# Diagonalization of Hermitian Matrices (part 2) | Summary and Q&A

6.3K views
January 19, 1970
by
IIT Madras - B.S. Degree Programme
Diagonalization of Hermitian Matrices (part 2)

## TL;DR

The lecture discusses the spectral theorem, which states that Hermitian matrices are unitarily diagonalizable, and the corollary for real symmetric matrices being orthogonally diagonalizable.

## Key Insights

• ❓ The spectral theorem states that Hermitian matrices are unitarily diagonalizable, whereas the corollary applies to real symmetric matrices being orthogonally diagonalizable.
• 👻 The proof of the spectral theorem utilizes Schur's theorem, which allows for the transformation of any matrix into an upper triangular matrix.
• 🫤 The diagonalization of a Hermitian matrix involves finding a unitary matrix that can transform it into a diagonal matrix with real numbers as entries.

## Transcript

Welcome to lecture 5 in week 5 of machine learning foundations. We are continuing the discussion on the linear algebra side of things. In this particular lecture, we will be able to prove the most important theorem that we wanted to get to which is the spectral theorem, which in essence states the following a Hermitian matrix is unitarily diagonali... Read More

### Q: What is the spectral theorem?

The spectral theorem states that a Hermitian matrix can be transformed into a diagonal matrix using a unitary matrix. For real symmetric matrices, they can be orthogonally diagonalized using an orthogonal matrix.

### Q: How is the proof for the spectral theorem derived?

The proof of the spectral theorem relies on Schur's theorem, which states that every matrix is upper triangularizable. By applying Schur's theorem, we can show that a Hermitian matrix can be transformed into an upper triangular matrix, which can then be further simplified to a diagonal matrix.

### Q: Are all unitarily diagonalizable matrices Hermitian?

No, the converse of the spectral theorem is not true. There are matrices that are unitarily diagonalizable without being Hermitian. The lecture provides an example of such a matrix, where the matrix is not equal to its Hermitian transpose but still has distinct eigenvalues.

### Q: How are real symmetric matrices related to the spectral theorem?

Real symmetric matrices are a special case of Hermitian matrices. The corollary to the spectral theorem states that real symmetric matrices can be orthogonally diagonalized using an orthogonal matrix, leading to a diagonal matrix with real entries.

## Summary & Key Takeaways

• The lecture introduces the spectral theorem, which states that a Hermitian matrix is unitarily diagonalizable, meaning it can be transformed into a diagonal matrix using a unitary matrix.

• This theorem also applies to real symmetric matrices, which can be orthogonally diagonalized using an orthogonal matrix.

• The lecture provides a proof for the spectral theorem using Schur's theorem, which states that every matrix is upper triangularizable.

• An example is given to demonstrate the diagonalization of a Hermitian matrix and a real symmetric matrix.