Documentation

LeanPool.ViscositySolutionTheory.ViscositySolns.Analysis.SemiconvexJensen.Jensen.VolumeEstimates

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

Jensen Contact-Set Theorem (VolumeEstimates) #

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.

theorem ViscositySolns.exists_pos_radius_ball_subset_neg_gradient_image_jensenContactSet {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta : ℝ} {G : Point n → Point n} (hr : 0 ≤ r) (hdelta : 0 < delta) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hfSphere : ContinuousOn f (Metric.sphere x0 r)) (hstrict : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x < f x0) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) :
∃ (rho : ℝ), 0 < rho ∧ Metric.ball 0 rho ⊆ (fun (z : Point n) => -G z) '' JensenContactSet f x0 r delta

The gradient image of the Jensen contact set contains a ball around 0.

In quantified mathematical form, under the hypotheses of the preceding contact-point selection theorem, if delta > 0, then there exists rho > 0 such that every q ∈ ball 0 rho can be written as q = -G z for some z ∈ JensenContactSet f x0 r delta.

theorem ViscositySolns.exists_pos_radius_ball_subset_neg_gradient_image_jensenGlobalContactSet {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta : ℝ} {G : Point n → Point n} (hr : 0 ≤ r) (hdelta : 0 < delta) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hfSphere : ContinuousOn f (Metric.sphere x0 r)) (hstrict : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x < f x0) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) :
∃ (rho : ℝ), 0 < rho ∧ Metric.ball 0 rho ⊆ (fun (z : Point n) => -G z) '' JensenGlobalContactSet f x0 r delta

The gradient image of the closed-ball global contact set contains a ball around 0.

In quantified mathematical form, under the same hypotheses as exists_pos_radius_ball_subset_neg_gradient_image_jensenContactSet, if delta > 0, then there exists rho > 0 such that every q ∈ ball 0 rho can be written as q = -G z for some z ∈ JensenGlobalContactSet f x0 r delta.

theorem ViscositySolns.ball_subset_neg_gradient_image_jensenGlobalContactSet_of_boundary_gap {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta gamma : ℝ} {G : Point n → Point n} (hr : 0 ≤ r) (hdelta : 0 < delta) (hgamma : 0 < gamma) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hgap : ∀ y ∈ Metric.closedBall x0 r, dist y x0 = r → f y ≤ f x0 - gamma) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) :
Metric.ball 0 (min (delta / 2) (gamma / (2 * (↑n * r + 1)))) ⊆ (fun (z : Point n) => -G z) '' JensenGlobalContactSet f x0 r delta

The explicit boundary-gap radius used in the smooth proof of Jensen's lemma.

In quantified mathematical form, assume every boundary point of closedBall x0 r satisfies f x ≤ f x0 - gamma, with gamma > 0. If delta > 0, then every vector in

ball 0 (min (delta / 2) (gamma / (2 * ((n : Real) * r + 1))))

is equal to -G z for some z ∈ JensenGlobalContactSet f x0 r delta.

theorem ViscositySolns.volume_pos_of_ball_subset_image_of_differentiableOn {n : ℕ} {S : Set (Point n)} {G : Point n → Point n} {c : Point n} {rho : ℝ} (hG : DifferentiableOn ℝ G S) (hrho : 0 < rho) (hsubset : Metric.ball c rho ⊆ G '' S) :

A measure-theoretic consequence used in the smooth part of Jensen's lemma.

In quantified mathematical form, let S ⊆ R^n, let G : R^n -> R^n, and assume that G is differentiable at every point of S, relative to S. If there are c ∈ R^n and rho > 0 such that ball c rho ⊆ G '' S, then S has positive Lebesgue measure.

The proof uses the standard theorem, already in mathlib, that a differentiable map from R^n to R^n sends every Lebesgue-null set to a Lebesgue-null set.

theorem ViscositySolns.volume_ball_div_le_of_ball_subset_image_of_image_le_mul {n : ℕ} {S : Set (Point n)} {G : Point n → Point n} {c : Point n} {rho : ℝ} {M : ENNReal} (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hsubset : Metric.ball c rho ⊆ G '' S) (himage : MeasureTheory.volume (G '' S) ≤ M * MeasureTheory.volume S) :

A quantitative image-measure lower bound.

In quantified mathematical form, let S ⊆ R^n, let G : R^n -> R^n, and let M ∈ [0,∞] satisfy M ≠ 0 and M ≠ ∞. If ball c rho ⊆ G '' S and volume (G '' S) ≤ M * volume S, then volume (ball c rho) / M ≤ volume S.

theorem ViscositySolns.volume_pos_of_ball_subset_image_of_image_le_mul {n : ℕ} {S : Set (Point n)} {G : Point n → Point n} {c : Point n} {rho : ℝ} {M : ENNReal} (hrho : 0 < rho) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hsubset : Metric.ball c rho ⊆ G '' S) (himage : MeasureTheory.volume (G '' S) ≤ M * MeasureTheory.volume S) :

A positive-measure consequence of the quantitative image-measure lower bound.

