Thesis Log

While I was writing my master thesis I used this side to keep track of what happened when. This side contains images and videos documenting my progress.

2007-05-18 update

While I currently don't have the time to be working on this project anymore, but in a quiet moment I still work on this project, hoping to have a new version ready by the end of the year, then performance should be much better, keep tuned.

2004-09-29 update

Today I finally presented my thesis to my advisor. So here are the presentation slides from that talk. Currently available in German only. English version is forthcoming.

2004-09-24 update

Finally it's done. My thesis is printed out and ready to be handed in. So it's finally time to update this page. Apart from writing on my thesis in these last two months quite some changes have been made to the reaction diffusion simulator. To name just a view of the many changes: transparencies are working again, a nasty bug was fixed on several GUI elements, the GUI has been updated so that only features are accessible, that can be executed in that context, and much more minor bug fixes in all parts of the program. So here is the updated version.

Since reading my thesis will help to understand what this project is about and just as important how to use it, here the PDF version of my thesis.

2004-07-27 update

Here a slightly changed version of the reaction diffusion simulator tools. There were some changes in the user interface. Furthermore the reaction plugins should now be configurable. This enables the integration of the reaction diffusion tools with other SBW modules. The main issue right now is the speed of execution. But in principle it is now finally possible to create a scene with the configuration unit then add reactions with JDesigner and finally simulate it using the simulation unit in unison with Jarnac .Amazing!

2004-07-23 update

Here finally a new version of my reaction diffusion simulator. There have been major changes in the user interface. Though I still expect some bugs to be present I think this version is fairly stable.

A tutorial and documentation about all the features will be forthcoming early next week. Until then please bear with the old how to document.

2004-07-22 update

Here the new version of the SBW - BioSPICE analyzers. Please make sure to have SBW, the SBW Modules and BioSPICE already installed before installing the analyzers. Be also sure to close BioSPICE before installing the analyzers.

2004-07-16 update

Time for a short update. These days I put a lot of effort into stabilizing the configuration and visualization unit. There are major improvements in the GUI design as well (including dockable windows and such). So I expect a new release to be ready by the end of next week. Till then here is a very interesting movie of a 100x100x100 simulation. That one ran continuously for 10 hours involving the Marching Cube plugin:

movie (you might need a new divx codec. Find it under DivX.com)

2004-07-07 update

A lot as happened since the last entry. Concerning the SBW Broker, a first version with remote capabilities is available.

Concerning the reaction diffusion project. The configuration file format is now finally SBML level2. Furthermore channels have been added to the configuration and simulation. Here an example of a nice 2D simulation involving two compartments that are only connected by one channel. The first picture was created using a texture generating visualization plugin, while the second picture uses a combination of the marching cube and point cloud plugin.

2004-06-15 update

Here is an updated version of the reaction diffusion simulator . With this version it is possible to save and continue a simulation with the visualization unit. Thus the simulation unit and visualization unit can be used alternately. (i.e. use the visualization unit to simulate the first steps, till you see the beginning of a pattern ... then save the simulation and continue it with the simulation unit).

Further changes introduce the feature of a clip-plane, that can be moved through the scene. This removes the problem that sometimes you couldn't see what you really wanted because some structure was in the way.

Finally this installer now includes all required DLLs so no additional download will be needed.

2004-06-14 update

Again a new version of the SBW Broker. The Installer includes the following modules:

• Inspector,
• Jarnac,
• JDesigner,
• libSBML - NOM,
• matlabTranslator and
• metatool

Changes :

• bug fix of a memory leak during registration of modules
• bug fix of registration process, should a module be registered more then once the old entry will be overwritten,
• added auto-registration process.

2004-06-09 update

New version of the SBW Broker including Jarnac, JDesigner.

2004-06-08 update

Here a new version of the C++ SBW Broker. (local functionality only). This time the release version should work fine. Here some interesting links for solving problems with release vs. debug version problems:

• Debug / Release
• Release Debugging
• Common Problems switching from Debug to Release Build

2004-06-07 update

Ok, here it is ... a first version of the reaction diffusion simulator. Admittedly there is still quite some work to be done. But the current version is stable enough to give you a first glimpse of it. The ZIP file contains a Win32 version of the programs. Please note the HOW-TO document. For questions or remarks mail: fbergman@kgi.edu.

In case you don't have the qt-mtnc321.dll you have it here just copy it to the bin directory of the installation - or somewhere in your PATH directories. Furthermore if you have not used OpenSceneGraph before you might need the OSG libraries. Find them here.    

2004-06-02 update

Here a current version of the C++ SBW Broker. (local functionality only). 

2004-05-28 update

Finally ... it's working. I removed a major flaw in the simulator regarding behavior around the membranes. (Thanks a lot Cameron for pointing that one out.) So here now two pictures ... the first one is of a 2D simulation of 100x100 points. (That is actually how the pictures from May 24th are supposed to look like.) And the second one is a 3D simulation with 100x100x100 points.

