Monday, May 27, 2019

100 days of github - take 2

2019-May-26, Sunday
Day 1
I had started #100DaysOfGithub once before, and it failed fast and hard.
So, here's take 2, and hoping I will stick around until the end.

I've been using edgelists to save graphs, not realizing I lose information about isolated nodes. I didn't have to worry about this earlier, because from a work point of view, I worked exclusively with connected graphs. But, now that I'm, um, branching out, I better start using adjlists.

https://github.com/malch2/networks/blob/master/EdgeListsAndAdjLists.ipynb

Wednesday, May 8, 2019

Getting Started With NetworKit

NetworkX is unable to handle large networks, with nodes ~100k.
So, let's see if other modules/packages can help me out. I turn to NetworKit.
For future me, this is how I set it up on my system:

1. Got CMake from here.
2. conda config --append channels conda-forge
3. conda install -c vgauthier networkit

Next, go here for an intro.
What I basically need is to find number of connected components. Check here for functions related to this.