Course syllabus

Course memo

FFR105 (FIM711), Stochastic optimization algorithms, SP1, HT22, 7.5p

The course is offered by the Department of Mechanics and Maritime Sciences. Lectures will be given on-site at the campus, they will not be live streamed. I will upload slides after the lectures so that anyone who is ill can still obtain the material from the lecture.

Contact details

During the course, we strive to be available as much as possible. You are welcome to ask questions at any time, either in class or at other times, and you may also ask questions via e-mail or telephone. You are always welcome at our offices, and personal visits are preferred to e-mails. 

Lecturer and examiner: 

Professor Mattias Wahde, Tel 772 3727, e-mail: mattias.wahde@chalmers.se

Course assistants:

Henrik Klein Moberg, e-mail: henrik.kleinmoberg@chalmers.se

Krister Blanch, e-mail: krister.blanch@chalmers.se

Finding our offices: Go to Hörsalsvägen 7, enter the building (nya M-huset), so that you have Café Bulten on your right as you enter. Then go up one flight of stairs, and enter the corridor (Vehicle Engineering and Autonomous Systems). If the door is locked, please dial the appropriate extension, as shown in the list beside the door (e.g. 3727 for Mattias).

Course purpose

The aim of the course is for the students to attain an understanding of methods in computer science that are inspired by natural processes such as evolution and cooperative behavior (for example, swarming), and also to be able to apply such methods in many different kinds of optimization problems. The methods studied in the course are relevant both in technical applications, for example in the optimization and design of autonomous systems, and for understanding biological systems, for example through simulation of evolutionary processes.

Schedule

The course schedule is given below, and the same information can also be found in TimeEdit (Links to an external site.). The page numbers refer to pages in the course book (see Course Literature below).

Note the time for the first lecture: It starts at 08.00 (sharp) on Aug. 30.

Date	Time	Room	Content
20220830	08.00-09.45	GD	Course introduction and motivation, (pp. 1-8), Classical optimization methods (introduction, pp. 8-12)
20220831	08.00-09.45	HB4	Classical optimization methods (i), (pp. 12-21, Appendix B.1, pp. 173-174)
20220902	10.00-11.45	KE	Classical optimization methods (ii), pp. 21-34, Handout of introductory programming problem
20220906	10.00-11.45	KE	Evolutionary algorithms: background and introduction, (pp. 35-45, 82- 83). Handout of home problem 1
20220906	18.00-21.00	MT9, MT11-13	Introduction to stochastic optimization algorithms in Matlab. Note: This session starts at 18.00, officially, but the rooms are available from 17.00, for those who want to start early. Teachers will be available from 18.00.
20220907	08.00-09.45	HB4	Evolutionary algorithms: components of EAs, (pp. 46-59)
20220909	10.00-11.45	HB3	Evolutionary algorithms: properties, (pp. 59-71), Appendix B.2, pp. 174- 183. Handin of introductory programming problem
20220913	10.00-11.45	HB1	Classical optimization methods and evolutionary algorithms (review, problem solving, Q&A, etc.)
20220914	08.00-09.45	HB4	Linear genetic programming and interactive evolutionary computation, pp. 72-81
20220916	10.00-11.45	KE	Neural networks (Appendix A, pp. 151-172), data analysis (Appendix C, pp. 193-204)
20220920	10.00-11.45	KE	Evolutionary algorithms: Applications I (various papers etc.), Handin of home problem 1
20220921	08.00-09.45	---	No lecture
20220923	10.00-11.45	KE	Evolutionary algorithms, Applications II (various papers etc.), Handout of home problem 2
20220927	10.00-11.45	KE	Ant colony optimization: background and introduction, pp. 99-106
20220928	08.00-09.45	HB4	Ant colony optimization: AS vs. MMAS, applications, properties of ACO, (pp. 107-116), Appendix B.3, (pp. 183-187)
20220930	10.00-1145	KE	Particle swarm optimization: Background and introduction, (pp. 117-124)
20221004	10.00-11.45	HA4	Particle swarm optimization: Properties of PSO, applications, (pp. 124-138)
20221005	08.00-09.45	HB4	Problem-solving class (various problems), review
20221007	10.00-11.45	---	No lecture
20221011	10.00-11.45	HB1	Performance comparison (EAs, ACO, PSO), (pp. 139-149)
20221012	08.00-09.45	HB4	Applications of stochastic optimization algorithms in autonomous robots, vehicles, and conversational agents. Handin of home problem 2
20221014	10.00-11.45	HB3	Ethics in AI and course summary
20221018	09.30-11.30	my office	Consultation (exam preparation)
20221019	08.00-11.45	---	No lecture

