Re: [Numpy-discussion] GSoC 2014 NumPy
Hello Ray, Thanks for your suggestion! I just read the links you provided and I think I can implement it as long as I do further research on zoom fft algorithm. So I wonder if this can be a GSoC project? Maybe I should extend this idea or combine it with other ideas? BTW, just for curiosity, why we need both scipy.linalg and numpy.linalg? Is implementing all functions in numpy a bad idea? Regards, Leo Mao
On Sun, Mar 2, 2014 at 5:12 AM, Leo Mao <lmao20001@gmail.com> wrote:
Hello Ray, Thanks for your suggestion! I just read the links you provided and I think I can implement it as long as I do further research on zoom fft algorithm. So I wonder if this can be a GSoC project?
By itself that's not enough for a GSoC project.
Maybe I should extend this idea or combine it with other ideas?
It's possible to come up with an interesting proposal in this area I think. An issue may be that the FFT code in numpy and scipy isn't very actively worked on at the moment, so finding a suitable mentor could be tricky.
BTW, just for curiosity, why we need both scipy.linalg and numpy.linalg? Is implementing all functions in numpy a bad idea?
The overlap is mostly due to historical reasons. The long-term plan is to remove duplicate functions from scipy.linalg. Ralf
On Tue, Mar 4, 2014 at 5:42 AM, Ralf Gommers <ralf.gommers@gmail.com> wrote:
It's possible to come up with an interesting proposal in this area I think. An issue may be that the FFT code in numpy and scipy isn't very actively worked on at the moment, so finding a suitable mentor could be tricky.
So should I choose another topic? Actually, I just read the thread "numpy gsoc topic idea" and I found that the idea "vector math library integration" really interests me! And I have a qeustion: how can I find a suitable mentor? Currently I'm digging into the source of numpy and trying to make a small pull request. Also I keep thinking how to write a proper proposal. Are these enough? Can I do more for now? I will be grateful for any advice. Thanks in advance. Regards, Leo Mao
On Wed, Mar 5, 2014 at 5:52 PM, Leo Mao <lmao20001@gmail.com> wrote:
On Tue, Mar 4, 2014 at 5:42 AM, Ralf Gommers <ralf.gommers@gmail.com>wrote:
It's possible to come up with an interesting proposal in this area I think. An issue may be that the FFT code in numpy and scipy isn't very actively worked on at the moment, so finding a suitable mentor could be tricky.
So should I choose another topic?
I suspect that that may be better.
Actually, I just read the thread "numpy gsoc topic idea" and I found that the idea "vector math library integration" really interests me!
If it's interesting, I suggest to dive in. It's Julian's idea, so probably he's interested to (co-)mentor a project on this topic. Julian?
And I have a qeustion: how can I find a suitable mentor?
That's a nontrivial question. Let me send a separate email to the list about that.
Currently I'm digging into the source of numpy and trying to make a small pull request. Also I keep thinking how to write a proper proposal. Are these enough? Can I do more for now?
Keep in mind that your proposal will need to go through a few rounds of feedback and rework before it's solid enough that you can submit it. The same usually goes for pull requests. Cheers, Ralf
I will be grateful for any advice. Thanks in advance.
Regards, Leo Mao
_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
participants (2)
-
Leo Mao -
Ralf Gommers