From ecroot@math.gatech.edu Thu Nov 12 12:23:43 2009 Date: Thu, 12 Nov 2009 12:23:42 -0500 (EST) From: Ernie Croot To: Blackili Rudi Milhose Subject: Re: grader... Dear Blackili, I have decided to have you only grade 2 problems for accuracy, and the rest for effort. The two problems to be graded for accuracy, out of 20 points each (so, 60 points left for effort), are the following: First, go to the course webpage at http://www.math.gatech.edu/~ecroot/3012/3012.html And then look where it says ``Homework 5''. There should be 2 links here -- these are the first and second parts of the assignment. Grade problem 1 from the first part of HW 5, and problem 2 on the second part of HW 5, for accuracy. Aim for a course median of at least 80. ---- Solutions. Here is the solution to problem 1 (in case you need it): The set A1 x B1 union A1 x B2 = A1 x (B1 union B2); likewise, A2 x B1 union A2 x B2 = A2 x (B_1 union B2). But now A1 x (B1 union B2) union A2 x (B1 union B2) = (A1 union A2) x (B1 union B2). And, the size of this cross product is clearly |A1 union A2| * |B1 union B2|. Now, applying inclusion-exclusion to each of these factors, we get |A1 union A2| = |A1| + |A2| - |A1 intersect A2| |B1 union B2| = |B1| + |B2| - |B1 intersect B2|. ANd we are done. (This problem is designed to test manipulations and properties of cross products -- several different properties were required to solve it.) Solution to problem 2 in the second set: First, suppose that A has only n elements. Then, there are only n! bijections f : A -> A, since bijections can be interpreted as permutations of the set A. So, if phi is any such bijection, then since the composition of two bijections is again a bijection, this implies that all phi, phi^2, phi^3, ... are bijections. And since there are only n! bijections f : A -> A, we must have that phi^i = phi^j, for some i,j with i < j; that is, for each x in A we have (phi^i)(x) = (phi^j)(x). Now, since phi is a bijection it is invertible, with inverse phi^{-1}; and so, we have that upon composing with phi^{-i} on the left on both sides, we get x = (phi^{-i} compose phi^i)(x) = (phi^{-i} compose phi^j)(x), for all x. So, phi^{j-i} is the identity map, since it sends x -> x, for all x in A. We are now done. (This problem was designed to test knowledge of the pigeonhole principle, along with various properties of maps, such as bijections from a finite set to itself.) Best, Ernie