Oh, I wasn't aware of this policy. Thanks for letting me know. I'll think 
about being a mentor in the next few days instead.

On Thursday, 9 March 2017 15:42:34 UTC-5, Dima Pasechnik wrote:
>
> Dear Jan,
> thanks for your message. It seems that you are not eligible to participate 
> in GSoC as a student any more. 
> Indeed, https://developers.google.com/open-source/gsoc/faq has 
>
> What are the eligibility requirements for participation?
> section that says:
>
>    - You have not already been accepted as a Student in GSoC more than 
>    once
>    
> However, you might consider becoming a mentor...
>
>
> On Thursday, March 9, 2017 at 11:20:08 AM UTC, Jan Gorzny wrote:
>>
>> 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