Florian Marquardt

Theoretical Physicist at [@MPI_ScienceOfLight](https://wisskomm.social/@MPI_ScienceOfLight ). This is the account of my research group. Topics include #machinelearning for (#quantum) #physics, #neuromorphic computing, artificial scientific discovery, topology, quantum computing, optomechanics, ... ELLIS fellow

Homepage: https://mpl.mpg.de/divisions/marquardt-division
YouTube channel: https://www.youtube.com/channel/UCv6Vor64ENpDyevHvrDQSLA
LinkedIn: https://www.linkedin.com/in/florian-marquardt-4320281a4/
Blog (all my Mastodon posts): https://fmarqua.pages.gwdg.de/blog/

Last Notes

2026-09-07 18:46:34 GMT

- reply

An interesting thought exercise related to this: what if someone else than Einstein would have derived the theory of relativistic gravitation, but in a rather mathematical, purely formal way, without the equivalence principle and the geometric interpretation? This seems to be another example where just having a solution (in this case: a model with predictive power) would not be the entire story and what you suggest (resolving the same problem in another way) would have paid off nicely.

2026-08-26 10:51:24 GMT

- reply

Breaking up US AI labs right now would probably realistically mean they are turned over to US government control, which I am afraid would be significantly worse, given the current government.
I think one of the best things to happen in this context would be a European frontier open source model, preferably not run by a company but by some EU research institution --- providing a welcome counterweight.
But I am disheartened by the failure of Mistral (which is a company, but at least located in France) to compete at the level of OpenAI or Anthropic, while Chinese models are catching up.

2026-08-26 09:37:04 GMT

- reply

"we'll need to decide between keeping math human and letting math go its own way" -- I never thought of it in this sense yet, but 'letting math go its own way' is a very interesting notion! (of course, the path taken will not be intrinsic to math, but depend to se degree on the judgement and preferences of the AI models, just like now it depends on the judgement and preferences of human mathematicians). What a future!

2026-02-22 13:46:25 GMT

- reply

That is a reasonable starting point for a discussion about AI use you are linking to. In the end, it will be the readers (rather than the platform) that have to make up their mind. Maybe they will go even more than before by name recognition and reputation if the vast majority of submissions become noise.

2025-04-06 09:35:07 GMT

Today I am returning from my scouting trip to the island of #Helgoland , where we will be organizing our conference 100 Years of Quantum Mechanics in June. The island in the North Sea is a beautiful spot! The ferry ride from Hamburg takes almost four hours. We are looking forward to a distinguished set of invited participants (see below). We will also be recording, for those who cannot be there.
#Quantum
@npub175x…tk6n
[@MPI_ScienceOfLight](https://wisskomm.social/@MPI_ScienceOfLight )
https://helgoland2025.yalepages.org
https://fediscience.org/system/media_attachments/files/114/290/309/750/397/291/original/3c57ba37f35ebe25.jpeg
https://fediscience.org/system/media_attachments/files/114/290/309/759/213/937/original/8abed84d5fe99841.jpeg

2025-02-19 22:00:34 GMT

- reply

In simple terms, how should I think of a Safeguarded AI system and the role of category theory in developing it?

2025-02-18 18:40:32 GMT

Today I gave an interview to #Deutschlandfunk German radio #DLF about why I signed the call
"Declaration To Defend Research Against U.S. Government Censorship"
https://docs.google.com/document/d/1cC6Vu_CAtgxSck8S9be_hmzSu-w2DsaJ02Nw5oeayAg/edit?tab=t.0
The interview probably/possibly airs tomorrow (Wednesday).
#DefendResearch

2025-01-29 09:33:29 GMT

Please share: Our Max Planck Institute recently left X and is present here on Mastodon. Give them a follow! Beautiful pictures from the science of light!
#Mastodon #Physics #Science #Light #Quantum #Optics #Photonics #Pictures
@npub175x…tk6n
From: [@MPI_ScienceOfLight](https://wisskomm.social/@MPI_ScienceOfLight )
https://wisskomm.social/@MPI_ScienceOfLight/113906463840724222