##Hypergraph representation
It would be extremely nice to have a HyperGraph class similar to other network classes in networkx.
However, I still don't have a really good idea of representing hypergraphs. Probably the best way would be to store nodes and hyperedges and add other info as some kind of surplus?
My current idea is to inherit after networkx Graph class and provide similar interface. Basically what it does now is representing hypergraph as clique on nodes adjacency level (I suspect that multigraph approach would be better) and as graph of hyper edges.
I need to write more methods and think about this multigraph approach because it sounds pretty intersting!
##Networkx tutorial thoughts
It was a very simple tutorial!
##Hypergraphs and matroids paper
What is incidence matrix?
Oh, this paper wasn't really good. But it used the book: Berge-Holland hypergraphs and it looks promising
I have this idea of writing a series of blogposts about hypergraphs using this book as reference material to understand it better and promote hypergraphs as a whole. Cool!
##Materials about markov chains from UW
It looks really promising!
Równanie Chapmana-Kołmogorowa - I have no idea, but when I was checking it with octave it didn't hold!
##Diffusion on markov chains
I really have this specific uncertainty about it. The best way to be sure is to check it. I'm reading now the ipython notebook on markov chains.
##Time to get wiser?
I downloaded a book about markov chains.
The article/lecture was helpful! It has a theorem with choosing the one element and powering matrix, yay.
I'll check it with my case. It prooves this theorem with degree of the node and it's stationary distribution pi!
It should be the case with my hyperedges "degree", it's easy to check.
It's again a few days/weeks later
##21-04-2014 Tests in hypergraph repository
I think that I should think more about what I do here. There is a lack of design. It's a bit too much bottom up work.
I have some distinct modules and notebooks. And this is awesome.
I would like to work more on those notebooks to make them tell complete stories. It's important for my research. I don't want to realize that, "well, I haven't thought about all those corner cases, and it was a premature generalization". It's important to think!
To ensure good design and that my software works - I want to introduce testing to the system.
Tests don't have to be too complicated. I think that python nose would be good enough.
But what do I want to test? Probably it would be awesome to have every module tested. I also have those "demo modules" and notebooks. And they should work well.