Final syllabus
Geof H. Givens, Jennifer A. Hoeting: Computational Statistics, 2nd Edition
ISBN: 978-0-470-53331-4.
In the link to the book web-page you also find datasets and R-scripts connected to examples/exercises in the book
- Chapter 1 - Background: Will only be referred to when needed
- Chapter 2 - Optimization: General methods. Everything except
- the details about convergence rates (all methods)
- Iteratively reweighted least squares (2.2.1.1)
- the Gauss-Newton method (2.2.3)
- 06.05.2019: Presentation for this chapter is slightly updated due to an error (wrong sign) in the Taylor expansion for the log-likelihood on page 3.
- Chapter 3 - Combinatorial optimization. Everything except
- section 3.4.2
- Chapter 4 - The EM algorithm
- Section 4.1: All
- Section 4.2: All but except 4.2.3.1 and 4.2.3.2
- 13.05.19: Some errors in the presentation for this chapter have been corrected:
- Page 9: theta^t -> theta in last line
- Page 15: (s) -> g(s) in denominator in last term.
- Chapter 6 - Monte Carlo methods
- Not section 6.3.1.1
- 20.05.19, page 19: Derivation corrected and star taken avay from hat mu.
- 20.05.19: Page 21: Corrected hat mu_IS^* to hat mu_IS
- 20.05.19: Part 2, page 3, 1/n removed from second estimate.
- Chapter 7 - Markov Chain Monte Carlo
- 20.05.19: Page 24: More clearly specified which index that grows when looking at limits.
- 20.05.19: Error on eps/2 term corrected on page 40
Storvik (2017): Sequential Monte Carlo for state space models.
Storvik (2019): The Stochastic gradient algorithm
- This note has now been updated.
- In section 4, only Lemma 1 and Theorem 1 (as well as their proofs) is part of the syllabus
- 13.05.19: Some errors with respect to maximization/minimization has been corrected. Further, in the remark to Theorem 1, A-3 have been corrected to A-2.
Hamiltonian Monte Carlo, presented in Slides from chapter 7.
Variational inference,presented in these slides
- 20.05.19: Page 2, c replaced by x in denominator
- 20.05.19: Page 8, z_j replaced by z_{-j} in the algorithm
All given exercises
- 21.05.19: Trial exam ex 5d) An error in the second derivative with respect to beta_1 is corrected
- 21.05.19: Exam 2017: In the formulae for the Poisson distribution, z and lambda is switched.
- 21.05.19: Extra exercise 4: Corrected to that N_2 do not communicate even if p is odd.
There are some misprints in the book. Some are given at through the book webpage. Some additional misprints found by the lecturer is given here. If you find more errors, please let me know.