- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes CoMNO, which constructs closure evidence by comparing prediction-before-downsampling with downsampling-before-prediction. A recurrent memory preserves this evidence, while spatially routed experts produce additive corrections to the base prediction. Experiments across five PDE systems report improved long-horizon accuracy and physical fidelity.
Critical strengths
- The method targets the persistent influence of unresolved spatial structures on long-horizon prediction.
- The commutator signal, recurrent memory, and spatially adaptive experts constitute a coherent correction pipeline, and the ablations indicate that each component contributes to rollout performance.
- Experiments span five PDE systems, spatial coarsening, high-frequency Fourier errors, module ablations, and frozen-checkpoint interventions.
Critical weaknesses
- The observable commutator is obtained by further coarsening , whereas the target closure concerns scales that are already lost when moving from to . In the absence of a comparison against an oracle high-resolution closure residual, the experiments establish the predictive utility of this signal, but not its interpretation as unresolved fine-scale influence.
- Although the neural-operator baselines are broad, no explicit learned closure model is included, which leaves the specific advantage of commutator-based closure over alternative closure formulations unclear.
- The retrained "w/o Operator Commutator" variant removes both the evidence and discrepancy losses, whereas the frozen-checkpoint intervention suppresses the evidence only at inference. Consequently, the effects of the online signal and of its training-time regularization are not fully disentangled.
Questions for authors
- How does the observable commutator, obtained by further coarsening , relate to the observation-grid closure residual induced by information already lost when moving from to ? Is there any direct comparison against, or correlation with, an oracle residual derived from high-resolution dynamics?
- Could the authors disentangle the respective contributions of the online commutator evidence and of ? The current retrained ablation removes both.
- What evidence substantiates the advantage of commutator-based closure over existing learned closure formulations? Are comparisons with explicit closure models available under a matched protocol?