Two boundary points of a ball of radius 1 are joined by a curve contained in the ball and having length less than 2. Prove that the curve is contained entirely within some hemisphere of the given ball.
Problem
Source: 1974 USAMO Problem 3
Tags: geometry, 3D geometry, sphere, function
13.03.2010 08:43
Well it is actually intuitively evident I guess. Consider a function $ f$ that gives the distance between two points lying on the sphere. (automatically consider a mapping from $ f$ to $ g$ where $ g$ gives the length of the curve between the two points). So, $ AB \le OA + OB$ ($ A,B$ are points on the sphere) and $ O$ is the center of the sphere. $ AB \le 2$. But then you can say that if you want $ f$ to attain its maximum, then automatically your $ A,B$ will tend to be diametrically opposite by the equality of the skew triangle's lengths or whatever (I dont know how it is called). But then, we are given that $ AB<2$, so we can say that for equality itself, the points are diametrically opposite at the maximum, so $ f$ will make sure that $ A,B$ lie entirely within one hemisphere.So we are done.
13.03.2010 10:52
Agr_94_Math wrote: Well it is actually intuitively evident I guess. Consider a function $ f$ that gives the distance between two points lying on the sphere. (automatically consider a mapping from $ f$ to $ g$ where $ g$ gives the length of the curve between the two points). So, $ AB \le OA + OB$ ($ A,B$ are points on the sphere) and $ O$ is the center of the sphere. $ AB \le 2$. But then you can say that if you want $ f$ to attain its maximum, then automatically your $ A,B$ will tend to be diametrically opposite by the equality of the skew triangle's lengths or whatever (I dont know how it is called). But then, we are given that $ AB < 2$, so we can say that for equality itself, the points are diametrically opposite at the maximum, so $ f$ will make sure that $ A,B$ lie entirely within one hemisphere.So we are done. That's not rigorous at all. Why would you want $ f$ to attain its maximum?
13.03.2010 12:45
Well it is closed continuous and bounded region over which $ f$ is defined. So automatically it attains its global maximum due to Weirstrass theorem.
23.02.2012 14:35
Well, let's consider the most disadvantageous case, when the curve is an arc of great circle. i.e. belonging to the intersection of the sphere with a plane passing through its center, let $A, B$ be its extremities and draw an equatorial plane containing $A$ and being perpendicular onto the plane OAB. One of the two hemispheres thus constructed contains $AB$ entirely, since the length of the arc $AB$ is lesser that a semicircle. Best regards, sunken rock
13.04.2023 08:30
I think you are all misunderstanding it as a simpler problem than it is. It is saying ANY curve of length <2 is contained in a hemisphere, but this is true because a hemisphere can contain anything that has maximum length 2