Home·The theorem·The proof in one reading·Architecture·Units·Manuscript·Glossary·FAQ·Work items·Data·Method·Provenance·Changes
← III.1 The block-spin step and the symmetric gauge fixingPart III·Detailed notesIII.3 Free lattice kernels and covariances →

III.2. Choosing the constants: the order of choice and the RG schedule

Contents

1. What this unit proves 2

2. What it uses 5

3. Strategy 6

4. Steps 6

5. Where the difficulty lies 6

6. Notation 7

7. What remains unexplained 7

Code dependencies 7

Lean source 7

Additional source locations 8

Key statements 11

Part III — Ultraviolet construction (renormalization group).

Explains the order in which constants and scales are chosen so that the renormalization estimates hold together.

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.1 · Next: III.3 →

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

Source inventory and conventions

Part: III, Ultraviolet construction (renormalization group). Lean: YM/Price/ (the constants and the level schedule, 406 lines), YM/Shoot/ (shooting, 228 lines), YM/Integration/RG05Preconstruction/, YM/Integration/NativeSelection/ and other files of YM/Integration/, YM/RG08/Calendar/, and parts of YM/RG08/, YM/RG12/, YM/FullVolume/ (in particular Roll/) and YM/TraceExtension/. Size: 18,156 lines in the source allocation.

1. What this unit proves

Proposition III.2.1 (order of choice). For simple \(G\), choose:

  1. A sheet: kernel \(K\) with branches on every patch; first stage \(j_0 \geqslant 4\); floors \(f_j \in \mathbb{N}\); stage records with a clip of relaxation \(r_j\), derivative orders \(k_j \geqslant 6\), spatial powers \(q_j \geqslant 2\), support exponents \(a_j > 0\), widths \(w_j > 0\), path counts \(\varpi_j > 0\), drift allowances \(\omega_j \geqslant 0\), column bounds \(\gamma_j \geqslant 1\), \(r_j w_j \geqslant 12\); nondecreasing \(k, q, a\).

  2. Nondecreasing odd \(\ell_j \geqslant \max (3, f_j)\), with, for \(j \geqslant j_0\),

\(\displaystyle \omega_j / \log \ell_j \leqslant \min (\kappa / 4, j^{- 2}), \qquad \log \mathcal{C}_j / \log \ell_j \leqslant j^{- 2}, \qquad \mathcal{C}_j = 1 + \varpi_j + w_j^{- 1} + \gamma_j + r_j + a_j .\)
  1. Positive nonincreasing \(\theta_j \leqslant 1 / j\) for \(j \geqslant j_0\).

  2. Unrestricted delays: real entry couplings \(\beta^{\text{ent}}_j\), real barrier-dependent entry scales \(\sigma^{\text{ent}}_j (\bar{\beta})\), natural shells \(S_j\) and exponents \(g_j\).

  3. A barrier \(\bar{\beta} > 1\), with \(\bar{\beta} \geqslant \beta^{\text{ent}}_{j_0}\), with \(x^{- 1 / 2} e^{(\log \log x)^2} \leqslant 1\) for every \(x \geqslant \bar{\beta}\).

  4. A wait: real readiness scales \(\sigma^{\text{rdy}}_j\), windows \(h_j : \mathbb{R} \to \mathbb{N}\) with \(h_j (s) / s \to 0\) as \(s \to \infty\) for each fixed \(j\), and natural \(n_{\min}\).

Delays and waits are ordered componentwise. Factors exist for every sheet; \(\theta\) can lie below any prescribed positive sequence; barriers exist above any prescribed real. Status: definitions and three existence theorems. Lean: Sheet, Rolls.exists_rolls, Protection.exists_margin, Bar.exists_above.

Proposition III.2.2 (schedule). For simple \(G\) and any III.2.1 data, there are \(J_m\), \(L_m = \ell_{J_{m + 1}}\), \(n_p = 4 (M + p + 1)\) with natural \(M\), retaining the prescribed data, such that:

Every positive integer divides \(n_p\) infinitely often. Status: theorem; divisibility variant in III.10. Lean: Wait.ready, Wait.engine.

Proposition III.2.3 (shooting). Let \(G\) be simple; fix a level system with \(\omega_{J_{m + 1}} \leqslant (\kappa / 4) \log L_m\) and a recipe prescribing paths \(P^{\mathcal{R}}_{N, m}\) and start, retention and transition tests. Admissibility requires these paths, the level budget point and clip, width \(w_{J_m}\), path count \(\varpi_{J_m}\), loop cap \(2 M_m^{\theta_{J_m}}\), allowances \(\hat{\beta}_m^{- 11 / 10}, e^{- M_m^{1 / 5}}\), and retention. Continuity requires continuous \(\beta, v\), fixed activity address sets, and differences tending to zero in summed budget masses of regular derivative bounds through budget order on the healthy domain and forced uniform bounds on all configurations.

