{Exp} (a) \text{Exp} (b)\) and \(\text{Exp} (- a) \text{Exp} (a) = 1\). For nonempty \(x \in R\), \(\text{Exp} (c \delta_x) = 1 + c \delta_x\). Every \(F (\emptyset) = 1\) has a unique zero-constant \(C\) with \(\text{Exp} (C) = F\); \(C\) depends continuously on \(F\). Status: theorem. Lean: Exponential.pass_add, connected_spec.

Proposition II.2.4 (tagging; repacking). For finite \(s\) and \(\tau : \iota \to R\), a nonzero tagged product requires pairwise disjoint tags with union in \(R\); it equals \((\prod_{i \in s} c_i) \delta_{\bigcup_i \tau (i)}\), and weights factor. If \(\sigma : \iota \to \kappa\), admissibility is any property of finite slot sets, and \(\sum_i \|F_i \|_w < \infty\), then

\(\displaystyle \sum_{\text{\scriptsize{$\begin{array}{c} s \text{ finite} : \sigma |_s \text{ injective}\\ \sigma (s) \text{ admissible} \end{array}$}}} \prod_{i \in s} F_i = \sum_{\text{\scriptsize{$\begin{array}{c} S \text{ finite}\\ S \text{ admissible} \end{array}$}}} \prod_{k \in S} \sum_{\sigma (i) = k} F_i,\)

with summable terms. Status: theorem. Lean: tagged_prod, Packets.repack.

Proposition II.2.5 (components). For arbitrary finite supports \(X_i\) and finite label set \(s\), overlap-chain classes partition \(s\) into nonempty connected sets, called sparks, with disjoint support-unions. Conversely, disjointly supported sparks recover exactly these classes. Removing \(i \in s\) from a spark \(s\) leaves components each meeting \(X_i\). Status: theorem. Lean: separates, excise.

Proposition II.2.6 (tree bound). Let \(M \in \mathbb{N}\), \(a, \gamma, f_i \geqslant 0\) and \((2 M + 1)^4 \gamma \leqslant 1\). If, at every site \(x\), \(\sum_{i : x \in X_i} f_i e^{(a + 1) \text{cov}_M (X_i)}\) is summable and at most \(\gamma\), then

\(\displaystyle \sum_{\text{\scriptsize{$\begin{array}{c} s \text{ spark}\\ x \in X (s) \end{array}$}}} e^{a \text{cov}_M (X (s))} \prod_{i \in s} f_i \leqslant e^{\gamma} - 1,\)

and this family is summable; labels are not repeated. Status: theorem. Lean: Ac.forest.

Proposition II.2.7 (two overlaps). Give each label finite mentions and walls, and mark some labels dirty. Within a selection, a label is hot when mention-connected to dirt; its fuel comprises mentions and, when hot, walls in disjoint copies of their ambient sets. Fuel components retain their internal hotness. They bijectively decompose finite selections into nonempty fuel-connected sets with disjoint mention-unions and disjoint wall-unions among dirty-containing components. If walls are lattice supports and mention overlap implies wall overlap inside such a connected set, it is wall-connected. Status: theorem. Lean: Kiln.undo, Kiln.attends.

Proposition II.2.8 (fusion). Fix a gauge group \(G\), four-dimensional periodic lattice \(\Lambda\), activity parameters and regular/forced addresses (forced inventories nonempty and contained in their supports, with extra tests) and coefficients \(E_i, K_i\), plus logarithms \(\ell_i\) on nonempty supports \(Y_i\), indexed by \(\mathbb{N}\). Patches are forced supports or \(Y_i\). Pellets are finite patch-connected selections containing a forced label, with disjoint forced supports; their reaches are \(K_i\) and \(e^{\ell_i} - 1\). Pellets admit a bijective enumeration. Fusion preserves regular activities and assigns each pellet its union support, union forced inventory, conjunction of forced eligibility tests, and eligibility indicator times its reach-product. Status: definitions; enumeration theorem. The density, bound, covariance and parametric continuity results belong to III.10. Lean: fused, encoding.

The digest condition requires unfused validity with bounds \(\delta, b\); logarithms supported on patches meeting the large-field set, uniformly summably bounded, measurable, gauge invariant, local in internal links and flags, and affine covariant. Fusion bounds require \(\kappa, \gamma \geqslant 0\), \((2 B + 1)^4 \gamma \leqslant 1\), and a configuration-independent \(f \geqslant 0\) with \(| \text{reach} | \leqslant e^{- \kappa} f\) on forced labels and \(| \text{reach} | \leqslant f\) otherwise, satisfying II.2.6's hit condition with cover size \(B\) and exponent \(e_P + 1\). Lean: Digests, Buys.

