What is the connection between $V(\vec x) = \frac{1}{2}(x_1^2+x_1x_2+x_2^2)$ and $V(\vec x) = \frac {1}{2} x^TPx $?

35 Views Asked by At

For example, $V(\vec x) = \frac{1}{2}(x_1^2+x_1x_2+x_2^2)$ has an equivalent representation $V(\vec x) = \frac {1}{2} x^TPx $ where $P$ is some matrix

Can someone make this connection clearer for me in the following three aspects:

  • What can we say about the definiteness of V and its matrix P (Does V and P always have the same definiteness?)

  • What are some properties of this matrix P (for any V quadratic not limited to the case at hand)?

  • Are there any physical intuition to quadratic forms? Does the polynomial and its associate matrix representation of some physical quantity?

1

There are 1 best solutions below

0
On BEST ANSWER

The matrix can be taken to be symmetric, with $P_{ii}$ the coefficient of $x_i^2$, and $P_{ij} = P_{ji}$ half the coefficient of $x_i x_j$, in the quadratic form. Thus in your example, $P = \pmatrix{1 & 1/2 \cr 1/2 & 1\cr}$. The matrix has the same definiteness (positive definite, positive semidefinite, indefinite, negative semidefinite, negative definite) as the quadratic form.