- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 3
Read review
Paper summary
This paper proposes LM-HNN, which decomposes the Hamiltonian energy model into shared local functions into multiple interaction neighborhoods. This construction reduces the dimensionality of the functions being learned while retaining dynamics induced via Hamiltonian energy functions. Spectral diagnostics show improved recovery of fine-scale energy variations and their derivatives. Authors tested LM-HNN across six conservative systems, and show a better performance compared with some prior works including HNN, SympNet, FS-HNN, and included improved performance in unseen-energy evaluations.
Critical strengths
- The idea is relevant in learning a compositional Hamiltonian system.
- The experiment and ablations are convincing at least to me, for given six dynamical systems and to understand if local interaction and decomposition help in learning a high dimensional energy modeling.
Critical weaknesses
- The claim that this model is a local multi-scale model is not convincing nor clear to me. In L180-206, authors claim the neighborhood construction and interaction within local variables but I didn't find different scale or (I saw but I am unsure if is identical to ). In appendix it seems three high dimensional cases are only in chain topology. It might be better if there exists a visualization to explain how the multiscale is being properly addressed in LM-HNN.
- Only conservative systems are considered, where observation noise or dissipative or general systems are ruled out, which is more general and practical if data are given via trajectory observations.
Questions for authors
- What is the need of multi-scale or neigobourhood setup for single pendulum, double pendulum and fast-slow oscillator? Why it is better than conventional methods even when there is no complex state space topology at all?
- Could you quantify the overlap between training and test distributions of local strains or neighborhoods? This would distinguish recombination of familiar interactions from extrapolation of the local interaction law itself. Would reported in appendix same as defined in Section 4.1 and how does that affect choice of respectively?
- Can one generalize the connectivity graph of given systems or use articulated chain models to test more general decompositional energy learning cases?
6 comments
W2. Robustness to Observation Noise and Dissipative Dynamics
We evaluate LM-HNN on the FPUT chain () under observation noise and energy dissipation. All experiments use parameter-matched HNN baselines, three independent seeds, and 120-step rollouts from unseen initial conditions.
Observation noise. We add Gaussian noise to the training data, with standard deviations corresponding to 5% and 10% of each coordinate's standard deviation.
| Noise level | Model | Rollout MSE | Peak err. |
|---|---|---|---|
| 0% | LM-HNN | 4.9e-3 | 0.83% |
| 0% | HNN | 2.0e-1 | 4.86% |
| 5% | LM-HNN | 1.0e-2 | 0.95% |
| 5% | HNN | 2.8e-1 | 5.22% |
| 10% | LM-HNN | 6.0e-2 | 1.64% |
| 10% | HNN | 6.4e-1 | 6.62% |
Although both models degrade with increasing noise, LM-HNN maintains substantially lower rollout and peak errors at every noise level, demonstrating robustness to noisy observations.
Dissipative dynamics. We further introduce linear damping with , for which the true system retains approximately 54% of its initial energy. When trained on the damped trajectories using their original conservative formulations, neither model can reproduce the energy decay. We therefore introduce the same learnable damping term into both models, following dissipative Hamiltonian formulations [3, 4]:
| Formulation | Model | Rollout MSE | Peak err. | Learned | |
|---|---|---|---|---|---|
| Conservative | LM-HNN | 1.4e-1 | 8.60% | 0.99 | — |
| Conservative | HNN | 2.6e-1 | 9.06% | 1.00 | — |
| With damping | LM-HNN | 4.1e-3 | 0.81% | 0.54 | 0.0999 |
| With damping | HNN | 1.1e-1 | 4.01% | 0.54 | 0.1011 |
Both dissipative extensions recover the damping coefficient and energy decay, but LM-HNN provides substantially more accurate trajectories. The dissipative extension leaves the local multiscale Hamiltonian representation unchanged and only adds the damping term.
These experiments show that LM-HNN is robust to observation noise and can accommodate dissipative dynamics through an explicit damping term. The results have been included in the revised appendix.
Q1. Role of Energy Decomposition and Multiscale Neighborhoods in Low-Dimensional Systems
The single pendulum, double pendulum, and fast–slow oscillator evaluate a different aspect of LM-HNN from the interacting many-body systems. In these low-dimensional benchmarks, the performance gains primarily arise from the structured decomposition of kinetic and potential energies, rather than spatial locality or multiscale neighborhood aggregation.
Unlike a conventional HNN that learns the entire Hamiltonian through an unconstrained network, LM-HNN separates kinetic and potential energies. For the pendulum systems, it explicitly encodes quadratic momentum dependence:
This reduces the approximation burden by requiring the network to learn configuration-dependent coefficients rather than an arbitrary function of positions and momenta. For the single pendulum, the structure captures quadratic kinetic energy and periodic potential; for the double pendulum, it additionally incorporates the configuration-dependent mass matrix and angular interactions. The fast–slow oscillator tests structured Hamiltonian learning under separated temporal scales, which are distinct from spatial neighborhood radii.
Thus, the low-dimensional benchmarks validate the benefits of kinetic–potential decomposition and structured momentum dependence, while the interacting-chain and lattice experiments demonstrate the additional advantages of locality and multiscale representations. The parameter-matched comparisons in Table 14 further support that the improvements are not simply due to increased model capacity.
Q2(a). Training–Test Overlap of Local Neighborhoods and Unseen-Energy Generalization
We analyze the FPUT chain with masses to quantify whether unseen-energy generalization arises from recombining familiar local interactions or extrapolating local interaction laws.
Models are trained on energies per mass , where is the total energy, and evaluated at unseen values . We compare training–test overlap for individual bond strains, local neighborhoods, and full states.
