Skip to content

Local nonlinear geometry and Haar realization · Part 7

Shadow Theory and the Wilson Yang–Mills Construction: A Source–Readout Reconstruction

For completeness the analytic product chart B(a, b) = log(exp a exp b) satisfies B(0, 0) = 0 and DB(0, 0)(a, b) = a + b by matrix multiplication and the local inverse relation. A bounded second derivative on a small closed ball yields

∥B(a, b) −(a + b)∥≤C∥(a, b)∥2,
∥B(a, b)∥≤C∥(a, b)∥.
Source page 100.
Searchable transcription
∥B(a, b) −(a + b)∥≤C∥(a, b)∥2,
∥B(a, b)∥≤C∥(a, b)∥.

The exponential recovers the product there. Coordinatewise application gives uniform finite-lattice sup-norm constants and a genuine left chart with this quadratic error. The fixed chart has

log 1 = 0, exp(log U) = U, ∥log U∥≤C chord(U).

The last bound follows from chart Lipschitz continuity and ∥mat(U) −1∥= cG chord(U). Choose the radius within both the Lipschitz and right-inverse neighborhoods. Set be = log Ue on requested edges and be = 0 elsewhere. Then ∥b∥∞≤Ch and its exponential agrees on the requested set; no global logarithm or continuous cutoffis needed. Finally, for unitary U,

UXU∗−X = (U −1)XU∗+ X(U∗−1).
Source page 100.
Searchable transcription
UXU∗−X = (U −1)XU∗+ X(U∗−1).

Frobenius unitary invariance and submultiplicativity bound this by 2∥U −1∥∥X∥; division by the color scale proves the physical adjoint bound. Induction using product triangle inequality and inversion invariance proves the path-chord bound, including repeated edges. The realized record retains local lookup sets, carrier words, signed endpoint shifts and chart domains. Recovery therefore preserves each local comparison and impulse bound without extending local hypotheses to the whole lattice.

The local logarithm construction has an explicit recovery domain. For its recorded edge set D and fixed chart radius ρ, let

(
log Ue,
e ∈D,
0,
e /∈D.
UD = {U : chord(Ue) < ρ (e ∈D)},
(LDU)e =
Source page 100.
Searchable transcription
(
log Ue,
e ∈D,
0,
e /∈D.
UD = {U : chord(Ue) < ρ (e ∈D)},
(LDU)e =

Choose ρ within both chart neighborhoods just specified. Its source domain is π−1(UD) and its source reader is LDπ. For the coordinate restriction RD and coordinatewise exponential E,

RDELDπ = RDπ,
LDπσ = LD.
Source page 100.
Searchable transcription
RDELDπ = RDπ,
LDπσ = LD.

These identities prove exact patch recovery and section recovery. The bound

∥LDπ(z)∥∞≤C max
e∈D chord(π(z)e)
Source page 100.
Searchable transcription
∥LDπ(z)∥∞≤C max
e∈D chord(π(z)e)

preserves the same constant. The construction is applied to the gauged physical links appearing in the local comparison. Their gauge and the set D remain recorded data. Consequently the recovered logarithm has precisely the stated local chart domain and the zero extension used by the source construction.

12.53 Curvature remainder and common non-packet envelopes

On a fixed sup-norm ball the local vertex estimate is

∥curvature(face(exp b)) −curl b∥≤C∥b∥2
∞,
Source page 100.
Searchable transcription
∥curvature(face(exp b)) −curl b∥≤C∥b∥2
∞,

with a lattice-independent constant. Indeed the analytic four-coordinate map

(a0, a1, a2, a3) 7−→curvature(exp a0 exp a1 exp(−a2) exp(−a3))

vanishes at zero and has derivative a0 + a1 −a2 −a3. Bounded-second-derivative Taylor expansion gives the quadratic remainder and a uniform linear-part bound. Substitute the actual face coordinates, each bounded by ∥b∥∞, including when periodicity identifies coordinates.

