ES.1802 September 21, 2026

Overview

The September 21, 2026 ES.1802 lecture develops matrix multiplication as row-by-column dot products, then connects eigenvalues and eigenvectors to whether a matrix has a complete set of independent directions. Examples contrast a defective matrix with the scalar matrix 2I, and use a complete eigenvector basis to derive the diagonalization formula A = BΛB⁻¹.

Key takeaways

A product of an r×c matrix and a c×s matrix is an r×s matrix, and each entry (AB)ᵢⱼ is the dot product of row i of A with column j of B.
Right multiplication by a diagonal matrix scales the columns of the matrix on its left, a pattern that makes the eigenvalue matrix equation AB = BΛ possible.
An n×n matrix is diagonalizable when it has n independent eigenvectors; repeated eigenvalues alone do not determine whether that condition holds.
The matrix [[2, 1], [0, 2]] has repeated eigenvalue 2 but only one independent eigenvector direction, making it defective.
The scalar matrix 2I has eigenvalue 2 in every direction, so vectors (1, 0) and (0, 1) give a complete eigenvector basis.
With eigenvectors as the columns of B and eigenvalues on the diagonal of Λ, AB = BΛ implies A = BΛB⁻¹.

Chapters

0:00 A Silent-Film Introduction to Matrix Multiplication

The class opens with a wordless, playful demonstration of matrix multiplication, using music suggestions including the Jaws theme.
The demonstration sets up the lesson’s central operation: multiplying matrices by extending the familiar matrix-by-vector process.

3:30 Matrix Products as Repeated Row-by-Column Dot Products

To multiply A by B, take each column of B and multiply it by A using the matrix-by-vector procedure.
Each resulting column becomes the corresponding column of AB, so a matrix product can be assembled one column at a time.
An individual entry in the product comes from dotting one row of A with one column of B.

6:30 Dimension Rules and the Entry Formula for AB

If A has dimensions rA × cA and B has dimensions rB × cB, the product AB exists only when cA = rB.
When the product exists, AB has dimensions rA × cB.
Using indexed notation, (AB)ᵢⱼ is row i of A dotted with column j of B; Aᵢⱼ denotes the entry in row i, column j.

10:20 Computing a 2×3 by 3×2 Matrix Product

The worked example multiplies a 2×3 matrix by a 3×2 matrix, producing a 2×2 result.
The entries are computed as row-column dot products; the lecture’s example yields [[6, 10], [13, 12]].
The indexed formula and the column-by-column method describe the same calculation.

15:20 How Diagonal Matrices Scale Matrix Columns

The 2×2 example considers multiplying [[1, 2], [3, 4]] by a diagonal matrix with entries a and b.
The first column is scaled by a, and the second column is scaled by b, illustrating how right multiplication by a diagonal matrix acts columnwise.
This column-scaling pattern will reappear when the lecture writes eigenvalue equations as a matrix equation.

17:50 Eigenvalues, Singular Matrices, and Lost Directions

For an n×n matrix, the characteristic equation has degree n, giving n eigenvalues when counted with algebraic multiplicity, though some may be repeated or complex.
A singular 2×2 transformation squashes at least one dimension; a direction that is collapsed has eigenvalue 0.
An eigenvector identifies a direction preserved by the transformation, while its eigenvalue gives the signed scaling along that direction.

22:00 The 2×2 Eigenvalue Shortcut and a Defective Matrix

For A = [[a, b], [c, d]], the lecture reviews the eigenvalue shortcut λ = m ± √(m² − det(A)), where m is the average of the diagonal entries.
For [[2, 1], [0, 2]], the discriminant is zero, so λ = 2 is a repeated eigenvalue.
Solving (A − 2I)v = 0 gives only one independent eigenvector direction, represented by (1, 0), so the matrix is defective rather than diagonalizable.

34:28 Why the Scalar Matrix 2I Has a Complete Eigenvector Set

Changing the example to [[2, 0], [0, 2]] again gives the repeated eigenvalue 2, but now A − 2I is the zero matrix.
Every vector in two-dimensional space is an eigenvector for 2I; (1, 0) and (0, 1) provide two independent directions.
The example distinguishes a repeated eigenvalue with one eigenvector direction from a repeated eigenvalue with a full basis of eigenvectors.

42:28 Building A = BΛB⁻¹ from Eigenvector Equations

For a complete n×n matrix, collect n independent eigenvectors as the columns of B and their eigenvalues along the diagonal of Λ.
The equations Avᵢ = λᵢvᵢ combine into AB = BΛ, because each column of AB is the matching eigenvector scaled by its eigenvalue.
Since the eigenvectors are independent, B is invertible; multiplying AB = BΛ by B⁻¹ gives A = BΛB⁻¹, the diagonalization of A.

Transcript excerpt

0:00 So, welcome to the um 21st of September and the phrase is I'm going to butcher this. Maybe we'll see how well I do. Okay, there's Jewish holiday so I'm just recording these lectures. Um Hi. What the heck are we going to do today? Um when I don't sleep. Oh, we're missing you. Yeah. Okay. Thank you. I know we're filming today. Yamar has something. I don't know the phrase, but it's a Jewish holiday. So, I'm going to reward people who are out. Um, yes. As the cool kids, don't just get there. Arrive. Yes. Um, when I don't sleep, I get equivalently intoxicated and I start

0:47 doing all kinds of trippy stuff like making these slides. So, let's go over this. Matrix Multiplication as a silent film. We need a soundtrack if this is going to be a silent film. >> So, someone give me a song that would make sense in the background of a silent film. >> It could be Mr. Magic by Grover Washington Jr., but recommendations. Will Baby Shark be the one that we do? >> Whenever I get onto an airplane, the very first thing I do is I look for people carrying their headphones and I latch on before they do and then I play um uh Baby Shark on their headphones. It's very fun. >> Yes. diamonds and pairs. All right. Okay. It's going to be the jaws theme if we don't get anyone saying otherwise. >> Okay. Matrix multiplication without words. You ready? Okay.

1:45 Okay. Part picture. Yes. [music] That's the other one I sent to people's

2:32 headphones is Oops, I farted Again, which is a parody of Oops, I did it again by Britney Spears. You're welcome. I should have kept the Jaws theme on loop. Oh dear. There's also the heavy

This is the start of the transcript. Save this video free for the whole transcript, searchable and time-coded.

Related notes

Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, ChristianAltoSax.