This category is for correspondence with students applying to or participating with Stan’s Google Summer of Code efforts.
Did you want to mentor someone? Last year, only Krzysztof volunteered. It’s a lot of paperwork for not much money, even with NumFOCUS doing most of the paperwork.
@Bob_Carpenter, I administer the R participation in Google Summer of Code. We pass the per-project money ($500 per project slot) to the R Foundation, not to the mentors. Google leaves it up to the org. whether to pass the project stipends to the mentors or not, we decided years ago that that money has more impact at the foundation level than at the individual mentor level.
Google Summer of Code is really about the code that comes out of the student projects, and the stipends Google pays to the students so that they can spend the summer writing open source code!
In any event, if Stan wanted to participate under the R umbrella, I would certainly welcome more Stan projects in R’s participation in GSoC. We had one student (@LuisDamiano) use Stan in GSoC last year, and he has submitted somewhat related work to Stancon 2018. I’d love to see students work on improving Stan and RStan during GSoC. Anything I can do towards that end I am happy to discuss.
Thanks for the clarification and invitation to work under the R umbreall, @braverock.
I didn’t even realize there was money going to the project! I meant the amount of money going to the student wasn’t that much given the amount of paperwork involved. I figured the real win would be in recruiting new developers.
NumFOCUS provides an umbrella under which we can sponsor students. Last year, only Krzysztof took them up on that. @bgoodri and @jgabry would know more about R projects—I live down at the C++ level.
@seantalts @Krzysztof_Sakrejda I am pursuing my masters from IIIT Bangalore, I would like to work on Protocol Buffer Data Transport Layer project in GSoC 2018.
I have completed the first steps in accordance to that.
I am really interested in working on this project. I am up for any more tasks you want to give me related to this.
Looking forward to working with you.
Hey Tarun, thanks for writing! According to this, student applications open March 12 and you should submit a proposal to us then. If you want high-level feedback on your proposal feel free to post a link to it or something.
Thank you @seantalts for the reply. By the end of this week, I will be posting my proposal link and I hope you will give your valuable feedback on that.
NumFOCUS has acted as our mentor organization, so someone needs to coordinate with them or fill out the paperwork to make Stan itself into a mentor organization if it doesn’t have to be a company.
If it’s R related, you could also follow up with @braverock (see above).
I’m coordinating with them - we submitted two potential projects Jan 23.
The organization applications for GSoC 2018 have already been handed in. We’re just waiting for Google to announce the selected organizations for 2018, then student applications will start.
Yes, any R-related Stan project would be welcome as a proposal under the R GSoC umbrella. We had one last year, and there is one proposed for this year as well. I’m happy to chat here or off-forum about the process. Basically, we’d need a project idea page, two mentors preferably from different institutions, and preferably with at least one mentor having prior GSoC experience.
@seantalts This isn’t quite right. contributions prior to the application are one of the few ways we really have to evaluate whether we want to commit our time to a student. With GSoC it would be pretty standard to lead @Tarun_Pandey through a straightforward (!) PR or two prior to the final decision on the application.
For R GSoC we also hope to see a prior history of engagement with the community, through mailing list activity, patches, etc. Those tend to be successful students in GSoC, and have a higher chance of sticking around post-GSoC. I’m also not a huge fan of even a successful student doing GSoC and then disapearing for six months before it is time to prepare for the next GSoC.
Thank you for addressing the issue.
I would also like to work on some tasks before submitting the final proposal.
Do you have some examples of your work on github or something like that so I could take a better guess at what to suggest for a first small project?
Thanks! Sorry I missed this, I have a different account at work and we had snow days (and the notifications don’t cross-post). I’ll check that out and make a suggestion.
For a start, could you look at the ‘writer’ interface here:
The usage here (particularly where it is used for ‘diagnostic’ and ‘sampler’ output:
One of the simple projects we’ve had kicking around is implementing a filtered csv writer from CmdStan so that may be a nice start. There alread is a filtered csv writer in rstan but it uses some R-specific things so it doesn’t make sense to take it wholesale. rstan’s version is here:
Could you read those files and see what you think about doing a filtered writer as a first PR. The PR could just be for the writer and tests so you wouldn’t have to worry about modifying the CmdStan interface yet. It’s possible this is either too simple or too complicated for a start so let me know if it is.
Some background, just FYI for now:
There is a protobuf version of the writer here:
proto files are here:
(More) carefully defining the proto files for this would be a part of the GSoC project.
@Tarun_Pandey, tagging in case you missed it.
@Krzysztof_Sakrejda Thank you for the inputs, will get back to you soon on this.