Documentation

LeanPool.ViscositySolutionTheory.ViscositySolns.Analysis.SemiconvexJensen.Jensen.MainTheorems

Ported for Lean Pool: imports relocated and code adapted to Lean/Mathlib v4.34.0.

Jensen Contact-Set Theorem (MainTheorems) #

Part of the localized Jensen contact-set theorem and the smooth convex approximation machinery used to prove it. Split from Jensen.lean; see the umbrella module docstring.

The continuous-extension convolution theorem.

In quantified mathematical form, suppose 0 < r < R, g is convex on closedBall x0 R, G is continuous on all of R^n, and G = g on closedBall x0 R. Choose rho = (R - r) / 2. The functions

g_m = (jensenMollifierBump n rho _ m).normed volume ⋆ G

converge uniformly to g on closedBall x0 r, are twice continuously differentiable, and are convex on closedBall x0 r.

Smooth convex approximation on interior closed balls, obtained by continuous extension and mollification.

The smooth approximation theorem needed for the mollification step in Jensen's lemma.

In quantified mathematical form, this says: if 0 < r, r < R, f : R^n -> R is coordinate semiconvex on closedBall x0 R with nonnegative constant lambda, and f x < f x0 for every x ∈ closedBall x0 r with x ≠ x0, then there are a number gamma > 0 and smooth functions f_m : R^n -> R such that:

  • every boundary point of closedBall x0 r satisfies f x ≤ f x0 - gamma;
  • f_m -> f uniformly on closedBall x0 r;
  • every f_m is twice continuously differentiable;
  • every x ↦ f_m x + (lambda / 2) |x|^2 is convex on closedBall x0 r.

This is the exact mollification input from the source proof, separated from the measure-theoretic and semijet arguments.

Equations
  • One or more equations did not get rendered due to their size.
Instances For

    Smooth convex approximation of the convexified function implies the smooth semiconvex approximation theorem used in Jensen's lemma.

    In quantified mathematical form, let g(x) = f(x) + (lambda / 2) |x|^2. If g can be approximated uniformly on closedBall x0 r by twice continuously differentiable functions g_m that are convex on closedBall x0 r, then f_m(x) = g_m(x) - (lambda / 2) |x|^2 approximates f uniformly on closedBall x0 r, each f_m is twice continuously differentiable, and each x ↦ f_m(x) + (lambda / 2) |x|^2 is convex on closedBall x0 r.

    theorem ViscositySolns.JensenGlobalContactSetLimsupApproximationPrincipleOnClosedBall.of_uniform_approximation {n : ℕ} (hdata : ∀ (f : Point n → ℝ) (x0 : Point n) (r : ℝ), 0 < r → (∃ (lambda : ℝ), CoordinateSemiconvexOn lambda (Metric.closedBall x0 r) f) → (∀ x ∈ Metric.closedBall x0 r, x ≠ x0 → f x < f x0) → ∀ (delta : ℝ), 0 < delta → ∃ (fseq : ℕ → Point n → ℝ) (c : ENNReal), 0 < c ∧ (∀ (eps : ℝ), 0 < eps → ∃ (N : ℕ), ∀ (m : ℕ), N ≤ m → ∀ z ∈ Metric.closedBall x0 r, |fseq m z - f z| ≤ eps) ∧ (∀ (m : ℕ), MeasurableSet (JensenGlobalContactSet (fseq m) x0 r (delta / 2))) ∧ ∀ (m : ℕ), c ≤ MeasureTheory.volume (JensenGlobalContactSet (fseq m) x0 r (delta / 2))) :

    It is enough to construct approximants which converge uniformly on the closed ball and whose closed-ball global contact sets have a common positive lower measure bound.

    In quantified mathematical form, suppose that in the situation of JensenGlobalContactSetLimsupApproximationPrincipleOnClosedBall, for every delta > 0 there are functions f_m : R^n -> R and a number c > 0 such that:

    • for every ε > 0, there exists N such that, for every m ≥ N and every y ∈ closedBall x0 r, |f_m(y) - f(y)| ≤ ε;
    • for every m, JensenGlobalContactSet f_m x0 r (delta / 2) is measurable;
    • for every m, the measure of JensenGlobalContactSet f_m x0 r (delta / 2) is at least c.

