View Single Post
juliette sekx
#5
May27-09, 02:25 PM
P: 31
Quote Quote by Hurkyl View Post
Well, the biggest problem is to figure out just what you really want to know. If you are able to figure that out, then that might give us some ideas about what calculation to do.
My calcualtions are still in their infancy right now, so I'm just trying things out. All 8900 matrices should (in theory) be exactly the same, but of course, they have small differences. I want to quantitatively describe how 'close together' all these matrices are. This way when I look at the 8900 matrices produced using a different theory, I can do the same calculation on those 8900 matrices. And the thoery that produces matrices that are 'more similar' or 'closer to being all the same' is the better theory. But to do this type of comparison, I need a robust way to measure 'how similar' 8900 points are to each other!

Quote Quote by Hurkyl View Post
the method is to write this calculation as a sum of dot products, and then expand and reorganize te terms.
I don't under stand what you mean by sum of dot products here :S
To find the Euclidean distance between vectors V and U, i can do sqrt [(U-V) dot (U-V)].
But what do you mean by SUM of dot products ??