PH-Dreamer: A Physics-Driven World Model via Port-Hamiltonian Generative Dynamics

arXiv 2026 56.3 method

TLDR

A physics-driven world model using Port-Hamiltonian dynamics to enforce physical priors, improving simulator fidelity and control efficiency.

Reasoning

The paper introduces a novel framework that integrates physical principles into world models, demonstrating clear improvements in reward alignment, energy consumption, and jerk. Strengths include a principled approach and empirical validation, but the abstract lacks details on benchmark diversity and comparison baselines.

Read-first score

Read-first score 56.3, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 56.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 29%

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Citation impact 18%

72.2

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.72150319

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=benchmark

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FrontierBridge

Rank sensitivity

Stability: volatile; rank range: 489.

Keyword Scores

world model

world simulator

generative world model

world dynamics prediction

model-based reinforcement learning world model

interactive world model

video world model

Deep Analysis

Innovations

Embedding implicit physical priors into recurrent transitions by modeling projected latent evolution as action controlled energy routing governed by flow and dissipation, biasing the PH phase space toward a more compact and physically structured representation.
Developing a kinematics aware energy world model that estimates the Hamiltonian and power balance from proprioceptive observations, providing an explicit physical signal for thermodynamic reasoning.
Establishing an energy guided Actor-Critic that uses Lagrangian multipliers to regularize policy optimization toward lower energy and smoother control.

Methodology

The paper proposes PH-Dreamer, a world model that integrates Port-Hamiltonian dynamics into recurrent state space architectures. It uses three components: implicit physical priors in latent transitions, a kinematics-aware energy estimator, and an energy-guided Actor-Critic with Lagrangian multipliers. The model is evaluated on visual control benchmarks, measuring asymptotic returns, reward alignment, phase space volume, energy consumption, and jerk.

Key Results

The model attains superior asymptotic returns, tighter lower variance alignment between imagined and real rewards, reduces latent phase space volume by 4.18-8.41%, energy consumption by up to 7.80%, and mean squared jerk by up to 9.38%.

Resources

Paper