equation for a graph relating to the Jacobi Algorithm

TheBeege

New member
Joined
Oct 26, 2007
Messages
1
This is a programming project for Calculus 3 for CS majors @ Georgia Tech. I'm using Matlab to program a random 5x5 matrix, A, to be run through the Jacobi Algorithm until the Off of the diagonalized matrix, B, is < 10^(-9). I've got the algorithm programmed, but I don't understand this next part of the assignment.

"If bk denote the value of ln(Off(B)) after the kth partial diagonalization, we will have the theoretical bound bk =< k ln (9/10) + ln (Off(A)) : (The value 9/10 comes from 1−2/(n^2 −2n) with n = 5). Plot your points (k; bk) together with a graph of the line y = x ln (9/10) + ln (Off(A)) : do this for 10 randomly generated matrices. How does the actual data compare with the theoretical bound?"

So I understand that for the theoretical bound, bk is to be substituted for ln(Off(B)), which refers to the diagonalized matrix, and k is the number of iterations necessary to achieve an Off(B) value < 10^(-9).

For the second equation, what do Y and X correspond to??? It seems that they've just replaced bk and k, respectively, but that wouldn't make sense. Also, do you think I would have to recalculate the theoretical bounds for each of those 10 matrices?

I'm so confused, and I'm stuck at a hotel for my cousin's wedding. Everyone expects me to attend the rehearsal dinner... ugh.
 
Top