The regular-log and forced-letter assembly uses one envelope independent of sector, forest and Gaussian variable. Fix ε below one quarter of the support exponent and one eighth of each stencil width, footprint weight v ≥0 and retained order l ≥4. Choose the block side for inherited footprints, then fix the positive Gaussian floor, matrix upper prices, chart constants and remaining real-sector parameters. Assume covariance-window error of rate −1. For −1/2 < b < 0 and any forced exponent u, common nonnegative weights satisfy

| bagi(U)| ≤wi,
| farLogp(z)| ≤yp,
|efarLogp(z) −1| ≤yp.
(NF5)
Source page 101.
Searchable transcription
| bagi(U)| ≤wi,
| farLogp(z)| ≤yp,
|efarLogp(z) −1| ≤yp.
(NF5)

Their weighted footprint sums have budgets βu F and βb F , respectively. Specified footprint exponentials, overlap prices, radius charges and retained derivatives are paid before reducing to these powers.

To prove the bookkeeping, combine inherited physical, finite local and covariance-window letters
into b(0)
p . Its budget is T(x) = βF (x)s + X(x) + N(x), with −1/2 < s < b and X, N of rate −1/2.
Thus T has rate s and eventually T ≤1, 2T ≤βb
F . Evaluate uniform estimates on an empty-seed
dummy sector at the same radius and cutoffand retain their background-independent weights. Its
hypotheses hold and the estimates quantify over all real backgrounds; the weights are chosen before
the integrated Gaussian value. Nonnegative weighted sales bounds imply b(0)
≤1 individually. The
p
inequality |eq −1| ≤2a for |q| ≤a ≤1 permits yp = 2b(0)
p ; the arranged factor-two reserve pays
its budget: multiplying the sales budget by two costs at most βb
F .
Forced letters retain their
independent price. Real localization factors in [0, 1] can only decrease these bounds, proving (NF5)
uniformly for logarithms and Mayer factors.
Searchable transcription
To prove the bookkeeping, combine inherited physical, finite local and covariance-window letters
into b(0)
p . Its budget is T(x) = βF (x)s + X(x) + N(x), with −1/2 < s < b and X, N of rate −1/2.
Thus T has rate s and eventually T ≤1, 2T ≤βb
F . Evaluate uniform estimates on an empty-seed
dummy sector at the same radius and cutoffand retain their background-independent weights. Its
hypotheses hold and the estimates quantify over all real backgrounds; the weights are chosen before
the integrated Gaussian value. Nonnegative weighted sales bounds imply b(0)
≤1 individually. The
p
inequality |eq −1| ≤2a for |q| ≤a ≤1 permits yp = 2b(0)
p ; the arranged factor-two reserve pays
its budget: multiplying the sales budget by two costs at most βb
F .
Forced letters retain their
independent price. Real localization factors in [0, 1] can only decrease these bounds, proving (NF5)
uniformly for logarithms and Mayer factors.
The actual finite local types are row-energy remainders indexed by signed orientation and row,
diagonal logarithmic corrections indexed by fine bond, and Haar-chart corrections indexed by active
fine bond. Alongside inherited logarithms, the five chain types are real-plate principal log chains
with coefficient −1/2, blank-plate chains with +1/2, covariance correction chains with +1/2, affine
first corrections and residual quadratic chains. Chain length is the extra charge; the affine bond
term has charge zero.
Supports are full receipts enlarged by lB, including every lookup.
The
lifted record retains all these tags, supports and parameter choices; readout recovers the uniform
weights before field evaluation. The local four-coordinate Taylor map is recovered before restricting
periodic coordinates, so its bound and (NF5) transport without treating coincident coordinates as
independent physical variables.
Searchable transcription
The actual finite local types are row-energy remainders indexed by signed orientation and row,
diagonal logarithmic corrections indexed by fine bond, and Haar-chart corrections indexed by active
fine bond. Alongside inherited logarithms, the five chain types are real-plate principal log chains
with coefficient −1/2, blank-plate chains with +1/2, covariance correction chains with +1/2, affine
first corrections and residual quadratic chains. Chain length is the extra charge; the affine bond
term has charge zero.
Supports are full receipts enlarged by lB, including every lookup.
The
lifted record retains all these tags, supports and parameter choices; readout recovers the uniform
weights before field evaluation. The local four-coordinate Taylor map is recovered before restricting
periodic coordinates, so its bound and (NF5) transport without treating coincident coordinates as
independent physical variables.

