Hi,

I thought I would take a minute to express my interest in the project 
mentioned in the title and describe myself a bit before the application 
procedure. I am a second year PhD student at the University of Waterloo 
(Waterloo, Ontario, Canada), studying graph algorithms and complexity. I am 
particularly interested in fixed-parameter analysis of graph problems and 
fine-grained complexity, though I am also interested in other areas of 
computer science. 

The project looks interesting to me as I am familiar with concepts 
mentioned in the project description. Modular decompositions, have been 
used along with the notion of a 'modular-width' of a graph, in order to 
parameterise several problems and obtain FPT results, as in e.g. [1]. 
Ideally, I'd also like to see where these notions can be applied to other 
FPT algorithms as well, but that's probably not the goal of this program. 
My PhD thesis will likely focus on FPT in some way, and this was one area 
of literature I was recently reading. Further, for my last MSc (I have two: 
one in graph algorithms, one in model driven software engineering), I 
characterised the end-vertices of LexBFS for bipartite permutation graphs 
(see [2]), so I am familiar with LexBFS quite well too. 

It looks like the aim of the project is to implement the algorithm for 
modular decomposition of [3], and that doesn't seem unreasonable, along 
with adding an efficient representation of LexBFS. Other (stretch) goals 
listed are split decompositions and skew partitions of graphs. I'm not as 
familiar with the most recent literature for these concepts: is [4] still 
the fastest for skew partitions? is [5] best for computing split 
decompositions?

Are there are other ideas that the users/developers of Sage might want that 
are related? If there are multiple students interested in the same project, 
as there appears to be, then that might also be of interest. I will try to 
think of some too. A possible idea (though it's probably 'low-hanging 
fruit' in the tree of things to do) is to implement other search algorithms 
in Sage. It looks clear that LexBFS is useful, but does Sage have an 
implementation of LexDFS [6]? (a quick search in the ticket tracker didn't 
show anything, but maybe that's because it's done)

Lastly, are there any ideas that mentors for this project think might 
require novel research? That is, developing *new* algorithms (and 
implementing them) for problems that have not yet been solved. As a PhD 
student, publications are always worth seeking out, though since that is 
not the aim of the summer of code, I certainly can't expect it. I have 
participated in the summer of code program twice before (once with the Java 
Pathfinder team, and once with a project of TU Vienna, though it is now 
apart of the AOSSIE), and in the latter case, my algorithms for proof 
compression lead to two publications (one accepted [7], and one submitted). 
I suspect such goals might also be of interest to mentors, so I thought I 
would ask.

Over the next few days, I will go through the suggested process of getting 
familiar with the Sage development process (opening tickets and the like). 
Until then, any questions and feedback to this post would help in the 
upcoming application procedure. 


Thanks,

Jan Gorzny


[1] https://arxiv.org/abs/1308.2858
[2] 
http://www.crypticcode.ca/jan.gorzny/wp-content/uploads/2015/09/2015-discrete-lbfs-ends.pdf
 
(to appear in DAM)
[3] http://www.cs.toronto.edu/~mtedder/TedderModular.pdf
[4] https://link.springer.com/chapter/10.1007%2F978-3-540-89550-3_11
[5] http://epubs.siam.org/doi/10.1137/10080052X
[6] http://epubs.siam.org/doi/abs/10.1137/050623498
[7] 
http://www.crypticcode.ca/jan.gorzny/wp-content/uploads/2015/09/2015-cade-folu.pdf

-- 
You received this message because you are subscribed to the Google Groups 
"sage-gsoc" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sage-gsoc.
For more options, visit https://groups.google.com/d/optout.

Reply via email to