A test sample counts as covered if it is as close to the training data as training samples are to other training trajectories. Specifically, its Euclidean nearest-neighbor distance to the training set must fall below the 95th percentile of nearest-neighbor distances between different training trajectories, with this threshold set separately for each representation. This feature-space distance measures distribution overlap and is distinct from the Chebyshev metric defining spatial neighborhoods.
| Representation | In-range test | Unseen-energy test |
|---|---|---|
| Individual strains within training range | 100% | 99.1% |
| Strain neighborhoods, | 97% / 94% / 93% | 87% / 75% / 72% |
| Position neighborhoods, | 97% / 96% / 94% | 89% / 80% / 76% |
| Full configuration | 93% | 66% |
| Full state | 94% | 40% |
The individual-strain row reports the fraction of test strains lying between the smallest and largest strains seen in training. All remaining rows use the nearest-neighbor criterion described above.
Despite the shift in global energy, 99.1% of individual strains remain within the training range, and local neighborhoods retain substantially greater coverage than full states. In particular, the full-state coverage drops to 40%, while radius-zero strain and position neighborhoods retain 87% and 89% coverage, respectively. The Toda and Morse chains show similar coverage of single strains and local neighborhoods, as reported in the newly added Table 22 of the updated manuscript.
These findings support the interpretation in Section 4.3 that LM-HNN can reuse familiar local interaction laws in globally unseen configurations. The reduced coverage at larger radii also shows that some local energy functions must extrapolate under more substantial energy shifts. This analysis has been included in Appendix C.10 and connected to the energy-generalization discussion in Section 4.3.
Q2(b). Relationship Between Individual Radius and Radius Set
As defined in W1, denotes a single neighborhood radius, whereas represents the set of radii combined by LM-HNN. For example, defines three separate neighborhoods , , and , whose learned energy contributions are aggregated through scale weights.
The coverage analysis in Q2(a) also explains the role of different radii. Smaller neighborhoods retain greater training–test overlap under energy shifts, whereas larger neighborhoods capture more extended interactions. Combining multiple radii allows LM-HNN to represent these complementary interaction scales within the same Hamiltonian. The notation and neighborhood selection have been clarified in Section 4.
Thank you for the constructive comments and suggestions. In response, we have clarified the mathematical definition and implementation of local multiscale neighborhoods, added experiments on two-dimensional lattices and articulated chains, and investigated robustness to observation noise, energy dissipation, and unseen energy levels. We address each weakness and question below.
W1. Definition, Visualization, and Effectiveness of Local Multiscale Neighborhoods
We clarify the distinction between individual neighborhood radii and the set of radii used by LM-HNN, together with their construction, aggregation, and relationship to physical interaction ranges.
Given a reference index set I with distance metric d, the radius-r neighborhood is
Nr(i)=j∈I(i,j)≤r.
Here, r denotes a single neighborhood radius, whereas R is the set of radii used simultaneously. For Cartesian lattices, we adopt the Chebyshev metric on fixed reference coordinates:
d∞(i,j)=∥xi−xj∥∞=max1≤k≤D∣xi,k−xj,k∣.
In one dimension, this gives a window of 2r+1 consecutive indices. In two dimensions, it gives a (2r+1)×(2r+1) square patch that includes diagonal neighbors. Near fixed boundaries, out-of-domain entries are zero-padded. These neighborhoods depend on the reference geometry rather than instantaneous displacements or added physical coupling edges.
The extracted variables depend on the energy component. For one-dimensional potentials, neighborhoods are defined over bond strains sb=qb+1−qb, giving sb(r)=(sb−r,…,sb+r). Kinetic branches instead use site-position neighborhoods and, when needed, local momenta. Two-dimensional models use the corresponding square patches.
At each radius, a separate network is shared across equivalent neighborhoods, and their contributions are aggregated as
Vθ(q)=∑r∈RπrV∑i∈IVvθ,r(ψi,rV(q)).
Here, πrV is the learned scale weight and ψi,rV extracts local potential variables. The kinetic branches follow Eqs. (10)–(12). For the original chains, R=0,1,2 corresponds to 1, 3, and 5 consecutive strains, with a shared width-64 network at each scale. Our additional two-dimensional experiments evaluate configurations including 1,2 and 1,2,3.
Interaction-range experiments. The receptive-field ablations in Appendix B and the new experiments on two-dimensional FPUT lattices demonstrate how the choice of radius relates to the underlying physical interactions.
We consider an L×L square lattice of unit masses with fixed boundaries, where each site a=(i,j) has scalar displacement qa and momentum pa. The Hamiltonian combines nonlinear nearest-neighbor FPUT bonds with distance-dependent harmonic interactions [1, 2]:
H=∑a21pa2+∑⟨a,b⟩ϕFPUT(qa−qb)+21∑∥a−b∥∞≤rc (a,b) not nearest neighborsg(dab2)αlr(qa−qb)2.
Here, ϕFPUT uses α=0.25 and β=0.10, while dab is the Euclidean distance and rc is the Chebyshev interaction cutoff. We evaluate a short-range setting (rc=1, g=0.3), with diagonal harmonic coupling, and an extended-range setting (rc=3, αlr=1, g=0.3), combining anharmonic nearest-neighbor interactions with longer-range harmonic couplings.
The following table reports rollout MSE on 4×4 lattices over three training seeds, with a parameter-matched HNN baseline.
LM-HNN substantially outperforms HNN under all tested radius configurations. A smaller radius is most accurate for short-range interactions, whereas broader multiscale neighborhoods improve accuracy when the interaction range increases. A potential term of radius r couples two sites only if they lie within Chebyshev distance 2r, so R=1 misses the couplings at distance 3 in the extended-range setting.
These results demonstrate that neighborhood selection has a direct physical interpretation and that combining multiple scales allows LM-HNN to capture interactions of different spatial ranges. We have clarified these in revised Section 4.