Assume, for every \(N, p\), a continuous admissible Wilson start family on \(\mathfrak{B}_N\) with coupling equal to its parameter, \(v = 0\), density equal to the Wilson weight at that coupling, passing the start test. Assume also that every continuous admissible level-\((m + 1)\) family on any \([l, u]\) with \(l \leqslant u\), \(m < N\), with couplings in \(\mathfrak{B}_{m + 1}\), admits a continuous admissible level-\(m\) family whose density is its block-spin push-forward at the input coupling, passing the transition test, with \(| \beta' - \beta - \kappa \log L_m | \leqslant \omega_{J_{m + 1}}\).

Then for all \(N, p\) a selected history exists: admissible representations in \(\mathfrak{B}_m\), Wilson top with \(v = 0\), these transitions, \(\beta_{N, p, 0} = \bar{\beta}\), paths independent of \(p\). For every \(\varepsilon > 0\), some \(m_0\) gives \(| \beta_{N, p, m} - \hat{\beta}_m | \leqslant \varepsilon s_m\) for all \(N \geqslant m \geqslant m_0\), all \(p\). Status: conditional theorem; hypotheses discharged in III.10. Lean: Recipe.races.

Proposition III.2.4 (joint purchase). Assume simple \(G\), kernel \(K\) branched on every patch, first stage \(\geqslant 4\), nondecreasing \(k_j \geqslant \max (6, j)\), \(q_j \geqslant \max (2, j)\), \(a_j \geqslant 100\), nondecreasing \(Q : \mathbb{N} \to \mathbb{N}\) with \(Q_j \geqslant 1\), records \(\Omega, \Delta\), degree-\(300\) tableau at \(\xi\) and contraction condition sphere at \((\xi, \omega_v, \omega_t)\), with \(\xi \geqslant 2\), \(0 < \omega_v \leqslant 1\), \(\omega_t > 0\), and prediction \(\pi\) prepaid at every sheet and price map.

There exist a sheet \(s\) preserving the prescribed kernel, first stage, derivative orders, powers and support exponents, price map \(T\), real sequence \(Y\), nondecreasing \(K' \geqslant 1\) and floor \(\tau \geqslant f^s\). For every floor \(k \geqslant \tau\), factors exist; for every block-factor sequence \(\ell\) for \(s\) with floors \(k\) there is \(U\); for every wage \(w \geqslant U\), postage rate and exchange datum there are post-\(L\) costs; for every positive \(\theta^0\) there is protection \(\theta \leqslant \theta^0\); for every menu, auxiliary record and delay \(T_0\) there is \(V \geqslant T_0\); for every \(V' \geqslant V\), barrier and wait \(W\) there is \(Z \geqslant W\); for every \(W' \geqslant Z\) and level system meeting III.2.2 with divisible periods there are a directed datum and factor \(X\) recording these choices, satisfying \(\pi (X)\).

Wages, exchanges, menus and auxiliary records equal the prescribed data; postage rates do not exceed the prescribed rates. The same factor satisfies the III.12 conditions Degree, Savvy, Roofing, PaysW, Spire, Wild, Tower.Sav with Canyon, Cadre, AccordV with \(Q\) (III.11), top kinetic data (III.9), amplification (III.1), and a bandage with extras \((\Omega, \Delta, b)\) (III.15). Support exponents are \(\geqslant 100\); \(e^{(a_j + 2) (2 \hspace{0.17em} \text{span}_{ij} + 13)^4} \leqslant \ell_i\) for \(j \leqslant i \leqslant j + 1\). The bandage also supplies the conditional family package Unlatched with a vacuous additional conclusion, under the family hypotheses of III.2.5. Status: theorem. Lean: richFaresWithPrediction.

Proposition III.2.5 (plentiful construction). Under III.2.4 and \(Q_i \geqslant 2\), its conclusion additionally ensures: every compatible trajectory, reference levels with \(d + 3 \leqslant J_{t (d)}\), section transfers and consecutive gauge doors satisfying family consistency, door compatibility and prescribed transfer bounds from level \(t (d)\) whenever \(t (d) \leqslant N\) give, for every extraction order, the complete weak-continuum requirement (III.11) and trace gallery (VII.2). Floors, delays and waits may increase; the prediction persists. Status: theorem. Lean: plentifulWithPrediction, Unlatched.

