Reduce the quadratic form 3x2 + 2y2 + 3z2 - 2xy - 2yz into a canonical
form using an orthogonal transformation.
The matrix of this quadratic form is
"\\begin{pmatrix}\n 3 & -1 & 0\\\\\n -1 & 2 &-1\\\\\n0 & -1 &3\n\\end{pmatrix}"
We have to find its eigenvalues and eigenvectors.
"\\det\\begin{vmatrix}\n 3-x & -1 & 0\\\\\n -1 & 2-x &-1\\\\\n0 & -1 & 3-x\n\\end{vmatrix}=(3-x)^2(2-x)-2(3-x)"
"=(3-x)((3-x)(2-x)-2)=(3-x)(x^2-5x+4)=(3-x)(x-1)(x-4)"
Therefore, the eigenvalues are 1, 3 and 4.
1) Consider the eigenvalue 1:
"\\begin{pmatrix}\n 3-1 & -1 & 0\\\\\n -1 & 2-1 &-1\\\\\n0 & -1 & 3-1\n\\end{pmatrix}\n\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 0\\\\\n 0\\\\\n0\n\\end{pmatrix}"
"\\begin{pmatrix}\n 2 & -1 & 0\\\\\n -1 &1 &-1\\\\\n0 & -1 & 2\n\\end{pmatrix}\n\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 0\\\\\n 0\\\\\n0\n\\end{pmatrix}"
"2x-y=0", "2z-y=0", therefore, the eigenvector is
"\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 2\\\\\n 1\\\\\n2\n\\end{pmatrix}"
Its norm is "\\sqrt{2^2+1^2+2^2}=3". Therefore, the normalized eigenvector, corressponding to the eigenvalue 1, is "\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 2\/3\\\\\n 1\/3\\\\\n2\/3\n\\end{pmatrix}".
2) Consider the eigenvalue 3:
"\\begin{pmatrix}\n 3-3 & -1 & 0\\\\\n -1 & 2-3 &-1\\\\\n0 & -1 & 3-3\n\\end{pmatrix}\n\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 0\\\\\n 0\\\\\n0\n\\end{pmatrix}"
"\\begin{pmatrix}\n 0 & -1 & 0\\\\\n -1 &-1 &-1\\\\\n0 & -1 & 0\n\\end{pmatrix}\n\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 0\\\\\n 0\\\\\n0\n\\end{pmatrix}"
"-y=0", "-x-y-z=0", therefore, the eigenvector is
"\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 1\\\\\n 0\\\\\n-1\n\\end{pmatrix}"
Its norm is "\\sqrt{1^2+0^2+(-1)^2}=\\sqrt{2}". Therefore, the normalized eigenvector, corressponding to the eigenvalue 3, is "\\begin{pmatrix}\n 1\/\\sqrt{2}\\\\\n 0\\\\\n-1\/\\sqrt{2}\n\\end{pmatrix}".
3) Consider the eigenvalue 4:
"\\begin{pmatrix}\n 3-4 & -1 & 0\\\\\n -1 & 2-4 &-1\\\\\n0 & -1 & 3-4\n\\end{pmatrix}\n\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 0\\\\\n 0\\\\\n0\n\\end{pmatrix}"
"\\begin{pmatrix}\n -1 & -1 & 0\\\\\n -1 &-2 &-1\\\\\n0 & -1 & -1\n\\end{pmatrix}\n\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 0\\\\\n 0\\\\\n0\n\\end{pmatrix}"
"-x-y=0", "-y-z=0", therefore, the eigenvector is
"\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}=\n\\begin{pmatrix}\n 1\\\\\n -1\\\\\n1\n\\end{pmatrix}"
Its norm is "\\sqrt{1^2+(-1)^2+1^2}=\\sqrt{3}". Therefore, the normalized eigenvector, corressponding to the eigenvalue 3, is "\\begin{pmatrix}\n 1\/\\sqrt{3}\\\\\n -1\/\\sqrt{3}\\\\\n1\/\\sqrt{3}\n\\end{pmatrix}".
By the construction we have
"\\begin{pmatrix}\n 3 & -1 & 0\\\\\n -1 & 2 &-1\\\\\n0 & -1 &3\n\\end{pmatrix}\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}="
"=\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}\\begin{pmatrix}\n 1 & 0 & 0\\\\\n 0 & 3 &0\\\\\n0 & 0 &4\n\\end{pmatrix}"
and
"3x^2+2y^2+3z^2-2xy-2yz=\\begin{pmatrix}\n x &y & z\n\\end{pmatrix}\\begin{pmatrix}\n 3 & -1 & 0\\\\\n -1 & 2 &-1\\\\\n0 & -1 &3\n\\end{pmatrix}\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}="
"\\begin{pmatrix}\n x &y & z\n\\end{pmatrix}\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}\\begin{pmatrix}\n 1 & 0 & 0\\\\\n 0 & 3 &0\\\\\n0 & 0 &4\n\\end{pmatrix}\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}^{-1}\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix}"
But the matrix
"\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}" is orthogonal, since it is composed with the orthonormal eigenvectors of the symmetric matrice. therefore
"\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}^{-1}=\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}^{T}"
"=\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}"
Put
"\\begin{pmatrix}\n 2\/ 3 & 1\/\\sqrt{2} & 1\/\\sqrt{3}\\\\\n 1\/3 & 0 &-1\/\\sqrt{3}\\\\\n2\/3 & -1\/\\sqrt{2} &1\/\\sqrt{3}\n\\end{pmatrix}\\begin{pmatrix}\n x\\\\\n y\\\\\nz\n\\end{pmatrix} = \\begin{pmatrix}\n X\\\\\n Y\\\\\nZ\n\\end{pmatrix}"
then
"3x^2+2y^2+3z^2-2xy-2yz=(X\\,Y\\,Z)\\begin{pmatrix}\n 1 & 0 & 0\\\\\n 0 & 3 &0\\\\\n0 & 0 &4\n\\end{pmatrix}\\begin{pmatrix}\n X\\\\\n Y\\\\\nZ\n\\end{pmatrix}=X^2+3Y^2+4Z^2"
And we reduce the given quadratic form to the canonical form by orthogonal transformation.
Comments
Leave a comment