Extreme Peeling Method: A Cross-Disciplinary Approach

• Extreme peeling method is a cross-disciplinary strategy that iteratively removes boundary layers based on problem-specific extremality criteria, applicable in stochastic geometry, adhesive mechanics, robotics, and statistical inverse problems.
• The method’s framework leverages localized reduction techniques—such as convex hull peeling and front propagation—to derive precise asymptotic scaling laws and energy balance relations.
• Practical implementations include optimized robotic trajectories for safe wound dressing removal, volumetric field design in machining, and recursive box shrinking in survival analysis.

Extreme peeling method denotes several technically distinct peeling constructions whose common feature is progressive removal, detachment, or isolation of a boundary-defined structure. In the cited literature, the label covers convex hull peeling of Poisson clouds in simple polytopes, high-force detachment of self-adherent or viscoelastic films and tapes, electro-capillary thin-film release, adhesion-aware robotic dressing removal, field-driven volume peeling in multi-axis machining, recursive box peeling for survival bump hunting, leaf peeling on finite trees, and energy-based peeling at extremal black hole horizons (Calka et al., 2024, Li et al., 2023, Liang et al., 2024, Dazard et al., 2015, Avdonin et al., 12 May 2025, Borthwick et al., 2023).

1. Scope and recurrent structure

The collected literature suggests that the phrase does not denote a single standardized algorithm. Instead, it refers to a family of procedures in which a system is simplified or advanced by repeatedly removing an outer layer, propagating a front, or shrinking a region according to a problem-specific extremality criterion. In stochastic geometry, the extremal set is the set of convex-hull vertices. In adhesive mechanics, it is a moving peeling front or a narrow contact zone. In recursive statistics, it is a slice of an axis-aligned box. In inverse problems on trees, it is a leaf sheaf. In black-hole analysis, peeling is a hierarchy of regularity and decay at a null hypersurface rather than material detachment (Calka et al., 2024, Wilting et al., 2020, Dazard et al., 2015, Avdonin et al., 12 May 2025, Borthwick et al., 2023).

Domain	Peeled object	Governing mechanism
Random geometry	Convex-hull layers	Removal of extreme points
Adhesion and fracture	Tape, sheet, or film fronts	Energy balance, viscoelasticity, capillarity, Maxwell stress
Robotics and machining	Dressing boundary or stock volume layers	MPC with fracture; vector/scalar field layers
Statistics	Covariate-space box	Recursive peeling by survival criteria
Graph inverse problems	Sheaves of leaves	Recalculation of TW or response data
Black-hole analysis	Regularity hierarchy at a horizon	Conformal compactification and energy fluxes

A common structural pattern is localization. The relevant dynamics are concentrated near vertices of a polytope, near a peeling front, near a boundary layer, or near a null hypersurface. A second recurrent pattern is reduction: each peeling step produces a smaller object of the same type, together with updated effective data.

2. Convex hull peeling and early layers in simple polytopes

In stochastic geometry, extreme peeling is explicitly identified with convex hull peeling or onion decomposition. For a finite point set $X\subset \mathbb{R}^d$, the layers are defined recursively by

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$

and, for $n\ge 1$,

$$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$

Each boundary $\partial_n(X)$ is the $n$-th layer. The paper "Limit theory for the first layers of the random convex hull peeling in a simple polytope" studies this construction for a homogeneous Poisson point process $P_\lambda$ in a fixed simple polytope $K\subset \mathbb{R}^d$, with intensity $\lambda \mathbf{1}_K(x)\,dx$, in the regime $\lambda\to\infty$ and fixed layer index $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$0 (Calka et al., 2024).

Two classes of functionals are central. The first is the $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$1-face count

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$2

where $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$3 denotes the number of $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$4-dimensional faces of a polytope $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$5. The second is the outer defect volume

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$6

The analysis uses score representations for both face counts and volume, allowing them to be written as sums of local contributions over Poisson points. The missing-volume score $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$7 yields

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$8

with parallel variance asymptotics and convergence in probability.

The main asymptotic statement is that the first few layers have the same logarithmic growth exponents as the first hull. For any fixed $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$9 and $n\ge 1$0, there exist constants $n\ge 1$1 such that

$n\ge 1$2

$n\ge 1$3

For the defect volume, there exist $n\ge 1$4 such that

$n\ge 1$5

$n\ge 1$6

Both families satisfy central limit theorems with explicit rates (Calka et al., 2024).