In quantified mathematical form, under the hypotheses of volume_ball_div_le_of_ball_subset_image_of_image_le_mul, if rho > 0, then S has positive Lebesgue measure.

theorem ViscositySolns.volume_ball_div_le_of_ball_subset_image_of_hasFDerivWithinAt_det_le {n : ℕ} {S : Set (Point n)} {G : Point n → Point n} {G' : Point n → Point n →L[ℝ] Point n} {c : Point n} {rho : ℝ} {M : ENNReal} (hSmeas : MeasurableSet S) (hG' : ∀ x ∈ S, HasFDerivWithinAt G (G' x) S x) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hdet : ∀ x ∈ S, ENNReal.ofReal |(G' x).det| ≤ M) (hsubset : Metric.ball c rho ⊆ G '' S) :

A quantitative lower bound from the Jacobian image-measure estimate.

In quantified mathematical form, let S ⊆ R^n, let G : R^n -> R^n, and let G' x be the Fréchet derivative of G at x relative to S for every x ∈ S. Suppose S is measurable, ball c rho ⊆ G '' S, and ofReal |det (G' x)| ≤ M for every x ∈ S, where M ≠ 0 and M ≠ ∞. Then volume (ball c rho) / M ≤ volume S.

theorem ViscositySolns.volume_explicit_ball_div_le_jensenGlobalContactSet_of_boundary_gap_det_le {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta gamma : ℝ} {G : Point n → Point n} {G' : Point n → Point n →L[ℝ] Point n} {M : ENNReal} (hr : 0 ≤ r) (hdelta : 0 < delta) (hgamma : 0 < gamma) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hgap : ∀ y ∈ Metric.closedBall x0 r, dist y x0 = r → f y ≤ f x0 - gamma) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) (hContactMeas : MeasurableSet (JensenGlobalContactSet f x0 r delta)) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hG' : ∀ z ∈ JensenGlobalContactSet f x0 r delta, HasFDerivWithinAt (fun (w : Point n) => -G w) (G' z) (JensenGlobalContactSet f x0 r delta) z) (hdet : ∀ z ∈ JensenGlobalContactSet f x0 r delta, ENNReal.ofReal |(G' z).det| ≤ M) :
MeasureTheory.volume (Metric.ball 0 (min (delta / 2) (gamma / (2 * (↑n * r + 1))))) / M ≤ MeasureTheory.volume (JensenGlobalContactSet f x0 r delta)

Quantitative closed-ball global contact-set lower bound with an explicit boundary-gap radius.

In quantified mathematical form, assume every boundary point of closedBall x0 r satisfies f x ≤ f x0 - gamma, with gamma > 0. Let rho = min (delta / 2) (gamma / (2 * ((n : Real) * r + 1))). If the closed-ball global contact set is measurable, if z ↦ -G z has derivative G' z at every point of this contact set relative to the contact set, and if ofReal |det (G' z)| ≤ M on the contact set, where M ≠ 0 and M ≠ ∞, then

volume (ball 0 rho) / M ≤ volume (JensenGlobalContactSet f x0 r delta).

theorem ViscositySolns.eventually_boundary_gap_of_uniformOn_closedBall {n : ℕ} {f : Point n → ℝ} {fseq : ℕ → Point n → ℝ} {x0 : Point n} {r gamma : ℝ} (hr : 0 ≤ r) (hgamma : 0 < gamma) (hgap : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x ≤ f x0 - gamma) (hunif : ∀ (eps : ℝ), 0 < eps → ∃ (N : ℕ), ∀ (m : ℕ), N ≤ m → ∀ z ∈ Metric.closedBall x0 r, |fseq m z - f z| ≤ eps) :
∃ (N : ℕ), ∀ (m : ℕ), N ≤ m → ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → fseq m x ≤ fseq m x0 - gamma / 2

Uniform convergence preserves a positive boundary gap after discarding finitely many terms.

In quantified mathematical form, assume every boundary point of closedBall x0 r satisfies f x ≤ f x0 - gamma, with gamma > 0, and assume f_m -> f uniformly on closedBall x0 r. Then there exists N such that for every m ≥ N and every boundary point x,

f_m x ≤ f_m x0 - gamma / 2.

theorem ViscositySolns.volume_pos_of_ball_subset_image_of_hasFDerivWithinAt_det_le {n : ℕ} {S : Set (Point n)} {G : Point n → Point n} {G' : Point n → Point n →L[ℝ] Point n} {c : Point n} {rho : ℝ} {M : ENNReal} (hSmeas : MeasurableSet S) (hG' : ∀ x ∈ S, HasFDerivWithinAt G (G' x) S x) (hrho : 0 < rho) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hdet : ∀ x ∈ S, ENNReal.ofReal |(G' x).det| ≤ M) (hsubset : Metric.ball c rho ⊆ G '' S) :

A positive-measure consequence of the Jacobian image-measure estimate.

