For a particular u in the range [k,k+1), where k is an integer, [u] = k, so that the first integral is the area of the isosceles right triangle with two legs equaling k. That area is k^2/2.
The integral on the right is that of a step function. The area involved consists of 1x1 squares. When k is 1, the number of squares is zero; when k is 2, the number of squares is 1. In general, the number (and therefore area) is Sigma{i=0 to k1} n, which is (k^2k)/2.
However, the step function is to be evaluated beyond k, by an additional uk to u, and the question involves how far to the right of k will result in the total area being equal to that of the isosceles right triangle of paragraph 1. To the right of x=k, the height of the step function is k. We need to gain the difference between k^2/2 and (k^2k)/2, which is k/2. As the height of the step function is a constant k, the base of the rectangle needs to be 1/2, to get the added area to be k/2, confirming u=[u]+1/2.
Is there a flaw in this logic?
Edited on March 20, 2007, 3:49 pm

Posted by Charlie
on 20070320 15:47:49 