I finally managed to make some experiments on my implementation of pso (available from here). I played a little with parameter on two different test: Iteration test - program tries 20 runs with in 2 dimensional space on problems introduced in last article (here). Dimension test - program tries 2 to 10 dimension on problems those three problems. Some nice plots went from this experimenting and I want to share them with you....

I’m working on particle swarm optimization as my project to Soft computing lectures. If you are interested you can find source code at my github. Here are some gifs I made playing with that: Sphere Sphere function as simple optimization problem. I used gbest topology here with population of 10 to see it clearly at plot. Rosenbrock function Rosenbrock or also banana function or valley function. More informations can be found here....

Recently I was struggling with slowness of vim while scrolling. I ran into this stackoverflow comment that explains how to debug in Python: :profile start profile.log :profile func \* :profile file \* " At this point do slow actions :profile pause :noautocmd qall! I’ve found out it slow downs mostly function Highlight_Match_Paren() and the fix can be found here But that is another story.

After few months after my graduation and therefore from submission of my bachelor thesis (A tool for computing Nash equilibria) I finally managed to release neng! There are few refactorings made to neng from version that was submitted as bachelor thesis. But the most important are features: Deals with noncooperative games in normal form. Computes all pure Nash equilibria - brute force. Computes all mixed Nash equilibria in two-players games nondegenerate games with Support enumeration algorithm....

I’ve finished reading book The Pragmatic Programmer and it has few interesting advices how to write better code and be better programmer. Here are some points that have great value to me (I’ve read it in czech so excuse if some term are badly double-translated :)): Take responsibility over your code, your work. Read technical book every month. Involve in community. No redundancy in code nor in documentation. Get the most orthogonality you can - every module should be independent of other modules....

Long time I really hated standard flow of IPython work: import mystuff obj = mystuff.myobject() obj.do_stuff() Then I would find out that I have to change code in mystuff and do: reload(mystuff) obj = mystuff.myobject() obj.do_stuff() But with autoreload extension there is no need to run those horrible and distracting commands again (my source is stackoverflow). Simply run in ipython: %load_ext autoreload %autoreload 2 And every time you now hit Enter key to confirm command the changes you made to the code will be reloaded....