12.54 Explicit Short letters and their local census

The row Short exponent at center o vanishes within 200B of seeds and otherwise is

−βθ [go(u) lighto(u) + extractedo(μ)],
0 ≤go(u) ≤1,
0 ≤θ ≤1.
(NF6)
Source page 101.
Searchable transcription
−βθ [go(u) lighto(u) + extractedo(μ)],
0 ≤go(u) ≤1,
0 ≤θ ≤1.
(NF6)

The diagonal term vanishes if either bond endpoint meets the 200B seed dome and otherwise equals

1
2
� joint(plateblank) −joint(plate)

e.
Source page 101.
Searchable transcription
1
2
 joint(plateblank) −joint(plate)

e.

The Haar term is the localized chart logarithm on active bonds outside the 304B seed dome, zero elsewhere. These are exponent signs; the activity is eShort −1. On an intact real sector, light equals the actual debit. Outside the dome the local slice is interior and the predictor alive. Clean plaquette bounds supply the kept-path test locally; outside, its exponential threshold is at least two, so the universal chord bound supplies it. Thus the exact row-energy split applies with the configuration’s existing choices.

A nonzero localization factor supplies the grafted-input variation bound, yielding |go lighto | ≤ J(x) of rate −3/2. Protected means bound each of at most two parent path values by A(x) of rate −1. The extracted term consists of at most two nonnegative half-squares, hence is at most A(x)2. Therefore W(x) = 2βF (x)(J(x) + A(x)2)

bounds the row contribution and has rate −1/2. Local flattening and principal-inverse comparison for real and blank plates with the same windows give a diagonal bound Ediag of rate −1/2. The Haar bound is CHaar fiery(x) of rate −1/2. Their nonnegative sum gives a common short-letter bound Eshort. The caption census and radius-(l + 1)B support containment give sales budget

ballCost((l + 1)B)(| Signed | spots(L) + 2 density(L))Eshort(x)ev ballCost(l+1).

It retains rate −1/2 and is independent of total volume. Before union, inherited regular activities use healthy-sector sup envelopes and vanish offthe healthy set. A forced activity is anchored at a point of its nonempty inventory. Incoming support budgets then imply anchored budgets. Inspected-footprint conversion pays overlap and charge weights once 4ε is below the incoming exponent and L exceeds the fixed threshold. This yields regular price βq F for q > −11/10 and forced price βu F for every prescribed u. A degree-k translated atom with coefficient rate a and aggregate magnitude rate b has envelope equal to coefficient

bound times fieryktimes fixed magnitude; its sales exponent can be any value strictly greater than
a−k/2+b. Finite signed orientations cost a fixed factor. Cubic and quartic coefficient rate one, and
loop quadratic coefficient rate zero, place all three below every exponent greater than −1/2, leaving
room for conversion and polynomial charges. True tails, including lower-scale tails, have arbitrary
power saving at positive retained cutofforder, with finite orientation and plaquette sums. Choose
an intermediate s < b for the six inherited families; the factor six is absorbed by βb
F /βs
F →∞.
This supplies the inherited part of (NF5). In the lift, seed domes, masks, chart domains, window
geometries, exact exponent signs and inherited family tags are retained. Literal readout recovers
the actual debit identity and each nonnegative census term before taking the uniform sales bound.
The inherited letters satisfy their own support estimates before they are joined to these local types.
Searchable transcription
bound times fieryktimes fixed magnitude; its sales exponent can be any value strictly greater than
a−k/2+b. Finite signed orientations cost a fixed factor. Cubic and quartic coefficient rate one, and
loop quadratic coefficient rate zero, place all three below every exponent greater than −1/2, leaving
room for conversion and polynomial charges. True tails, including lower-scale tails, have arbitrary
power saving at positive retained cutofforder, with finite orientation and plaquette sums. Choose
an intermediate s < b for the six inherited families; the factor six is absorbed by βb
F /βs
F →∞.
This supplies the inherited part of (NF5). In the lift, seed domes, masks, chart domains, window
geometries, exact exponent signs and inherited family tags are retained. Literal readout recovers
the actual debit identity and each nonnegative census term before taking the uniform sales bound.
The inherited letters satisfy their own support estimates before they are joined to these local types.

