I'm trying to show that: $$\forall x,y\in\mathbb{R}^n, d(x,y)=\left(\sum_{i=1}^n(x_i-y_i)^2\right)^{1/2}$$

is a distance. However I have not proved Cauchy-Schwarz yet and I'm pretty sure I wouldn't know how to use it in this case. Can someone give me a hint on how to solve this?

Thanks a lot for helping!!!

  • $\begingroup$ I think you might be able to prove it using induction over the dimension $n$ using just the Pythagorean Theorem and an argument that any line in $\mathbb{R^{n-1}}$ is orthogonal to the axis along the coordinate in $\mathbb{R^{n}}$ that is not in $\mathbb{R^{n-1}}$ (i.e. the $n$-th coordinate). E.g. lines in $xy$-plane are all orthogonal to $z$-axis. $\endgroup$ – Marconius Jul 29 '15 at 18:45
  • $\begingroup$ @Marconius Thanks for your comment, but I am still not sure how to do it. Would you be so kind to explain your idea a little more? Thanks! :D $\endgroup$ – Guilherme Salomé Jul 29 '15 at 18:55
  • 1
    $\begingroup$ @GuilhermeSalomé,have you seen Cauchy -Schwarz proof?en.wikipedia.org/wiki/Cauchy%E2%80%93Schwarz_inequality $\endgroup$ – haqnatural Jul 29 '15 at 19:03
  • 1
    $\begingroup$ Are you allowed to use $||\mathbf{v}|| = \sqrt{\langle\mathbf{v},\mathbf{v}\rangle}$? $\endgroup$ – Marconius Jul 29 '15 at 19:28
  • 1
    $\begingroup$ see e.g. 1, 2 $\endgroup$ – 6005 Jul 29 '15 at 19:38

$$ \sum_i (x_i + \lambda y_i)^2\geqslant 0 $$ $$ \sum_i x_i^2+\lambda x_iy_i +\lambda x_iy_i+\lambda^2y_i^2 \geqslant0 $$ Let $\lambda=-(\sum_jx_jy_j)/(\sum _jy_jy_j)$ then: $$\sum_i x_i^2-\frac{(\sum_ix_iy_i)(\sum_jx_jy_j)}{(\sum _jy_jy_j)}-\frac{(\sum_ix_iy_i)(\sum_jx_jy_j)}{(\sum _jy_jy_j)}+(\frac{\sum_jx_jy_j}{\sum _jy_j^2})^2 \sum_iy_i^2 \geqslant0$$

Simplifying and adjusting the terms on both sides of the inequality yields: $$(\sum_i x_iy_i)^2 \leqslant \sum_i x_i^2 \sum_iy_i^2 \tag{cqfd!}$$ EDIT Now to the triangle inequality:

$$\sum_i(x_i+y_i)^2=\sum_i x_i^2 +2\sum_i x_iy_i + \sum_iy_i^2 \leqslant \sum_ix_i^2 +2\sqrt{\sum_ix_i^2 \sum_iy_i^2} +\sum_iy_i^2$$ $$=(\sqrt{\sum_ix_i^2} + \sqrt{\sum_iy_i^2} )^2 \tag{cqfd!}$$

| cite | improve this answer | |
  • $\begingroup$ Thanks for your answer! But how do I go from this to Minkowski? $\endgroup$ – Guilherme Salomé Jul 29 '15 at 20:22
  • $\begingroup$ In other words, how do I go from your inequality to the triangle inequality for that distance? $\endgroup$ – Guilherme Salomé Jul 29 '15 at 20:30
  • $\begingroup$ @GuilhermeSalomé I've written the triangle inequality, which is a particular case of the Minkowski inequality (to my knowledge, it needs the Holder inequality to be proven) $\endgroup$ – GeorgSaliba Jul 29 '15 at 20:56

Put $\mathbf{v} = \mathbf{x} - \mathbf{y}$


$d(x,y) = \|\mathbf{x}-\mathbf{y}\| = \|\mathbf{v}\|$


$\langle{\mathbf{v},\mathbf{v}}\rangle = \langle{\mathbf{x}-\mathbf{y},\mathbf{x}-\mathbf{y}}\rangle = \displaystyle{\sum\limits_i^n(x_i-y_i)^2}$

and so $\|v\| = \sqrt{\langle\mathbf{v},\mathbf{v}\rangle}$ implies the desired result.

| cite | improve this answer | |

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.