Mehmet Süzen
banner
msuzen.bsky.social
Mehmet Süzen
@msuzen.bsky.social
Physicist | Scientist | #TheoreticalPhysics #Maths #Philosophy #Cosmology #StatisticalPhysics #Quantum #Gravity #DataScience #AI #MachineLearning
Own views.

Blog: http://science-memo.blogspot.com
mathstodon: https://mathstodon.xyz/@msuzen
Pinned
M83 VAR! plate was a historic moment. Our understanding of who we are as human beings and where do we live. This was probably a defining moment in human history
#hubble #cosmology

carnegiescience.edu/about/histor...
Hubble's Famous M31 VAR! plate
carnegiescience.edu
I am pleased to see that even after 10 years later, R ecosystem is superbly reproducible in a field like statistical physics.

#R #R-lang #statisticalmechanics
October 24, 2025 at 1:30 PM
Modern science has been a voyage into the unknown, with a lesson in humility waiting at every stop. Many passengers would rather have stayed home. - Carl Sagan
October 24, 2025 at 1:27 PM
Great progress for R community. Now R ecosystem catching up with Python in terms of usage.

rig github.com/r-lib/rig R installation manager (this is similar to ‘pyenv’

cc @posit.co well done to talented team.
cc @ramikrispin.bsky.social
GitHub - r-lib/rig: The R Installation Manager
The R Installation Manager. Contribute to r-lib/rig development by creating an account on GitHub.
github.com
September 13, 2025 at 5:42 PM
M83 VAR! plate was a historic moment. Our understanding of who we are as human beings and where do we live. This was probably a defining moment in human history
#hubble #cosmology

carnegiescience.edu/about/histor...
Hubble's Famous M31 VAR! plate
carnegiescience.edu
September 13, 2025 at 5:24 PM
Differentiating Ergodicity of von Neumann-Birkhoff and Gibbs-Boltzmann is now resolved

Ergodicity detection algorithms
arxiv.org/html/2508.08...
Code:
github.com/msuzen/resea...

#statisticalmechanics #boltzmann #gibbs #ergodicity

cc @stevenstrogatz.com @stephenwolfram.bsky.social
research/ergodicity_symbolic_dynamics at main · msuzen/research
Memo's research works. Contribute to msuzen/research development by creating an account on GitHub.
github.com
August 14, 2025 at 8:07 AM
Circular Reasoning Fallacy: If I have a tail, I will chase it, but If I chase it, I will have my tail. #fallacies #logic
July 11, 2025 at 9:55 PM
Reposted by Mehmet Süzen
My latest for @nytimes.com -- please repost so your followers can see this for free. www.nytimes.com/interactive/...
How a Puzzle About Fractions Got Brain Scans Rolling (Gift Article)
A story of bowling pins, patterns and medical miracles.
www.nytimes.com
June 30, 2025 at 10:33 AM
Mathematical research, is it invention or discovery?

If there there is a natural phenomenon having mathematical description or a mathematical description corresponds to a natural phenomenon, it is discovery. Others are inventions.
Sometimes, inventions turn out to be discoveries.

#maths #ai
July 1, 2025 at 5:44 AM
Reposted by Mehmet Süzen
(2/2) This weekend, I took this template to the next level, fully cloning my local terminal settings, which include the following features:
🔹 Setting zsh and on-my-zsh
🔹 Mounting my local zsh history
🔹 Installing my local CLI tools

Template repo: github.com/RamiKrispin/...
GitHub - RamiKrispin/vscode-python-uv-template: A Python Environment Template for VScode with UV
A Python Environment Template for VScode with UV. Contribute to RamiKrispin/vscode-python-uv-template development by creating an account on GitHub.
github.com
June 22, 2025 at 12:38 AM
Similarity of real matrices with different shapes: deep learning and brain networks.

Mixed random matrix ensembles to find equivalent deep architectures.
arxiv.org/abs/2006.13687

#MatrixSimilarity #NeuronalNetworks
Equivalence in Deep Neural Networks via Conjugate Matrix Ensembles
A numerical approach is developed for detecting the equivalence of deep learning architectures. The method is based on generating Mixed Matrix Ensembles (MMEs) out of deep neural network weight matric...
arxiv.org
June 19, 2025 at 5:05 PM
Reposted by Mehmet Süzen
New paper: "Large Language Models and Emergence: A Complex Systems Perspective" (D. Krakauer, J. Krakauer, M. Mitchell).