12.55 Ordered-chain census and exact chart localization

A route consists of a starting fine bond and an ordered list of successors.
Allowed hops have
endpoints within coarse distance d, so centers are within d + 2.
The successor count is Pd =
ballCost(d + 2) density(L), where density(L) = L4| Axis |. Induction gives P n
d routes of n hops,
retaining repetitions. Attach the mB halo of both endpoints at every node. If d+2 ≤mB, successive
halos meet and their union has B-cover number at most (n + 1) steady(m), with steady(m) =
ballCost(m+1). Each receipt point lies within (n+1)(mB +1) of the start. For a letter with n+u
hops, charge n and magnitude Cqn on allowed routes, the number of starting bonds for a receipt
containing a fixed x is at most ballCost((n+u+1)(mB+1)) density(L). Using n+u+1 ≤(u+1)2n
Source page 102.
Searchable transcription
A route consists of a starting fine bond and an ordered list of successors.
Allowed hops have
endpoints within coarse distance d, so centers are within d + 2.
The successor count is Pd =
ballCost(d + 2) density(L), where density(L) = L4| Axis |. Induction gives P n
d routes of n hops,
retaining repetitions. Attach the mB halo of both endpoints at every node. If d+2 ≤mB, successive
halos meet and their union has B-cover number at most (n + 1) steady(m), with steady(m) =
ballCost(m+1). Each receipt point lies within (n+1)(mB +1) of the start. For a letter with n+u
hops, charge n and magnitude Cqn on allowed routes, the number of starting bonds for a receipt
containing a fixed x is at most ballCost((n+u+1)(mB+1)) density(L). Using n+u+1 ≤(u+1)2n

gives the four-dimensional factor 16n. Including e2εn and the footprint price, the order-n sum is at

most HC(qR)n, where
Source page 103.
Searchable transcription
most HC(qR)n, where
H = ballCost((u + 1)(mB + 1)) density(L)P u
d ev(u+1) steady(m),
R = 16Pdev steady(m)+2ε.
Source page 103.
Searchable transcription
H = ballCost((u + 1)(mB + 1)) density(L)P u
d ev(u+1) steady(m),
R = 16Pdev steady(m)+2ε.

For qR ≤1/2 the sales bound is 2HC, uniformly in site and period. Every endpoint and full lookup halo is counted before shortening. Finite route sums followed by the nonnegative order sum justify the bound, as does bounding every finite subset. This independently recovers the ordered-chain support estimate used in the matrix assembly. Record the exact localization identities

γa(u, e) = χinner(at, ∥ue∥),
Γa(u, Y ) =
Y
e∈Y
γa(u, e).
Source page 103.
Searchable transcription
γa(u, e) = χinner(at, ∥ue∥),
Γa(u, Y ) =
Y
e∈Y
γa(u, e).

Both are in [0, 1]. For at > 0, ∥ue∥≤at implies γa = 1; γa ̸= 0 implies ∥ue∥≤2at and hence γ2a = 1. Thus nonzero Γa bounds every selected variation and gives Γ2a = 1 on the same set. For Haar density J, the localized logarithm is γ2a log J. If γa ̸= 0 and 2at lies in the chart, exponentiating this logarithm gives J exactly. For its all-argument bound, if γ2a = 0 the localized logarithm is zero; otherwise ∥ue∥≤4at. Positivity, analyticity and J(0) = 1 give | log J(ue)| ≤C∥ue∥≤4Cat once 4at is in the fixed neighborhood. Multiplication by the cutoffpreserves the bound, proving the uniform Haar contribution of (NF6) on one common tail. This size bound holds at all arguments, without assuming the argument is in the chart. The lift retains ordered route coordinates, complete receipts, selected edge sets and chart thresholds. Literal recovery preserves these exact cutoff implications and the density identity, as well as every term of the nonnegative census sum.

For a precise source domain, interpret the localized logarithm as the piecewise reader