The paper emphasizes a layer-independence phenomenon: for each fixed $n\ge 1$7, the exponents $n\ge 1$8 for face counts and $n\ge 1$9 for defect volume do not depend on the layer, even though the leading constants do depend on $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$0. The geometric mechanism is localization near vertices. A sandwiching theorem places the first $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$1 layers with high probability inside a thin floating-body shell

$$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$2

with

$$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$3

Near each vertex of the simple polytope, logarithmic rescaling sends the local configuration to a limiting Poisson process on $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$4 with intensity $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$5, and the limiting constants are expressed as integrals of limiting scores and two-point correlations.

The same work frames early-layer peeling as an extension of the classical smooth-versus-polytope dichotomy. For smooth convex bodies, first-layer face counts scale polynomially; for simple polytopes, they scale logarithmically. The cited results show that this dichotomy persists through the first few peeling layers, while the Calder–Smart regime addresses much deeper layers whose index scales like $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$6 (Calka et al., 2024).

3. Adhesive, viscoelastic, and ultratough peeling regimes

In mechanics, extreme peeling refers to nonclassical front interactions and strongly amplified peeling forces. A canonical example is the self-adherent loop studied in "How to unloop a self-adherent sheet" (Wilting et al., 2020). A tape folded sticky-side to sticky-side forms a loop with two contact lines, one at $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$7 and one at $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$8, separated by a contact length $$\mathrm{conv}_n(X):=\mathrm{conv}\Bigl(X\setminus \bigcup_{i=1}^{n-1}(\partial_i(X)\cap X)\Bigr),\qquad \partial_n(X):=\partial \mathrm{conv}_n(X).$$9. Pulling does not immediately open the loop. Instead, once $\partial_n(X)$0 becomes small, the contact lines interact, the loop size $\partial_n(X)$1 decreases, and a tenuous moving contact persists until a critical loop size $\partial_n(X)$2 is reached. The naive energetic condition for a vanishing contact region is

$\partial_n(X)$3

where $\partial_n(X)$4 and $\partial_n(X)$5 are the curvatures at the rebonding and peeling fronts. To capture rupture at finite $\partial_n(X)$6, the paper introduces an interaction energy

$\partial_n(X)$7

leading to coupled propagation conditions. The loop-side curvature satisfies $\partial_n(X)$8, rupture occurs when $\partial_n(X)$9, and the critical loop size obeys

$n$0

The measured critical size scales as $n$1, consistent with the theoretical exponent near $n$2 (Wilting et al., 2020).

A second line of work concerns viscoelastic amplification at low peeling angles. In "On the Afferrante-Carbone theory of ultratough peeling", the governing energy balance for an elastic tape on a viscoelastic substrate is written as

$n$3

The extreme regime occurs near critical speeds for which

$n$4

Then the quadratic elastic term is nearly canceled by viscoelastic dissipation, the peeling law becomes Rivlin-like,

$n$5

and the apparent toughness can become very large at small $n$6. The same paper stresses that this is an angle-sensitive on–off mechanism and that experimental or numerical verification is needed (Ciavarella, 2020).

A related transient setting appears in "Modelling the non-steady peeling of viscoelastic tapes", which studies V-shaped peeling of a viscoelastic tape on a rigid substrate under constant front velocity, constant force, or constant tip velocity (Ceglie et al., 2024). The geometry is governed by

$n$7

and the power balance is

$n$8

Adhesion is modeled by

$n$9

The results show that, for a given applied force, peeling can be prevented, start and stop after some while, or endlessly propagate, depending on the original undeformed tape geometry. The analysis targets the regime $P_\lambda$0, where the near-front response is effectively elastic with modulus $P_\lambda$1, and fast tip pulling produces a transient force maximum above the eventual steady force (Ceglie et al., 2024).

The influence of interacting fronts also appears in periodic multiple V-peeling on a compliant substrate of thickness $P_\lambda$2, periodicity $P_\lambda$3, and detached length $P_\lambda$4 (Menga et al., 2019). The total potential energy

$P_\lambda$5

is no longer translationally invariant because substrate-mediated interaction depends on $P_\lambda$6, $P_\lambda$7, and front location. A critical detached length $P_\lambda$8 can prevent unstable peeling under fixed load, while for fixed $P_\lambda$9 the critical load can be minimized by tuning $K\subset \mathbb{R}^d$0. The paper presents this as a route to either enhanced adhesion or low peel adhesion in biomedical devices such as wound dressings and transdermal drug delivery systems (Menga et al., 2019).

