October meetup - Githubissues

betatim commented 7 years ago

Let's try to schedule a meetup towards the end of October:

@wolfv - xtensor
@markusbaden Modern pension fund analytics using pandas, jupyter and dash
@kmader - introduction to dash

Date

Thursday, 05.10.2017

Location

ETH HG E22

betatim commented 7 years ago

What related libraries do we know that are similar to xtensor or would make a nice theme? Some thoughts:

Pybind11
cython
numba
something on profiling

are all kind of in the field of "faster numerical python"

jakeret commented 7 years ago

Some updates on the Python JIT compilers would be interesting. Next to Numba we have

Pyjion
Pyston
GT-py

wolfv commented 7 years ago

I am hoping to integrate xtensor with jet at some point :)

Also, there is pythran! https://github.com/serge-sans-paille/pythran

betatim commented 7 years ago

So plenty of ideas ... we invite one from each project and have an "epic rap battle of JIT'ing python"? :rofl:

SylvainCorlay commented 7 years ago

Serge Guelton is also coming over to pydata paris to talk about pythran in sep :)

jakeret commented 7 years ago

@wolfv @markusbaden could you provide us a short outline of your talk (1-2 sentence on the content and about you) that we can post on the meetup page?

wolfv commented 7 years ago

Hi @jakeret,

this is the long outline for the xtensor talk:

xtensor is a C++ template tensor algebra library supporting NumPy-style broadcasting and universal functions, aiming at feature parity with NumPy with a native feel. In this talk we will present the highlights of the expression system. Then we will show how xtensor can be used to create NumPy-aware Python extension modules with the xtensor-python project and the xtensor-cookiecutter template.

In this presentation, I will focus on the authoring of NumPy-aware Python extensions with xtensor.

First, we present a general overview of the xtensor expression system and its main features:

lazy broadcasting and universal functions.
an API following the idioms of the C++ standard library.
tools to manipulate array expressions and build upon xtensor.

Then we show how the expression-based system can be used to symbolically manipulate array expressions based on other data structures, such as a numpy arrays thanks to Python's buffer protocol, but also a database or the file system.

The special case of the bindings with NumPy is handled by the xtensor-python project which makes it extremely easy to author Python extension operating inplace on NumPy arrays, with a modern STL-compliant API. We conclude the talk with a demonstration of the xtensor-cookiecutter template project for a Python extension defining a NumPy ufunc, and exposing other C++ functionalities.

If you want shorter:

xtensor is "NumPy" for C++, with added laziness and speed! It offers an extremely easy way to write high-performance extensions for Python in C++, which natively operate on NumPy buffers (zero copy overhead). Wolf has just finished his master in Robotics, Systems and Controls at ETH, and hopes to contribute to the scientific Python stack through xtensor and accompanying libraries in the future, as well as to the Jupyter ecosystem.

Cheers, and hope to see you soon!

Wolf

markusbaden commented 7 years ago

Just to avoid confusion, is the meetup Thursday Oct 5 (as in #27) or Friday Oct 6 (as above)?

jakeret commented 7 years ago

It's the 5th. Apologize the confusion

markusbaden commented 7 years ago

I just noticed an (ambitious) typo in my talk title. It should be Modern pension fund analytics using pandas, jupyter and dash.

So dash not dask, although we have been thinking about dask there is no real use case yet, as data is not really big :smiley:

betatim commented 7 years ago

Asked ETH about hosting, waiting for reply.

Fixed the dask -> dash typo.

betatim commented 7 years ago

@kmader we briefly discussed a dash talk: would you be interested in giving a short (20min max) intro level talk about dash and how it compares to shiny?

To all speakers: can we aim at 20min of talk plus 10min of questions each? Maybe have more material in your slides for people to follow up later or selectively skip things based on what the audience is interested in/what level they are at for the topic? Basically, can we all work together so that we won't end with two hours of lecturing 😃

betatim commented 7 years ago

Also: I am pretty excited about this meetup!

jakeret commented 7 years ago

@betatim I created a draft for the meetup. Could you have a look?

betatim commented 7 years ago

Should we make the lightning talks and the third talk an either or?

We could start having an intro slide advertising events. Put the slide up as a "screen saver" while people arrive. For this meetup:

euroscipy hosting discussion: contact XXX
Swiss Python Summit: early bird tickets and call for papers

markusbaden commented 7 years ago

20 + 10 min works for me. If anybody wants to chime in on dash that would be great. We just started using it and would love to hear other people's experience. Even a discussion or lightning talk would be nice

betatim commented 7 years ago

Room is booked: HG E22, @jakeret time to announce it?

wolfv commented 7 years ago

awesome, 20 + 10 mins is perfect for me.

kmader commented 7 years ago

sure a dash talk on 5 October works for me

jakeret commented 7 years ago

Awesome! Could you post a 2 sentence abstract that I can put on the meetup page

jakeret commented 7 years ago

@kmader are you still in for the talk? Can I announce you on the meetup page? Do you have a short outline?

kmader commented 7 years ago

@jakeret sure I can still give that talk. So @betatim's idea for the title was Where Shiny Shines and Dash Disappoints: How to get started with reactive web GUIs in python

It will start off with a basic introduction in web GUIs and why the reactive design pattern is even useful.
It will then go into the specifics of how Shiny for R works and how Dash in Python work.
I can then talk about how we are using it and the tricky points for getting started.

Does that work?

jakeret commented 7 years ago

Sounds great. I copy&pasted it to the meetup site. Looking forward

betatim commented 7 years ago

Welcome slide: https://docs.google.com/presentation/d/1a81dypPkY5TEKkIUYm-FKG81NAAnJPerQ2a40QJuLkw/edit?usp=sharing collecting events and other things worth advertising. Feel free to add a small banner/note to the slide.

wolfv commented 7 years ago

are you collecting slides and we present from one laptop?

jakeret commented 7 years ago

Usually each speaker plug in his/her notebook. Typically, we ask the speakers to upload their slides after the talk to our GitHub repo

weisro commented 7 years ago

Great talks, once again, and to who ever payed my unsupervised beer today: big thanks, approach me on the next meetup for a revenge:)

Am 05.10.2017 10:26 schrieb "Joel Akeret" notifications@github.com:

Usually each speaker plug in his/her notebook. Typically, we ask the speakers to upload their slides after the talk to our GitHub repo

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pydatazrh/meetup/issues/28#issuecomment-334395247, or mute the thread https://github.com/notifications/unsubscribe-auth/AUSgtep-Wv_R3051PMhDuuUKuTe5dU0Mks5spJKjgaJpZM4PAi4D .

betatim commented 7 years ago

@markusbaden @wolfv @kmader could you add your slides/links to material to this repo in the folder Meetup8/?

Thanks everyone for the nice talks.

jakeret commented 7 years ago

Great talks, once again, and to who ever payed my unsupervised beer today: big thanks, approach me on the next meetup for a revenge:)

I'll do ;-)

pydatazrh / meetup

October meetup #28

Date

Location