Implement the shifting algorithm for 3 and 4-connectivity

[chaoxu]

Implement the fast shifting algorithm for connectivity. It include a fast algorithm for both 3 and 4-connectivity. (the algorithms also works for higher connectivity, but not as fast.)

The following two papers contain the idea of the algorithm:

​http://www.caam.rice.edu/caam/trs/90/TR90-15.pdf Rajan, A. (1987). Algorithmic applications of connectivity and related topics in matroid theory. Northwestern university.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​a81e5ce

add testing for 4 connectivity

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​4aa5dc2

The algorithm seems to fail when 2*r(E) > E, so take the dual

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​8b7c032

Fixed the bug

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​bef6a56

Only use edges in a spanning tree for computing 3 connectivity

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​14cdee9

use reduced representation for linear matroids

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​e5f5195

optimized 4_connected_shifting

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​cfc248d	should use non-zero instead of 1
​3dfdb43	Merge branch 'develop' into t/18687/shifting_algorithm
​95edcb0	fixed the bug

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​dd7b56f	linear time for single shifting
​28ebcb8	fixed bug: the representation might change between calls
​378fb37	polish up, add tests

[yomcat]

Needs rebasing.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​c54b4b3

Merge branch 'develop' into t/18687/shifting_algorithm

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​d4cde3e

put shifting into lean_matrix

[Stefan]

Something fishy is going on with the commits. When I click the green link above, it looks like the branch is removing all of the contents of matroid.pyx. The "git trac try" command seems to work fine, so I don't quite know what's going on here.

One thing that definitely needs fixing is that some makefiles get added by mistake.

Some more questions / issues:

• the shifting algorithm is not mentioned as an option in the is_3connected docstring
• the _is_3connected_shifting docstring mistakenly has a 4 in its first line
• there is no user-facing is_4connected that can benefit from the shifting algorithm.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​8a57453	Merge branch 'develop' into t/18687/shifting_algorithm
​acfe0f1	clean up

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​9fc24ef	rm build files
​095e372	update docs

[chaoxu]

I made a new merge, that strange problem seems to go away.

Fixed other problems.

[Stefan]

I'm happy with this code now. Positive review from me.