In projective geometry, a line can be viewed by its projective line, which has only one point at infinity, and if a point keeps moving to $+\infty$ it will return to $-\infty$. On the other hand many integrals are taken over the real line. I wonder if there is a connection between two concepts. Would it be an integral over a circle or something? Would this be affect if the integral is replaced with a sum over $\mathbb Z$? I understand the function under integration is a member of $\mathbb R^\mathbb R$, and the series under summation is a member of $\mathbb R^\mathbb Z$

  • $\begingroup$ Yes from $[x:y ] \mapsto [y:x]$ you see the topology of $\mathbb{P}^1(\mathbb{R})$ is of a circle and you can define an integral accordingly, for example with $\int_{\mathbb{P}^1(\mathbb{R})} f(u)du = \int_{-\pi/2}^{\pi/2} f([\tan x:1])dx$ $\endgroup$ – reuns Nov 6 '17 at 8:41

The issue is choosing a metric (alternatively, a measure) on the real projective line with respect to which the integration is to be performed. It is natural to identify the real projective line with a circle in which case the natural metric is simply the least length of an arc joining the two points, i.e., the subtending angle $\theta$ in radians.

The way to define a distance between two points $A,B$ in the projective plane is to take the representing lines $a,b$ through the origin in the plane and set the distance between them be the angle between vectors $\alpha,\beta$ spanning $a,b$ respectively. Thus one would set the distance between $A$ and $B$ to be $\arccos \alpha \cdot \beta$ where "$\cdot$" is the scalar product, if $\alpha$ and $\beta$ are unit vectors.

  • $\begingroup$ Oh, so the model of projective line depends on how the metric is defined on it? $\endgroup$ – Ooker Nov 6 '17 at 13:05
  • 1
    $\begingroup$ In projective geometry there is no distance; only relations of collinearity, concurrence, etc. If you want to integrate you have to add an additional piece of information to the picture, namely the metric and/or measure. In some cases there are fairly natural ones, as in the case of the circle. $\endgroup$ – Mikhail Katz Nov 6 '17 at 13:13
  • $\begingroup$ In the case of the metric induced from the euclidean $l^2$ norm, what exactly the circle we integrate? Can you give me an example? $\endgroup$ – Ooker Nov 6 '17 at 16:24
  • $\begingroup$ The Fourier series works for periodic functions, which can be imagined as functions in circle. Is there a connection between the two? $\endgroup$ – Ooker Nov 8 '17 at 15:07
  • 1
    $\begingroup$ To answer your question "what exactly the circle", the simplest way of putting it is to say that it is the circle parametrized by the angle $\theta$ appearing as $\arccos \alpha\cdot\beta$ as above, as $\theta$ runs from $0$ to $\pi$. $\endgroup$ – Mikhail Katz Nov 8 '17 at 15:14

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.