On MathOverflow I saw this inequality. Let $E$ is a normed linear space.

$$ \|x+y\|+\|y+z\|+\|z+x\|\le\|x\|+\|y\|+\|z\|+\|x+y+z\|,\qquad\forall x,y,z\in E $$

Apparently this is always true if $E = \mathbb{R}^n$. I always thought Triangle inequality was the only foundation inequality for normed spaces. Does Hlawka inequality follow as a consequence?

The same MO question discusses potential counterexamples. Why is the space of $2 \times 2$ self-adjoint matrices not Euclidean with the following norm?

$$ ||A|| = \tfrac{1}{2}|\mathrm{tr}( A)| + \tfrac{1}{\sqrt{2}}\big|\big| A - \mathrm{tr}(A)I_{2\times 2} \big|\big|_2$$

The weighted sum of the trace and the Hilbert-Schmidt norm.

  • 2
    $\begingroup$ For references, see also here. $\endgroup$ – Dietrich Burde Nov 3 '14 at 21:20
  • 3
    $\begingroup$ The Hlawka Identity takes care of it, $\displaystyle (\|x\|+\|y\|+\|z\|-\|x+y\|-\|y+z\|-\|z+x\|+\|x+y+z\|)(\|x\|+\|y\|+\|z\| + \|x+y+z\|) = \sum\limits_{cyc} (\|x\|+\|y\| - \|x+y\|)(\|z\| - \|x+y\| + \|x+y+z\|)$ $\endgroup$ – r9m Nov 3 '14 at 21:28

This is not true for any finite dimensional liner space greater than or equal to 3.

Counterexample:in $\mathbb{R}^3$, with norm defined by $(x,y,z)_p=(|x|^p+|y|^p+|z|^p)^{1/p}$, $p\ge 3$. Take $x=(1,1,-1)$, $y=(-1,1,1)$, $z=(1,-1,1)$. Clearly the inequality holds for $p=1$ and $p=2$. It was proved that any finite two dimesional space satisfies this inequality (and so is Hlawka space).


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.