Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm just beginning to learn topology, and there's something that I realize has been nagging at me since roughly the eight grade.

Why does the Euclidean distance in N-dimensional space involve a bunch of squaring and square-rooting?

I understand it, mind you.

  1. In 1-space the "distance" is $\sqrt{\Delta x^2}$ is which is just $\Delta x$.
  2. In two-space, you're doing the Pythagorean theorem.
  3. In 3-space, I visualize it like this: you want to find the distance $\Delta x, \Delta y, \Delta z$ so you:

    • let $a = \sqrt{\Delta x^2 + \Delta y^2}$, $a$ is the distance along the $x, y$ plane
    • the total distance is $\sqrt{a^2 + \Delta z^2}$
    • the above expands to $\sqrt{\Delta x^2 + \Delta y^2 + \Delta z^2}$

To generalize, faced with $n$ dimensions, you pick two, find the distance along those two dimensions, plop a point down there, and repeat (until you have only one dimension left, at which point you're done.)

However, I don't understand why distance in n-space requires repeated squaring. Shouldn't there be a distance formula that requires cubing and the cube-root in three-space, or quading (is that a thing) and quad-roots in four-space, and so on? Just in the interest of symmetry, it seems weird that squares get special treatment.

I know this is a very philosophical question, but is there a way to find Euclidean distances in n-space that involves taking the nth power and nth root instead of repeatedly projecting down a dimension?

share|improve this question

2 Answers 2

up vote 11 down vote accepted

There are other metrics than the Euclidean one. In particular, you can define the "$\ell^p$ norm" for any $p \ge 1$, where the distance from $(x_1, x_2, \ldots, x_n)$ to $(y_1, y_2, \ldots, y_n)$ is $(|x_1 - y_1|^p + |x_2 - y_2|^p + \ldots + |x_n - y_n|^p)^{1/p}$. What is special about the Euclidean metric is that it allows rotations through arbitrary angles.

share|improve this answer
    
Cool! Are there any uses to such alternative definitions? Also, what do you mean by saying that one allows rotation while the other does not? How can a distance metric prevent rotation? –  Nate May 21 '12 at 5:50
3  
What Robert means is that the Euclidean distance of two points will be the same before and after you rotate them. This is not necessarily the case for other metrics. –  Miha Habič May 21 '12 at 6:01

An explanation with some algebraic flavor is through linear transformations and inner products; these inner products $\langle\cdot,\cdot\rangle$ are fundamental objects for vector spaces as they can quantitatively measure linear independence through bilinearity. On the assumption of a basis for our space we can define the inner product to be the dot product $x\cdot y=\sum_ix_iy_i$; this is the unique inner product for which $\langle e_i,e_i\rangle=1$ for each basis vector $e_i$. However, being linear in both arguments means that scalar multiplication gives $\langle \lambda v,\lambda v\rangle=\lambda^2\langle v,v\rangle$. In order to force this to be scalar multiplicative we need to set $\|v\|=\sqrt{\langle v,v\rangle}$, so that $\|\lambda v\| = |\lambda|\,\|v\|$.

So ultimately, this way of explanation boils down to: a translation-invariant metric is equivalent to having a norm on the space, and the Euclidean norm will be a special case of "measuring" the linear independence of a vector with itself, in the sense of an inner product, so the presence of the distinguished power/root of two is because one and one (a vector and its copy) make two.

share|improve this answer

Your Answer

 
discard

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.