(
γ2a(u, e) log J(ue),
γ2a(u, e) ̸= 0,
0,
γ2a(u, e) = 0.
la(u, e) =
Source page 103.
Searchable transcription
(
γ2a(u, e) log J(ue),
γ2a(u, e) ̸= 0,
0,
γ2a(u, e) = 0.
la(u, e) =
On the common tail 4at is within the positive chart, so the first branch is defined wherever selected.
This defines the zero extension without requiring evaluation of the chart logarithm outside its
domain. With u = pζ, put lZ
a (ζ, e) = la(pζ, e) and define γZ
a , ΓZ
a by the same composition. Then
Source page 103.
Searchable transcription
On the common tail 4at is within the positive chart, so the first branch is defined wherever selected.
This defines the zero extension without requiring evaluation of the chart logarithm outside its
domain. With u = pζ, put lZ
a (ζ, e) = la(pζ, e) and define γZ
a , ΓZ
a by the same composition. Then
lZ
a (jv, e) = la(v, e),
γZ
a (ζ, e) ̸= 0 =⇒elZ
a (ζ,e) = J((pζ)e).
Source page 103.
Searchable transcription
lZ
a (jv, e) = la(v, e),
γZ
a (ζ, e) ̸= 0 =⇒elZ
a (ζ,e) = J((pζ)e).
Finite products on the recorded edge set recover Γa componentwise. The same implication and
bound |lZ
a | ≤4Cat therefore hold on the whole realized tangent domain. These identities preserve
the actual localized density factor used in the correction exponent.
Source page 103.
Searchable transcription
Finite products on the recorded edge set recover Γa componentwise. The same implication and
bound |lZ
a | ≤4Cat therefore hold on the whole realized tangent domain. These identities preserve
the actual localized density factor used in the correction exponent.

12.56 Exact warning factors and quiet-box witnesses

The five detector addresses are a fine plaquette, two distinct coarse-bond types, a coarse site and a fine site. Their raw factors are respectively fine cutoff, predictor-noise cutoff, pin-frame mass, site-frame mass and contracted flag charge. A fixed district contains each focus and lies within one coarse step of it; warning support is its lB halo. A sixth type is the fine-bond logarithmic-cutoff complement with its two-endpoint support. A detector whose district meets the 300B seed dome is compulsory and its optional letter is zero. Otherwise the letter is exactly

Γ2a(u, plaza(haloB(district)))(rawStorm(u) −1).
(NF7)
Source page 103.
Searchable transcription
Γ2a(u, plaza(haloB(district)))(rawStorm(u) −1).
(NF7)

Fine-bond warnings are γa(u, e) −1 on active bonds outside the 304B dome, zero otherwise. Put Cstorm = 1 + | FrameTag | + | Points |. Raw detectors lie between zero and this constant: cutoffs are in [0, 1], frame masses are bounded by their finite summand counts and contracted charges are zero or one. Hence all warnings have absolute value at most Cstorm. For u = μ + β−1/2z, quietness of the radius-D box about o means

|ze,α| ≤ξt
p
β
if dist(o, addr(e)) ≤D.
(NF8)
Source page 104.
Searchable transcription
|ze,α| ≤ξt
p
β
if dist(o, addr(e)) ≤D.
(NF8)
If ∥μe∥≤Mt, the Euclidean norm bounded by the coordinate absolute sum gives ∥ue −μe∥≤nGξt
and ∥ue∥≤(M + nGξ)t, without coordinate independence.
For background guide bound A > 0, choose first τ = √qdir/32, then L so that 2(A/L2 +
τ)/√qdir < 1/4, then W so that 2(A + τ)/
√
k < W. For uniform row price Y , a concrete choice is
Source page 104.
Searchable transcription
If ∥μe∥≤Mt, the Euclidean norm bounded by the coordinate absolute sum gives ∥ue −μe∥≤nGξt
and ∥ue∥≤(M + nGξ)t, without coordinate independence.
For background guide bound A > 0, choose first τ = √qdir/32, then L so that 2(A/L2 +
τ)/√qdir < 1/4, then W so that 2(A + τ)/
√
k < W. For uniform row price Y , a concrete choice is
!
(nG + 1)−1,
τ
3nG scale−1 |Y | + 1
ξ = min
> 0.
Source page 104.
Searchable transcription
!
(nG + 1)−1,
τ
3nG scale−1 |Y | + 1
ξ = min
> 0.

