As of February 28, 2018, this page is provisorial and may be subject to small modifications !
TUGonline links for Mathematics:
VO
and
UE
and
KV
Mathematic Students: please register also for the KV !
TUGonline links for Computer Science:
VO
and
UE
Contents of the course
Entropy, conditional entropy, Kullback-Leibler divergence, related
properties and formulas
Entropy rate, rate of stationary processes and Markov chains
Asymptotic equipartition property, AEP for iid random variables and
ergodic Markov chains, typical sets and data compression
Codes, prefix codes and entropy, proof of the optimality of Huffman codes
Noisy channels, channel capacity, Shannon's channel coding theorem
Entropy of continuous random variables and related issues (Math only)
The
exam for the course is oral; date to be fixed in contact with W. Woess.
Course language: English, exam language: English or German according to
preference of candidate.
Modalities of the course (IMPORTANT)
Please read
modalities.pdf
carefully.
Exercise classes
Registration for the exercises via TUG online: links for
Mathematics
and
Computer Science.
In case of de-registration: observe deadlines!
Weekly exercises
Below you find links to the exercises to be prepared for the respective class.
Every week an exercise sheet (pdf) is made available online.
The solved current exercises have to be marked at the latest on the evening before the exercise classes
by Tuesday 11 pm. in the online checking system.
Participaton is required, later marking is NOT possible!
Some solutions
Grading of the exercises
The Online checking system is applied.
For each exercise, one of those participants who have marked it are called to explain the solution at the
blackboard. For this, "blackboard points" are awarded, maximum number of points according to
the exercise sheet. If it becomes apparent that the exercise was not prepared, there are
point detractions.
The following table gives a overview:
blackboard presentation |
blackboard points |
extraordinary or very good presentation of the solution |
all points of the presented example |
correct solution |
between 0 and the maximal number of possible points |
person is not able to explain the example or makes mistakes |
minus points: up to the number of made crosses of the current week |
absence of the called person (first time) |
minus doubled number of made crosses of the current week |
absence of the called person (second time) |
negative evaluation |
Only in exceptional cases of illness, participants - after notification of the group leader by email
before the exercise class -
can submit written versions of the solved exercises and set the corresponding crosses within a week after the exercise class during the office hours
of G. Wiegel, resp. A. Fuchs. In this case, a
medical certificate has to be delivered, and the careful presentation of some
of the marked exercises at the office blackboard will be required.
The exercise modalities do not support mass desertion caused by other exams
or similar causes coming from other courses or seminars. In this case,
participants are asked to inform the person responsible for the other event
about the problem and to ask her or him to adapt the schedule accordingly.
In any case, the available points are sufficient for obtaining
even the best mark while missing an exercise class.
The success quotient is given by:
100 * (number of made ticks + achieved blackboard points)/ possible number of ticks
The ticks are not weighted by the possible black board points or anything else.
final mark |
success quotient |
1 |
≥90% |
2 |
≥80% |
3 |
≥70% |
4 |
≥60% |
Online checking system
The website of the online checking system can be found here:
At the first login, a password has to be required, which is sent to the email address
resulting from TUGonline. The solved exercises of the current week have to be marked online
until one day before the respective exercise class (Tuesday) before 11:00 p.m.
The list of participants is generated automatically from the TUGonline inscriptions.
Erstellt am 2.3.2017, modifiziert Feb. 2018.
Impressum:
Für den Inhalt verantwortlich:
Wolfgang Woess, Gundelinde Wiegel
Institut für Diskrete Mathematik
TU Graz, Steyrergasse 30, A-8010 Graz