Numero
Veckobladet om forskning,
undervisning och
administration, m.m.
Nummer 26· 26 augusti 2004 · Årgång 34      
NADA

ram

Docentföreläsning

Lars Engebretsen, sal E3,
måndagen den 13 september klockan 15.

Valfrihetens pris

Ett kombinatoriskt optimeringsproblem ses oftast som en statisk enhet, där det gäller att maximera någon viss, given, målfunktion under vissa bivillkor. Problemet att lösa optimeringsproblem ses ibland som en kamp mellan två personer: en fiende som försöker
konstruera en “svår” målfunktion och “svåra” bivillkor, och en hjälte som försöker räkna ut svaret.
Detta synsätt passar dock inte uppenbart in på problem som härrör från ekonomiska system eller andra stora komplexa system med många
aktörer. Dessa system kännetecknas istället av att varje spelare i systemet har sin egen, subjektiva, uppfattning om vad som är bäst för just honom eller henne. Det naturliga optimeringsproblemet blir då att varje spelare försöker maximera sin egen nytta, enligt sin egen uppfattning om vad som är bäst. Märkligt nog finns det trots detta i många fall ett objektivt mått som kombinerar spelarnas individuella åsikter; detta mått tolkas sedan vanligtvis som priset för en viss produkt eller tjänst på en marknad.
Att det finns ett objektivt pris är tilltalande i sig, men det vore än mer tilltalande om det gick att räkna ut det. Det visar sig att man i vissa fall faktiskt kan räkna ut de objektiva priserna med rimlig ansträngning, men att det i de flesta realistiska fall är svårt att räkna ut dem.

Lars Engebretsen

 

Pseudo-Empirical Studies
Magnus Boman
Thursday Aug 26, 13.00-14.00
Room 502, 5th floor, Forum building

Magnus Boman is a professor in Intelligent Software Services at the Royal Institute of Technology His professor’s chair is placed at DSV, where he is also a Stockholm University docent since 1997. He also works at SICS, where he is the leader of the HUMLE research laboratory.Kicking of the new seminar series
“Current Research Topics at DSV”
see http://www.dsv.su.se/~mab/CRTvt04.htm

Abstract
Results from empirical studies are sometimes used for prediction purposes. For instance, the spread of previous year’s flu may be studied with the purpose of predicting some of the epidemiological characteristics of this year’s flu. For phenomena with less regularity than flu, however, this method has several drawbacks. For some objects under study, historical data is meagre, such as when a new pathogen appears, for example. For most objects under study, the society or more generally the environment in which earlier empirical studies were carried out differ considerably from the society subject to the predictions. For all objects under study, some non-trivial mapping from previous studifes onto the future must be made. The mapping must be made with dynamics of time as well as space in mind, and usually also under new and uncertain policy conditions. That this is hard goes without saying, but as there have been no good alternatives, this is how predictive results have been achieved in the last century. In this research, a new method in aid of the traditional one is presented. Since it subjects artificial populations (i.e., populations of artificial individuals mapping 1:1 to real populations) to empirical study, we name it pseudo-empirical studies. [One could read pseudo-empirical studies as “empirical studies of surrogates (pseudo-individuals)” but since the empiricism itself here is somewhat synthetic, we stick with the shorter term, which also is in keeping with the dictionary definition “not genuine but having the appearance of”.

 

Extra Exjobbspresentation i datalogi

Bookmarks and Content in a Concept Browser
Predonescu, Ioana

Tisdagen den 31 augusti 2004
kl 10.15 i rum 4523

Ämne: Datalogi Program: D, Uppdragsgivare: Cid, Nada
Handledare: Ambjörn Naeve och examinator: Yngve Sundblad

 

Welcome to Wireless Seminar


Success Factors in Building Research Networks – The Experience of the National Capital Institute of Telecommunications in Canada
Speaker:Dr. Maike L. Miller,Canada
Monday 6 September at 13.00
Venue: Wireless@KTH, Isafjordsgatan 30b,Kista
Read more: http://www.wireless.kth.se


Statistisk rådgivning till
doktorander och forskare vid
naturvetenskaplig fakultet

Behöver du hjälp med att analysera dina insamlade data? Eller med att planera en undersökning eller ett experiment? Då ska du kontakta Statistiska forskningsgruppen (SFG).
SFG hör till avdelningen för matematisk statistik vid Stockholms universitets matematiska institution och bedriver statistisk
konsult-verksamhet gentemot forskare, företag och myndigheter.
Du som är doktorand eller forskare vid SU:s naturvetenskapliga fakultet erbjuds kostnadsfritt 10 timmar statistisk konsultation under år 2004. För konsultation utöver detta är kostnaden 100 kr/timme. Detta erbjudande har blivit möjligt tack vare ett särskilt anslag från
naturvetenskapliga fakulteten.
För att anmäla intresse av konsultation, kontakta SFG:s föreståndare Jan-Olov Persson på telefon eller e-post.
Mer information om SFG:s verksamhet finner du på www.math.su.se/matstat/sfg/.