Course literature

Wahde, M., Biologically Inspired Optimization Methods: An Introduction, WIT Press, 2008

Note: The book is sold by Chalmers' bookstore. It is also possible to buy the book online (at Amazon and many other bookstores). Note that Chalmers' bookstore offers the book at a reduced price, but has only a limited supply. The book is also available at Chalmers' library, both in physical form and as an eBook.

 

Course design

The course consists of a sequence of lectures, usually three per week (but note that there are some exceptions; see below or in TimeEdit (Links to an external site.)), as well as a Matlab session in the evening of 20190910. The students must also solve a small, introductory programming problems and two sets of home problems (see Examination below). The course ends with a final exam (see Examination below).

 

Changes made since the previous course (2021)

The changes from 2021 are relatively minor, except that, this year, the course will be given entirely on-site, not remotely. The home problems have been updated (modified), and some of the slides have also been updated, particularly regarding applications of stochastic optimization algorithms.

Learning outcomes

After completion of the course the student should be able to...

• Implement and use several different classical optimization methods, e.g.
  gradient descent and penalty methods.
• Describe and explain the basic properties of biological evolution, with emphasis
  on the parts that are relevant for evolutionary algorithms.
• Describe and explain fundamental properties of cooperative behavior (e.g.
  swarming).
• Define and implement (using Matlab) different versions of evolutionary
  algorithms, particle swarm optimization, and ant colony optimization, and apply
  the algorithms in the solution of optimization problems.
•  Compare different types of biologically inspired computation methods and
  identify suitable algorithms for a variety of applications.

 

Examination

The examination consists of one separate (small) introductory programming problem (mainly to learn the coding standard), two sets of home problems, and an exam at the end of the course. The programming language used (for the home problems) will be Matlab (no exceptions allowed).

Introductory programming problem: Even though many students are probably used to programming in Matlab (and other programming languages), some students are not. In order to make sure that all students reach an acceptable level of programming knowledge, you will have to begin by solving a (simple) programming problem, making sure to follow the coding standard. This problem (and the coding standard) will be made available on 20220902, and the solution should be handed in on (or before) 20220909. In order to get a passing grade for this assignment (which is required), you should make sure that your program (i) solves the problem, and (ii) follows the coding standard.

Home problems:
The problem sheets will be made available on 20220906 (set 1) and 20220923 (set 2), and should be handed in no later than 20220920 (set 1) and 20221012 (set 2). Maximum total score (sets1+2): 25p.

Each set will contain both mandatory problems (that must be solved satisfactorily in order to pass the course) and voluntary problems (that are necessary to solve for students aspiring to receive a high grade).

Make sure to submit your solutions (via Canvas) on time! Penalties for delays: 0-6 hours: 0p, 6-24 hours: -1p, 24-48 hours: -2p, > 48 hours: -3p. 

Exam:
Date: 20221026 14.00-18.00. Maximum total score: 25p.
Note: In order to attend the exam, you must sign up for it, no later than 20221009. More information about the exam will follow later.

Grade requirements:
The minimum requirements for a passing grade (grade 3 at Chalmers, grade G at GU) are to

1.  ... obtain at least 10 p on the exam and ...
2.  ... generate and submit a satisfactory solution to the introductory programming problem and ...
3. ... generate and submit satisfactory solutions to the mandatory home problems. 

The additional requirements for the various grades are as follows: (the numbers refer to the sum of the exam result and the home problems, maximum 50p in total)

Chalmers:
5 Total score in [42,50]
4 Total score in [33,41.5]
3 Total score up to 32.5 (i.e. just the minimum requirements; see above)

GU:
VG: Total score in [39,50]
G: Total score up to 38.5 (i.e. just the minimum requirements; see above)

ECTS:
ECTS grades are offered to Erasmus Mundus students, using the same ranges as for
Chalmers grades (i.e. A=5, B=4 etc.)