panthema / 2006
Thumbnail of a network in the NetFundamentals Technical Report

NetFundamentals Seminar - Technical Report Finished

Posted on 2006-12-19 18:19 by Timo Bingmann at Permlink with 0 Comments. Tags: netfundamentals university

Today the technical report on Robert Gallager's Minimum Delay Routing Algorithm Using Distributed Computation was finished. It is part of my work for the NetFundamentals seminar organized by Decentralized Systems and Network Services Research Group at the Institute of Telematics. In the seminar six of the computer science papers most worth reading are reviewed and presented. The technical report was written by Dimitar Yordanov and myself and a presentation will follow.

The technical report can be downloaded as PDF (470 KB)

Abstract

One of the computer science papers most worth reading is Gallager's algorithm for minimum delay routing. The merit of Gallager's paper is its rigorous mathematical approach to a problem, which is more often taken care of using heuristics. The approach is founded on a well designed mathematical network model, which is custom-tailored to describe the minimum total delay routing problem. Mathematical observations on the model lead to two conditions for achieving global optimization, which are based on the marginal delay of links and neighbors. From these observations and conditions an iterative, distributed routing algorithm is naturally derived. Gallager finishes his analysis by proving in detail that the algorithm achieves total minimum delay routing. Algorithm and model are reviewed and illustrated in detail in this technical report.

QtSqlView Screenshot 1

QtSqlView 0.8.0 Released

Posted on 2006-10-10 12:56 by Timo Bingmann at Permlink with 0 Comments. Tags: qtsqlview c++

Released the first version 0.8.0 of QtSqlView under the GPL.

QtSqlView is a simple and easy to use SQL database browser written in Qt 4.x using the excellent QtSql components. Using QtSql drivers it can natively connect to MySQL, PostgreSQL and SQLite databases. Furthermore other database systems may be accessed using their ODBC drivers. QtSqlView is released under the GNU General Public License: source code and win32 binary may be downloaded here.

This short program was initially written for a set of windows users, who need to access and edit a PostgreSQL database. All this is possible with M$ Access and ODBC, but the configuration of PostgreSQL's ODBC driver and the ODBC DSN is far too complicated for the average database editor. Thus problem-free access of open-source databases was top priority for QtSqlView.

QtSqlView boasts the following features:

You may download the source code for Linux/OSX or a setup package for Windows.

More screenshots are available as well.


SDIOS06 Shell Screenshot

SDIOS06 - Source Code and Ready-To-Run Image

Posted on 2006-09-14 09:26 by Timo Bingmann at Permlink with 2 Comments. Tags: sdios06 university c++

As promised the source code to SDIOS06 was released under the GPL.

SDIOS06 is a toy operating system developed during the practical course SDI (System Design and Implementation) at the Systems Architecture Group of the University of Karlsruhe. It was designed and written by Timo Bingmann, Matthias Braun, Torsten Geiger and Andreas Mähler. Two games were ported to a custom SDL implementation using the VMware "hard"-ware: SDLjump and SuperTux. For more information and screenshots see my blog entry 20060727-SDI-Demo.

The source code archive was published on the L4Ka.org page: http://www.l4ka.org/86.php

A local copy of the source archive (7.4 MB) is available as well. The README file contains a great deal of information about SDIOS06's design and modules. The complete source code can be browsed on the web.

To make demonstration as easy as possible a ready-to-run binary vmware image (3.8 MB) can be downloaded. The image contains SDIOS06 installed on a virtual vmware disk image. The VMware image can be run using the free VMware Player.


Mandelbrot Example

sdlfractal 0.1

Posted on 2006-08-09 21:33 by Timo Bingmann at Permlink with 0 Comments. Tags: sdlfractal c++

Sdlfractal is a port of a simple little fractal generator which I wrote some five years ago. It is not supposed to replace wonderful tools like xfractint so it is very simple and not very fast. The main goal of this project was to learn SDL and it turned out that fltk was also required. The main construction is an SDL surface extended by a canvas class which simulates a high-resolution coordinate system. On this coordinate system the fractals can be drawn.