Jan-Olov Persson

 

Guest lecture on generation of
emotion in music 30th of August

Place: lecture hall 510 in the Forum-building
in Kista.Date: 30th of August, 10.15 - 12
We are proud to invite you all to a guest lecture by PhD Roberto Bresin from the Speech, Music and Hearing departmenet at KTH.

Part I: The science of music performance
The most important techniques for measuring a performance
The acoustical cues of importance for communicating expressivity
How the use of acoustical cues can influence the performance style
A system for the synthesis of music performance
Part II: Emotion in music performance
The theories presented in Part I will be expanded to the field of emotionally expressive music performance. Examples will be provided as well as possible applications. Among the latter, the use of expressive ringtones in mobile phones, and an application for visual display of emotion in music performance.


Kristina Höök & Petra Fagerberg

 

Masters Thesis Presentation


VoIP quality aspects in 802.11b networks
Candidate: Victor Nunes, Instutito Superior Tecnico (IST), Lisbon
Opponent: Sherif Yousef, IMIT, KTH
Advisor: Ian Marsh, SICS
Examinator: Olof Hagsand, LCN, KTH

Place: KTH Forum, Kista, LCN Seminar room, 8th Floor, Lift B
Date: Friday 27th August 2004, 10.00

Abstract:
IEEE 802.11b is increasingly being used in public and office networks. Therefore it must cope with both voice and data traffic. In this
thesis we assess the suitability of 802.11b networks to carry real-time voice traffic using the IP protocol.
Our measurements focused on the three major IP metrics: loss, delay and jitter. We have looked at 802.11b networks in both ad-hoc and infra-structure mode. We have also recorded the environmental factors such as the distance between the nodes and information about obstacles where possible.
We have found that 802.11b networks can support the requirements of real-time voice communication when there is clear line of sight to the peer node or the communicating nodes are close. We have recorded and made available an extensive repository of the voice sessions we captured for further study.

 

Master thesis seminar

Tuning of Lead- and Lag-filters
Speaker: Jonas Gärd
(NOTE THAT THE PRESENTATION WILL BE IN SWEDISH):
Seminar date: Friday August 27 at 9.15
in seminar room, 3rd floor,
Osquldas väg 10, KTH

The work describes the creation of an adjustment algorithm that will replace today’s manual adjustment of Lead- and Lag-filters at ABB Robotics industrial robots. The adjustment is mainly for the filter parameters in the control system. Three different algorithms have been created to replace the manual adjustment. One of them is based on the criteria from the manual adjustment and another one on optimization of the sensitivity function. The last one is a combination of the previous two. In the report you can follow the creation of the algorithms and how they behave during simulations and tests on real robots. All three algorithms are multiple times faster than
today’s adjustment and present equivalent results at the same time.
WELCOME!


The Automatic Control group at S3, KTH


ram

 

 

Numero är institutionstidningen vid Nada - institutionen för numerisk analys och datalogi, KTH. Numero utkommer normalt på torsdags-förmiddagar under teminstid. Manus måste lämnas in före kl.12 på onsdagar.Manus, tips, förslag och andra bidrag till Numero kan lämnas på något av följande sätt:

Bidrag för artiklar och notiser bör i största möjliga mån vara färdigformulerade och korrekturlästa. Varje Numeronummer utkommer i två former:

Numeroredaktionen består av Rickard Buch. Ansvarig utgivare är Ingrid Melinder. Numeros innehåll uttrycker inte institutionens officiella ståndpunkt annat än då detta anges.

 

ram
ram
Kalendarium
ram

ram

Thursday 26 Aug

13.00-14.00

Pseudo-Empirical Studies
Magnus Boman, Room 502, 5th floor, Forum building

ram
ram

Friday 27 August

9.15



10.00






master thesis seminar
Tuning of Lead- and Lag-filters, Jonas Gärd
in seminar room, 3rd floor, Osquldas väg 10, KTH

Masters Thesis Presentation
VoIP quality aspects in 802.11b networks
Victor Nunes, Instutito Superior Tecnico (IST), Lisbon, KTH Forum, Kista, LCN Seminar room, 8th Floor, Lift B

ram
ram

Mon 30th August

10.15 - 12.00

kl 15.00

Guest lecture on generation of emotion in music
PhD Roberto Bresin. Lecture hall 510 in the Forum-building in Kista.

Exjobbseminarium i Numerisk Analys.
Stepsize Adaptation and Turning Point Calculation for Continuation Methods. Claes Robertson. rum 4523
Handledare: Michael Hanke (NADA)


Tisdag 31 augusti

kl 10.15

Exjobbspresentation i datalogi,
Bookmarks and Content in a Concept Browser. Predonescu, Ioana, rum 4523

ram
ram

Monday 6 Sept

13.00

Success Factors in Building Research Networks
Dr. Maike L. Miller,Canada. Venue: Wireless@KTH, Isafjordsgatan 30b, Kista

ram
ram

Måndag 13 sept

kl 15.00

Docentföreläsning: Valfrihetens pris
Lars Engebretsen. sal E3

ram
ram

 

ram
ram