Canonical: https://knowlpedia.com/convex-analysis/subquadratic-absorption/

# Absorbing a subquadratic power

A power below two can be bounded by any positive multiple of the square plus a controlled constant.

For \(0<\alpha<2\), \(A,b\ge0\), and \(\delta>0\),
\[
bA^\alpha\le\delta A^2+C_\alpha\delta^{-\alpha/(2-\alpha)}b^{2/(2-\alpha)}.
\]
This is a **subquadratic absorption estimate**, a rescaled form of [Young's inequality](https://knowlpedia.com/convex-analysis/youngs-inequality/).

## Derivation and radius factors

Apply Young with conjugate exponents \(2/\alpha\) and \(2/(2-\alpha)\), rescaling the two factors so the first term is \(\delta A^2\). Equivalently, maximize \(bA^\alpha-\delta A^2\) over \(A\ge0\). If \(b=R^{-1}\) and \(\alpha=3/2\), the remainder is \(C_\delta R^{-4}\). The strict inequality \(\alpha<2\) is what permits an arbitrarily small coefficient in front of \(A^2\).
