The project is to be done in teams (team-building instructions) and is worth 2 higher-education credits (ECTS credits). The main objective of the project is to exercise further the theoretical knowledge gained in the lectures, but on a problem that you propose and that gets our approval.
Let an approach be either a MiniZinc model for the entire problem, or a script (find a Python-based way of doing that at Resources) with pre-processing, solving (possibly on a pipeline of multiple models), and post-processing. We are not just interested in a sufficiently complete and efficient approach, but also in explanations and experimental evaluations, hence a report is also required for the project and the quality of the report determines the project score.
Six help sessions are scheduled for the project.
PhD students do either the assignments (possibly in a duo team), or just the project (solo; possibly in connection with their PhD research; but they are highly encouraged to do the assignments nevertheless), or both: contact the head teacher.
A project proposal of three problems, listed by decreasing preference, each answering two questions (what? which third-party instance data?) must be submitted at Studium (as a single file in .pdf format; all other formats will be rejected; state the team number and the names of the team members): the head teacher and assistants may interactively (via Studium) negotiate revisions with you during the subsequent seven days until ideally our approval (if approved, then pass, else fail). The problems can be your own or drawn from the following websites (in the latter case, it suffices to name the problem):
Let us know if you think some problem ought to be taken off this list, for instance because an approach with MiniZinc or OPL can be found on the internet. If your approach turns out to be competitive, then you can even contribute it to the annual MiniZinc Challenge or to CSPlib!
An approved project proposal entitles to continuation (deadlines and schedule in the tables below):
Initial report: See the slides numbered 79 and 80 of Topic 1. If your initial report is submitted on time at Studium and (1) follows the imposed structure of the demo report, which is longer than the one for assignments (LaTeX source, if you use LaTeX: comment away line 15 and uncomment line 16), (2) cites the source of the problem and every included third-party picture, (3) has a description of the problem in your own words, (4) discusses a single approach (see above for the definition of approach), and (5) every model follows the structure of the skeleton model, then you pass, else you fail and are not entitled to a final report. One or two students are assigned at Studium for written peer review(s) of your initial report: see below.
The demo report has an additional task for project reports, and here are the instructions. Correctness: How do you argue for the correctness of the approach? For example, did you use a checker based on another model, say a model by another team or a translation of a MIP model given somewhere? Did you use the same new instances as another team? If it is a satisfaction problem: did you compare the numbers of solutions with numbers reported somewhere or by another team? If it is an optimisation problem: did you compare objective values proven optimal before timing out (respectively objective values known when timing out) with optimal objective values (respectively best-known objective values) reported somewhere or by another team?
Written peer review: If you (1) attach (like this) a peer review to the initial report of your assigned team on time at Studium (so that the other team can see your review), and (2) submit your peer review on time also at Studium to the assignment "Project Peer Review" (so that the teachers can grade it), and (3) your peer review is individually written (a duo team writes two reviews) and (4) constructive and (5) follows the imposed form (instructions, plus: Author = team number of the initial report you review, and: Advisor = your name), then you pass, else you fail.
Final report: The written peer review(s) and feedback by the teachers will help you prepare your final report (including model and data files, if not other code; there is no imposed filename): it will get a score \(p \in 0..10\), under (1--5) the same conditions as the initial report, but (6) it is rewritten where relevant and not just expanded incrementally.
| Deliverable or Event | Deadline or Date |
|---|---|
| Proposal | Wed 13 Sep at 13:00 |
| Approval (pass/fail) | Wed 20 Sep at 13:00 |
| Help a | Fri 22 Sep |
| Help b (also for Assignment 3) | Thu 28 Sep |
| Help c (also for Assignment 3) | Mon 02 Oct |
| Help d (also for Assignment 3) | Wed 04 Oct |
| Help e | Wed 11 Oct |
| Help f | Thu 12 Oct |
| Initial Report (pass/fail) | Fri 13 Oct at 13:00 |
| Peer review (pass/fail) | Wed 18 Oct at 13:00 |
| Final Report (score \(p \in 0..10\)) | Mon 30 Oct at 13:00 |
The project proposal, initial and final reports, and peer review must be submitted electronically via Studium, whose clock may be different from yours. Submission deadlines are hard. Exceptions must be negotiated in due time during work hours with the head teacher, upon reporting a convincing case of force majeure. Grading will only start after a deadline, so you can submit multiple times until then.
The considerations on ethics of the assignment reports also apply to the project report.
We reserve the right to give different project scores to the teammates of a team, depending for example on the performance at the peer review.
Each higher-education credit (ECTS credit) translates under Swedish university law into an expected 26.67 hours of work for the average student. The project is worth 2 credits: discounting about half of the 21 hours spent in period 1 on the lectures (the other half being similarly discounted on the 3 assignment credits), the project and peer review should reflect 49.5 hours of work by the average student, for each teammate. This does not clash with other courses you are taking, as university studies are legally defined to take 400 hours of work per study period (normally 10 weeks), and the standard 15 credits targeted in a study period are calibrated to reach that total.
The project grade is determined by a published scale, depending on your project score \(p \in 0..10\).
The overall grade for Course 1DL451 is determined by a published scale, depending on the modelling assignment scores and the project score \(p\).
The overall grade for Course 1DL442 is determined by a published scale, depending on the modelling assignment scores, the project score \(p\), and the programming assignment scores.
These rules are effective as of Mon 28 Aug 2023. The head teacher reserves the right to modify them at any moment, should special circumstances call for this.
