UofM - MATH2740 - Lecture 09 - Part 1 - SVD (an example by hand)

Value of the Information & Strength of the Argument

The value of the information is high for a student learning SVD: it provides a concrete, fully worked example that illustrates the algorithm. The argumentation is solid: each step is justified with linear algebra principles, such as the properties of block diagonal matrices and the orthogonality of eigenvectors. The instructor also points out common pitfalls, like sign errors, and encourages verification. However, the example is simple and does not cover more complex cases or applications, which limits its depth.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is good: the computations are correct and the reasoning is clear. The instructor references a theorem from the course notes about orthogonality of eigenvectors, but no external sources are cited. The title accurately reflects the content: it is a lecture on SVD with a hand-worked example. The video is part of a university course (MATH2740), which adds credibility. However, the lack of references to textbooks or papers is a minor weakness.

The title accurately describes the content: a lecture on SVD with a worked example by hand.

8/10

The lecture is a clear, step-by-step manual computation of the Singular Value Decomposition (SVD) of a small matrix. The instructor demonstrates a block diagonal trick for eigenvalues, computes eigenvectors, and constructs the SVD components. The content is mathematically correct and well-explained, though it is a basic example and does not discuss applications or proofs in depth.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction and statement of the problem: compute SVD of matrix A = [[1,1,0],[0,0,1]].
• 00:30Computation of A^T A.
• 02:00Observation that A^T A is block diagonal and review of block diagonal determinant and inverse properties.
• 05:00Computation of eigenvalues of the blocks, yielding eigenvalues 2, 1, and 0.
• 07:00Determination of singular values: sqrt(2), 1, and 0.
• 09:00Computation of eigenvectors for each eigenvalue.
• 12:00Normalization of eigenvectors to form orthogonal matrix V.
• 15:00Construction of matrix Sigma with singular values.
• 18:00Computation of left singular vectors u1 and u2.
• 21:00Final assembly of U, Sigma, V and verification of A = U Sigma V^T.