diagonalization of quadratic form
A quadratic form may be diagonalized by the following procedure:
-
1.
Find a variable x such that x2 appears in the quadratic form. If no such variable can be found, perform a linear change of variable so as to create such a variable.
-
2.
By completing the square, define a new variable x′ such that there are no cross-terms involving x′.
-
3.
Repeat the procedure with the remaining variables.
Example Suppose we have been asked to diagonalize the quadratic form
Q=x2+xy-3xz-y2/4+yz-9z2/4 |
in three variables. We could proceed as follows:
-
•
Since x2 appears, we do not need to perform a change of variables.
-
•
We have the cross terms xy and -3xz. If we define x′=x+y/2-3z/2, then
x′2=x2+xy-3xz+y2/4+9z2/4-3yz/2 Hence, we may re-express Q as
Q=x′2-yz/2 -
•
We must now repeat the procedure with the remaining variables, y and z. Since neither y2 nor z2 appears, we must make a change of variable. Let us define z′=z+2y.
Q=x′2-y2-yz′/2 - •
Title | diagonalization of quadratic form |
---|---|
Canonical name | DiagonalizationOfQuadraticForm |
Date of creation | 2013-03-22 14:49:34 |
Last modified on | 2013-03-22 14:49:34 |
Owner | rspuzio (6075) |
Last modified by | rspuzio (6075) |
Numerical id | 7 |
Author | rspuzio (6075) |
Entry type | Algorithm |
Classification | msc 15A03 |
Related topic | DiagonalQuadraticForm |