    Then the global-contact limsup approximation principle follows.

    theorem ViscositySolns.JensenGlobalContactSetLimsupApproximationPrincipleOnClosedBall.of_uniform_approximation_tail {n : ℕ} (hdata : ∀ (f : Point n → ℝ) (x0 : Point n) (r : ℝ), 0 < r → (∃ (lambda : ℝ), CoordinateSemiconvexOn lambda (Metric.closedBall x0 r) f) → (∀ x ∈ Metric.closedBall x0 r, x ≠ x0 → f x < f x0) → ∀ (delta : ℝ), 0 < delta → ∃ (N : ℕ) (fseq : ℕ → Point n → ℝ) (c : ENNReal), 0 < c ∧ (∀ (eps : ℝ), 0 < eps → ∃ (M : ℕ), ∀ (m : ℕ), M ≤ m → ∀ z ∈ Metric.closedBall x0 r, |fseq m z - f z| ≤ eps) ∧ (∀ (m : ℕ), N ≤ m → MeasurableSet (JensenGlobalContactSet (fseq m) x0 r (delta / 2))) ∧ ∀ (m : ℕ), N ≤ m → c ≤ MeasureTheory.volume (JensenGlobalContactSet (fseq m) x0 r (delta / 2))) :

    It is enough to prove the uniform-approximation form of the global-contact principle after discarding finitely many approximating functions.

    In quantified mathematical form, suppose that in the situation of JensenGlobalContactSetLimsupApproximationPrincipleOnClosedBall, for every delta > 0 there are a natural number N, functions f_m : R^n -> R, and a number c > 0 such that:

    • for every ε > 0, there exists M such that, for every m ≥ M and every y ∈ closedBall x0 r, |f_m(y) - f(y)| ≤ ε;
    • for every m ≥ N, JensenGlobalContactSet f_m x0 r (delta / 2) is measurable;
    • for every m ≥ N, the measure of JensenGlobalContactSet f_m x0 r (delta / 2) is at least c.

    Then the global-contact limsup approximation principle follows by replacing f_m with the sequence f_{N+m}. This matches the source proof, where the smooth estimates are required only for sufficiently small mollification parameters, equivalently for sufficiently large integer indices.

    The smooth approximation theorem implies the global-contact approximation principle used to prove Jensen's lemma.

    In quantified mathematical form, suppose 0 < r, r < R, f : R^n -> R is coordinate semiconvex on closedBall x0 R, and f x < f x0 for every x ∈ closedBall x0 r with x ≠ x0. If mollification provides smooth functions f_m converging uniformly to f on closedBall x0 r such that, for some nonnegative semiconvexity constant lambda, every x ↦ f_m x + (lambda / 2) |x|^2 is convex on closedBall x0 r, then the smooth closed-ball contact-set estimate applies to all sufficiently large m. Consequently the closed-ball global contact sets for the functions f_m have a common positive lower measure bound, and their limsup is contained in the Jensen contact set of f.

    theorem ViscositySolns.JensenContactSetLimsupApproximationPrincipleOnClosedBall.of_eventually {n : ℕ} (hdata : ∀ (f : Point n → ℝ) (x0 : Point n) (r : ℝ), 0 < r → (∃ (lambda : ℝ), CoordinateSemiconvexOn lambda (Metric.closedBall x0 r) f) → (∀ x ∈ Metric.closedBall x0 r, x ≠ x0 → f x < f x0) → ∀ (delta : ℝ), 0 < delta → ∃ (N : ℕ) (fseq : ℕ → Point n → ℝ) (c : ENNReal), 0 < c ∧ (∀ (m : ℕ), N ≤ m → MeasurableSet (JensenContactSet (fseq m) x0 r (delta / 2))) ∧ ⋂ (k : ℕ), ⋃ (m : ℕ), ⋃ (_ : k ≤ m), JensenContactSet (fseq m) x0 r (delta / 2) ⊆ JensenContactSet f x0 r delta ∧ ∀ (m : ℕ), N ≤ m → c ≤ MeasureTheory.volume (JensenContactSet (fseq m) x0 r (delta / 2))) :

    It is enough to prove the approximation principle for all sufficiently large indices.

