FluidDyn

FluidDyn is a framework for research and teaching in fluid dynamics.

Follows us on #Fediverse:

[FluidDyn @pyfluiddyn]

Final call to influence the use of #Nek5000 via #Python. Awaiting the feedback of all Nek5000 and CFD folks.

https://t.co/woYlc8xKCY

We have also released the updated #Snek5000 v0.9.0 🚀

[FluidDyn @pyfluiddyn]

Give #Snek5000 a try now

https://t.co/mxsuhHe9p9

We welcome comments from #Nek5000 and the #CFD community on the future of #Snek5000 and #Fluidsim

#Python #Fluiddyn #OpenSource #OpenScience

💫 cheers!
🐘[email protected]
🐘[email protected]

[FluidDyn @pyfluiddyn]

Snek5000 0.8.0 & Fluidsim 0.7.0 

After a long code sprint, improved versions of computational fluid simulation codes are out #Snek5000 and #FluidSim

https://t.co/I1XvQXIrmr

[FluidDyn @pyfluiddyn]

Here is an beautiful simulation produced from #Snek5000 using `snek5000-cbox` a convective box solver by 🐘[email protected] and Arman. Find more such examples here:

https://t.co/cOQXkGjFbP

#Nek5000 #CFD #Python
A 2D fluid simulation with vert…

[FluidDyn @pyfluiddyn]

New wave of #FluidDyn packages just landed this September

- #Transonic v0.5.1: gets a tiny bug fix release
- #FluidFFT v0.3.3: its build system is now compatible with editable wheels and Python versions up to 3.10
-… https://t.co/0knAavIOA2

[FluidDyn @pyfluiddyn]

🕯️ New releases of FluidDyn projects this season 🕯️

#Transonic 0.4.12 bug fix in Jupyter notebooks
#FluidSim 0.5.1 with a Taylor Green forcing for ns3d solvers
#FluidDyn 0.4.1 compatible with pyfftw 0.13.0
#FluidImage 0.1.3 compatible fixes for UVmat, OpenCV and PyQT5

[FluidDyn @pyfluiddyn]

And while using #Python, several options exist to achieve performance: use a different #Python implementation (#PyPy for instance) or write a compiled extension a.k.a. glue code.

[FluidDyn @pyfluiddyn]

Brett Cannon (@brettsky) has made similar recommendations [1] as we did in the recent reply [2] to a Nature article. Performance considerations are best made at the end of the software life-cycle.

[1]: https://t.co/klMHrkpM4r
[2]: https://t.co/imaeuDH7UX

[FluidDyn @pyfluiddyn]

1. #PyPy, #Numba, #Pythran & #Transonic with some changes in the original code improves performance remarkably.
2. Reduction in time to solution compensates for increased power requirement for multithreading.

Article: https://t.co/7M1naNq5Wx
Code:… https://t.co/89k7ful8iG

[FluidDyn @pyfluiddyn]

A common misconception is that #Python is slow and cannot be fixed.

In this new #Nature #Astronomy article we reassert Python can be performant via awareness of and use of efficient compilers. Read more here:

https://t.co/7M1naNHGO5

[FluidDyn @pyfluiddyn]

#Transonic is available via #ArchLinux AUR now, for convenience.

https://t.co/AmkIz2simj

You can always install it using #pip and #conda too.

https://t.co/cwpbBB5XGi

https://t.co/rekmBmE1GM

[FluidDyn @pyfluiddyn]

What if we told you #Python + #Numpy + #Transonic + #Pythran) could run blazingly fast, even when compared to native codes in #Cpp and #Fortran?

https://t.co/csgCGwbY0d

Note: We made a similar benchmark vs a Fortran code in our #FluidSim paper.
Energy consumption vs Time to s…

[FluidDyn @pyfluiddyn]

There is an array of interesting talks lined up on #PyConSE today. The track "Software Development & Performance" looks particularly interesting with talks on #testing, #profiling, and accelerating #Python.

https://t.co/tSCGifiSPW

[FluidDyn @pyfluiddyn]

📦 #fluidsim 0.3.3 is out after nearly a year’s worth of hard work. Notable changes:

“New” time stepping algorithm substituting phase-shifting instead of dealiasing

Faster, flexible installation

#CFD

[FluidDyn @pyfluiddyn]

#transonic at v0.4.5, turns 2 years old this month 🎂

Also on this, #ThrowbackThursday, a shout out to last year's #PyCon and #EuroScipy where #transonic landed:

- https://t.co/DUahiFduSQ
- https://t.co/TANR6SDd6U
- https://t.co/ApxUUyYC6c

📦https://t.co/cwpbBB5XGi

[FluidDyn @pyfluiddyn]

Confused about whether, when and how to optimize #Python code? You might find your answer with this talk from @pyconse:

https://t.co/ApxUUyYC6c

#transonic

[Juhis @Hamatti]

"The past, present, and future of extending Python" by @teoliphant is a must watch from this conference. His hour-long talk about extending #Python provides great insights into the topic.

https://t.co/EcflCLJwEp

[FluidDyn @pyfluiddyn]

@el_hult How about https://t.co/Mp5MIsbqEi

[indygreg @indygreg]

Mercurial is now passing all tests on Python 3.6 and Python 3.7 on Linux. Finally.

[Pierre Augier @pierre_augier]

@NathanShammah @Lasofi @condaforge @pypyproject @numba_jit @pyconfr @asso_python_fr @EuroSciPy Authors of packages of the #Python scientific ecosystem, did you read our note https://t.co/CxEDpdnoQY?

We describe problems (perf for numerical kernels, incompatible accelerators, ...) and possible improvements.

What do you think about these ideas?

RT needed🙂! #numpy #scipy