A cute problem I found the other month that I didn't get to share here was the following: Consider a spherical loaf of bread of radius $r$ which is cut into $n$ even slices of equal thickness, $2r/n$. Which slice has the most bread? Which slice has the most crust?
Oh, this is one of the famous properties characterizing the sphere. It's the only surface of revolution with the property that the surface area between parallel planes (perpendicular to the axis of revolution) depends only on the distance between the planes.
I've actually used this property in talks to middle-school and high-school students.
Suppose I'm using contour integration to perform some integral
The integral contains say, 20 poles on the real axis
say, 18 of these are roots of unity by virtue of some Bose-Einstein distribution type thing, and 2 of them are not
Suppose now I'm taking one contour around the 18 roots of unity
My book claims this is exactly equal to the minus the sum of the two other poles, plus an integral over the real axis of the function I'm integrating over, excluding this distribution
At any rate, I'll throw out the comment that you can often sum all the residues in the plane by computing a single residue at $\infty$. Maybe that trick is involved here — if the residue at $\infty$ happens to be $0$.
Yes sorry this is poorly phrased, what I'm trying to say is that some distribution has poles at the values 2\pi m of my integration variable, so it's like 1/(1-exp(i*x)) for example
@s.harp I get that, but shouldnt it contain $\pi i$ times the other residues? I'm integrating over the real axis now. Should I argue that I can shift these poles slightly away from the real axis?
i think what ted remakred upon is what you have to do here @1010011010 , you may also see your contour around your line as a countour around the two poles outside of the line and around the pole at ininity en.wikipedia.org/wiki/Residue_at_infinity
I could shift my distribution function by assuming the imaginary part was infinitesimally positive, then I would automatically enclose the other poles, and integrate along the real axis :-)