English
Let g be a density-like function on E with symmetry and regularity properties. Then the measure of the sublevel set {x : g(x) < 1} equals a gamma-normalized integral: μ({g < 1}) = ∫ e^{-g^p} / Γ(… ).
Русский
Пусть g — функция плотности на E с симметрией и регулярностью. Тогда мера подмножества {x: g(x) < 1} равна гамма-нормализованному интегралу.
LaTeX
$$$\\mu\\{ x : g(x) < 1\\} = \\mathrm{OfReal}\\left( \\frac{ \\int_E e^{-(g(x))^p} d\\mu(x) }{ \\Gamma( \\dim(E)/p + 1 ) } \\right)$$$
Lean4
theorem measure_lt_one_eq_integral_div_gamma {p : ℝ} (hp : 0 < p) :
μ {x : E | g x < 1} = .ofReal ((∫ (x : E), Real.exp (-(g x) ^ p) ∂μ) / Real.Gamma (finrank ℝ E / p + 1)) := by
-- We copy `E` to a new type `F` on which we will put the norm defined by `g`
letI F : Type _ := E
letI : NormedAddCommGroup F :=
{ norm := g
dist := fun x y => g (x - y)
dist_self := by simp only [_root_.sub_self, h1, forall_const]
dist_comm := fun _ _ => by rw [← h2, neg_sub]
dist_triangle := fun x y z => by convert h3 (x - y) (y - z) using 1; simp [F]
edist := fun x y => .ofReal (g (x - y))
edist_dist := fun _ _ => rfl
eq_of_dist_eq_zero := by convert fun _ _ h => eq_of_sub_eq_zero (h4 h) }
letI : NormedSpace ℝ F :=
{ norm_smul_le := fun _ _ ↦ h5 _ _ }
-- We put the new topology on F
letI : TopologicalSpace F := UniformSpace.toTopologicalSpace
letI : MeasurableSpace F := borel F
have : BorelSpace F :=
{ measurable_eq := rfl }
-- The map between `E` and `F` as a continuous linear equivalence
let φ :=
@LinearEquiv.toContinuousLinearEquiv ℝ _ E _ _ tE _ _ F _ _ _ _ _ _ _ _ _
(LinearEquiv.refl ℝ E : E ≃ₗ[ℝ] F)
-- The measure `ν` is the measure on `F` defined by `μ`
-- Since we have two different topologies, it is necessary to specify the topology of E
let ν : Measure F := @Measure.map E F mE _ φ μ
convert (measure_unitBall_eq_integral_div_gamma ν hp) using 1
· rw [@Measure.map_apply E F mE _ μ φ _ _ measurableSet_ball]
· congr!
simp_rw [Metric.ball, dist_zero_right]
rfl
· refine @Continuous.measurable E F tE mE _ _ _ _ φ ?_
exact @ContinuousLinearEquiv.continuous ℝ ℝ _ _ _ _ _ _ E tE _ F _ _ _ _ φ
· -- The map between `E` and `F` as a measurable equivalence
let ψ :=
@Homeomorph.toMeasurableEquiv E F tE mE _ _ _ _
(@ContinuousLinearEquiv.toHomeomorph ℝ ℝ _ _ _ _ _ _ E tE _ F _ _ _ _ φ)
-- The map `ψ` is measure preserving by construction
have : @MeasurePreserving E F mE _ ψ μ ν :=
@Measurable.measurePreserving E F mE _ ψ (@MeasurableEquiv.measurable E F mE _ ψ) _
rw [← this.integral_comp']
rfl