#numericalcomputing — Public Fediverse posts

🎉 Successfully conducted a webinar on #Scilab and #Xcos as part of the FOSSEE project, IIT Bombay.

Sessions like these help bridge the gap between academia, industry professionals, and accessible computational tools. It is encouraging to see growing interest in Free and Open Source Software for numerical computing and engineering applications.

▶️ Session recording:
https://www.youtube.com/live/-1ujPyB5KVo

📩 [email protected]

#FOSSEE #IITBombay #OpenSource #FLOSS #EngineeringEducation #NumericalComputing

Interested in numerical computing using open-source software?

Join our webinar on Scilab to learn about modelling, simulation, optimization, and data visualization for engineering applications.

📅 16 May 2026 | 4:00 PM IST
🔗 https://zoom.us/webinar/register/WN_-0CUKvimSxOfpaU06v4Yqw#/registration

#Scilab #OpenSource #NumericalComputing #Engineering #FOSSEE #IITBombay #Webinar

GNU Octave 11 introduce miglioramenti significativi per prestazioni, precisione numerica, GUI e compatibilità MATLAB. Scopri tutte le novità della nuova versione. #GNUOctave #OpenSource #NumericalComputing #Linux

https://www.linuxeasy.org/gnu-octave-11-tutte-le-novita-della-nuova-release/?utm_source=mastodon&utm_medium=jetpack_social

I recently dumped my notes on modulus of convergence for hypergeometric functions on my website. I also had some thoughts on numerical accuracy. Not very valuable thoughts, but thoughts nonetheless.

For those who are wondering why anyone would care, many math libraries such as GSL, boost, and so on, suck. If you are trying to do intense calculations with any sort of decent accuracy, these libraries have dusty corners that fail. And they don't document where those corners are. Or they don't have routines for complex arguments. I write my own routines for these cases.

#Math #RealAnalysis #NumericalComputing

https://www.skewray.com/articles/numerical-accuracy-of-generalized-hypergeometric-series

A few years ago (mid 2023), I wrote up some research notes regarding the modulus of convergence of the generalized hypergeometric functions. This month I wrote up and posted a series of articles that are those notes, cleaned up a bit, and this article is the wrap-up of the series.

#Math #RealAnalysis #NumericalComputing

https://www.skewray.com/articles/modulus-of-convergence-for-generalized-hypergeometric-functions

We all get the feeling that, day by day, the world is converging towards disaster. But what tells us how fast? The Modulus of Convergence does!

#Math #RealAnalysis #NumericalComputing

https://www.skewray.com/articles/modulus-of-convergence

In mathematics, we say that a function is bounded if can restrict its image. Oddly, we never seem to 'bind' a function, though. I can find bounds on the the remainder of generalized hypergeometric functions, and I never used the word 'bind' either. Maybe 'bounding' refers to bunnies and deer?

#math #RealAnalysis #NumericalComputing

https://www.skewray.com/articles/remainder-of-the-conway-maxwell-poisson-distribution-normalization-factor

The generalized hypergeometric series are ubiquitous in the world of computing special functions, for certain amounts of ubiquity. Turns out the speed of convergence is related to the obscure Conway-Maxwell-Poisson distribution, which no one has ever heard of - pretty much the opposite of ubiquity.

#math #RealAnalysis #NumericalComputing

https://www.skewray.com/articles/bounding-the-remainder-of-generalized-hypergeometric-series

Accurately computing generalized hypergeometric functions is hard. How many terms do we need? Guess we need a general expression for the size of the terms in the series. Oh, wait, I've got one right here!

#math #RealAnalysis #NumericalComputing

https://www.skewray.com/articles/estimating-the-terms-of-generalized-hypergeometric-series

I use generalized hypergeometric functions a lot. That means I can type "generalized hypergeometric function" and not make a typo. Is it possible to compute these exactly? Sometimes!

#math #NumericalComputing

https://www.skewray.com/articles/computing-generalized-hypergeometric-function-from-series

“I Don’t Like NumPy”, ‘Dynomight’ (https://dynomight.net/numpy/).

Via HN: https://news.ycombinator.com/item?id=43996431

#NumPy #Python #Math #LinearAlgebra #NumericalComputing #Rants

Python for Mathematics

#Python #Math #NumericalComputing #CompMath

via @vinceknight

https://fosstodon.org/@vinceknight/114533757984575722

Solving physics-based initial value problems with unsupervised machine learning

🔗: https://journals.aps.org/pre/abstract/10.1103/PhysRevE.111.055302
(Open access)

#physics #machinelearning #NumericalComputing #ai #SciML

A Python implementation of David Deutsch’s Constructor Theory framework

🔗: https://github.com/gvelesandro/constructor-theory-simulator

#physics #math #python #programming #NumericalComputing

MATLAB has had an ahead of time codegen backend (MATLAB Coder) for a long time now.

But today I learned that someone added a new target to it to generate WebAssembly and JavaScript bindings (via Emscripten).

So if you've been compiling FORTRAN to Wasm until now for your numeric kernels and simulations, there's now a more user-friendly option.

https://www.mathworks.com/matlabcentral/fileexchange/69973-generatejavascriptusingmatlabcoder

#webassembly #matlab #matlabcoder #numericalcomputing

BLIS/FLAME support for @risc_v is now part of the official release! Includes support for the riscv vector spec 1.0.

https://github.com/flame/blis/releases/tag/1.0

#hpc #scientificcomputing #riscv #risc_v #numericalcomputing

New Breakthrough Brings Matrix Multiplication Closer to Ideal.

By eliminating a hidden inefficiency, computer scientists have come up with a new way to multiply large matrices that’s faster than ever.

Quanta Magazine https://www.quantamagazine.org/new-breakthrough-brings-matrix-multiplication-closer-to-ideal-20240307/

#CompSci #ComputerScience #NumericalComputing

Very nice:

“BLIS: BLAS-like Library Instantiation Software Framework”, Science Of High-Performance Computing (née FLAME) (https://github.com/flame/blis).

Via HN: https://news.ycombinator.com/item?id=39122286

#BLIS #BLAS #HPC #NumericalComputing #LinearAlgebra #Performance #FLAME