4. Fluid-assisted and electro-capillary peeling of films and sheets

In fluid-mediated peeling, the peeling front is controlled by lubrication stresses and elastic deformation of the overlying sheet. "Viscous control of peeling an elastic sheet by bending and pulling" studies a viscous fluid injected beneath a thin elastic sheet resting on a rigid substrate with a pre-wetting film of thickness $K\subset \mathbb{R}^d$1 (Lister et al., 2013). In the small-deflection regime $K\subset \mathbb{R}^d$2, the governing evolution is

$K\subset \mathbb{R}^d$3

with blister radius $K\subset \mathbb{R}^d$4 determined by a local bending-controlled peeling wave. The front speed obeys

$K\subset \mathbb{R}^d$5

and the similarity law is

$K\subset \mathbb{R}^d$6

For large deflections $K\subset \mathbb{R}^d$7, stretching produces membrane tension, and both peeling by bending under tension and peeling by pulling lead to

$K\subset \mathbb{R}^d$8

The paper explicitly frames these propagation laws as elastic analogues of capillary wetting laws (Lister et al., 2013).

Electro-capillary peeling replaces mechanical loading by Maxwell stress in an electrolyte-assisted interface. In "Electro-capillary peeling of thin films", a plasma-bonded thin film on ITO glass is detached by driving liquid to percolate into the bonding layer under a DC voltage as low as $K\subset \mathbb{R}^d$9–$\lambda \mathbf{1}_K(x)\,dx$0 V (Li et al., 2023). The mechanism combines electric-field-induced Maxwell stress,

$\lambda \mathbf{1}_K(x)\,dx$1

with capillary spreading and competition against adhesive stress and film tension. The critical peeling voltage scales as

$\lambda \mathbf{1}_K(x)\,dx$2

and the experimentally summarized peeling-length relations are

$\lambda \mathbf{1}_K(x)\,dx$3

The method is reported to achieve a much lower deformation and strain of film, reaching $\lambda \mathbf{1}_K(x)\,dx$4 relative reduction, with measured maximum strain $\lambda \mathbf{1}_K(x)\,dx$5. For 1.5, 2.5, 3.5, and 4.5 V, the average peeling rates are $\lambda \mathbf{1}_K(x)\,dx$6, $\lambda \mathbf{1}_K(x)\,dx$7, $\lambda \mathbf{1}_K(x)\,dx$8, and $\lambda \mathbf{1}_K(x)\,dx$9 mm/s, respectively (Li et al., 2023).

These two fluid-assisted settings differ in actuation but share a front-controlled structure. In one case, lubrication and elastic curvature determine the front law; in the other, Maxwell stress opens the wedge and enables interfacial spreading. Both use a thin interfacial layer to regularize detachment and both support explicit scaling relations for front propagation.

5. Robotic and machining implementations

In robotics, extreme peeling becomes a safety-critical control problem. "AutoPeel: Adhesion-aware Safe Peeling Trajectory Optimization for Robotic Wound Care" models dressing removal by a particle-based multi-soft-body system with XPBD, where adhesion is represented by breakable springs between dressing and skin particles (Liang et al., 2024). A spring with energy

$\lambda\to\infty$0

fractures when $\lambda\to\infty$1, after which $\lambda\to\infty$2. The current peeling front is approximated by the first and second adhesive boundary layers $\lambda\to\infty$3 and $\lambda\to\infty$4. The MPC objective is designed to stretch $\lambda\to\infty$5 while protecting $\lambda\to\infty$6: $\lambda\to\infty$7 This is combined with a non-penetration penalty based on a signed distance function and a trajectory smoothness term. On two wound-dressing phantoms, the method reduced maximum and mean deformations relative to vertical and arc baselines. For WD1, $\lambda\to\infty$8 was reduced to $\lambda\to\infty$9 mm and $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$00 to $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$01 mm; for WD2, $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$02 was reduced to $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$03 mm and $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$04 to $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$05 mm (Liang et al., 2024).

In manufacturing, peeling is volumetric rather than interfacial. "Vector Field Based Volume Peeling for Multi-Axis Machining" represents the removable volume

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$06

by a tetrahedral mesh and seeks a scalar field $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$07 whose iso-surfaces

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$08

serve as working surfaces for rough machining (Dutta et al., 2023). The construction begins with an optimized tetrahedron-wise vector field $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$09 aligned with user-specified anchor vectors, using

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$10

To make the field realizable as a scalar gradient, the method iteratively enforces irrotationality through Helmholtz–Hodge decomposition and the discrete Poisson problem

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$11

The integrability defect is measured by

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$12

The pipeline also detects and repairs singularities of four types to avoid inaccessible floating volumes. Physical machining experiments were reported on a mannequin, cup, torus, freeform surface, and impeller. Compared with planar roughing, the curved peeling strategy reduced maximum depth and average depth variation; for the mannequin, the reported values were $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$13 mm versus $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$14 mm for maximum depth and $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$15 mm versus $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$16 mm for average variation (Dutta et al., 2023).

