Identify the quadratic form of the given equation

lesdes
Messages
12
Reaction score
0
<Moderator's note: Moved from a technical forum and thus no template.>

Hello I am given the following problem to solve.

Identify the quadratic form given by ##-5x^2 + y^2 - z^2 + 4xy + 6xz = 5##.
Finally, plot it.

I cannot seem to understand what I have to do. The textbook chapter on quadratic forms simply just forms the matrix of the given equation. Is that what I am expected to do?

It would be nice if someone could point me in the correct direction or link me to a relevant document/article.
 
Last edited by a moderator:
Physics news on Phys.org
lesdes said:
Hello I am given the following problem to solve.

Identify the quadratic form given by ##-5x^2 + y^2 - z^2 + 4xy + 6xz = 5##.
Finally, plot it.

I cannot seem to understand what I have to do. The textbook chapter on quadratic forms simply just forms the matrix of the given equation. Is that what I am expected to do?

It would be nice if someone could point me in the correct direction or link me to a relevant document/article.
I assume you should say what it is: a hyperboloid, a sphere or whatever which form it has. The matrix is a first step, to bring it into a form without mixed terms is the next.
 
  • Like
Likes lesdes
fresh_42 said:
I assume you should say what it is: a hyperboloid, a sphere or whatever which form it has. The matrix is a first step, to bring it into a form without mixed terms is the next.

Okay so let's see if I am on the right path. What I have done is that I have written ##−5x2+y2−z2+4xy+6xz=5## in the form
$$
(x y z)
\begin{pmatrix}
-5 & 2 & 3 \\
2 & 1 & 0 \\
3 & 0 & -1
\end{pmatrix}
\begin{pmatrix}
x \\
y \\
z
\end{pmatrix}
= 5
$$

as suggested by the book:
1.png

is then
2.png
.

Next, I used the Principal Axes Theorem
3.png

and formed the determinant
##
\begin{vmatrix}
-5 -\lambda & 2 & 3 \\
2 & 1 -\lambda & 0 \\
3 & 0 & -1 -\lambda
\end{vmatrix}
##

which then gives the eigenvalies ##\lambda_1 = -7##, ##\lambda_2 = 2## and ##\lambda_3 = 0##. Then, the eigenvectors are ##\vec v_1 = (-4, 1, 2)##, ##\vec v_2 = (1, 2, 1)##, and ##\vec v_3 = (1, -2, 3)##. Then, I formed the matrix

##
P =
\begin{pmatrix}
-4 & 1 & 1 \\
1 & 2 & -2 \\
2 & 1 & 3
\end{pmatrix}
##

Then, I calculated ##P^{T}AP## such that
##
\begin{pmatrix}
-147 & 0 & 0 \\
0 & 12 & 0 \\
0 & 0 & 0
\end{pmatrix}
##.
And finally,
##
(x_1 y_1 z_1)
\begin{pmatrix}
-147 & 0 & 0 \\
0 & 12 & 0 \\
0 & 0 & 0
\end{pmatrix}
\begin{pmatrix}
-x_1\\
y_1 \\
z_1
\end{pmatrix}
##

gave me ##-147x_1^2 + 12y_1^2 = 5##. Am I correct so far?

And what next? Do I have to say what this last equation is geometrically? How do I find that out?
 

Attachments

  • 1.png
    1.png
    4.7 KB · Views: 335
  • 2.png
    2.png
    2.2 KB · Views: 365
  • 3.png
    3.png
    26.4 KB · Views: 410
Last edited:
I haven't gone through the details yet, but the plot looks o.k. But you still need a ##z-##component. It cannot be only a two dimensional object, as it was a three dimensional at the start. And isn't there a sign error in the eigenvalue matrix?
 
  • Like
Likes lesdes
fresh_42 said:
I haven't gone through the details yet, but the plot looks o.k. But you still need a ##z-##component. It cannot be only a two dimensional object, as it was a three dimensional at the start. And isn't there a sign error in the eigenvalue matrix?

Yes there is a sign error in the eigenvalue matrix. But I only made the error when typing it in this thread. I will correct that now.

You do not have to check whether the eigenvalues, eigenvectors and the multiplication ##P^TAP## is correct. I let Mathematica do the computations.
But where do I get the third component from? As it is not there anymore due to the last multiplication. Also in the book example, after applying this procedure the last equation was two dimensional. Is this a mistake?
 
Last edited:
lesdes said:
You do not have to check whether the eigenvalues, eigenvectors and the multiplication ##P^TAP## is correct. I let Mathematica do the computations.
But where do I get the third component from? As it is not there anymore due to the last multiplication. Also in the book example, after applying this procedure the last equation was two dimensional. Is this a mistake?

The third component is hidden. It's ##0 {z_1}^2## since there's a zero eigenvalue. It's a type of cylinder.
 
