Canonical: https://knowlpedia.com/functional-analysis/three-dimensional-sobolev-inequality/

# Three-dimensional Sobolev inequality

The L6 norm of a whole-space H1 function is bounded by its L2 gradient norm.

For \(u\in H^1(\mathbb R^3)=W^{1,2}(\mathbb R^3)\),
\[
\|u\|_6\le C\|\nabla u\|_2.
\]
The norms are whole-space [Lebesgue norms](https://knowlpedia.com/measure-theory/lp-space/). The same estimate holds for a vector field, with a dimensional constant.

## An elementary route

For real \(f\in C_c^1(\mathbb R^3)\), the fundamental theorem of calculus gives \(|f(x,y,z)|\le A_1(y,z)=\int|\partial_1f(s,y,z)|\,ds\), and analogously \(A_2(x,z),A_3(x,y)\). Hence \(|f|^{3/2}\le(A_1A_2A_3)^{1/2}\). Integrating first in \(x\) and using Cauchy–Schwarz, then in \((y,z)\) and using it again, yields
\[
\|f\|_{3/2}\le\prod_{j=1}^3\|\partial_jf\|_1^{1/3}\le C\|\nabla f\|_1.
\]
Apply this to \(f=|u|^4\). Hölder gives \(\|u\|_6^4\le C\|u\|_6^3\|\nabla u\|_2\); divide if \(u\ne0\). Smooth compact-support approximation extends the inequality to \(H^1\).

## References

- [Hunter, Sobolev Spaces, §§3.5–3.7](https://www.math.ucdavis.edu/~hunter/pdes/ch3.pdf).