Both robotic and machining formulations turn peeling into a constrained optimization problem. In the robotic case, the state space is deformable adhesion with fracture thresholds. In machining, it is a volumetric field-design problem with accessibility and topological constraints. In both, the front or layer is not merely observed but explicitly planned.

6. Recursive peeling in survival analysis and inverse problems on trees

In statistics, peeling is an axis-aligned region-shrinking procedure. "Cross-validation and Peeling Strategies for Survival Bump Hunting using Recursive Peeling Methods" defines a nested box sequence

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$17

where each step removes a small slice from one face of the current box in covariate space (Dazard et al., 2015). For survival data with right censoring, the peeling choice is guided by a rate of increase

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$18

where $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$19 is a survival surrogate. The paper implements three peeling criteria: the log-rank test statistic, the Nelson–Aalen cumulative hazard summary, and the log-hazard ratio from a Cox model. It also introduces two cross-validation strategies, Averaged Cross-Validation and Combined Cross-Validation, and emphasizes replicated cross-validation. The recommended optimizer in high-dimensional settings is the concordance error rate. In the WIHS clinical example, the cross-validated optimal peeling length was $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$20, and the final high-risk subgroup was characterized by

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$21

with highly significant survival separation and stepwise log-rank permutation p-values $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$22 from steps 1 through 5 (Dazard et al., 2015).

In graph inverse problems, peeling is combinatorial and recursive. "On inverse dynamical and spectral problems for the wave and Schrödinger equations on finite trees. The leaf peeling method" studies a finite metric tree $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$23 with boundary $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$24 equal to the set of leaves (Avdonin et al., 12 May 2025). The spectral problem is

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$25

on each edge, with continuity and Kirchhoff conditions at internal vertices and Dirichlet conditions on $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$26. The associated Titchmarsh–Weyl matrix $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$27 maps boundary values to outward derivatives. The dynamical problem is

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$28

with boundary control on $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$29 and response operator

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$30

The peeling unit is not a single edge but a sheaf: a pre-sheaf in which all but one edges are boundary edges. After identifying a sheaf, the algorithm removes its boundary edges and recalculates inverse data for the smaller tree. In the spectral version, the updated TW matrix $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$31 is obtained from the original $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$32 by solving local Cauchy problems on the peeled edges and applying explicit update formulas such as

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$33

In the dynamical version, the new response operator is recovered from convolution identities, singular–regular decompositions of kernels into delta and delta-derivative parts, and stepwise Volterra equations. The method requires a finite tree, standard Kirchhoff conditions, and sufficiently large observation time in the dynamical setting (Avdonin et al., 12 May 2025).

These two recursive settings have a shared logic. A large, heterogeneous object is reduced by repeated elimination of peripheral structure, and each reduction step is accompanied by effective data update rather than full recomputation from scratch.

7. Peeling at extremal black hole horizons and comparative perspective

In geometric analysis, peeling denotes a hierarchy of decay and regularity rather than a physical removal process. "Peeling at extreme black hole horizons" studies solutions of the wave equation near extremal horizons and recasts peeling as finiteness of energy fluxes of derivative hierarchies through a null hypersurface (Borthwick et al., 2023). On the exterior of extreme Reissner–Nordström, one considers the conformally rescaled metric $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$34, $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$35, and the rescaled field $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$36. Future null infinity $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$37 becomes a degenerate Killing horizon of $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$38, and the Couch–Torrence inversion

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$39

is an isometry of the compactified exterior $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$40, exchanging the future horizon $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$41 with $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$42.

For the conformal wave equation, peeling at order $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$43 at $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$44 is characterized by finiteness of fluxes of $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$45 with $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$46; by the inversion, the same criterion defines peeling at order $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$47 at the extremal horizon. The horizon version is expressed through energy fluxes across $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$48 and $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$49, and equivalent initial-data norms involve the operator

$$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$50

The paper extends the result to massive fields and to a class of spherically symmetric extremal horizons satisfying a condition $$\mathrm{conv}_1(X):=\mathrm{conv}(X),\qquad \partial_1(X):=\partial \mathrm{conv}_1(X),$$51 in the image metric (Borthwick et al., 2023).

Across the cited domains, the word peeling therefore has multiple precise meanings. In stochastic geometry it is literal removal of extreme points; in adhesive mechanics it is front propagation under fracture, capillarity, or viscoelasticity; in robotics and machining it is controlled trajectory or layer generation; in statistics it is recursive box shrinkage; in graph inverse problems it is removal of leaf sheaves; and in black-hole analysis it is an energy-based regularity hierarchy. The shared mathematical theme is iterative boundary reduction under a local extremality law. A plausible implication is that “extreme peeling method” is best treated as a cross-disciplinary schema rather than a single theory.