In quantified mathematical form, under the hypotheses of volume_ball_div_le_of_ball_subset_image_of_hasFDerivWithinAt_det_le, if rho > 0, then S has positive Lebesgue measure.

theorem ViscositySolns.measure_limsup_le_of_forall_measure_le {α : Type u_1} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {E : ℕ → Set α} {B K : Set α} {c : ENNReal} (hEmeas : ∀ (m : ℕ), MeasurableSet (E m)) (hBfinite : μ B ≠ ⊤) (hsubB : ∀ (m : ℕ), E m ⊆ B) (hlimsup : ⋂ (n : ℕ), ⋃ (m : ℕ), ⋃ (_ : n ≤ m), E m ⊆ K) (hlower : ∀ (m : ℕ), c ≤ μ (E m)) :
c ≤ μ K

Measure lower bounds pass from a sequence of sets to its limsup.

In quantified mathematical form, let E m be a sequence of measurable subsets of a measure space, and suppose all E m are contained in a set B with finite measure. If a set K contains

⋂ n, ⋃ m, ⋃ (_ : n ≤ m), E m,

and if c ≤ μ (E m) for every m, then c ≤ μ K.

theorem ViscositySolns.measure_pos_of_limsup_subset_of_forall_measure_ge {α : Type u_1} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {E : ℕ → Set α} {B K : Set α} {c : ENNReal} (hc : 0 < c) (hEmeas : ∀ (m : ℕ), MeasurableSet (E m)) (hBfinite : μ B ≠ ⊤) (hsubB : ∀ (m : ℕ), E m ⊆ B) (hlimsup : ⋂ (n : ℕ), ⋃ (m : ℕ), ⋃ (_ : n ≤ m), E m ⊆ K) (hlower : ∀ (m : ℕ), c ≤ μ (E m)) :
0 < μ K

Positive measure passes from a sequence of uniformly positive measurable sets to any set containing their limsup.

In quantified mathematical form, under the hypotheses of measure_limsup_le_of_forall_measure_le, if 0 < c, then 0 < μ K.

theorem ViscositySolns.volume_pos_jensenContactSet_of_limsup_contact_sets {n : ℕ} {f : Point n → ℝ} {fseq : ℕ → Point n → ℝ} {x0 : Point n} {r delta : ℝ} {c : ENNReal} (hc : 0 < c) (hmeas : ∀ (m : ℕ), MeasurableSet (JensenContactSet (fseq m) x0 r delta)) (hlimsup : ⋂ (n_1 : ℕ), ⋃ (m : ℕ), ⋃ (_ : n_1 ≤ m), JensenContactSet (fseq m) x0 r delta ⊆ JensenContactSet f x0 r delta) (hlower : ∀ (m : ℕ), c ≤ MeasureTheory.volume (JensenContactSet (fseq m) x0 r delta)) :

The limsup measure passage specialized to Jensen contact sets.

In quantified mathematical form, let f_m : R^n -> R be a sequence of functions and let f : R^n -> R. Suppose there exists c > 0 such that, for every natural number m, the set JensenContactSet (f_m) x0 r delta is measurable and has Lebesgue measure at least c. If

⋂ n, ⋃ m, ⋃ (_ : n ≤ m), JensenContactSet (f_m) x0 r delta

is contained in JensenContactSet f x0 r delta, then JensenContactSet f x0 r delta has positive Lebesgue measure.

theorem ViscositySolns.volume_pos_jensenContactSet_of_differentiableOn_neg_gradient {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta : ℝ} {G : Point n → Point n} (hr : 0 ≤ r) (hdelta : 0 < delta) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hfSphere : ContinuousOn f (Metric.sphere x0 r)) (hstrict : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x < f x0) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) (hG : DifferentiableOn ℝ (fun (z : Point n) => -G z) (JensenContactSet f x0 r delta)) :

Positive measure of the Jensen contact set from differentiability of the gradient image map.

