Michaël Defferrard

@nperraud, can you take a look?

Thanks for the review @nperraud. I'll address it soon. While we're at it, I'm thinking about integrating knn and radius graphs. I'd remove the `kind` parameter, and rename `k` to...

Thanks! Can you fix the tests?

Thanks for the kind words @gcuendet. I mostly need to address review comments. I would also like to [merge knn and radius graphs](https://github.com/epfl-lts2/pygsp/pull/43#issuecomment-467938475). I'll probably get to this in the...

* [x] rebase or merge? Too long history for rebase, and would make things even more difficult for sphere-graphs and learned-graph. * [ ] `NNGraph` is a bad name: Graph...

* Let's go with `NearestNeighbors` then. :) * Agreed. Some synonyms for ideas: `spread`, `extent`, `size`, `width`, `span`, `stretch`. Maybe we should think of it as a step before the...

Yes the [first coefficient is doubled in the PyGSP](https://github.com/epfl-lts2/pygsp/blob/70f47eaf5d4b521f3247c4f5f10e7df1f39b2b47/pygsp/filters/approximations.py#L103). In the ML code I wrote I did not double it indeed. The code in the PyGSP is a [straight translation...

Thanks for the historical pointer. :) I don't think there's a mistake. It's a matter of definition. A factor of two must appear for the first coefficient. The coefficient is...

> The pi/2 in eq. 52 and the 1/2 in eq. 53 should explain the convention used in the code. Agreed. > Now I have not gone deeply to know...

Indeed, we use [`2/n` as a factor](https://github.com/epfl-lts2/pygsp/blob/70f47eaf5d4b521f3247c4f5f10e7df1f39b2b47/pygsp/filters/approximations.py#L52) instead of `2/pi`. As for #74, the definition of the coefficients came from the [matlab implementation](https://github.com/epfl-lts2/gspbox/blob/a7d9aac5e239f1bcb37a9bb09998cc161be2732f/utils/gsp_cheby_coeff.m#L102). I believe that is intentional and correct,...