    In quantified mathematical form, suppose that in the situation of JensenContactSetLimsupApproximationPrincipleOnClosedBall, for every delta > 0 there exist a sequence f_m, a natural number N, and a number c > 0 such that the measurability and common lower-measure bound for the contact sets with perturbation radius delta / 2 hold for every m ≥ N, and the limsup of the full sequence is contained in the original contact set with perturbation radius delta. Then the approximation principle follows by replacing f_m with the shifted sequence f_{N+m}.

    The nonsmooth approximation principle implies the localized Jensen contact-set theorem.

    In quantified mathematical form, if every strict closed-ball maximum satisfying the coordinate semiconvexity hypothesis admits the approximating sequence and uniform positive lower measure bound described in JensenContactSetLimsupApproximationPrincipleOnClosedBall, then the Jensen contact set has positive measure at every strict local maximum.

    The global-contact approximation principle implies the localized Jensen contact-set theorem.

    In quantified mathematical form, if every strict closed-ball maximum satisfying the coordinate semiconvexity hypothesis admits the approximating sequence of closed-ball global contact sets and uniform positive lower measure bound described in JensenGlobalContactSetLimsupApproximationPrincipleOnClosedBall, then the Jensen contact set has positive measure at every strict local maximum.

    The source-aligned interior global-contact approximation principle implies the localized Jensen contact-set theorem.

    In quantified mathematical form, assume that the approximation principle is available whenever the contact-set radius r is strictly smaller than a radius R on which the function is semiconvex. If f is semiconvex on closedBall x0 R and has a strict local maximum at x0, choose a positive radius r ≤ R / 2 on which the maximum is strict. Then r < R, so the interior approximation principle gives approximating closed-ball global contact sets on closedBall x0 r with a common positive lower measure bound. The limsup measure lemma gives positive measure for JensenContactSet f x0 r delta, and monotonicity in the radius gives positive measure for JensenContactSet f x0 R delta.

    Smooth convex approximation on interior closed balls implies the localized Jensen contact-set theorem.

    In quantified mathematical form, assume that every convex function on a closed ball closedBall x0 R can be uniformly approximated on each smaller closed ball closedBall x0 r, with 0 < r < R, by twice continuously differentiable functions which are convex on closedBall x0 r. Then Jensen's contact set has positive Lebesgue measure for every semiconvex function with a strict local maximum.

    The localized Jensen contact-set theorem for semiconvex functions.

    In quantified mathematical form, if f : R^n -> R is semiconvex in a neighborhood of a strict local maximum point x0, then for every delta > 0 the set of points where some linear perturbation with slope of norm less than delta has a local maximum has positive Lebesgue measure.

    theorem ViscositySolns.JensenContactSetPositiveMeasureOnClosedBallTheorem.of_forall_closedBall_strict {n : ℕ} (hsmall : ∀ (f : Point n → ℝ) (x0 : Point n) (r : ℝ), 0 < r → (∃ (lambda : ℝ), CoordinateSemiconvexOn lambda (Metric.closedBall x0 r) f) → (∀ x ∈ Metric.closedBall x0 r, x ≠ x0 → f x < f x0) → ∀ (delta : ℝ), 0 < delta → 0 < MeasureTheory.volume (JensenContactSet f x0 r delta)) :

    It suffices to prove the localized Jensen contact-set theorem for closed balls on which the center is strictly larger than every other point.

    In quantified mathematical form, suppose that the following statement has been proved: for every f, x0, and r > 0, if f is coordinate semiconvex on closedBall x0 r and f x < f x0 for every x ∈ closedBall x0 r with x ≠ x0, then JensenContactSet f x0 r δ has positive measure for every δ > 0. Then the full localized Jensen contact-set theorem follows for every strict local maximum point.