To control fractal parameters the generator displays a second window using the fltk engine. From this dialog the currently displayed fractal and it's parameters can be changed. This requires a dual engine event loop in the program.

By dragging the mouse on the drawing canvas you can zoom into all fractals. If fractal drawing takes too long (and nothing is shown), then just click the canvas and the generator will stop.

The fractal generator can also save high-resolution PNG files. The following images are some examples created by the fractal generator.

The source code to sdlfractal 0.1 can be downloaded in a tar.bz2 archive (121 KB). It is also browsable on the web.

Sdlfractal was designed to the portable to Win32 using SDL and fltk compiled with MinGW. A compiled version which should run out of the box on most windows. Download the zip archive(267 KB) containing the executable.

MD5sums of the source and binary archives:
538da60a5ef2d427fbb901d6080e631e sdlfractal-0.1.tar.bz2
5bab1ccb93e170f5c42d995a6761ca7a sdlfractal-0.1-win32.zip

Mandelbrot snow storm

This section of the mandelbrot fractal is so beautiful that it is my current wallpaper. It can be downloaded at 800x600 (242 KB), 1024x768 (372 KB) or 1280x960 (552 KB).

This blog entry continues on the next page ...

Front page of slides

Talk about Study Thesis - Slides and Movie

Posted on 2006-08-02 19:43 by Timo Bingmann at Permlink with 0 Comments. Tags: compsci study thesis graphviz university talk

At the University of Karlsruhe it is common to give a concluding talk after finishing a study thesis (Studienarbeit). I presented my study thesis "Visualisation of Very Large Graphs" today at the mid-day seminar of the Theoretical Computer Science Group. The slides of the talk are available as PDFs below:

studythesis-talk-visualisation.pdf (1.6 MB) - also available with two slides per page (1.6 MB)or four slides per page (1.6 MB).

During the talk I showed a movie, in which a r-tree of the autobahnen of Germany is constructed incrementally. I actually produced two movies for this purpose: in the first the edges are added in a given sequential order and in the second the order is randomized. Both movies are encoded using transcode and the XviD codec.

Sequential Movie (3.6 MB)
Randomized Movie (5.1 MB)

The talk was received very well, and I may continue in this field with my diploma thesis.

The visualisation library is used by a route planning algorithm developed at the Theoretical Computer Science Group. By integrating my study thesis it was possible to create a Java web applet which animates the route planning algorithm. The Java applet is currently available at http://algo2.iti.uni-karlsruhe.de/schultes/hwy/demo/.

This blog entry continues on the next page ...

SVG Logo der Universität Karlsruhe

Posted on 2006-08-01 19:22 by Timo Bingmann at Permlink with 2 Comments. Tags: university

Für meinen morgigen Vortrag zur Studienarbeit, habe ich (viel zu) lange nach einem vektor-gezeichneten Logo der Univerisität Karlsruhe (TH) gesucht. Es war nur sehr schwer eines zu finden. Aus einem nach langen Stöbern gefundenen PDF habe ich dann folgende SVG, PDF und EPS Grafiken mit Inkscape erzeugt, die bestimmt jemand anderem etwas Mühe ersparen:

UniKa Logo Farbig SVG SVG (Inkscape) EPS PDF
UniKa Logo Graustufen SVG SVG (Inkscape) EPS PDF

Hinweis: Die Grafiken sind keineswegs offiziell oder sonst wie von der Univerität bereitgestellt.


Presented Practical Work for SDI (System Design and Implementation)

Posted on 2006-07-27 20:00 by Timo Bingmann at Permlink with 0 Comments. Tags: sdios06 university

Today was the final demo presentation day for the practical course SDI (System Design and Implementation) at the Systems Architecture Group of the University of Karlsruhe. Goal of the course was to implement a simple operating system on top of the L4 microkernel, which is developed at the Group.

