digitalcourage.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
Diese Instanz wird betrieben von Digitalcourage e.V. für die Allgemeinheit. Damit wir das nachhaltig tun können, erheben wir einen jährlichen Vorausbeitrag von 1€/Monat per SEPA-Lastschrifteinzug.

Server stats:

823
active users

#computing

36 posts18 participants1 post today
Replied in thread

@GPJohnston Fantastic. Honestly, it feels a bit like magic. There's a real freedom that come with knowing that I don't ever have to waste money on a new computer again. I'd guess there are many millions of computers that need not go into the recycling process for years. For those of us that are concerned with the #climate, this is an action anyone can take.

Most Americans I know pretend they are helpless to enact change on a personal level. That's not true.

Today in thinking-sideways news:

"NaN-Propagation: A Novel Method for Sparsity Detection in Black-Box Computational Functions", Peter Sharpe 2025
arxiv.org/abs/2507.23186

The abstract is already amazing.

"By systematically contaminating inputs with NaN and observing which outputs become NaN, the method reconstructs conservative sparsity patterns that eliminate a major source of false negatives."

"... advanced strategies such as NaN payload encoding via direct bit manipulation enable faster-than-linear time complexity, yielding speed improvements"

Code: github.com/peterdsharpe/nan-pr

arXiv logo
arXiv.orgNaN-Propagation: A Novel Method for Sparsity Detection in Black-Box Computational FunctionsWhen numerically evaluating a function's gradient, sparsity detection can enable substantial computational speedups through Jacobian coloring and compression. However, sparsity detection techniques for black-box functions are limited, and existing finite-difference-based methods suffer from false negatives due to coincidental zero gradients. These false negatives can silently corrupt gradient calculations, leading to difficult-to-diagnose errors. We introduce NaN-propagation, which exploits the universal contamination property of IEEE 754 Not-a-Number values to trace input-output dependencies through floating-point numerical computations. By systematically contaminating inputs with NaN and observing which outputs become NaN, the method reconstructs conservative sparsity patterns that eliminate a major source of false negatives. We demonstrate this approach on an aerospace wing weight model, achieving a 1.52x speedup while uncovering dozens of dependencies missed by conventional methods -- a significant practical improvement since gradient computation is often the bottleneck in optimization workflows. The technique leverages IEEE 754 compliance to work across programming languages and math libraries without requiring modifications to existing black-box codes. Furthermore, advanced strategies such as NaN payload encoding via direct bit manipulation enable faster-than-linear time complexity, yielding speed improvements over existing black-box sparsity detection methods. Practical algorithms are also proposed to mitigate challenges from branching code execution common in engineering applications.

I got new stickers from @fsf so I fixed the outside of my 2012 Mini to match the inside which runs Trisquel GNU/Linux.

Many of us spend a lot of time using computers everyday and we can do more to make sure we use an OS and software that is built to respect our freedom, privacy and security.

Bonus: It runs great on old computers, is climate and resource friendly, saves money.

I got a start in upgrading 7 donated laptops. 4 are done. GNU/Linux Mint with Xfce desktop installed in about 30 minutes per computer. Another 10-15 minutes to run updates. Specs: 4 to 8GB of memory, hard drives (rather than SSDs). Instead of being recycled they are ready for continued use. Totally usable for day-to-day general use. Anyone can do this kind of direct action, community mutual aid that conserves energy and resources.

#GNU#Linux#Endof10

#AI #AlanTuring #TuringInstitute #Turing #Computing #ArtificialIntelligence #UK #Technology #ComputerScience #Science
It’s quite ironic that an organisation established in commemoration of one of our greatest geniuses may fail because its leaders appear to be too incompetent to run it…

Staff fear UK's Turing AI Institute at risk of collapse bbc.com/news/articles/c24zz2vd

British Secretary of State for Science, Innovation and Technology Peter Kyle departs a cabinet meeting at Downing Street in London. He has short, sharp grey hair and is wearing a smart suit.
www.bbc.comStaff fear UK's Turing AI Institute at risk of collapseA complaint raises concerns about the funding risks and a "toxic" culture at the Alan Turing Institute.

I have been thinking about distributed reactive computation. There is way too much plumbing involved in building apps.

Have landed on an elegant model, with which, expressing a multi-user, secure, high-performance todo app takes 20 lines of code (including a basic reactive UI).

The new algorithm for directed cheapest routes "slices the graph into layers, moving outward from the source like Dijkstra’s. But rather than deal with the whole frontier at each step, it uses the Bellman-Ford algorithm to pinpoint influential nodes, moves forward from these nodes to find the shortest paths to others, and later comes back to other frontier nodes. It doesn’t always find the nodes within each layer in order of increasing distance, so the sorting barrier doesn’t apply. And if you chop up the graph in the right way, it runs slightly faster than the best version of Dijkstra’s algorithm. It’s considerably more intricate, relying on many pieces that need to fit together just right. But curiously, none of the pieces use fancy mathematics."

quantamagazine.org/new-method-

A still from a video of nodes being connected by a white line
Quanta Magazine · New Method Is the Fastest Way To Find the Best Routes | Quanta MagazineA canonical problem in computer science is to find the shortest route to every point in a network. A new approach beats the classic algorithm taught in textbooks.

"Choreographic programming is a paradigm for concurrent and distributed software, whereby descriptions of the intended communications (choreographies) are automatically compiled into [decentralized code.
It is difficult to combine w/higher-order functions]: compilation is not modular (editing a part might require recompiling everything) and the generated code can perform unexpected global synchronisations."
portal.findresearcher.sdu.dk/e

University of Southern DenmarkModular Compilation for Higher-Order Functional Choreographies