We look at claims of "emergent capabilities" & "emergent intelligence" in LLMs from the perspective of what emergence means in complexity science.

arxiv.org/pdf/2506.11135
arxiv.org
June 16, 2025 at 1:15 PM
'anc'
directory can be used to host code or notebooks for reproducibility on arXiv. This is superb, while we can host code on arXiv too, as a supplement in a compact manner.

info.arxiv.org/help/ancilla...

#arXiv #reproducibility
Ancillary Files (data, code, images) - arXiv info | arXiv e-print repositorycontact arXivsubscribe to arXiv mailingsReport an issue
info.arxiv.org
June 9, 2025 at 12:01 PM
Causality encoded in statistical models would not make statistical technique suddenly causal.

#CausalInference #StatisticalModel
June 3, 2025 at 4:35 AM
#ReinforcementLearning #RL Congrats to Barto & Sutton. Actually this. one is also physics domain. I wouldn't surprise if they are nominated. for Physics Nobel too.

awards.acm.org/about/2024-t...
Andrew Barto and Richard Sutton are the recipients of the 2024 ACM A.M. Turing Award for developing the conceptual and algorithmic foundations of reinforcement learning.
Andrew Barto and Richard Sutton as the recipients of the 2024 ACM A.M. Turing Award for developing the conceptual and algorithmic foundations of reinforcement learning. In a series of papers beginning...
awards.acm.org
May 15, 2025 at 5:06 PM
Yes. Zero is even

#maths
April 26, 2025 at 7:15 PM
Entia non sunt multiplicanda praeter necessitatem
Occam

#Overfitting #DataScience #MachineLearning
April 10, 2025 at 2:07 PM
The concept of Occam's gap

Resolution of misconception of overfitting: Differentiating learning curves from Occam curves
science-memo.blogspot.com/2023/04/Occa...

#Overfitting #MachineLearning #DeepLearning #AI
Resolution of misconception of overfitting: Differentiating learning curves from Occam curves
Preamble   Occam (Wikipedia) A misconception that overfitted model can be identified with the  amount of  generalisation gap  between model'...
science-memo.blogspot.com
April 7, 2025 at 4:15 PM
Reposted by Mehmet Süzen
Using JWST, researchers have monitored an isolated exoplanet for 2.5 of its rotation periods. The object’s variable IR spectrum confirmed previous hints of patchy clouds and provided new evidence of hot spots in the upper atmosphere and fluctuating carbon chemistry.
An Isolated Exoplanet Has a Patchy Atmosphere
Observations by the JWST observatory of a star-free super-Jupiter reveal large-scale inhomogeneities in its atmosphere.
physics.aps.org
April 1, 2025 at 6:15 PM
Reposted by Mehmet Süzen
Very excited about the upcoming Natural Philosophy Symposium in Baltimore, May 29-31. There still some room to register!

www.naturalphilosophyhopkins.org/natural-phil...
Natural Philosophy Symposium 2025 - Johns Hopkins Natural Philosophy Forum
The Johns Hopkins Natural Philosophy Forum will be hosting an international symposium on all aspects of natural philosophy, featuring talks from philosophers and scientists in a variety of disciplines...
www.naturalphilosophyhopkins.org
March 28, 2025 at 4:12 PM
"A mathematician is a machine for turning coffee into theorems"
Paul Erdös

#LongCoffee #Maths #AppliedMaths #physics
March 26, 2025 at 5:58 PM
In-Context Learning meets Chaos

In-context learning is quite promising but limitations appear akin to chaotic dynamics.

A new Kind of AutoML:
memosisland.blogspot.com/2024/05/llm-...

#llm #reasoning #InContextLearning #chaos #ai
#ChainOfThought #LLMstability #SymbolicAutoML
A new kind of AutoML: On the fallacy of many-shot in-context learning with LLMs and PLMs
Extension of this post, appear as a short article with a conjecture: Mehmet Süzen. In-context learning as a new kind of symbolic-AutoML: Lya...
memosisland.blogspot.com
March 23, 2025 at 5:58 PM