SDIJump

Our practical course group was one of the two groups to demonstrate a working system on the last day of the semester. The presented operating system contained following basic components:

Furthermore we utilized the VMWare video "card" to enable graphics output. The "hardware" is well documented in the source code to the X11 video driver contributed by VMWare.

The frame buffer and keyboard input was linked up to a ported SDL implementation. After adding a lot of POSIX library function calls like fopen, opendir and getenv, it was possible to compile our favourite games: SDLjump and SuperTux. Both run very well on our operating system from scratch project.

Some screenshots of the shell, sdijump and supertux are available. In future the whole source code and compiled binary images may become available

A big thank you to my friends Matthias Braun, Torsten Geiger and Andreas Mähler, who made this practical work possible and especially fun!

This blog entry continues on the next page ...

Front Page of HelpPC HTML version

Nostalgic HTML Conversion of the HelpPC Shareware Reference

Posted on 2006-07-10 15:00 by Timo Bingmann at Permlink with 4 Comments. Tags: helppc fun

Browse the converted HelpPC Reference Library

The HelpPC Shareware Reference is an collection of reference information from 1991 concerning the PC and PS/2. It was created by David Jurgens and includes a hypertext viewing program for DOS. The reference library may be old, but many parts are still applicable to present-day PCs.

During some nostalgic moments I remembered the fun and motivation I received out of this reference during my early programming years. I found out that only a very crude transcription into HTML existed on the web at http://heim.ifi.uio.no/~stanisls/helppc/and decided to create one which is more true to the source.

Main Goal for the new converter was to retain the look and feel of the old DOS hypertext program: the use of CSS including the pseudo-class "hover" allowed me to create a HTML page with remarkable similarity. The special ASCII character are not included in the standard HTML entities, however the can be found in the Unicode character table. The converter uses a carefully picked set of Unicode dingbats and line symbols to replace the undisplayable control characters. Thus the HTML pages contain no images and can be displayed by any browser utilizing Unicode fonts. All current-day browsers (Firefox 1.5, IE 5, Opera 9.0) display these symbols well.

You may browse the converted files and can download an offline version: as a zip archive (1.2 MB) or a tar.bz2 archive (257 kB).

The converter itself is a Perl script which requires the Template Toolkit and the original HelpPC distribution. All output files are valid XHTML 1.0 Strict. To run the converter download the following files from on of the locations:

Original HelpPC distribution (255 kB) from Simtel.NET local copy
Converter Perl Script (15 kb) download display code

Extract the zip file and run the Perl script in the path containing the .TXT files. The script will create a subdirectory "html" with 1015 XHTML files. You can modify the nostalgic.css or the embedded text templates to fit your taste.


Studienarbeit "Visualisierung sehr großer Graphen" fertiggestellt

Posted on 2006-06-20 17:31 by Timo Bingmann at Permlink with 0 Comments. Tags: compsci study thesis graphviz university c++

Heute habe ich meine Studienarbeit "Visualisierung sehr großer Graphen" am Institut für Theoretische Informatik fertig gestellt und abgegeben. Die Studienarbeit ist eine 3-4 monatige wissenschaftliche Arbeit und dient als Vorbereitung auf die Diplomarbeit.

Die Studienarbeit kann zum Durchlesen als PDF (2,4 MB) herunter geladen werden. Auf diesen Abschnitt folgenden die deutsche und englische Zusammenfassung der Arbeit. Weiter gibt es eine Version zum Ausdrucken (PDF 15,3 MB) mit einer hochauflösenden Vektorgraphik im Anhang.

Zusammenfassung

In dieser Studienarbeit wird untersucht, mit welchen Methoden sehr große Graphen wie ein Straßennetzwerk von Europa effizient und komfortabel visualisiert werden können. Als Ausarbeitung entsteht ein C++ Rahmenwerk für die Datenhaltung eines Graphen mit Attributen und ein Java Applet, das mit dem Datenhaltungs-Server mittels CORBA kommuniziert. Das Rahmenwerk kann leicht in bestehende Graphanwendungen integriert werden, um deren Algorithmen zu animieren.

