Treat Dask and other libs in the Scientific Libraries lecture?

[jstac]

trafficstars

Should dask be treated in this lecture: https://lectures.quantecon.org/py/sci_libs.html ?

Any other libraries we're not treating but should at least mention? Jax?

I'm rehashing this question from Slack, where @AakashGfude says: dask is awesome and easy to use. I think we can. It has got a lot of traction as well, with 4.9k stars in github.

@mmcky @shizejin @qbatista @cc7768

Any library we treat needs an application to illustrate it. Thanks in advance for your thoughts!

[mmcky]

I would vote for a dask lecture for sure! I think it would be a really fun lecture ... happy to help with this once the new build pathway is up and running etc.

[mmcky]

Another library to consider would be xarray. I haven't used it much myself but it would be nice to take a closer look as it is becoming a more important foundational library.

http://xarray.pydata.org/en/stable/#

[jstac]

Sounds good. We just need use cases, I guess.

[jstac]

Perhaps we should drop Cython from this lecture and just recommend that our users stick to Numba.

It's caused execution errors on various machines and it's probably worse on Windows (for our users).

[mmcky]

hey @jstac I think this is a good idea, particularly given a large number of users are on windows. This is an excellent point and not well tested.

[mmcky]

This might be useful in the future (as it is currently in active development)

• [ ] https://modin.readthedocs.io/en/latest/

[duncanhobbs]

Dask is used in this QuantEcon Notes example to compute proximity matrices.

It might be possible to use Dask to import the HS trade data and do some of the dataframe manipulations beforehand (currently done with Pandas) as well.

[jstac]

Thanks @duncanhobbs --- this could be a good example.