  • Like
Likes lesdes
lesdes said:
I let Mathematica do the computations.
I guess Mathematica also needs a correct matrix, but maybe you have a special version.
 
  • Like
Likes lesdes
lesdes said:
Then, the eigenvectors are ##\vec v_1 = (-4, 1, 2)##, ##\vec v_2 = (1, 2, 1)##, and ##\vec v_3 = (1, -2, 3)##. Then, I formed the matrix

##
P =
\begin{pmatrix}
-4 & 1 & 1 \\
1 & 2 & -2 \\
2 & 1 & 3
\end{pmatrix}
##
You need to normalize the eigenvectors before forming the matrix P. The diagonalized matrix will then have the eigenvalues on the diagonal. In fact, the theorem you quoted stated this. So once you found the eigenvalues, you knew enough to figure out the geometric shape.
 
  • Like
Likes scottdave and lesdes
lesdes said:
Yes there is a sign error in the eigenvalue matrix. But I only made the error when typing it in this thread. I will correct that now.

You do not have to check whether the eigenvalues, eigenvectors and the multiplication ##P^TAP## is correct. I let Mathematica do the computations.
But where do I get the third component from? As it is not there anymore due to the last multiplication. Also in the book example, after applying this procedure the last equation was two dimensional. Is this a mistake?

Something is wrong: for the matrix
$$A = \begin{bmatrix} -5 & 2 & 3 \\ 2 & 1 & 0 \\3 & 0 & -1 \end{bmatrix}$$
I let Maple do the computations, and got eigenvalues ##-7, 2, 0## and the associated eigenvectors. You need to remember that when forming the matrix ##P## you must use normalized eigenvectors (that is, forming an orthonormal set). Doing that in Maple I get
$$P = \begin{bmatrix}
-4/\sqrt{21} & 1/\sqrt{6} & 1/\sqrt{14} \\
1/\sqrt{21} & \sqrt{2/3} & -\sqrt{2/7} \\
2/\sqrt{21} & 1/\sqrt{6} & 3/\sqrt{14}
\end{bmatrix}
$$
This gives
$$P^T A P = \begin{bmatrix} -7 & 0 & 0 \\ 0 & 2 & 0 \\ 0 & 0 & 0 \end{bmatrix}$$
as it should.
 
  • Like
Likes lesdes
  • #10
Ray Vickson said:
Something is wrong: for the matrix
$$A = \begin{bmatrix} -5 & 2 & 3 \\ 2 & 1 & 0 \\3 & 0 & -1 \end{bmatrix}$$
I let Maple do the computations, and got eigenvalues ##-7, 2, 0## and the associated eigenvectors. You need to remember that when forming the matrix ##P## you must use normalized eigenvectors (that is, forming an orthonormal set). Doing that in Maple I get
$$P = \begin{bmatrix}
-4/\sqrt{21} & 1/\sqrt{6} & 1/\sqrt{14} \\
1/\sqrt{21} & \sqrt{2/3} & -\sqrt{2/7} \\
2/\sqrt{21} & 1/\sqrt{6} & 3/\sqrt{14}
\end{bmatrix}
$$
This gives
$$P^T A P = \begin{bmatrix} -7 & 0 & 0 \\ 0 & 2 & 0 \\ 0 & 0 & 0 \end{bmatrix}$$
as it should.

Ah yes now I see. I have the same result through mathematica.
1.png


So now I have ##-7x^2 + 2y^2 = 5##. This result should be equivalent to ##−5x^2+y^2−z^2+4xy+6xz=5## as I understand it? Hence if I graph both in mathematica, then I should receive the same graph?
 

Attachments

  • 1.png
    1.png
    15.7 KB · Views: 854
  • #11
So now I've plotted both equations in mathematica and I have the following results
3.png


It might be that I did not understand the quadratic form purpose. I thought the purpose was to write a more complicated equation into a simpler form. I am a little bit confused now.
 

Attachments

  • 3.png
    3.png
    42.6 KB · Views: 437
  • #12
lesdes said:
Ah yes now I see. I have the same result through mathematica.
View attachment 220952

So now I have ##-7x^2 + 2y^2 = 5##. This result should be equivalent to ##−5x^2+y^2−z^2+4xy+6xz=5## as I understand it? Hence if I graph both in mathematica, then I should receive the same graph?
The variables will be different: using
$$\pmatrix{x'\\y'\\z'} = P \pmatrix{x\\y\\z}, $$
your equation in the "primed" variables will be
$$-7 x'^2 + 2 y'^2 = 5,$$
which is a cylinder parallel to the ##z'##-axis in ##(x',y',z')##-space. That will be a rotated cylinder in ##(x,y,z)##-space.
 
  • Like
Likes lesdes
Back
Top