Søren Fuglede Jørgensen

@fuglede
282 Followers
128 Following
355 Posts
Quantum algorithm developer at Kvantify, wind farm optimizer at Ørsted before that, quantum topologist in academia before that, maths PhD, nerd.
Website 🌐https://fuglede.dk
Code 👨‍💻https://codeberg.org/fuglede

@sorenhave @h0gh Det føles i hvert fald en smule hacket at skulle lave særregler for flertalsregeringer.

Så altså, noget i stil med https://fuglede.codeberg.page/kongeligeundersoger/

Så mangler vi bare en liste over samtlige politiske emner. Men det må Troels så stå for.

Den kongelige undersøger

Hvis jeg var kongen, ville jeg droppe det der med brevene og i stedet bede MF'erne opremse alle stridspunkter i mulige regeringsgrundlag, mulige løsninger til de punkter, hvordan de hver især ville have det med hver løsning, og så løse et eller andet grimt heltalsprogram, der producerer et regeringsgrundlag sammen med det mindst utilfredse flertal.
Matematikere i hobetal boykotter Trumps USA: Ikke et hak bedre end Rusland

2.200 forskere vil boykotte kongres, hvor 'matematikkens nobelpris' gives. USA er for usikkert, lyder kritikken, der også har en dansk afsender.

Videnskab.dk

LinkedIn launch release post: https://www.linkedin.com/posts/fuglede_launching-httpsitsidcloud-many-of-you-share-7445009135210979328-2uj1

Edit: Alright, April 1 is over; we're closing down.

Edit: Alright, April 1st is over, unfortunately the company's shutting down. ---- Launching https://itsid.cloud! Many of you will recall the clever use of large language models to “de-copyright” the… | Søren Fuglede Jørgensen | 19 comments

Edit: Alright, April 1st is over, unfortunately the company's shutting down. ---- Launching https://itsid.cloud! Many of you will recall the clever use of large language models to “de-copyright” the works of Studio Ghibli, or more recently, the reimplementation of chardet that enabled a change in licensing. Indeed, avoiding the need to properly attribute works to their original authors - sometimes a costly affair - has emerged as a notable use case of AI. In all such cases, however, the generated work never quite matches the original. Large language models are prone to hallucinations and non-deterministic behavior. To address these challenges, I’m happy to announce the launch of Itsid - the first large language model to provide mathematically guaranteed reproduction of its inputs. For more information, including a free demo, visit: https://itsid.cloud To celebrate the launch, I’m offering a bonus for my LinkedIn network: comment with a piece of text you’d like to test, and I’ll reply with the model’s output! | 19 comments on LinkedIn

LinkedIn

You will all recall the clever use of LLMs to “de-copyright” the works of Studio Ghibli. Indeed, LLMs are great for not having to attribute works to their authors.

In such cases, however, the generated work never quite matches the original. LLMs are prone to hallucinations and non-deterministic behavior.

Enter Itsid! The first LLM to provide mathematically guaranteed reproduction of its inputs: https://itsid.cloud

#ai #llm

Edit: Alright, April 1 is over; we're closing down.

Itsid - Thanks for the Laughs!

A while ago, I wrote about Welter's game: a simple combinatorial game where two players take turns moving coins down a line until no moves are possible. It has some curious connections to coding theory, but for our purposes, it's just a little game to play over the holiday.

However, the game setup implicitly discriminates against people who don't have any coins, and we can't have that, so here's a web version: https://welter.fuglede.dk/

#maths #combinatorics #games #combinatorialgames

Welter's Game

A combinatorial strategy game: move coins along a number line and force your opponent into a losing position.

Here's some stuff we've been up to at work lately: One of the main bottlenecks when using variational quantum algorithms to solve problems in quantum chemistry is that you need to call your quantum computer a lot of times.

A ton of fixes have been suggested in the past; we gather many of them in a single framework, show how to combine them and how to compare what comes out: https://arxiv.org/abs/2603.20029

And here's an animation that illustrates what the problem is about: https://kvantify-assets.s3.eu-west-1.amazonaws.com/anim-arxiv-2603-20029.mp4

Variance reduction methods in the estimation of Pauli sums

Accurately estimating expectation values of quantum observables with as few measurements as possible is crucial to many quantum computing applications. We introduce a framework that covers many of existing measurement strategies and introduce heuristics that can be used to enhance randomized schemes, including those based on Pauli grouping with inverse probability weighting and variants of the classical shadow algorithm. We show how to maximize information gain from such schemes, while carefully optimizing the distribution of possible measurements, and show that simple grouping algorithms can get close to, and in some cases exceed, state-of-the-art accuracy for unbiased estimation of expectation values on a standard quantum chemistry benchmark. We show how these randomized methods may be compared to more recent measurement schemes, such as shadow grouping, derandomized shadow, and overlapped grouping measurement, we show how the same strategies can be used to augment these schemes, and we demonstrate that we can reduce measurement costs by up to a factor of two by allowing Clifford measurement circuits for otherwise Clifford-less methods.

arXiv.org

So, all users of Microsoft's GitHub Copilot (Free/Pro/Pro+) should now expect to have their intellectual property and personally identifiable information stolen, unless they explicitly opt-out. Feedback is … not great:

https://github.com/orgs/community/discussions/188488

Regarding personally identifiable information specifically, here's your recurring reminder that in the EU, users have the right to file complaints over GDPR infringements (in Denmark, that would be at at https://www.datatilsynet.dk/borger/klage)

FAQ: Privacy Statement update on Copilot data use for model training (Free/Pro/Pro+) · community · Discussion #188488

Hello GitHub Community👋 We’re sharing an update to our Privacy Statement and Terms of Service about how we use personal data to develop, improve, and secure GitHub products and services, including ...

GitHub
In my online undergraduate P5.js course, students are about to begin the module on motion and physics, including a bit of physics simulation using Matter.js. It suddenly occurred to me that I had never seen anybody put together this particular demo before, and I realized it had to be done. Messy source code at https://editor.p5js.org/isohedral/full/vJa5RiZWs.

Elpriserne i DK1 i morgen:
Laveste pris: 0.59 kr. /kWh (2026-02-17 13:30:00)
Højeste pris: 1.00 kr./kWh (2026-02-17 18:45:00)
Gennemsnitspris: 0.73 kr./kWh

Elpriserne i DK2 i morgen:
Laveste pris: 0.62 kr. /kWh (2026-02-17 04:00:00)
Højeste pris: 6.81 kr./kWh (2026-02-17 17:45:00)
Gennemsnitspris: 1.37 kr./kWh