Beskjeder

Takk til alle studentene i dette kurset som kom gjennom den vanskelige situasjonen i v?r, og gjorde en veldig bra prestasjon i eksamen. Her er noen l?sningsforslag. God sommer til alle sammen!

Thanks to all the students in this course who came through the difficult situation this spring, and did a great job in the exam. Here are suggested solutions. Have a great summer!

Welcome to the zoom lecture today.

The Zoom ID is the same as before, and will continue to be the same for the rest of the course. See the announcement in canvas.

Eksamen avvikles som en 7-dagers individuell hjemmeeksamen med alle hjelpemidler tillatt. Oppgavene vil likne p? en obligatorisk oppgave. Samarbeid vil ikke ansees som fusk, men du m? selv ha formulert og skrevet besvarelsen som leveres inn, og den skal gjenspeile din forst?else av pensum. Karakteren vil v?re best?tt/ikke best?tt, og grensen for ? best? vil v?re 40 % (som E ved ordin?r eksamen). Eksamen publiseres i Inspera ved opprinnelig eksamenstidspunkt og skal leveres inn som ¨¦n PDF-fil i Inspera innen samme tidspunkt 7 dager senere. Fakultetet jobber med l?sninger for studenter som evt. ikke har tilgang til datamaskin/nett.

The exam will be organized as a 7-day individual home exam and any help you can get is allowed. The exam will be similar to a compulsory assignment. Collaboration will not be considered cheating, but you must have formulated and written the answer that is submi...

It's now available from the schedule. Deadline 23 April (one week later than orginally planned). Good luck!

See announcement in canvas.

Dear students,

I've now understood that there will be no lectures or tutorials this week in any form (digital included).

But I'll try next week to use zoom for the lecture on Tuesday, 1015-1200.

Best wishes,

Mike

I've corrected all the first assignment deliveries. Many of them were very well answered. Only a handful were too weak, mainly on the L1 regression, and are "incomplete". The new deadline is Thurs 26. Please try to complete the L1 part. If it is now difficult to get access to matlab or python, you should at least be able to write down A, b, c for the LP problem.

If you have any questions or difficulties, please just email me.

Best wishes,

Michael Floater

Dear students, please just try to study on your own this week, following this week's schedule. Study Sections 6-8 of the "Mini-introduction to convexity". The main things are Theorem 5, Cor 6, Proposition 7 (but not the proofs), and Sections 7 and 8 (Fourier-Motzkin). Then have a go at Exercises 1-9. I'll post answers later this week.

I might be able to give a digital lecture next week, but I need to get the necessary equipment first.

Kj?re studenter, jeg h?per alle ha det bra med "hjemmekontor"! Jeg har n? lagt ut l?sningsforslag til ukens oppgaver (matrisespill), se timplanen.

Dear students, I hope all is well with the "homeoffice"! I have now posted suggested answers to this week's exercises (matrix games), see the weekly schedule.

Beste hilsener til alle!

Best wishes to all!

Mike

Kj?re studenter,

Som kjent blir det ingen gruppetime idag men jeg skal legge ut l?sningsforslag til ukens oppgaver (11.3, 11.2, 11.5, 11.6) i l?pet av idag eller imorgen.

Hilsen, Mike

Fra i morgen, torsdag 12. mars, gis det ikke klasseromsundervisning i emnet. Dette gjelder b?de forelesninger og gruppeundervisning.  

Det blir lagt til rette for et utvidet digitalt undervisningstilbud i emnet. Mer informasjon om dette vil bli lagt ut p? semestersiden senest 23. mars.

Here is test_simplex.m, which gives some examples of how to use simplex.m.

The use of a tolerance for the two sign tests makes a big difference in robustness for larger LP problems. The new version of simplex.m seems to work correctly in all test examples I've run, including linear L1 regression with up to 1000 points.

Here's a new version of pivot.m.

It's now essentially only four lines of matlab code, and it seems to make simplex.m about twice as fast.

Dear students, Please find a new version of simplex.m. You can use it for the first compulsory assignment.

Please note that the new simplex.m calls pivot.m, so you need both routines. Note also that the new simplex.m returns the optimal dictionary which might sometimes be useful.

Drop 6.3. Just try 6.1, 6.6, and 7.1.

The first compulsory assignment is now available (go to "timeplan"). The deadline is Thursday 12 March.

These are now available, by going to the schedule "timeplan".

You can now find answers to the exercises in week 1 in the weeky schedule "timeplan". There is also matlab code there for making a single pivot, with examples, in case you want to check your calculations.

1. Ex 2.2 is easier than Ex 2.1, so start with 2.2.

2. I've made a link to a matlab routine you can download and use to test your answers.

See the schedule, "timeplan".

f?rste forelesning, tirsdag 14. jan, 1015-1200, VB Aud 2.