Proposition II.2.9 (switching). For finite \(\iota\), \(\text{val}_x (g) (r) = \sum_{i \subseteq r} x^{|i|} g (i)\) is invertible for \(x > 0\). A model satisfies \(\text{val}_x (g_x) = f_x\) on \([0, 1]\) and \(g_x \to g_0\); its jet is \(g_0\). A family is paid at cost \(d\) when models share a bound \(|g_x^u (i) | \leqslant b (i)\) with \(b \geqslant 0\), \(\sum_i b (i) \leqslant d\). Costs add under finite sums and multiply under finite products; summable costs permit infinite sums and commuting jets with sums. If modeled \(f_x (r) = f_x (r \setminus \{v\})\) for every \(r\) and \(x \in [0, 1]\), then \(\text{jet} (f) (i) = 0\) whenever \(v \in i\); pointwise measurability of paid families passes to jets. If \(\| \rho_v \|_w \leqslant 1\), then \(\sum_i \text{jet} (f^u) (i) \prod_{v \in i} \rho_v\) has a common coefficient majorant of weighted norm at most \(d\). Status: theorem. Lean: entire, Accent.crystal.

Proposition II.2.10 (Gaussian calculus). For complex symmetric \(P, Q\), define \(\mathcal{G}_Q p = p (X + \partial_Q) 1\) and \(\mathbb{E}_{Q, m} p = (\mathcal{G}_Q p) (m)\). These satisfy Wick recursion, \(\mathcal{G}_P \mathcal{G}_Q =\mathcal{G}_{P + Q}\), and covariance transport \(Q \mapsto AQA^T\) under substitutions with finitely supported rows. A real random vector \(\xi\) under a probability law with all polynomial moments and \(\mathbb{E} [\xi_v q (\xi)] =\mathbb{E} [(\partial_{Q_{v \cdot}} q) (\xi)]\) for every \(v, q\) has expectation \(\mathbb{E} [q (m + \xi)] =\mathbb{E}_{Q, m} q\) for every real \(m\). If \(B \geqslant 1\), \(|m_v |, |Q_{vw} | \leqslant B\), \(\deg p \leqslant d\), then \(|\mathbb{E}_{Q, m} p| \leqslant \|p\|_1 \pi (B, d)\). Two such pairs differing entrywise by at most \(E \geqslant 0\) have expectation difference at most \(\|p\|_1 E \pi (B, d)\). Status: theorem. Lean: warms, payment, Common.differs.

Proposition II.2.11 (polynomial steps). Let \(Q\) be symmetric and \(H_v\) linear. If \(C \geqslant 1\), \(\|H_v \|_1, |Q_{vw} | \leqslant C\), \(\deg p \leqslant d\), then \(T_{H, Q} p = (\mathcal{G}_Q p) \circ H\) has degree at most \(d\) and norm at most \(\|p\|_1 (2 (d + 2) C)^d\). Two such pairs differing by at most \(\varepsilon \geqslant 0\) in these norms have output difference bounded by \(\|p\|_1 \varepsilon (2 (d + 2) C)^d\). Zero cross-covariance between variables of \(P, R\) gives \(T (PR) = T (P) T (R)\). Status: theorem. Lean: crux, crossing, bridge.

Weighted coefficient quantities \(\sum_{\beta} r^{w \cdot \beta} |c_{\beta} |\), for \(w : V \to \mathbb{N}\), \(r \geqslant 0\), are submultiplicative, and norms for \(r > 0\). If \(0 \leqslant \varepsilon \leqslant 1\), \(|Q_{vw} | \leqslant \varepsilon\), \(\deg p \leqslant d\), then \(\|\mathcal{G}_Q p - p\|_1 \leqslant \|p\|_1 \varepsilon (2 (d + 2))^d\). For linear \(f_v\) and polynomials \(A_v\) without terms below degree two, \(\|f_v \|_1, \|A_v \|_1 \leqslant B\), \(B \geqslant 1\), \(0 < t \leqslant 1\), symmetric \(K\) with \(0 \leqslant \varepsilon \leqslant 1\), \(|K_{vw} | \leqslant \varepsilon\), and \(\deg P \leqslant n\),

