Canonical: https://knowlpedia.com/real-analysis/solenoidal-localization/

# Localization through a vector potential

Taking the curl after inserting a cutoff preserves divergence freedom and produces an explicit cutoff correction.

Let \(A\) be a smooth [vector potential](https://knowlpedia.com/real-analysis/vector-potential/) on \(U\subseteq\mathbb R^3\), with \(u=\nabla\times A\), and let \(\chi\) be a smooth cutoff with compact support in \(U\). Define
\[
v=\nabla\times(\chi A)
=\chi u+\nabla\chi\times A,
\]
extending \(\chi A\) by zero outside \(U\). Then \(v\) is smooth, compactly supported, and [divergence free](https://knowlpedia.com/real-analysis/divergence-free-field/) on \(\mathbb R^3\).

## Plateau and correction

Where \(\chi=1\) on a neighborhood, \(v=u\). The term \(\nabla\chi\times A\) is the cutoff correction and is supported where the cutoff varies. The identity follows by the componentwise product rule for curl, while \(\nabla\cdot v=0\) follows from the divergence-of-curl identity.

## Why the potential is needed

Multiplication alone gives \(\nabla\cdot(\chi u)=\nabla\chi\cdot u\), which is not usually zero. The construction assumes a potential on the region being localized; existence of a global potential can depend on the domain.

## Infinite asymptotic correction families

[Solenoidal asymptotic summation](https://knowlpedia.com/real-analysis/solenoidal-asymptotic-summation/) applies the same potential construction term by term with shrinking cutoffs and quantitative derivative tails.
