MATHLIBANNEX / CANONICAL DECLARATION CARD

Mollification by a normalized smooth kernel

MathlibAnnex.Mollification.mollify

def

Defines convolution with a nonnegative smooth kernel of integral one.

Statement

Let and , with their coordinate sup norms and Lebesgue measure on . For a real parameter and a map , define , where is the normalized bump described below.

Definition

For , take the specified smooth bump centered at , with inner radius and outer radius . For , the definition instead uses radii and . Set The denominator is positive. This normalization followed by convolution is the complete definition. The fallback defines the operation at every real parameter; approximation uses only .

Assumptions

The definition accepts every , every real , and every function . It does not assume continuity, measurability, or integrability. Analytic properties of the convolution require their own hypotheses.

Conclusion

The output is the function . For locally integrable this is the ordinary convolution with the compactly supported continuous kernel. For arbitrary the displayed integral is understood with the source library’s total Bochner-integral convention (zero when Bochner integrability fails).

For , the same normalized bump satisfies In particular its topological support, , is contained in . The definition fixes these bumps; no formula identifying them as dilates of a single kernel is asserted. Smoothness of and convergence are separate results.

Main citations

Lean source signature (exact)

noncomputable def mollify {n N : ℕ} (ε : ℝ)
    (f : (Fin n → ℝ) → (Fin N → ℝ)) : (Fin n → ℝ) → (Fin N → ℝ) :=
  standardMollifier ε ⋆[ContinuousLinearMap.lsmul ℝ ℝ, volume] f
In the source Mathematical meaning
{n N : ℕ} The domain and output dimensions of and , both with coordinate sup norms; either may be zero.
(ε : ℝ) (f : (Fin n → ℝ) → (Fin N → ℝ)) The real parameter and arbitrary map . This definition imposes neither nor regularity or integrability of .
standardMollifier ε The normalized bump . The cited definition uses radii for , and otherwise.
ContinuousLinearMap.lsmul ℝ ℝ, volume Scalar multiplication of a vector in by a real number, and Lebesgue measure on .
In the source Mathematical meaning
standardMollifier ε ⋆[ContinuousLinearMap.lsmul ℝ ℝ, volume] f The entire RHS: convolution . It uses the total Bochner integral, which is zero when Bochner integrability fails.
: (Fin n → ℝ) → (Fin N → ℝ) The output is the function . Approximation results require their own hypotheses.
Exact surrounding binder context (separate excerpt)
noncomputable section
open Set MeasureTheory Filter
open scoped Convolution Topology ENNReal NNReal Pointwise
namespace MathlibAnnex
namespace Mollification
Exact content identity

Declaration: MathlibAnnex.Mollification.mollify

Accepted content SHA-256: b4e9e1dfaeaea191863548272df300ed393323e9048f55d7f6a64aa6e32cb4b9

Accepted source guide SHA-256: c17d8730519abe8c478ed14b412685fe5ef985644d950fc028c307151ec31063

Source commit: 437e6e46228bbb8e91211ebded349d7a30020e73

Back to top ↑