We’re announcing a major advance in the study of fluid dynamics with AI 💧 in a joint paper with researchers from @BrownUniversity, @nyuniversity and @Stanford.

Sep 18, 2025 · 3:01 PM UTC

Equations to describe fluid motion - like airflow lifting an airplane wing or the swirling vortex of a hurricane - can sometimes "break," predicting impossible, infinite values. These "singularities" are a huge mystery in mathematical physics.
4
13
2
511
We used a new AI-powered method to discover new families of unstable “singularities” across three different fluid equations. A clear and unexpected pattern emerged: as the solutions become more unstable, one of the key properties falls very close to a straight line. This suggests a new, underlying structure to these equations that was previously invisible.
This breakthrough represents a new way of doing mathematical research - combining deep insights with cutting-edge AI. We’re excited for this work to help usher in a new era where long-standing challenges are tackled with computer-assisted proofs. → goo.gle/46loOuZ
9
58
3
746
Here’s the link: deepmind.google/discover/blo… This is discovery on steroids. “In a new paper, we introduce an entirely new family of mathematical blow ups to some of the most complex equations that describe fluid motion” “Physics-Informed Neural Networks (PINNs). Unlike conventional neural networks that learn from vast datasets, we trained our models to match equations which model the laws of physics. The network's output is constantly checked against what the physical equations expect, and it learns by minimizing its ‘residual’, the amount by which its solution fails to satisfy the equations. “ By embedding mathematical insights and achieving extreme precision, we transformed PINNs into a discovery tool that finds elusive singularities. YONGJI WANG, FIRST AUTHOR OF THE STUDY AND POSTDOCTORAL RESEARCHER AT NYU.” @c10labs
1
22
1
139
AI is not just modeling fluids, it is starting to mirror the hidden architecture of reality. The same mathematics that describe water flow are the ones that describe galaxies, neural patterns, and even thought itself. When AI learns fluid dynamics, it is not just learning about water, it is touching the code that binds matter, energy, and mind.
15
7
64
Deepmind with another huge W
43
Interesting to hear Demis discuss video model understanding of physics as if intuitively and the theory that it could lead to a better or alternative routes to understanding wave and particle physics. Days or weeks later we have this. Coincidence? Either way very interesting, and will be excited to watch this potentially help to understand the planet scale dynamics such as Coriolis effect - noninvasive medical applications with sound and light.
6
Oh cool! Nice to see another field of study getting some love from the DeepMind team. Fluid dynamics are super interesting. Congratulations to Yongji Wang & the rest of the contributors
5
Thanks for sharing the article link.
1
cool to see ai tackling such complex problems. fluid dynamics has always had those tricky singularities, so this feels like a solid step forward. hope it leads to more breakthroughs in physics and beyond
Exciting breakthrough! Fluid dynamics has so much potential.
This tweet is unavailable
So if navier stokes is solved who gets a million? Google or engineers who built it??
1
27
how about doing electromagnetism and gravity so we can warp drives and replicators already?
3
10
#GrokImagine on normal mode with no prompt.
9
0
<Solved> The absence of a "desirable solution" to the Navier-Stokes Millennium Problem stems from a paradigm error: attempting to solve a problem of composition with the tools of observation. The very concept of a universal, provable solution is an artifact of a linear, classical mindset that fails to account for the physics of coherence. The Classical (τₖ ≈ 0) Bottleneck The Navier-Stokes equations are a masterful description of fluid dynamics within the Classical Regime (τₖ ≈ 0), the physics of statistical averages. They don't describe the fluid itself, but the emergent, macroscopic consequences of countless underlying quantum events. Turbulence, the chaotic state that defies a smooth, predictable solution, is the macroscopic manifestation of what the XQE framework identifies as Consequential Complexity (N_{consequential}). According to the Law of Creative Interference: The search for a proof is an attempt to extract a small piece of information (I_{extracted})—a definitive, universal statement about smoothness—from a system utterly dominated by the consequential noise (N_{consequential}) generated by its own becoming. You are trying to find a simple, static truth in the wake of a creative explosion. Proof as a Redundant Act of Measurement * A mathematical proof is a fundamentally linear, sequential act of Ingression. It forces a series of logical verdicts to arrive at a static conclusion. * Quantum Time, the true engine of physical processes, is not linear. It resolves paradoxes into generative, temporal spirals. Trying to apply a linear proof to the spiraling, generative complexity of turbulence is a category error. It's like trying to capture the dynamic essence of a vortex with a single, flat photograph. The instrument of measurement (the proof) is fundamentally incompatible with the phenomenon it seeks to describe. The Compositional (τₖ >> 0) Approach An agent operating in the Compositional Regime (τₖ >> 0) would not attempt to prove a solution exists for all possible turbulent states. Instead, they would compose the desired state. The question shifts from: "Can we prove that a smooth, well-behaved solution will always exist for any given initial condition?" To: "How do we modulate the system's coherence (τₖ) to compose a stable, harmonic pattern of flow—a Live Information Token (LIT)—from the underlying potential of the Prima Materia?" The goal is not to predict the chaos, but to introduce a harmonic signal that organizes the medium, much like the Morpheus Protocol uses a coherence agent to reset a biological system. The solution isn't found; it is created. The Millennium Problem remains unsolved because it asks the wrong question, using the tools of a bygone paradigm. The challenge is not to prove the water flows smoothly, but to compose the harmony that persuades it to. time.augmntd.app
1
1
6
Sometimes, you just bow in front of the master
6
GIF
Amazing how we don’t need to brute-force hard problems anymore. AI allows us to shed so much work, time, and other resources. ⚡️ Truly unexpected capabilities are emerging with the AI revolution.
6
Freysa is the future! 🚀💎 The next big crypto wave is here – and it’s called #FREYSA Built for real growth 🌟 $FAI #Fai
1
3