2004-05-24 update

Ok, integration of simulation and visualization unit is now completed. So the next step would be an integration of the configuration unit. This would enable the application to accept user input at runtime and so the simulation could be influenced. 

Btw ... here a picture of a 2D simulation with membranes on the upper and lower side. It is interesting to see that patterns appear parallel to these membranes. At a later stage these lines begin to break up into dots.  

2004-05-19 update

Here the most current updates. I finally switched over to using palettes instead of figuring out the color to display by fancy formulas. This finally gets rid of the overall 'pinkness'. There will be several palette - files so that the user can decide how the concentrations will be displayed.

As far as the Broker is concerned, now the registry is working too. Thus startup of requested programs is now working.   

2004-05-17 update

Finally a prototypic version of the C++ Broker is working. Thus designing a model with JDesigner and simulating it with Jarnac finally works. Next steps would be to implement registry support and optimize it. But these tasks have lower priority.

While porting the Broker I kept updating my thesis document so that I should now finally be prepared to continue on the thesis program. Next steps here are building the integrated version I talked about on 2004-04-14. And to add a legend to the visualization plugins. 

2004-04-29 update

I guess it's time for an update. Currently I'm involved in porting the SBW Broker from Java to C++. A very interesting problem. How does this help me with my thesis? Well basically since the three units configuration / simulation / visualization will be SBW enabled it helps to be thoroughly acquainted with the system. Currently about 10% of the port is done.

2004-04-14 update

Finally the work on the configurator progressed so far, that moving and deleting of elements is efficiently possible. Also the work on the insertion tools is done for this stage (later on a new insertion tool will be added to be able to add channels). 

So for the next couple of days I'll focus on revising the simulator. And by the end of the week I should be able to focus on creating an integrated version - configuration-,  simulation-,  and visualization-unit in one application. 

2004-04-10 update

Here some information about what I've been doing the last weeks. The plugin mechanisms for different reaction schemes and for different visualization modes have been revised. The visualization plugins now work in a different thread. This way a visualization that takes too much time (happens when too many data points are to be displayed) can now be cancelled.

Since that is working now I focused again on the configuration unit. First a new Configuration object was introduced that follows a more object oriented approach.

And right now I focus again on the insertion tools ... but that will take some more time.

2004-03-25 update

It's been quite some time since I recorded updates here. So I just want to mention the current progress. Marching Cube and MarchingThetrahedron Algorithms were implemented and are working. So I started looking into membranes. Here are some pictures of a 100x100x1 simulation that I found rather interesting:

I found these pictures rather convincing. Sadly the effects you see on the membranes look rather bad in the 3D case. What happens here is that the Marching Cube / MarchingThetrahedron algorithms will just create a surface "around" the membrane ... like this: 

2004-03-14 movie

Ok ... after several days of hard work, I now got the marching cube visualization working. Furthermore I rewrote the simulator so that the minimum, maximum and mean for each iteration and the whole simulation will be calculated.

The results seem rather promising. Though there are some minor flaws in the plugin.  

this was a simulation of a 50x50x50 grid using the marching cube visualization.

2004-03-11 movies

Well ... it might be, that I was a little to optimistic two days back. Calculating the histogram on the fly will take too much time. Thus the low frame rate (2.0 fps) in the movie files you see below. But there are several ways how to fix this ... I might be able to calculate the histogram while doing the simulation.

Furthermore the histogram so far is based on one iteration at a time. Which is not the right thing to do. Since it is not possible anymore to see the diffusion progress as it is ... anyway here are the movies:

this was a simulation of a 50x50x50 grid with the point-cloud visualization.

this was a simulation of a 50x50x50 grid with the simple-cube-visualization.

2004-03-09 new images

... it seems like I start thinking in the right direction now. No use in displaying all data points every time. So I added two possibilities either let the user choose a threshold level restricting the data points, or calculate one dynamically. Here is what it looks like if I choose a dynamic approach for a 50x50x50 grid (should work equally well with a 300x300x300 grid yet to be tested though):

 

2004-03-05 presentation

Today I created a short presentation for the research retreat tomorrow. You can find the slides here. For the movie file on the last slide you will need the divx codec as found on www.divx.com

2004-03-03 New Simulation Results

ok, I changed the visualization module. Instead of cubes I display now only points. Thus larger grid sizes become possible. Though in the 3d case it is hard to keep track of the individual concentrations. Here are two short movies:

this was a simulation of a 80x80x1 grid.

this was a simulation of a 50x50x50 grid.

this was a simulation of a 100x100x1 grid.

2004-02-26 First Simulation Results

well, so far the simulation went rather well, so theoretically I would have lots of data.
On the down side, I can't evaluate the data, since so far I have no means to visualize
a grid larger then 20x20x10 and be able to move through the scene:

Here some first pictures of a diffusion in a *sic* 10x10x10 grid: