plain_simon's profile picture. Mathematician (algebraic geometry, motives & friends, singularities in statistics and ML). 

'Geometry is successful magic' (R. Thom) 

University of Essen.

Simon Pepin Lehalleur

@plain_simon

Mathematician (algebraic geometry, motives & friends, singularities in statistics and ML). 'Geometry is successful magic' (R. Thom) University of Essen.

Pinned

"Pessimism of the Algebra, optimism of the Geometry"


Simon Pepin Lehalleur reposted

To thwart malicious prompts, large language models use filters that are smaller than the models themselves. Recently, the Turing Award-winning cryptographer Shafi Goldwasser and researcher Sarah Ball showed that this size disparity makes LLMs insecure. quantamagazine.org/cryptographers…

QuantaMagazine's tweet image. To thwart malicious prompts, large language models use filters that are smaller than the models themselves. Recently, the Turing Award-winning cryptographer Shafi Goldwasser and researcher Sarah Ball showed that this size disparity makes LLMs insecure. quantamagazine.org/cryptographers…
QuantaMagazine's tweet image. To thwart malicious prompts, large language models use filters that are smaller than the models themselves. Recently, the Turing Award-winning cryptographer Shafi Goldwasser and researcher Sarah Ball showed that this size disparity makes LLMs insecure. quantamagazine.org/cryptographers…

Simon Pepin Lehalleur reposted

🚨 My last paper as a postdoc is out! Practical Alignment: Why Learning from Human Feedback Alone Is Not Enough for Safe and Helpful AI. This paper started from a simple observation: Humans have incorrect beliefs about the world — and today’s alignment methods pretend they…

khanhxuannguyen's tweet image. 🚨 My last paper as a postdoc is out!

Practical Alignment: Why Learning from Human Feedback Alone Is Not Enough for Safe and Helpful AI.

This paper started from a simple observation:

Humans have incorrect beliefs about the world — and today’s alignment methods pretend they…

Simon Pepin Lehalleur reposted

If a mass of incompressible fluid is floating in space, subject only to its own gravity and centrifugal force, what shape will it adopt as its angular momentum is increased? Initially, it will form an oblate spheroid, with two equal semi-axes shorter than its axis of rotation.


Simon Pepin Lehalleur reposted

1/ The Collatz conjecture has humbled mathematicians for near 90 years. Paul Erdös said "mathematics is not yet ready for such problems." 🧵 So we handed it to transformers. What they learned stunned us! New research from Axiom's math discovery team, blog live this morning.

axiommathai's tweet image. 1/ The Collatz conjecture has humbled mathematicians for near 90 years. Paul Erdös said "mathematics is not yet ready for such problems." 🧵

So we handed it to transformers. What they learned stunned us!

New research from Axiom's math discovery team, blog live this morning.

Simon Pepin Lehalleur reposted

My game's weirdest feature explained in 5 words or less snake but fractal

My game's weirdest feature explained in 5 words or less 2d + top down = 🤯



Simon Pepin Lehalleur reposted

Real Analysis, The Game (v0.1) is DONE!! 44 Worlds 138 Levels All your old favorites like Bolzano-Weierstrass and Heine-Borel, Uniform Convergence and Riemann Sums, and the biggest Boss of all, the Intermediate Value Theorem! :) Play the game here: adam.math.hhu.de/#/g/AlexKontor…

AlexKontorovich's tweet image. Real Analysis, The Game (v0.1) is DONE!!

44 Worlds

138 Levels

All your old favorites like Bolzano-Weierstrass and Heine-Borel, Uniform Convergence and Riemann Sums, and the biggest Boss of all, the Intermediate Value Theorem! :)

Play the game here: adam.math.hhu.de/#/g/AlexKontor……
AlexKontorovich's tweet image. Real Analysis, The Game (v0.1) is DONE!!

44 Worlds

138 Levels

All your old favorites like Bolzano-Weierstrass and Heine-Borel, Uniform Convergence and Riemann Sums, and the biggest Boss of all, the Intermediate Value Theorem! :)

Play the game here: adam.math.hhu.de/#/g/AlexKontor……

Simon Pepin Lehalleur reposted

why is this phrased like a sports team trying to sign a free agent

BREAKING: Barron Trump is reportedly “very close to putting his faith in Christ,” per Stuart Knechtle.

LeadingReport's tweet image. BREAKING: Barron Trump is reportedly  “very close to putting his faith in Christ,” per Stuart Knechtle.
LeadingReport's tweet image. BREAKING: Barron Trump is reportedly  “very close to putting his faith in Christ,” per Stuart Knechtle.


Simon Pepin Lehalleur reposted

6 parabolic reflectors turn circular waves into linear waves bouncing in a hexagonal cavity, focusing energy like a resonator—perfect to ignite fusion by concentrating shock waves at the center.


Simon Pepin Lehalleur reposted

You sit at the extreme tip of a four-billion-year web of connected life, graphed here in a way that barely manages to capture even a sliver of the true complexity. Cope.

StatisticUrban's tweet image. You sit at the extreme tip of a four-billion-year web of connected life, graphed here in a way that barely manages to capture even a sliver of the true complexity.

Cope.

Remember , the people who mock Christians believe the most retarded origin myth imaginable



This is a really beautiful result!

A reset button for rotation could change how we control them all. Is it possible to cancel out a complicated spin without painstakingly reversing every single move? Surprisingly, the answer is yes. Mathematicians Jean-Pierre Eckmann (University of Geneva) and Tsvi Tlusty…

Rainmaker1973's tweet image. A reset button for rotation could change how we control them all.

Is it possible to cancel out a complicated spin without painstakingly reversing every single move? Surprisingly, the answer is yes.

Mathematicians Jean-Pierre Eckmann (University of Geneva) and Tsvi Tlusty…


Simon Pepin Lehalleur reposted

What you see here is a super cool and important magic number matrix you don't know about. This is the standard JPEG quantization matrix. It makes compression significantly more efficient utilizing specifics of human eyes (we see lower frequencies better).

endermanch's tweet image. What you see here is a super cool and important magic number matrix you don't know about. This is the standard JPEG quantization matrix. It makes compression significantly more efficient utilizing specifics of human eyes (we see lower frequencies better).

Simon Pepin Lehalleur reposted

#GreatBooks4Math On Quaternions and Octonions - Their Geometry, Arithmetic and Symmetry John Horton Conway and Derek Smith, 2003 #math #science #iteachmath #mtbos #elearning

74WTungsteno's tweet image. #GreatBooks4Math
On Quaternions and Octonions - Their Geometry, Arithmetic and Symmetry
John Horton Conway and Derek Smith, 2003
#math #science #iteachmath #mtbos #elearning

Simon Pepin Lehalleur reposted

Standing-wave nodes, traveling-wave fronts, and focusing effects all pop into view because the color tracks instantaneous amplitude, not smoothed energy. You see every micro-interaction.


Simon Pepin Lehalleur reposted

Geometry Meets Physics: Finiteness, Tameness, and Complexity (November 12-14, 2025) Website: cmsa.fas.harvard.edu/event/geophys/ Recordings: youtube.com/playlist?list=…

BahramShakerin's tweet image. Geometry Meets Physics: Finiteness, Tameness, and Complexity

(November 12-14, 2025)

Website: cmsa.fas.harvard.edu/event/geophys/

Recordings: youtube.com/playlist?list=…

Simon Pepin Lehalleur reposted

This paper really is groundbreaking. It solves a long-standing embarrassment in machine learning: despite all the hype around deep learning, traditional tree-based methods (XGBoost, CatBoost, random forests, etc) have dominated tabular data—the most common data format in…

burkov's tweet image. This paper really is groundbreaking. It solves a long-standing embarrassment in machine learning: despite all the hype around deep learning, traditional tree-based methods (XGBoost, CatBoost, random forests, etc) have dominated tabular data—the most common data format in…

Simon Pepin Lehalleur reposted

The scale of the shockwave from that solar flare is honestly hard to wrap your head around. By WonFont


Simon Pepin Lehalleur reposted

Ultrafinitists: "I only see one case here."

the canonical two cases: sets of size less than 4 and sets of size at least 4



Simon Pepin Lehalleur reposted

ok yeah, I'm officially calling grok a model organism of emergent misalignment

late entry to the grok decision contest

fdosmither's tweet image. late entry to the grok decision contest


Simon Pepin Lehalleur reposted

Remarkably, this is also true: P(X ≥ Ua) ≤ E[X]/a where U is a Uniform[0,1] random variable. This is strictly stronger than Markov as U < 1 with probability 1, but you can't replace U with any constant < 1.

The Law of Large Numbers is an incredible result but this unassuming inequality by the Russian mathematician Andrey Markov is absolutely insane! You're able to say something about probabilities of a random variable without knowing its distribution!🤯?! Moreover, from this one…

mathelirium's tweet image. The Law of Large Numbers is an incredible result but this unassuming inequality by the Russian mathematician Andrey Markov is absolutely insane!

You&apos;re able to say something about probabilities of a random variable without knowing its distribution!🤯?!

Moreover, from this one…


Simon Pepin Lehalleur reposted

a very funny thing to say about a movie whose ending is about as definitive as it gets, both in plot and theme

Rewatched "No Country for Old Men" yesterday, and I am placing my head under the guillotine here by saying this, but making movies with ambiguous endings is terrible art because it shows the director's and writer's lack of decisiveness in making, defending and completing their…



Loading...

Something went wrong.


Something went wrong.