Indeed, you can't construct a convex polyhedron with no stable face: consider expanding a sphere gradually out from the centre of mass until it touches a face for the first time. It must be tangent to that face, and hence if the solid rests on that face then the CoG is directly above the point of tangency and the face is stable.
If the polyhedron isn't convex, then that proof doesn't work because the first face encountered by the expanding sphere needn't be tangent to it. And indeed, non-convex polyhedra obviously do exist which can't come to rest on any of their faces; any of the standard stellations of the dodecahedron or icosahedron is a good example. In that situation, replace the polyhedron by its convex hull (equivalent for stability purposes), and you find a stable orientation if not a literal face.
Another way to look at it is to observe that gravity will attempt to get the die's centre of mass as low as possible, so a face can only be unstable if there's some other orientation available with the CoM lower. So a polyhedron can only be astable if there's no face which places the CoM lowest – i.e. there must be an infinite sequence of faces each of which makes the CoM lower than the previous one. (You probably could construct such a shape, come to think of it; but (a) it wouldn't be a polyhedron in the strict sense any more, and (b) Bolzano-Weierstrass would dictate that there was some subsequence of those faces which converged to a limiting orientation, which would then necessarily be stable.)
no subject
Date: 2007-07-06 03:37 pm (UTC)If the polyhedron isn't convex, then that proof doesn't work because the first face encountered by the expanding sphere needn't be tangent to it. And indeed, non-convex polyhedra obviously do exist which can't come to rest on any of their faces; any of the standard stellations of the dodecahedron or icosahedron is a good example. In that situation, replace the polyhedron by its convex hull (equivalent for stability purposes), and you find a stable orientation if not a literal face.
Another way to look at it is to observe that gravity will attempt to get the die's centre of mass as low as possible, so a face can only be unstable if there's some other orientation available with the CoM lower. So a polyhedron can only be astable if there's no face which places the CoM lowest – i.e. there must be an infinite sequence of faces each of which makes the CoM lower than the previous one. (You probably could construct such a shape, come to think of it; but (a) it wouldn't be a polyhedron in the strict sense any more, and (b) Bolzano-Weierstrass would dictate that there was some subsequence of those faces which converged to a limiting orientation, which would then necessarily be stable.)