Physically Native World Models: A Hamiltonian Perspective on Generative World Modeling
> Paper: Physically Native World Models: A Hamiltonian Perspective on Generative World Modeling > Authors: Sen Cui, Jingheng Ma > arXiv: 2605.00412 | 2026-04-29
1. The AI That "Knows Pixels, Not Physics"
Imagine an AI predicting the next frame of a video:
Problems with existing world models:
- They only see pixel changes
- They don't understand the physics behind them
- Predictions may violate physics:
- A ball passing through a wall
- Objects vanishing from thin air
- Violations of energy conservation
- A mathematical framework of classical mechanics
- Describes systems via energy (the Hamiltonian)
- Predicts system evolution
- Guarantees energy conservation
- It captures the underlying laws of the physical world
- Energy conservation → predictions cannot "create from nothing"
- Reversibility → understanding causality
- Structure → better generalization
- Low sample efficiency: countless videos needed to learn "balls fall"; humans have innate physical intuition.
- Poor generalization: failure on scenarios absent from training data; physics priors help generalize.
- Unphysical predictions: violating common sense is dangerous in robotics and autonomous driving.
- Sample efficiency: strong inductive bias means less data is needed.
- Reliable predictions: no physically impossible outputs — safer in critical applications and verifiable.
- Interpretability: the model's "thinking" maps to physical quantities humans can inspect and debug.
- Nature follows concise mathematical laws
- Understanding these laws = understanding nature
- AI understanding physics = AI understanding the world
Three separate research routes: 1. 2D video generation: focused on visual future synthesis 2. 3D scene models: focused on spatial reconstruction 3. JEPA-style implicit models: focused on representation learning
But all three lack a key element: physics.
2. Hamiltonian Mechanics: The Mathematical Heart of the Physical World
What is Hamiltonian mechanics?
Why it suits world models:
3. Physically Native World Models
The paper proposes:
> World models should have physics priors built in, not merely learn physics from data.
Technical approach:
1. Hamiltonian structure — internal representations = physical state; evolution rules = Hamiltonian equations; prediction = physical simulation. 2. Energy constraints — generation is constrained by conservation of energy, eliminating physically impossible scenarios and making predictions more reliable. 3. Cross-modal unification — 2D video, 3D scenes, and implicit representations all fit a single physical framework. 4. Physical interpretability — internal states correspond to physical quantities (position, velocity, energy), making the model interpretable and debuggable.
The analogy: traditional world models learn physics only from video (like an infant), while physically native models are born knowing physics (like a physicist) — a higher starting point and better learning.
4. Why Physics Priors Matter
Problems with purely data-driven models:
Advantages of being physically native:
5. A Feynman-Style Judgment: Physics Is Nature's Language
Feynman said:
> "The laws of physics are the grammar of nature."
Applied to world models:
> "A world model without physics is like learning a language without grammar — it can say many words but nothing truly meaningful. Physics priors let world models speak nature's language."
This reflects a core belief of physics:
6. Takeaways
If you are building world models or predictive systems, ask yourself:
1. Does my model have built-in physics priors? 2. Do predictions obey physical conservation laws? 3. Could physical structure improve sample efficiency and generalization? 4. Does Hamiltonian mechanics apply to my scenario?
The core insight: a world model should not be just a pixel predictor — it should be a physics simulator.
When AI has physics "common sense" built in, it can make reasonable predictions in scenarios it has never seen. In the future of embodied intelligence, being physically native may be the key step from "looking intelligent" to "being intelligent."
In the art of predicting the world, physics is the oldest and most profound teacher.