Proposition III.2.6 (prediction). Let \(\pi\) be a prediction. For simple \(G\), a sheet with the perceived-estimate package of detailed Lemma 4.16 at fixed tariff and scope, and \(\tau > 0\), suppose: for every choice of targets with the canonical chart, expenses with deficit \(- 7 / 16\), programs, initial wages and baton there is a floor \(u\); for every floor \(l \geqslant u\), factors, wages, duration bounds, register, postage, scouts, protection, step records with \(\varepsilon \leqslant 1 / 4\) on transitions with output stage at least the first stage, post-\(L\) costs and delay \(V\), there is \(W \geqslant V\); for every later delay, barrier, wait, pricing record, directed datum and remaining factor data, shelter and step conditions and divisible periods imply \(\pi\) on the assembled factor. Then the complete native purchase of detailed Lemma 4.17 retains \(\pi\) on the emitted factor, on level systems with divisible periods, maximizing its floor with \(u\) and inserting \(W\) before its booth and final delay. Status: theorem; used with vacuous prediction. Lean: company, Prepaid.

Proposition III.2.7 (complete run). Given simple \(G\), penalty \(b > 0\), records \(\Omega, \Delta\), custody request \(\mathcal{H}\), eventually paid \(\Lambda\), and a demand assigning every sheet and price map nonempty choices \(c\), floors \(e_c : \mathbb{N} \to \mathbb{N}\) and real arrays \(F_c (n, i), F'_c [\ell] (n, i)\), there is a run with a campaign, penalty \(b\), preselection \(s, q, c\), and floor \(n \geqslant \max (f^s, e_c)\). It uses \(s\) with floors \(n\) and the price map derived from \(q\); both arrays are paid as in III.2.8(b), using actual \(\ell\), and \(F'_c [\ell] (0, J_{m + 1}) \leqslant \hat{\beta}_m\) for all \(m\). Orders satisfy \(k_j, q_j \geqslant j\); \(\Lambda\), treasury funding, common-period funding, pillars, the funded, news and read-out records, custody and sky hold. Any universally prepaid prediction can persist. Status: theorem. Lean: selectRunWithTreasuryPeriod_preselected_quote_orders_postnumeric.

Proposition III.2.8 (persistent payments).

(a) Eventually paid means: for every sheet, factors, protection and delay \(V\), some \(X \geqslant V\) ensures the property for every delay \(W \geqslant X\), every barrier and wait, and every system meeting their entrance conditions. Conjunction preserves it.

(b) For every real array \(F (n, i)\), eventually paid are \(1 \leqslant 1 + \log \log \hat{\beta}_m\) at all levels and: for every \(n\), eventually in \(m\), \(|F (n, i) | \leqslant 1 + \log \log \hat{\beta}_m\) for all \(i \leqslant J_{m + 1}\). For every real sequence \(F\), \(F (J_{m + 1}) \leqslant \hat{\beta}_m\) at all levels is eventually paid.

(c) Every countably encoded family of requests (stage floors and costs depending on the transition, block geometry and stage record \(X_{J_m}\) at step \(m\)) has a common floor \(\text{pre}\) fixed before factors. If \(\ell_j \geqslant \text{pre} (j)\) at every stage, every delay can be enlarged so every later delay, any barrier and wait, and every admitted system satisfy, for \(\text{code} (\alpha) \leqslant J_n\), \(m \geqslant n\), \(\text{pre}_{\alpha} (J_{m + 1}) \leqslant L_m\) and step cost \(\leqslant 1 + \log (2 + \log \hat{\beta}_m)\).

(d) For simple \(G\) and every kernel, a plan fixes charge names and dictionaries of finite words and spectra for every request before factors. Above its floor, delays can be enlarged persistently to fund requests and the inverse charge from level zero. Status: definition in (a); theorems. Lean: EventuallyPaid, eventuallyPaid, minDelay, buy.

Proposition III.2.9 (analytic estimates). Norms are componentwise exponentially weighted displacement \(\ell^1\) norms, summing colour indices for multilinear kernels; colour spaces are fixed and finite, and constants may depend on them. Crosses map coarse fields to fine fields at block size \(L\). Vanishing orders concern symbols at zero momentum.

Further estimates concern the post-block Hessian (design), cubic and quartic coefficients (tanker), and trace insertion (cloche). Status: theorems under the displayed hypotheses; only statements of these supplementary estimates are included in the notes.