In quantified mathematical form, assume the hypotheses which give contact points for all sufficiently small linear perturbations, and assume that f is differentiable on ball x0 r with gradient G. If the map z ↦ -G z is differentiable at every point of JensenContactSet f x0 r delta, relative to that contact set, then JensenContactSet f x0 r delta has positive Lebesgue measure.

theorem ViscositySolns.exists_pos_radius_volume_ball_div_le_jensenContactSet_of_neg_gradient_det_le {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta : ℝ} {G : Point n → Point n} {G' : Point n → Point n →L[ℝ] Point n} {M : ENNReal} (hr : 0 ≤ r) (hdelta : 0 < delta) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hfSphere : ContinuousOn f (Metric.sphere x0 r)) (hstrict : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x < f x0) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) (hContactMeas : MeasurableSet (JensenContactSet f x0 r delta)) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hG' : ∀ z ∈ JensenContactSet f x0 r delta, HasFDerivWithinAt (fun (w : Point n) => -G w) (G' z) (JensenContactSet f x0 r delta) z) (hdet : ∀ z ∈ JensenContactSet f x0 r delta, ENNReal.ofReal |(G' z).det| ≤ M) :
∃ (rho : ℝ), 0 < rho ∧ MeasureTheory.volume (Metric.ball 0 rho) / M ≤ MeasureTheory.volume (JensenContactSet f x0 r delta)

Quantitative Jensen contact-set lower bound from a determinant bound for the negative-gradient map.

In quantified mathematical form, assume the hypotheses which give contact points for all sufficiently small linear perturbations, and let G be the gradient of f on ball x0 r. Suppose that JensenContactSet f x0 r delta is measurable, that z ↦ -G z has derivative G' z at every contact point relative to the contact set, and that ofReal |det (G' z)| ≤ M at every contact point, where M ≠ 0 and M ≠ ∞. Then there exists rho > 0 such that volume (ball 0 rho) / M ≤ volume (JensenContactSet f x0 r delta).

theorem ViscositySolns.exists_pos_radius_volume_ball_div_le_jensenGlobalContactSet_of_neg_gradient_det_le {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta : ℝ} {G : Point n → Point n} {G' : Point n → Point n →L[ℝ] Point n} {M : ENNReal} (hr : 0 ≤ r) (hdelta : 0 < delta) (hfClosed : ContinuousOn f (Metric.closedBall x0 r)) (hfSphere : ContinuousOn f (Metric.sphere x0 r)) (hstrict : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x < f x0) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) (hContactMeas : MeasurableSet (JensenGlobalContactSet f x0 r delta)) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hG' : ∀ z ∈ JensenGlobalContactSet f x0 r delta, HasFDerivWithinAt (fun (w : Point n) => -G w) (G' z) (JensenGlobalContactSet f x0 r delta) z) (hdet : ∀ z ∈ JensenGlobalContactSet f x0 r delta, ENNReal.ofReal |(G' z).det| ≤ M) :

Quantitative Jensen lower bound for the closed-ball global contact set from a determinant bound for the negative-gradient map.

In quantified mathematical form, assume the hypotheses which give closed-ball maximizers for all sufficiently small linear perturbations, and let G be the gradient of f on ball x0 r. Suppose that JensenGlobalContactSet f x0 r delta is measurable, that z ↦ -G z has derivative G' z at every global contact point relative to the global contact set, and that ofReal |det (G' z)| ≤ M at every global contact point, where M ≠ 0 and M ≠ ∞. Then there exists rho > 0 such that volume (ball 0 rho) / M ≤ volume (JensenGlobalContactSet f x0 r delta).

theorem ViscositySolns.exists_pos_radius_volume_ball_div_le_jensenGlobalContactSet_of_continuous_gradient_det_le {n : ℕ} {f : Point n → ℝ} {x0 : Point n} {r delta : ℝ} {G : Point n → Point n} {G' : Point n → Point n →L[ℝ] Point n} {M : ENNReal} (hr : 0 ≤ r) (hdelta : 0 < delta) (hf : Continuous f) (hG : Continuous G) (hstrict : ∀ x ∈ Metric.closedBall x0 r, dist x x0 = r → f x < f x0) (hderiv : ∀ z ∈ Metric.ball x0 r, HasFDerivAt f (gradientLinearMap (G z)) z) (hM0 : M ≠ 0) (hMtop : M ≠ ⊤) (hG' : ∀ z ∈ JensenGlobalContactSet f x0 r delta, HasFDerivWithinAt (fun (w : Point n) => -G w) (G' z) (JensenGlobalContactSet f x0 r delta) z) (hdet : ∀ z ∈ JensenGlobalContactSet f x0 r delta, ENNReal.ofReal |(G' z).det| ≤ M) :

Quantitative Jensen lower bound for the closed-ball global contact set when the gradient is continuous.

In quantified mathematical form, under the smooth hypotheses used in Jensen's lemma, suppose Df = G on ball x0 r, the map G is continuous, the map z ↦ -G z has derivative G' z at every global contact point relative to the global contact set, and ofReal |det (G' z)| ≤ M on the global contact set, where M ≠ 0 and M ≠ ∞. Then there exists rho > 0 such that volume (ball 0 rho) / M ≤ volume (JensenGlobalContactSet f x0 r delta).