[Users] Announcing a new Python package for post-processing simulation data: kuibit

[Erik Schnetter]

Gabriele

This package looks quite interesting. Would you be interested in
giving a show-and-tell presentation at some point, demonstrating how
you use the package? I would be interested in attending.

-erik

On Thu, Jan 14, 2021 at 12:18 AM Gabriele Bozzola
<bozzola.gabriele at gmail.com> wrote:
>
> Hello,
>
> I developed a new package to analyze Einstein Toolkit simulations, kuibit [0,1].
> kuibit is a Python3.6+ code that I built from scratch following the same design
> (and in various instances, implementation details too) of Wolfgang Kastaun's
> PostCactus.
>
> kuibit provides high-level data types to easily work with grid functions, time
> and frequency series, gravitational waves, and so on. It also has readers to
> effortlessly access simulation data with full support for HDF5 and ASCII output
> (1D, 2D, 3D grid data, scalar data, reductions, horizon data, ...). You can find
> a reasonably comprehensive list of features in the documentation [2] or a
> high-level summary in the frontpage of the docs [3].
>
> One of the main reasons I wrote this code is for other people to use it.
> Our group (University of Arizona) is a young one and we don't have any sophisticated
> toolchain to analyze simulation data. Without suitable tools, post-processing
> simulations can be a daunting task for those that are new to the Einstein Toolkit.
>
> Given that I want other people to use kuibit, I made the effort to make the code user
> and developer-friendly. For users, there is documentation [4] with examples and
> small tutorials. Also, the package is on PyPI so it can be easily installed and updated.
> For developers, the entire codebase has unit tests and continuous integration [5],
> there are extensive comments, and the style of the code is rather verbose
> to help developers understand what is going on. The continuous integration also
> lints the code, performs static analysis, and generates the documentation,
> reducing the maintenance costs.
>
> kuibit takes care of all the low-level details need to deal with simulation data, so
> it greatly lowers the entry barrier in using the Einstein Toolkit. I believe that this,
> along with the care I put in making the code accessible to other developers,
> makes kuibit a good candidate for inclusion in the Einstein Toolkit.
>
> The main problem with kuibit is that it is a new code: regardless of all the
> tests I wrote, there will be bugs, unergonomic interfaces, and performance issues.
> kuibit needs to be tested with several real-world projects and cross-checked with
> other codes.
>
> I am happy to give a short introduction to kuibit during a weekly call if there's
> interest. In the meantime, the code is available here:
> https://github.com/Sbozzolo/kuibit
>
> Best regards,
> Gabriele Bozzola
>
> [0] https://github.com/Sbozzolo/kuibit
> [1] https://github.com/Sbozzolo/kuibit#what-is-a-kuibit
> [2] https://sbozzolo.github.io/kuibit/features.html
> [3] https://sbozzolo.github.io/kuibit/#summary-of-features
> [4] https://sbozzolo.github.io/kuibit/
> [5] https://github.com/Sbozzolo/kuibit/actions
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users



-- 
Erik Schnetter <schnetter at cct.lsu.edu>
http://www.perimeterinstitute.ca/personal/eschnetter/