Integrals against peak functions #
A sequence of peak functions is a sequence of functions with average one concentrating around
a point x₀. Given such a sequence φₙ, then ∫ φₙ g tends to g x₀ in many situations, with
a whole zoo of possible assumptions on φₙ and g. This file is devoted to such results. Such
functions are also called approximations of unity, or approximations of identity.
Main results #
tendsto_setIntegral_peak_smul_of_integrableOn_of_tendsto: If a sequence of peak functionsφᵢconverges uniformly to zero away from a pointx₀, andgis integrable and continuous atx₀, then∫ φᵢ • gconverges tog x₀.tendsto_setIntegral_pow_smul_of_unique_maximum_of_isCompact_of_continuousOn: If a continuous functioncrealizes its maximum at a unique pointx₀in a compact sets, then the sequence of functions(c x) ^ n / ∫ (c x) ^ nis a sequence of peak functions concentrating aroundx₀. Therefore,∫ (c x) ^ n * g / ∫ (c x) ^ nconverges tog x₀ifgis continuous ons.tendsto_integral_comp_smul_smul_of_integrable: If a nonnegative functionφhas integral one and decays quickly enough at infinity, then its renormalizationsx ↦ c ^ d * φ (c • x)form a sequence of peak functions asc → ∞. Therefore,∫ (c ^ d * φ (c • x)) • g xconverges tog 0asc → ∞ifgis continuous at0and integrable.
Note that there are related results about convolution with respect to peak functions in the file
Mathlib/Analysis/Convolution.lean, such as MeasureTheory.convolution_tendsto_right there.
General convergent result for integrals against a sequence of peak functions #
If a sequence of peak functions φᵢ converges uniformly to zero away from a point x₀, and
g is integrable and has a limit at x₀, then φᵢ • g is eventually integrable.
If a sequence of peak functions φᵢ converges uniformly to zero away from a point x₀ and its
integral on some finite-measure neighborhood of x₀ converges to 1, and g is integrable and
has a limit a at x₀, then ∫ φᵢ • g converges to a.
Auxiliary lemma where one assumes additionally a = 0.
If a sequence of peak functions φᵢ converges uniformly to zero away from a point x₀ and its
integral on some finite-measure neighborhood of x₀ converges to 1, and g is integrable and
has a limit a at x₀, then ∫ φᵢ • g converges to a. Version localized to a subset.
If a sequence of peak functions φᵢ converges uniformly to zero away from a point x₀ and its
integral on some finite-measure neighborhood of x₀ converges to 1, and g is integrable and
has a limit a at x₀, then ∫ φᵢ • g converges to a.
Peak functions of the form x ↦ (c x) ^ n / ∫ (c y) ^ n #
If a continuous function c realizes its maximum at a unique point x₀ in a compact set s,
then the sequence of functions (c x) ^ n / ∫ (c x) ^ n is a sequence of peak functions
concentrating around x₀. Therefore, ∫ (c x) ^ n * g / ∫ (c x) ^ n converges to g x₀ if g is
integrable on s and continuous at x₀.
Version assuming that μ gives positive mass to all neighborhoods of x₀ within s.
For a less precise but more usable version, see
tendsto_setIntegral_pow_smul_of_unique_maximum_of_isCompact_of_continuousOn.
If a continuous function c realizes its maximum at a unique point x₀ in a compact set s,
then the sequence of functions (c x) ^ n / ∫ (c x) ^ n is a sequence of peak functions
concentrating around x₀. Therefore, ∫ (c x) ^ n * g / ∫ (c x) ^ n converges to g x₀ if g is
integrable on s and continuous at x₀.
Version assuming that μ gives positive mass to all open sets.
For a less precise but more usable version, see
tendsto_setIntegral_pow_smul_of_unique_maximum_of_isCompact_of_continuousOn.
If a continuous function c realizes its maximum at a unique point x₀ in a compact set s,
then the sequence of functions (c x) ^ n / ∫ (c x) ^ n is a sequence of peak functions
concentrating around x₀. Therefore, ∫ (c x) ^ n * g / ∫ (c x) ^ n converges to g x₀ if g is
continuous on s.
Peak functions of the form x ↦ c ^ dim * φ (c x) #
Consider a nonnegative function φ with integral one, decaying quickly enough at infinity.
Then suitable renormalizations of φ form a sequence of peak functions around the origin:
∫ (c ^ d * φ (c • x)) • g x converges to g 0 as c → ∞ if g is continuous at 0
and integrable.
Consider a nonnegative function φ with integral one, decaying quickly enough at infinity.
Then suitable renormalizations of φ form a sequence of peak functions around any point:
∫ (c ^ d * φ (c • (x₀ - x)) • g x converges to g x₀ as c → ∞ if g is continuous at x₀
and integrable.