\(\displaystyle \| \Phi_t (P) - (\mathcal{G}_K P) \circ f\|_1 \leqslant \|P\|_1 (ntB (2 B)^n + \varepsilon (2 (n + 2))^n B^n),\)

where \(\Phi_t : X_v \mapsto f_v + t^{- 1} A_v (tX)\). A measurable \(g \in [0, 1]\) with \(\int (1 - g) d \mu \leqslant \eta\) is an \(\eta\)-guard. For probability \(\mu\) and measurable \(f\) with \(\int |f|^2 d \mu \leqslant B < \infty\), \(| \int gf \hspace{0.17em} d \mu - \int f \hspace{0.17em} d \mu |\) is at most \(\varepsilon' B + \eta / \varepsilon'\) for every \(\varepsilon' > 0\). Under II.2.10's probability-law hypotheses and entry bounds, \(\int |p (m + \xi) |^2 \leqslant \|p\|_1^2 \pi (B, 2 d)\). For finite \(V\), \(d \in \mathbb{N}\), \(\delta > 0\), there exist finite nonempty \(T \subseteq [- \delta, \delta]^V\) and \(C \geqslant 0\) with \(\|p\|_1 \leqslant C \max_{x \in T} |p (x) |\) for every \(\deg p \leqslant d\). Lean: LiftMark.stock, escaped, Spindle.molds.

Proposition II.2.12 (scales). For \(s > e\), put \(y = \log \log s\), \(\Pi = e^{y^2}\), \(\theta = s^{- 1 / 2} \Pi\), \(h = e^{y^{3 / 2}}\), \(j = h^{1 / 5}\), \(M = \lceil (\log s)^2 \rceil\), \(W = M^2\). For every fixed \(a, p > 0\),

\(\displaystyle y^2 \ll j \ll h \ll \Pi^a \ll s^p, \qquad W \sim (\log s)^4 .\)

For fixed \(a > 0\), \(k \in \mathbb{N}\), \(c \in \mathbb{R}\), eventually \(W^k, h \leqslant \Pi^a\), \(c \log W \leqslant ay^2\), \(y (s)^2 \leqslant aW (s / 2), aj (s / 2)\) and \(j (s) \leqslant h (s / 2)\). Eventually \(s \theta^2 = \Pi^2\), \(s \theta^3, s \theta^4 \leqslant 1\), and \(2 \Pi^v \theta^2 \leqslant 1\) for every \(v \leqslant 1\). Status: theorem. Lean: uniform_loop, rate_fire.

Proposition II.2.13 (numerical feasibility). Fix \(m, Q \in \mathbb{N}\), \(e, U, V \geqslant 0\), \(a > 0\), \(\alpha \geqslant 1\), \(L > 1\), and \(G (u) = e^{- ah (u)^2}\). Eventually, uniformly for \(u, v \in [s / 2, s]\), \((2 W (u) + 1)^4 \Pi (s)^{- 1} \leqslant 1\), \(G (u) \leqslant \Pi (s)^{- 3 / 4} e^{- 7 j (v) / 10}\) and \((\alpha + 1)^{\lceil 10 \log M (s) / \log L \rceil} \Pi (s)^{- 3 / 4} \leqslant \Pi (s)^{- 1 / 2}\). For fixed \(k, n \in \mathbb{N}\), \(A \geqslant 1\), \(\phi = 1 / (4 (nk + k + 1))\), eventually

\(\displaystyle n \Pi^{- 2} (A + \Pi^{- 2})^n \pi (\Pi^{\phi}, nk) + A^n \Pi^{- 2} \pi (\Pi^{\phi}, nk) \leqslant \Pi^{- 3 / 2} .\)

Status: theorems and numerical definitions. Lean: on_time, collected.

For numerical records with \(t, g, H_0 \in \mathbb{N}\), \(b, c, f, F \in \mathbb{R}\), \(K \geqslant 1\), \(c \geqslant 1\), \(F \geqslant 0\), \(f + c \leqslant - 8\), \(g \geqslant 6\), \(U = V\), \(f = 1 / 16 - U\), \(2 V + 2 \leqslant g - 1\), put \(C_d = (2 (d + 2) K)^d\). If \(b \geqslant 0\), \(L > 1\) and \(\log L \geqslant 2 (b + 2) \log q\) for \(q \in \{\alpha + 2, C_t, C_{2 t} \}\), then at \(H = \lceil (b + 2) y^2 / \log L \rceil\) eventually \((\alpha + 2)^H, C_t^H, C_{2 t}^H, H \leqslant \Pi\) and \(\Pi^{2 c + 1} + \Pi^{2 c + 2} \leqslant \Pi^{2 c + 3}\). Consequently \((\alpha + 2)^r \Pi^{- g} \leqslant \Pi^{- 4}\) for \(0 \leqslant r \leqslant H\). Such block-size bounds hold for all sufficiently large \(L\); parameters precede \(L\), which precedes \(s\). Lean: racetrack.

