Du befindest dich hier: FSI Informatik » Prüfungsfragen und Altklausuren » Hauptstudiumsprüfungen » Lehrstuhl 10 » anla-2016-03-02 (Übersicht)
Examiner: Prof. Rüde
Grade: 1.0
Note: There were no exam protocols for this subject before my exam, so I created my own trainer for ANLA with 138 questions. It covers the lectures 1-15, as these are crucial for understanding the rest of the book. Take a look at it at http://kilians.net/trainers/anla/.
Atmosphere: Prof. Rüde was very kind, it was a relaxed atmosphere. He directly started with the first question. He did not focus on formulas, but on the general understanding of the concepts.
Questions:
Gerschgorin’s Theorem (from exercises)
- He gave me a real symmetric matrix and asked me, what the Gerschgorin’s Theorem says about the location of the eigenvalues.
- What can I say about them regarding the symmetry of the matrix?
- Draw a diagram with the regions of the eigenvalues.
- Can eigenvalues lie on the edge of the disks?
- What can you say about eigenvalues of singular matrices?
Norms
- What three conditions must hold for a norm?
- How is a matrix norm defined and what is its meaning?
- What fourth condition holds for matrix norms?
Least Squares
- What is a Least Squares problem?
- What are the three ways to solve it?
- How do the three ways compare regarding their complexity?
- How do the three ways compare regarding their numerical stability?
Eigenvalue Algorithms (the questions all referred to the lectures “Eigenvalue Problems” to “Reduction to Hessenberg Form” in the Book)
- What are the general steps to compute the eigenvalues of a matrix?
- Why can’t we directly compute the eigenvalues of a matrix that is not in Hessenberg form?
- Explain the simple version of the QR algorithm.
- How did we modify the QR Algorithm to get better results?
Preparation for the exam: I read the book (“Numerical Linear Algebra” by Trefethen and Bau) from lecture 1 to 31 a couple of times. I tried to fully understand the chapters 1-15 and 24-27. I understood the basic concepts of the lectures 16-23 and 28-31. Although Prof. Rüde covered 1-38 in the lectures, this was the optimal preparation.