2. What it uses

The coupling shift with error \(\omega\) is an input to shooting, supplied by III.10 (Graces.winnow). Positivity of \(\kappa\) comes from slopeConstant_pos.

3. Strategy

Choose stage data before factors, then costs, protection, delays, barrier and wait. Maxima combine floors; persistent increases combine delays and waits. Long plateaux provide comparison windows; exponential side growth pays volume demands. Shooting narrows continuous Wilson families into successive coupling bands until the level-zero coupling equals \(\bar{\beta}\). The shooting windows are \([\bar{\beta} + (\kappa / 2) s_m, \bar{\beta} + 2 \kappa s_m] \subset \mathfrak{B}_m\).

4. Steps

  1. Exhaustion. Long plateaux meet entry and window thresholds; one period offset pays the finite first plateau. Program.exhaust, Wait.ready.

  2. Narrowing. Endpoint drift inequalities straddle the next window; first and last hitting points give a subinterval mapping entirely into it. Iterate downward. gates, narrow, steps.

  3. Running law. The coupling error is bounded by \(\sum_{k = 1}^m \omega_{J_k} = o (s_m)\) using \(J_k \to \infty\). trajectory.

  4. Diagonal payment. Raise entry couplings above double exponentials of finite partial sums of array entries. Since \(J_m \to \infty\), every fixed row is eventually included. eventuallyPaid.

  5. Composition. Combine sheet and native suppliers, joint and trace purchases and persistent payments by maxima of floors and transitivity of delay and wait bounds. Choose the schedule only after these increases, then assemble the run. sailsW, plentifulWithPrediction, engageWithCustodyFactor.

  6. Analytic gains. Move differences from vanishing kernels onto crosses: their two bounds give \(L^{- 2}\), fine-site summation gives \(L^4\), and \(d\) differences give \(L^{- d}\). Reflection cancels odd terms; local projection leaves exponentially small wrapping contributions. pullback_gain, mirror_price, unique_part.

5. Where the difficulty lies

Factor-dependent costs require later persistent thresholds. Shooting needs full-family continuity and band membership throughout each interval. The \(\kappa / 4\) drift bound separates endpoints; \(j^{- 2}\) controls accumulated error. Each row is paid eventually; the zeroth post-\(L\) row needs separate all-level payment.

6. Notation

\(m = 0\) is coarsest, \(N\) finest; \(p\) indexes volume, \(j\) stages. \(s_m = \sum_{i < m} \log L_i\), \(\kappa = 11 C_G / (24 \pi^2)\), \(\hat{\beta}_m = \bar{\beta} + \kappa s_m\), \(\mathfrak{B}_m = [\hat{\beta}_m / 2, 2 \hat{\beta}_m]\), \(M_m = e^{(\log \log \hat{\beta}_m)^2}\), \(t_m = \hat{\beta}_m^{- 1 / 2} M_m\), \(B_m = \lceil (\log \hat{\beta}_m)^2 \rceil\), \(R_m = (J_{m + 1} + 3) L_m\). In read-out, \(n, B, a_{\ast}\) are side, cover size, support exponent; \(K_{F, H}\) depends on the frame and holonomy tests. \(\text{span}_{ij}\) is transition support; a prediction is a predicate on the factor recording all choices. Stage records, protection and admitted entrance conditions are Stage, Protection and Paid.

7. What remains unexplained

Purchased predicates need comparison with their owning units. Underlying analytic suppliers and dictionaries lack independent verification here; drift normalization against III.9 remains unresolved. Odd-cancellation scenes, coefficient and readability interfaces and auxiliary parameters need fuller mathematical accounts. Supplementary Hessian, coefficient-extraction and trace-insertion proofs remain unexplained.

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, III.10, VII.2, III.12, III.7, VII.1, III.16, III.6, II.1, III.3, III.9, III.11, III.4, III.15, II.2, VIII.1, III.5, VII.6, VII.4, IV.1.

Referenced by: III.12, III.13, III.10, III.9, III.11, VIII.1, III.4, VII.1, III.15, III.6, III.14, III.1, VII.4, III.7, V.2, VII.2, II.1.

Lean source

Source allocation: 18,156 attributed lines; 2,085 line-counted declarations; 2,536 assigned declaration names. Of the assigned names, 2,491 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

← III.1 The block-spin step and the symmetric gauge fixingPart III·Detailed notesIII.3 Free lattice kernels and covariances →