Als Basis-Graphstruktur wird ein Adjazenz-Array verwendet und um Strukturen erweitert, die zu jedem Knoten und jeder Kante beliebig viele Attributwerte speichern. Zwei der Knotenattribute werden als Zeichenkoordinaten verwendet. Der Grundgraph und die Datenhaltung der Attributwerte wird auf möglichst kompakte Art und Weise gelöst. Graphanwendungen können eine Liste von temporären Änderungen erzeugen, die mit dem großen globalen Graphen zusammengeführt werden können. Um das Vorgehen der Graph-Algorithmen zu visualisieren, werden deren Funktionsaufrufe in einer Änderungsfolge kodiert, welche als Animation zum Java Client übertragen wird.

Um die Geschwindigkeit einer Ausschnittsanfrage zu erhöhen, wird die mehrdimensionale Indexstruktur R-Tree verwendet. Diese ermöglicht Anfragezeiten, die linear zur Anzahl der zurückgelieferten Kanten und unabhängig vom gewählten Ausschnitt sind. Es können komplexe Filterausdrücke aus Vergleichsbedingungen mit boolschen und arithmetische Operatoren verwendet werden, um die angezeigten Kanten in einem Visualisierungsauschnitt einzuschränken und so komfortabel bestimmte Aspekte der Anwendungs-Algorithmen zu untersuchen oder hervorzuheben.

Als Referenzanwendung wird das Rahmenwerk von der am Institut für Theoretische Informatik in Karlsruhe entwickelten Routenplanungsanwendung zur Visualisiserung mittels Web Applet verwendet.

Abstract

This study thesis investigates and implements methods used to efficiently visualize very large graphs like a street network of Europe. A C++ server framework is designed, which implements a data management library for graphs with attributes. A Java applet communicates with the data server via CORBA and draws sections of the graph. The graph data management library can easily be integrated into existing graph application to visual and animate calculations.

The data management library uses the adjacency array structure for representing the base graph. It is extended by similar data structures to hold an arbitrary number of attributes for each vertex and edge. The data structures are optimized towards highest storage efficiency. To modify the static global graph an application may construct a list of changes, which are efficiently merged into permanent data storage. To visualize the workings of an algorithm its function call sequence can be recorded into a list of graph changes. This change time line can be transfered to the Java client and displayed as an animation.

To accelerate access to arbitrary sections of the graph the spatial index structure R-Tree is used. It enables query times to linearly increase with the number of returned edges and be independent of the section size or position.

Furthermore complex filter expressions including comparisons and arithmetic operators can be applied to limit the displayed edges. This enables the client to explore the graph's details comfortably and highlight interesting aspects of an algorithm.

The graph data management library is used in a route planning application, which is being developed in a research group of the University of Karlsruhe. It will be used to visualize the route using a web applet.


Thumbnail of a slide from the Name Service Design talk

Talk in SDI Lab: Naming Service Design in a Multi-Server Operating System

Posted on 2006-06-01 17:00 by Timo Bingmann at Permlink with 0 Comments. Tags: sdios06 university talk

As part of my work for the SDI Lab work this summer, I held a short talk about how to design a name service in a multi-server operating system like L4.

The slides are available here:

1 Slide per page: NameServiceDesign.pdf 163 kB
2 Slides per page: NameServiceDesign-1x2.pdf 163 kB
4 Slides per page: NameServiceDesign-2x2.pdf 158 kB

Here the table of contents:

  1. Goals
  2. Name Catalogs
    1. Object Representation
    2. Catalogs
  3. IDL Interfaces
    1. Resolve Interface
    2. Bind Interface
  4. Extension / Ideas

RSS 2.0 Weblog Feed Atom 1.0 Weblog Feed Valid XHTML 1.1 Valid CSS (2.1)
Copyright 2005-2014 Timo Bingmann - Impressum