Put \(A_Q = (2 Q + 1)^4\) and \(R_Q (z) = e^{zA_Q}\). For \(B, \varepsilon, w \geqslant 0\), \(D \in \mathbb{N}\), define

\(\displaystyle q_0 = w \pi (B, D), \quad d_0 = wDB^D \varepsilon,\)
\(\displaystyle q_1 = d_0 + \varepsilon w^2 \pi (B, 2 D) + \varepsilon + w \varepsilon \pi (B, D),\)
\(\displaystyle q_2 = 2 q_0 d_0 + \varepsilon w^4 \pi (B, 4 D) + 2 \varepsilon + w^2 \varepsilon \pi (B, 2 D), \quad \mathcal{Q}= q_1 + q_2 + (q_0 + q_1) q_1 + q_1 q_0 .\)

For these records, if \(F \geqslant g + 10 Z + 10\), \(Z = 2 c + 8 t + 13\), eventually \(4 R_Q (e + 2) \mathcal{Q} (\Pi, \Pi^{- F}, \Pi^{2 c + 3}, 2 t) \leqslant \Pi^{- g}\). Lean: funding.

2. What it uses

3. Strategy

Square-zero markers encode allowed products; their exponential separates connected contributions. Connected-component decomposition and a Kotecký–Preiss induction control polymer sums. Switching transfers scalar coefficient bounds to markers. Algebraic Gaussian convolution controls polynomial comparisons, and separated scales absorb their fixed-degree losses.

4. Steps

  1. Construct the connected polynomial by \(C_{k + 1} = C_k + F - \text{Exp} (C_k)\): each correction removes another marker order. rolls.

  2. Remove one root per component. The extra unit in the hit exponent pays each attachment; the cardinality bound closes induction. lookup, Ac.forest.

  3. Restrict hotness to fuel components to recover the two-overlap decomposition. hot_comp, Kiln.undo.

  4. Invert switching triangularly and pass convergent models to zero. Models.exact.

  5. Apply Wick recursion inductively to moments and differences. width, differs.

  6. Compare logarithms of the scale functions and intersect the eventual inequalities, uniformly on \([s / 2, s]\). mint.

5. Where the difficulty lies

Cover subadditivity requires intersecting unions. Hotness depends on the selection. Switching needs convergence, beyond boundedness. Numerical feasibility supplies no analytic trajectory estimates.

6. Notation

\(P = (r, q, e_P, B, t_P)\) has \(r, q, B \in \mathbb{N}\), \(r \geqslant 6\), \(q \geqslant 2\), \(B \geqslant 1\), \(e_P > 0\), \(0 < t_P \leqslant 1\); \(e_P\) is its support exponent. Polynomials are complex; degrees are total and lie in \(\mathbb{N}\). \(\partial_b X_v = b_v\). \(X (s) = \bigcup_{i \in s} X_i\); \(\text{cov}_M\) is the least number of radius-\(M\) balls whose union is connected and contains the support. \(\|p\|_1\) sums absolute coefficients; \(\pi (B, d) = (2 B (d + 1))^d\). \(f \ll g\) means \(f / g \to 0\).

7. What remains unexplained

III.11 must construct restarted representations and verify analytic hypotheses along trajectories. The auxiliary regrouping and switching-cost declarations' full role remains to be traced.

Detailed notes: the full account of this unit, with proof sketches for every step, is on the detailed notes page.

Code dependencies

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, III.1.

Referenced by: III.11, III.12, VII.6, VII.4, III.10, III.13, III.2, III.15, VII.2, III.4, III.5, III.9, VII.1, VIII.1, III.8.

Lean source

Source allocation: 3,691 attributed lines; 793 line-counted declarations; 1,161 assigned declaration names. Of the assigned names, 1,097 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.

Additional source locations

Further pinned locations for the definitions and comparisons discussed in this unit.

Key statements

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.

All units · Overview of the proof · Repository

← II.1 Field labels and their lattice representativesPart II·Detailed notesIII.1 The block-spin step and the symmetric gauge fixing →