Sign up ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $Q$ be the cube $[-1,1]^{3}$ and $\pi$ be a plane in $\mathbb{R}^{3}$ that contains the origin but doesn't contain any vertex of $Q$. Suppose that $A$ is an invertible linear transformation from $\mathbb{R}^{3}$ to $\mathbb{R}^{3}$ and $A\left(Q\cap\pi\right)$ is contained in $Q$. Do there always exist vectors $v_1\in\mathbb{R}^{3}\setminus\pi$ and $v_2\in\mathbb{R}^{3}$ such that

$$\left\Vert v+tv_1\right\Vert_{\infty}\geq\left\Vert Av+tv_2\right\Vert_{\infty}.$$

share|cite|improve this question
The equations are not very readable, perhaps someone can fix? –  Igor Rivin Dec 12 '10 at 23:20
@Igor Rivin: fixed. @OP: re-tagged. –  Willie Wong Dec 12 '10 at 23:25

1 Answer 1

You are essentially asking whether a non-expanding linear map $A:(\pi,\|\cdot\|_\infty)\to(\mathbb R^3,\|\cdot\|_\infty)$ can be extended to a non-expanding linear map $B:(\mathbb R^3,\|\cdot\|_\infty)\to(\mathbb R^3,\|\cdot\|_\infty)$.

(A linear map $f:X\to Y$ between normed spaces is non-expanding if $\|f(x)\|_Y\le \|x\|_X$ for all $x\in X$. Given $v_1$ and $v_2$ as in the question, the map $B$ sending every vector of the form $v+tv_1$, where $v\in\pi$, to $Av+tv_2$, is non-expanding.)

The answer is yes, and the key is that the norm on the target space is of $\ell_\infty$ type.

Let $P_i:\mathbb R^3\to\mathbb R$ denote the $i$-th coordinate projection, $i=1,2,3$. Consider linear functions $A_i:\pi\to\mathbb R$ given by $A_i(x)=P_i(Ax)$, $x\in\pi$. Since $A(Q\cap\pi)\subset Q$, we have $|A_i(x)|\le\|x\|_\infty$ for all $x\in\pi$. By Hahn-Banach, there exists an extension $B_i:\mathbb R^3\to\mathbb R$ such that $|B_i(x)|\le\|x\|_\infty$ for all $x\in\mathbb R^3$. Let $B:\mathbb R^3\to\mathbb R^3$ be the linear map whose coordinate functions are $B_1,B_2,B_3$. Then $\|Bx\|_\infty\le\|x\|_\infty$ for all $x\in\mathbb R^3$ and $B=A$ on $\pi$.

Now let $v_1$ be any vector from $\mathbb R^3\setminus \pi$, define $v_2=Bv_1$, and we are done.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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