​Hi Yan, I'm glad you're interested. I'll check your code as soon as possible. Thanks for doing the "competency test". I would say we want both: speed and accuracy. It is not supposed to be a demo implementation, but real code that will be used by real industrial project. That's one of the aim of Boost in general. I hope this answers to your question. The next step is for you to write your proposal and develop it well on the point that you think are the most important. As you saw in the discussions on the mailing-list, the topic of having a solver is quite hot because it's one of the main purpose of such a library. This year, we have a lot of very good students so the competition will be hard. As soon as your proposal is ready, send it to me ​so that I can help you improving it. Best of luck ! David On Tue, Mar 10, 2015 at 8:03 AM, Yan Ni wrote:

rgraduate student in 4th year. I am interested in working for Boost in GSoC 2015 and found the uBLAS project very interesting. I've read some the codes on github and has written a little code (including a simple test) of Gaussian Elimination Algorithm as the competency test: https://github.com/leckie-chn/ublas. I've also read about Raj's post about the algorithms need to be implemented. And I think that the focus on this project are both accuracy (using some numerical algorithm instead of classical math methods)and speed (regarding cache & memory utilization). Please let me know your preference. Thank you.