This ensures nGξ ≤1 and noise row response at most τt/3. The other two row-bill contributions, nonlinear-to-linear error and protected row at μ, have rate −1 and are eventually at most τt/3 each since t has rate −1/2. Thus actual bills are at most τt. Adding the fine guide At/L2 bounds weighted curvature by (A/L2 + τ)t; curvature-to-chord conversion and the fine quota give chord below t/4. The predictor guide At and penalty k similarly give mismatch chord below Wt. Small-field tag identities make both frame masses exactly one. These statements hold outside the 30B dome on quiet radius-B boxes with the actual predictor choices. If the 2B box about a detector focus is quiet, a compulsory letter is already zero. Otherwise the focus is outside the 300B dome. Every point within B is outside the 30B dome and has a quiet B box. Thus fine and noise cutoffs are on unit plateaus and frame masses equal one. A nontrivial contracted flag would give a touched plaquette of chord larger than t whose parent is in that same neighborhood, contradicting its chord below t/4. Its charge is consequently one too, proving (NF7) vanishes. The fine-bond complement vanishes when mean and noise put its variation on the unit plateau. Therefore any nonzero optional warning has a coordinate violating (NF8) in its fixed enlarged support. This holds for each processed warning regardless of other cutoffvalues; no product of independent warning probabilities is used. The source lift retains detector type, district, compulsory mask, actual predictor tags and the coordinate witness. Literal recovery preserves each raw factor and the quiet-box implication, supplying the same witness to the joint Gaussian catastrophe estimate.

The witness transport can be stated as a set inclusion. For a processed warning i, let Ai be its recorded finite set of eligible bond–color coordinates in the enlarged support, and write Wi for its actual optional coefficient, including its compulsory mask. With all the preceding mean and threshold hypotheses retained, define

Ee,α = {z : |ze,α| > ξt
p
β}.
Source page 104.
Searchable transcription
Ee,α = {z : |ze,α| > ξt
p
β}.
The vanishing argument gives
{z : Wi(μ + β−1/2z) ̸= 0} ⊆
[
(e,α)∈Ai
Ee,α.
Source page 104.
Searchable transcription
The vanishing argument gives
{z : Wi(μ + β−1/2z) ̸= 0} ⊆
[
(e,α)∈Ai
Ee,α.

Realize the full noise input and let p read its physical variation. Taking preimages under p preserves this inclusion and the finite union exactly; composing with j recovers it because pj = I. Thus the witness address is recovered with its actual coordinate threshold. This is an inclusion for each warning coefficient, available before any joint Gaussian integration or selection of several warnings.

12.57 Explicit inherited support-hit payment

For completeness retain the source’s explicit normalization of the preceding anchor conversion. Let
p have fibers of cardinality at most P, with anchors oi, nonnegative weights bi and integer sizes di
satisfying P
oi=y bieAdi ≤γ at every fine site. Assume
Source page 105.
Searchable transcription
For completeness retain the source’s explicit normalization of the preceding anchor conversion. Let
p have fibers of cardinality at most P, with anchors oi, nonnegative weights bi and integer sizes di
satisfying P
oi=y bieAdi ≤γ at every fine site. Assume
x ∈Xi ⇒dist(x, p(oi)) ≤W(1 + di),
covB(Xi) ≤v + v′di,
Source page 105.
Searchable transcription
x ∈Xi ⇒dist(x, p(oi)) ≤W(1 + di),
covB(Xi) ≤v + v′di,

and 0 ≤gi ≤Cbi(1 + di)k, a ≥0, ε > 0, av′ + 2ε ≤A. Then, with x fixed,

X
giea covB(Xi) ≤eavγCP(2W + 1)4ε−(4+k)(4 + k)!eε
1 −e−ε
e−εm.
(NF9)
i:x∈Xi
di≥m
Source page 105.
Searchable transcription
X
giea covB(Xi) ≤eavγCP(2W + 1)4ε−(4+k)(4 + k)!eε
1 −e−ε
e−εm.
(NF9)
i:x∈Xi
di≥m

At degree j, there are at most P(2W + 1)4(1 + j)4 possible anchors. Sum their anchored budgets and use giea covB(Xi) ≤bieAdiC(1 + di)keav−2εdi.

One term of the exponential series at ε(1 + j) yields (1 + j)4+k ≤ε−(4+k)(4 + k)!eεeεj. The degree-j bound is therefore the numerator in (NF9) times e−εj. The resulting geometric tail proves (NF9). Every finite address subset obeys it, so infinitely many variants are allowed by nonnegativity. For block ratio L, parent and four residue coordinates uniquely specify a fine site, giving P = L4. For halo-lB enlargement of the parent image of Yi, put di = covM(Yi), M ≤2B. Then

v = (2(l + 6) + 1)4,
covB(Xi) ≤v + 10v
L di,
W = (l + 20)B.
Source page 105.
Searchable transcription
v = (2(l + 6) + 1)4,
covB(Xi) ≤v + 10v
L di,
W = (l + 20)B.

