cs.AIAug 24, 2026

Correcting a learned physical invariant improves world-model rollouts

Authors: Richard Bao

Organizations: Independent Researcher

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

World models can predict video without learning dynamics that they reliably preserve. We test whether a frozen DreamerV3 trained only on pendulum video learns a scalar that its own latent transition treats as approximately conserved. A label-free search recovers the same energy-like invariant across independently trained conservative models, while the same procedure finds no comparable invariant in matched damped models. During autonomous rollouts, this quantity drifts. Projecting the latent state back toward its initial level set reduces rollout error in all three conservative models, whereas matched random constraints usually increase it. These results distinguish a dynamically meaningful invariant from a merely decodable correlate and reveal a concrete failure mode: a world model can learn a physical constraint from pixels yet violate that constraint when it imagines forward.

Figures & tables

Figure 1: Each point represents one model. (a) The three trained conservative models recover energy correlations between 0.9670.967 and 0.9750.975 , while six untrained models span 0.170.17 – 0.910.91 . (b) The recovered scalar changes very little within observation-conditioned trajectories for conservative models and much more for damped models. (c) Rollout error as a function of projection strength α\alpha , relative to no edit. Enforcing the recovered constraint lowers error; enforcing a matched random polynomial constraint usually raises it; applying the recovered constraint to a damped model changes little.
statisticconservativedampedmedian ratio
∣ρ∣E|\rho|_{E}0.967–0.9750.018–0.09020×20\times
drift of CC0.70.7 – 1.3×10−41.3\times 10^{-4}0.203–0.2662414×2414\times
held-out invariance ratio0.0027–0.01210.983–0.994282×282\times
pairing residual0.829–0.8650.934–0.957n/a
Table 1: The same extraction pipeline applied to conservative and damped models. Each range covers three seeds. No conservative seed overlaps any damped seed on any row.
Figure 2: Each row compares one model with its own null distribution. The 20 small marks are norm-matched random degree-4 constraints, using the same draws across models; the diamond is that model’s recovered constraint. Labels count the random constraints with a better dose-response slope than the recovered one.
Appendix figures & tables2 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 3: Each point is one of the 12 extracted directions, ranked within its own model; the dotted line marks equal ranks. (a) Variance rank against damage rank at a 40-step horizon. (b) Damage rank against the magnitude of the invariant correction along that direction.
Figure 5
More in this topic

World Models

World models represent and predict how environments change, helping agents plan, act, and reason about possible futures.

See all 360 papers

Explore similar work

CardsList

  1. Imagined Rollouts are Kinematic, Not Dynamic: A Diagnosis of Long-Horizon World-Model Failure

    Jul 7, 2026Finn Rasmus Schäfer, Korbinian Moller, Yuan Gao +3World ModelsWorld Model Evaluation

    Autonomous Vehicle Systems Lab Technical University of Munich, Garching b. München, Germany · Data Analytics and Machine Learning Group Technical University of Munich, Garching b. München, Germany

  2. PhysiCalWorld: In-Context Physics Calibration for World Models

    Oct 21, 2026Yunhao Luo, Wentao He, Yinpei Dai +5World ModelsAction-Conditioned World Models

    University of Michigan

  3. IMPLY: Physically Anchored Consistency for World-Model Rollouts

    Sep 11, 2026Aman Mehta, Riya BaviskarWorld ModelsAction-Conditioned World Models

    Independent Researcher