1
$\begingroup$

When is it possible to assume that for any vector $\vec{x}=(x_1,\dots, x_n)\in\mathbb{R}^n$;

$$|x_{i}| \leq \|\vec{x}\|$$

For the case in $\mathbb{R}^2$:

$$|x_2| \leq \|\vec{x}\|$$

$$\left|x_2\right| \leq x_2 \cdot \sqrt{\left(\frac{x_1}{x_2}\right)^2+1}$$

Define $\alpha = \sqrt{\left(\frac{x_{1}}{x_2}\right)^2+1}$

Then $|x_i| \leq \|\vec{x}\| \iff \alpha >1$.

Is there a more general result ? Does this reasoning hold ?

$\endgroup$

2 Answers 2

2
$\begingroup$

Let $X=\mathbb{R}^{N}$. Then, for $x\in X$, the (square of the) $2$-norm is defined as: \begin{equation*} \|x\|_{2}^{2}=\sum_{i=1}^{N}|x_{i}|^{2}\geq |x_{i}|^{2} \text{ for a given index } i \qquad x=(x_{1},\ldots,x_{N}) \end{equation*} so taking the square root gives $\|x\|_{2}\geq |x_{i}|$ for all $i\in\{1,\ldots,N\}$. Next, it is a standard result that every norm on a finite-dimensional space (like $X$ in this case) is equivalent, so given another norm, say $\|\cdot\|'$, there are constants $c_{1},c_{2}>0$ such that \begin{equation*} c_{1}\|x\|'\leq\|x\|_{2}\leq c_{2}\|x\|' \end{equation*} for all $x\in X$. In particular, it would follow that \begin{equation*} |x_{i}|\leq\|x\|_{2}\leq c_{2}\|x\|' \end{equation*} so you can bound individual components by the overall norm for any norm on $X$, up to a constant dependent on the norm (with this constant being $1$ for the $2$-norm).

$\endgroup$
2
$\begingroup$

Assuming the notation $ \Vert \vec x \Vert $ denotes the standard norm in $ \mathbb R^n $, the answer to your first question is always. The argument is as follows. Let $ x $ be in $ \mathbb R^n $ and $ x_i $ be its $i^{th} $ component. Then, \begin{align*} | x_i | & = \sqrt{ |x_i|^2 } \\ & \le \sqrt{ |x_i|^2 + \sum_{j \neq i} |x_j|^2} \\ & = \Vert \vec x \Vert . \end{align*}

The key step here is the inequality above, which relies on the fact that the function $ f(t) = \sqrt{t} $ is increasing.

The argument you provide is a good place to start for the $ \mathbb R^2 $ case, but it requires some fixes: (1) exclude the $ x_2 = 0 $ case in order to divide by $ x_2 $, (2) replace the factor $ x_2 $ with $|x_2|$ on the right (remember: $ \sqrt{z^2} = |z| $ for real $ z $), (3) replace $ \alpha > 1 $ with $ \alpha \ge 1 $, and (4) state the connection between the statements you provide. After implementing these fixes, you can then establish that the inequality is always true in $ \mathbb R^2 $ by observing that $ \alpha \ge 1 $ always holds. This final bit follows also from the fact that $ f(t) = \sqrt{t} $ is increasing: \begin{align*} \alpha & = f\left( 1 + \left(\frac{x_1}{x_2}\right)^2 \right) \\ & \ge f(1) \\ & = 1 . \end{align*}

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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