Each point is within W(1+di) of the parent of any chosen anchor in Yi. Absorb the extra tariffe2εdi into the old weight first, replacing exponent δ by δ−2ε. The exact reserve condition is consequently

a(10v/L) + 2ε ≤δ −2ε.
Source page 105.
Searchable transcription
a(10v/L) + 2ε ≤δ −2ε.

This retains the L4 child factor and polynomial-degree payment explicitly. For the direct geometric proof, with L the block ratio and the periodic maximum-coordinate distance used throughout, centered residues place a site within L/2 of its embedded parent, while embedded coarse distances scale by L. Thus L dist(p(x), p(y)) ≤dist(x, y) + L. (NF10)

Centered representatives ensure this also holds on winding paths. The lift retains residues, anchors, degree and old exponent as separate data; recovery therefore preserves the two successive 2ε reserves and the explicit payment rather than replacing them by an unspecified constant.

12.58 Realizing the geometric input to (NF9)

Choose a realizing M-cover A of Y with |A| = d = covM(Y ). Its centers are connected at range 2M +1. Starting at one center, repeatedly attach an unvisited center along an edge from the visited connected set and insert an out-and-back detour of length at most 2(2M + 1). This constructs a unit-step walk visiting every center with N ≤d(4M + 2). Treat the empty support separately. Sample at times LBj, including zero. Parent samples form D with LB|D| ≤N + LB. By (NF10), consecutive samples and any intervening walk-point parent versus its preceding sample are within 2B. A point of Y is within M of a cover center, so its parent is within 3B of that center’s parent for M ≤2B, B ≥1. Thus halolB p(Y ) lies within (l + 5)B of D with the required connected centers.

To replace this larger-radius cover by a B-cover, tile each radius-mB ball with the coordinate grid spaced by B in the four coordinate directions, with |zi| ≤m + 1. Integer division puts every point within B of a grid point. Grids and their center connections give a connected union of size at most |D|(2(m + 1) + 1)4; passage to torus images only reduces cardinality. For m = l + 5,

covB(halolB p(Y )) ≤|D|(2(l + 6) + 1)4.
Source page 106.
Searchable transcription
covB(halolB p(Y )) ≤|D|(2(l + 6) + 1)4.

Since 4M + 2 ≤10B, multiplication by L and the sample count yield

L covB(halolB p(Y )) ≤(L + 10d)(2(l + 6) + 1)4.
Source page 106.
Searchable transcription
L covB(halolB p(Y )) ≤(L + 10d)(2(l + 6) + 1)4.

The traversal also gives dist(x, y) ≤d(4M + 2) + 2M for x, y ∈Y . Combine this with (NF10) and the halo radius to obtain the anchor-distance bound with W = (l+20)B, completing the geometric input to (NF9). The lifted record carries the realizing center set, traversal and sampling times, not merely the resulting number. Readout recovers their torus images and incidences, so connectedness and both cardinality bounds survive periodic identifications and repeated samples.

For clarity, the map p in (NF9)–(NF10) is the lattice parent map; write it as ppar in the recovery identities here. It is distinct from the tangent readout used for variations. The finite geometric record contains fine addresses x, y, their residues and parents, the centers A, the traversal and the sample times. Realization retains those discrete data, so its address readout R satisfies

R(ppar(x)) = ppar(R(x)), R(halolB ppar(Y )) = halolB ppar(R(Y )).

The identity-frame section recovers the original record. In particular P = L4, v = (2(l + 6) + 1)4

and W = (l + 20)B are identical on recovery. First reserving the charge and then applying (NF9) consequently transports precisely the condition

δ ≥4ε + 10av/L,
Source page 106.
Searchable transcription
δ ≥4ε + 10av/L,

which is the displayed two-reserve condition rewritten without changing its order of use. This also preserves the full factorial and geometric-tail payment in (NF9), including its degree parameter k.

Source: Shadow Theory and the Wilson Yang–Mills Construction: A Source–Readout Reconstruction. Both papers belong to one Zenodo record. No external funding.