[Haskell-cafe] GSoC Project Introduction: Accelerate Automatic Differentiation

[Charles Blake]

Dear all,


I'm a student interested in the GSoC programme and I wanted to introduce myself and make contact with those involved in potentially mentoring for the 'Implementing Automatic Differentiation in Accelerate' project (https://summer.haskell.org/ideas.html#accelerate-automatic-differentiation). If it would be more appropriate for me to contact those individuals directly then please let me know.


My name is Charlie Blake, and I'm currently coming to the end of my CS undergrad degree. Much of my study of late has been focused on functional programming and parallelism - so when I saw this project proposal it jumped out at me! I was expecting to be assisting research at my university (St Andrews University, Scotland) in this field over the summer, but this arrangement has recently fallen through. Hence why I'm unfortunately sending this message rather close to the deadline. Nevertheless, I'm now hoping very much to get on the GSoC program, to which I am very much committed.


With regards to the project itself, I'm currently in the process of drawing up a proposal. Although I'm making every effort to read as much as I can about the 'ad' and 'accelerate' libraries, it's still quite hard to get a good sense of the scope of the project. Would, for instance, it be a reasonable goal to expect to be able to implement all of the core 'ad' functionality in my own parallel library in the time-frame? This seems like the natural objective, but I don't want to propose an unrealistically large project if this is not feasible. If anyone has answers to these questions and might further be willing to discuss a draft project proposal with me over the next few days that would be greatly appreciated.


Many thanks,

Charlie Blake
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-cafe/attachments/20180323/776fa744/attachment.html>