91Ó°ÊÓ

A companion matrix is a special type of square matrix associated with a polynomial. It plays a crucial role in understanding the roots of the polynomial. For a given polynomial \( p(x) = x^n + a_{n-1}x^{n-1} + \cdots + a_0 \), the companion matrix captures the coefficients in a well-defined structure. The important property of this matrix is that its eigenvalues are exactly the roots of the polynomial. This makes the companion matrix an invaluable tool for finding polynomial roots through eigenvalue computations.

• The matrix is typically sized \( n \times n \) for a polynomial of degree \( n \).
• It has ones directly below the main diagonal (the lower sub-diagonal).
• The top row contains the negative coefficients of the polynomial, excluding the leading coefficient.

When you construct the companion matrix for \( p(x) = x^3 - 5x^2 + x + 1 \), it looks like this:\[C(p) = \begin{pmatrix}0 & 0 & -1 \ 1 & 0 & -1 \ 0 & 1 & 5\end{pmatrix}\]
Using this matrix allows for the application of numerical methods to approximate the polynomial's roots.

Eigenvalues are fundamental in various areas of mathematics, representing the "scalars" associated with a matrix transformation. Specifically, for a square matrix \( A \), an eigenvalue \( \lambda \) satisfies the equation \( A\mathbf{v} = \lambda\mathbf{v} \), where \( \mathbf{v} eq 0 \) is the corresponding eigenvector. In the context of companion matrices, these eigenvalues correspond to the roots of the polynomial, which provides a bridge between polynomial theory and linear algebra.

• To find the eigenvalues of a matrix, one typically solves the characteristic equation \( \det(A - \lambda I) = 0 \).
• The companion matrix of a polynomial typically results in eigenvalues that are exactly the roots of the polynomial when calculated correctly.
• These eigenvalues can often be real or complex numbers, depending on the polynomial and its coefficients.

The calculation of these eigenvalues is central to tasks like finding polynomial roots, as shown in the exercise where we approximate the root closest to a specific value using methods such as the shifted inverse power method.

The Rayleigh Quotient is a powerful tool in numerical linear algebra that helps refine the approximation of an eigenvalue. Given an approximate eigenvector \( \mathbf{v} \), the Rayleigh quotient \( R(A, \mathbf{v}) \) for a matrix \( A \) is defined as:
\[R(A, \mathbf{v}) = \frac{\mathbf{v}^\top A \mathbf{v}}{\mathbf{v}^\top \mathbf{v}}\]
This quotient is used to estimate the eigenvalue corresponding to \( \mathbf{v} \). It is particularly useful in iterative methods, like the shifted inverse power method.

• It provides a way to converge to the nearest eigenvalue by using successive approximations.
• In practical computing, the Rayleigh Quotient can achieve high precision in identifying the closest eigenvalue to an initial guess or shift.
• It helps in adjusting the shift in the inverse power iteration process to ensure convergence to the desired eigenvalue.

In the exercise, using the Rayleigh quotient ensures that we get an accurate measure of the eigenvalue (and, consequently, the polynomial root) nearest to our chosen point \( \alpha = 5 \).

Polynomial roots are the values of \( x \) for which a given polynomial equalizes to zero. Finding these roots can often be complex, especially for higher-degree polynomials. Each root corresponds to an eigenvalue of the polynomial's companion matrix. Approximating these roots effectively involves linking polynomial theories with linear algebra techniques.

• For a polynomial \( p(x) = x^3 - 5x^2 + x + 1 \), roots are the solutions to \( p(x) = 0 \).
• Roots can be real or complex numbers.
• Traditional algebraic methods for finding roots include factoring, using the quadratic formula, or synthetic division, but these can become complex or impractical for high-degree polynomials.

In advanced mathematics, methods like the shifted inverse power method paired with the companion matrix provide efficient ways to approximate these roots when direct methods are inefficient or infeasible. This approach is practical and widely used in computational mathematics. By identifying the roots through the computation of eigenvalues from the companion matrix, we translate the problem into a sequence of manageable tasks.