Meseret Gebre » Research http://meseretgebre.com Fri, 04 Jun 2010 07:15:05 +0000 http://wordpress.org/?v=2.9.2 en hourly 1 Muse Code Generator http://meseretgebre.com/archives/muse-code-generator/ http://meseretgebre.com/archives/muse-code-generator/#comments Wed, 25 Mar 2009 16:25:53 +0000 mez http://meseretgebre.com/?p=116 My latest work on research has been the development of code generation tool that basically sets up all the needed directories for a muse simulation to work correctly. It also gives users the option to generate agent,state, or event classes which are subclasses from muse base classes. The language I’ve chosen to create this tool is ‘kick ace’  python. Python is probably is my favorite scripting language. Once the tool is complete it will be documented on the muse site @ musesimulation.org

]]> http://meseretgebre.com/archives/muse-code-generator/feed/ 0 Research Work Issues http://meseretgebre.com/archives/research-work-issues/ http://meseretgebre.com/archives/research-work-issues/#comments Tue, 24 Mar 2009 09:31:13 +0000 mez http://meseretgebre.com/?p=106 I had a freaking break down today. There are just so many things going on with research and at times can feel very over whelming. I took this day off to get a hold of myself and to reorganize my current attack plan. Things on my list that have to be done for research include:

  • Thesis write up
  • muse benchmarking
  • muse website

My issue today was with testing.  Ignoring the fact that I am not a c++ expert (in my opinion the shear complexities of c++ requires many years of development to master, and the word expert should not be used to lightly),  dealing with code that runs in parallel is just a pain the a**. There is no easy way to test, no easy way to unit test. When a bug is discovered, well, good luck because hunting that bug down will consume valuable time. I have managed to make good progress on the muse website musesimulation.org. However my work on the thesis writeup has been at a stand still for the last week and a half because I have been developing simulations for my benchmarking.

Benchmarking is a whole other ball game. I found the best way to find bugs in the kernel is to create different types of simulations and see if they work. Everytime I create a benchmark simulation I find an error. This got me thinking, why is there no good way to test parallel code? However I am determined to create something useful and I will give it all I got :-)

Anyways, this is my rant for the day and if you have some tips about parallel development please feel free to leave some hints.

]]> http://meseretgebre.com/archives/research-work-issues/feed/ 0