Show that the tridiagonal matrix is positive definite

In summary, the conversation discusses proving that a tridiagonal matrix is positive definite using a hint involving the sum of squares and equality only if all squares are zero. The conversation goes through the steps of writing the matrix as a sum of squares and concludes that the matrix is positive definite if the vector is not zero and equal to zero if the vector is zero. The conversation ends with a nod and a thank you.
  • #1
mathmari
Gold Member
MHB
5,049
7
Hey! :eek:

We have the tridiagonal matrix $A=\begin{pmatrix}2 & 1 & \ldots & 0 \\ 1 & 2 & 1 & \ldots \\ \ldots & \ldots & \ldots & \ldots \\ 0 & \ldots & 1 & 2\end{pmatrix}$. I want to show that it is positive definite.

For that it is given the following hint:
1) $\langle x, Ax\rangle \geq 0$
2) $\langle x, Ax\rangle =0 \Rightarrow x=0$ I have done the following:

The $i$-th component of the vector $Ax$ is \begin{equation*}(Ax)_i=x_{i-1}+2x_i+x_{i+1} , \ i=1, 2, \ldots , n \ \text{ with } x_0=x_{n+1}=0\end{equation*}
Then we have the following: \begin{equation*}\langle x, Ax\rangle=\sum_{i=1}^nx_i(Ax)_i =\sum_{i=1}^nx_i\left (x_{i-1}+2x_i+x_{i+1}\right )=\sum_{i=1}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^n x_ix_{i+1}\end{equation*}

Is everything correct so far? (Wondering)

Now we have to show that it is positive if the vector $x$ is not the zero vector and equal to $0$ if the vector is the zero vector, right? But how can we do that? (Wondering)
 
Mathematics news on Phys.org
  • #2
mathmari said:
Hey! :eek:

We have the tridiagonal matrix $A=\begin{pmatrix}2 & 1 & \ldots & 0 \\ 1 & 2 & 1 & \ldots \\ \ldots & \ldots & \ldots & \ldots \\ 0 & \ldots & 1 & 2\end{pmatrix}$. I want to show that it is positive definite.

For that it is given the following hint:
1) $\langle x, Ax\rangle \geq 0$
2) $\langle x, Ax\rangle =0 \Rightarrow x=0$ I have done the following:

The $i$-th component of the vector $Ax$ is \begin{equation*}(Ax)_i=x_{i-1}+2x_i+x_{i+1} , \ i=1, 2, \ldots , n \ \text{ with } x_0=x_{n+1}=0\end{equation*}
Then we have the following: \begin{equation*}\langle x, Ax\rangle=\sum_{i=1}^nx_i(Ax)_i =\sum_{i=1}^nx_i\left (x_{i-1}+2x_i+x_{i+1}\right )=\sum_{i=1}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^n x_ix_{i+1}\end{equation*}

Is everything correct so far? (Wondering)

Now we have to show that it is positive if the vector $x$ is not the zero vector and equal to $0$ if the vector is the zero vector, right? But how can we do that? (Wondering)

Hey mathmari!

How about trying to write it as a sum of squares?
That is, something like $(x_1+x_2)^2 + (x_2 + x_3)^2 + ...$.
Squares are always $\ge 0$ aren't they?
With equality only if all squares are $0$? (Wondering)
 
  • #3
Klaas van Aarsen said:
How about trying to write it as a sum of squares?
That is, something like $(x_1+x_2)^2 + (x_2 + x_3)^2 + ...$.
Squares are always $\ge 0$ aren't they?
With equality only if all squares are $0$? (Wondering)

Ahh so we have the following, don't we?
\begin{align*}\langle x, Ax\rangle&=\sum_{i=1}^nx_i(Ax)_i \\ & =\sum_{i=1}^nx_i\left (x_{i-1}+2x_i+x_{i+1}\right )\\ & =\sum_{i=1}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^n x_ix_{i+1} \\ & =x_1 x_{0}+\sum_{i=2}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^{n-1} x_ix_{i+1} +x_nx_{n+1} \\ & =\sum_{i=2}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^{n-1} x_ix_{i+1} \\ & = \sum_{i=1}^{n-1}x_i x_{i+1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^{n-1} x_ix_{i+1} \\ & = 2\sum_{i=1}^{n-1}x_i x_{i+1}+2\sum_{i=1}^n x_i^2 \\ & = \left (\sum_{i=1}^{n-1} x_i^2+x_n^2\right )+2\sum_{i=1}^{n-1}x_i x_{i+1}+\left (x_1^2+\sum_{i=2}^n x_i^2\right ) \\ & = \sum_{i=1}^{n-1} x_i^2+2\sum_{i=1}^{n-1}x_i x_{i+1}+\sum_{i=2}^n x_i^2+x_1^2+x_n^2 \\ & = \sum_{i=1}^{n-1} x_i^2+2\sum_{i=1}^{n-1}x_i x_{i+1}+\sum_{i=1}^{n-1} x_{i+1}^2+x_1^2+x_n^2 \\ & = \sum_{i=1}^{n-1} \left (x_i^2+2x_i x_{i+1}+ x_{i+1}^2\right )+x_1^2+x_n^2 \\ & = \sum_{i=1}^{n-1} \left (x_i+ x_{i+1}\right )^2+x_1^2+x_n^2 \end{align*}
So we get $$\langle x, Ax\rangle \geq 0 \ \text{ and } \ \langle x, Ax\rangle=0 \iff x_i=0 \ \forall i \iff x=0$$

(Wondering)
 
  • #4
mathmari said:
Ahh so we have the following, don't we?
\begin{align*}\langle x, Ax\rangle&=\sum_{i=1}^nx_i(Ax)_i \\ & =\sum_{i=1}^nx_i\left (x_{i-1}+2x_i+x_{i+1}\right )\\ & =\sum_{i=1}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^n x_ix_{i+1} \\ & =x_1 x_{0}+\sum_{i=2}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^{n-1} x_ix_{i+1} +x_nx_{n+1} \\ & =\sum_{i=2}^nx_i x_{i-1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^{n-1} x_ix_{i+1} \\ & = \sum_{i=1}^{n-1}x_i x_{i+1}+2\sum_{i=1}^n x_i^2+\sum_{i=1}^{n-1} x_ix_{i+1} \\ & = 2\sum_{i=1}^{n-1}x_i x_{i+1}+2\sum_{i=1}^n x_i^2 \\ & = \left (\sum_{i=1}^{n-1} x_i^2+x_n^2\right )+2\sum_{i=1}^{n-1}x_i x_{i+1}+\left (x_1^2+\sum_{i=2}^n x_i^2\right ) \\ & = \sum_{i=1}^{n-1} x_i^2+2\sum_{i=1}^{n-1}x_i x_{i+1}+\sum_{i=2}^n x_i^2+x_1^2+x_n^2 \\ & = \sum_{i=1}^{n-1} x_i^2+2\sum_{i=1}^{n-1}x_i x_{i+1}+\sum_{i=1}^{n-1} x_{i+1}^2+x_1^2+x_n^2 \\ & = \sum_{i=1}^{n-1} \left (x_i^2+2x_i x_{i+1}+ x_{i+1}^2\right )+x_1^2+x_n^2 \\ & = \sum_{i=1}^{n-1} \left (x_i+ x_{i+1}\right )^2+x_1^2+x_n^2 \end{align*}
So we get $$\langle x, Ax\rangle \geq 0 \ \text{ and } \ \langle x, Ax\rangle=0 \iff x_i=0 \ \forall i \iff x=0$$

(Wondering)

(Nod)
 
  • #5
Klaas van Aarsen said:
(Nod)

Thank you! (Yes)
 

1. What is a tridiagonal matrix?

A tridiagonal matrix is a square matrix with non-zero elements only on the main diagonal, the diagonal above the main diagonal, and the diagonal below the main diagonal. All other elements are zero.

2. How do you determine if a tridiagonal matrix is positive definite?

To show that a tridiagonal matrix is positive definite, we need to prove that all of its leading principal minors (determinants of the top left submatrices) are positive. This can be done using the Sylvester's criterion or the Cholesky decomposition.

3. What is the significance of a positive definite tridiagonal matrix?

A positive definite tridiagonal matrix has many useful properties, such as being invertible, having all positive eigenvalues, and being stable. It is also commonly used in numerical methods for solving linear equations and optimization problems.

4. Can a tridiagonal matrix be both positive definite and symmetric?

Yes, a tridiagonal matrix can be both positive definite and symmetric. In fact, all symmetric tridiagonal matrices with positive diagonal entries are positive definite.

5. How is the positive definiteness of a tridiagonal matrix related to its diagonally dominant property?

A tridiagonal matrix that is diagonally dominant (meaning that the absolute value of the diagonal element is larger than the sum of the absolute values of the off-diagonal elements) is always positive definite. However, the converse is not always true - a positive definite tridiagonal matrix may not be diagonally dominant.

Similar threads

  • General Math
Replies
8
Views
1K
Replies
10
Views
909
Replies
1
Views
762
Replies
1
Views
400
  • General Math
Replies
2
Views
1K
Replies
8
Views
1K
Replies
32
Views
1K
  • General Math
Replies
8
Views
2K
Replies
7
Views
1K
Replies
2
Views
3K
Back
Top