- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper presents a model named ADE-Solver based on observation conditions, which is used to predict the future state of partial differential equations (PDEs). This method incorporates time differences in historical observations, combines multi-scale U-Net with spatial and temporal attention mechanisms, and generates sampling offsets, value corrections, and adaptive gating structures for specific targets. Each future state is constructed by bilinear sampling at the latest observed field learned coordinates, and gated corrections are added to build it. All target states share the same observation anchor points, enabling parallel evaluation of their final construction results. During the training process, the loss function of the solution is adopted, and regularization is applied to displacement smoothness, displacement amplitude, and value correction.
This method was evaluated on three simulated PDE benchmark datasets and a real-world cylindrical flow dataset, using a uniform protocol of 20 observations for 20 targets. The paper reports performance improvements compared to nine baseline models and includes ablation experiments such as persistence, residual prediction, reconstruction, branch removal, and sequential anchoring.
Critical strengths
- A clear observation-based prediction mechanism. This method directly reuses the latest observation field, assigning different operational roles to coordinate sampling and value correction. This is a specific inductive bias that can retain the flexibility of model value changes while reducing the burden of reconstructing existing spatial structures. The shared anchoring expression also clearly supports the parallel construction of multiple future states.
- ADE-Solver achieved the best RMSE values and relative errors reported on all four evaluation datasets (including measured cylinder flow data). The performance improvement was particularly significant in diffusion-reaction and shallow water problems.
Critical weaknesses
- Insufficiently controlled ablations. Section 4.3 defines w/o DRL as an ordinary residual predictor:
Appendix A.7 defines w/o Spatial Transfer as a gated residual predictor:
Although both omit spatial transfer, their NS RMSE increases are approximately 42% and 4.95%, respectively. The manuscript does not clearly specify their prediction-head architectures and parameter counts, or whether the ordinary residual receives regularization comparable to that applied to the state-variation field. These details are needed to explain the substantial difference between the two residual controls and determine how much of the reported gain comes from spatial transfer, residual parameterization, or regularization.
- No free-running rollout evaluation. Evaluation is limited to direct 20-to-20 prediction, leaving accuracy and stability under repeated model application unverified.
Questions for authors
- How does ADE-Solver perform in the free-running rollout with a prediction window of more than 20 steps?
- Besides complexity analysis, could you provide a measured comparison of inference time under the same conditions of hardware, precision, batch size, resolution and prediction duration?
- The paper has already stated that 7 experiments have been run. Could you report the mean ± standard deviation of the seven runs and other statistical indicators in more detail?
5 comments
Thank you to the author for supplementing the details of the ablation experiment and the results of the 80-step rollout. The author's response addressed our related concerns. But I still have the following two questions:
- The author claims in Table 1 of the paper that . If this statement refers to the ADE-Solver against the most competitive baseline model and the t-test conducted based on the seven runs given in the rebuttal, then based on the mean and standard deviation disclosed by the author, I calculated that the metrics of SW, the metrics of NS, and the RMSE of CF seem cannot meet . Here, we take the RMSE of the NS problem as an example, that is, to consider:
| Method | Mean | Standard deviation |
|---|---|---|
| ADE-Solver | 0.00085 | 0.00027 |
| SAOT | 0.001141 | 0.000660 |
Since the correlation coefficient r must be less than or equal to 1, the following equation holds:
\overline D & =\overline Y-\overline X =0.001141-0.000850=0.000291. \\ r &=\frac{{\operatorname{Cov}}(X,Y)}{s_Xs_Y}\leq1,\\ s_D^2 &= s_Y^2+s_X^2-2r\,s_Xs_Y \geq(s_Y-s_X)^2,\\ s_D&\geq|s_Y-s_X|=0.000390. \end{aligned}$$ Based on this, the t-statistic is calculated, resulting in: $$t_{\mathrm{paired}} =\frac{\overline D-0}{s_D/\sqrt n} \leq\frac{0.000291}{0.000390/\sqrt7} \approx1.974, \qquad \mathrm{df}=n-1=6.$$ And this corresponds to 0.096, which cannot reach 0.01. In conclusion, could the author please explain in detail how the t-test is conducted, such as the models being compared and the testing method? 2. The computational efficiency provided by the author is impressive. However, in the computational complexity analysis, a result of $O(N^2)$ was given for the spatial attention. For this reason, I examined the code disclosed in the author's anonymous repository and found that in the actual implementation of ADE-Solver, the U-Net style model uses the linear Transformer while the coarsest level uses a standard Transformer(lines 522 and 524 of `model/unet.py`); The standard attention calculation is shown in lines 331, 352, and 356, while linear attention is used in all other layers (lines 514 and 534 of `model/unet.py`; lines 254, 255, 258, and 260), which does not conform to what is disclosed in the article (line 161 on page 3 of the paper). May I ask if the author can clearly specify which specific attention mechanism was used? In addition, according to the code in the anonymous repository, in the experiment, The default configuration does not handle grids of different sizes differently (for the fixed-scale configuration, see line 30 of `configs/diff_react/ade.yaml` and line 26 of `configs/ns_incom/ade.yaml` ` configs/shallow_water/ade. Yaml ` line 30, ` configs/cylinder/ade.yaml ` line 24; The downsampling structure is shown in lines 167, 168, 471, 472, and 516 of `model/unet.py`). Then, when the mesh increases, for example, $128 \times 128$, with the same downsampling ratio, The coarsest layer of the grid will reach $32 \times 32$, which is 1024 tokens. At this point, can ADE-Solver still maintain the current computational efficiency? Due to the above issues, I choose to maintain my rating of 2 in this round.Thank you again for your careful review. We regret the confusion caused by the incorrect standard deviations in our previous response. We have corrected those values, recomputed the \(p\)-values, clarified the attention implementation, and benchmarked inference at two resolutions, including the setting with 1,024 bottleneck tokens. We appreciate your help in making the manuscript clearer.
- Statistical significance in Table 1
Thank you for carefully checking the reported statistics and for sharing your calculation for Incompressible NS. We traced the discrepancy to an error in the sample standard deviation formula in the script used to prepare our previous rebuttal. We corrected the formula and recomputed the standard deviations from the same seven runs per method that produced the means in Table 1.
| Method | DR RMSE | DR Rel. L2 | SW RMSE | SW Rel. L2 | NS RMSE | NS Rel. L2 | CF RMSE | CF Rel. L2 |
|---|---|---|---|---|---|---|---|---|
| DeepONet | 8.94e-3 | 6.29e-2 | 2.88e-3 | 2.75e-3 | 2.45e-3 | 3.93e-2 | 5.37e-3 | 5.65e-2 |
| FNO | 5.96e-6 | 6.38e-5 | 3.52e-5 | 3.51e-5 | 7.37e-4 | 8.25e-3 | 4.26e-4 | 5.61e-3 |
| CNO | 3.86e-5 | 3.13e-5 | 2.43e-5 | 4.33e-5 | 5.13e-5 | 6.72e-4 | 2.77e-4 | 4.78e-4 |
| Transolver | 4.49e-4 | 4.73e-3 | 3.87e-3 | 3.33e-3 | 4.28e-3 | 9.26e-3 | 2.61e-2 | 2.97e-2 |
| HPM | 3.15e-4 | 4.07e-3 | 1.77e-3 | 1.60e-3 | 4.21e-4 | 6.00e-3 | 1.52e-3 | 1.94e-3 |
| WDNO | 6.82e-3 | 9.27e-3 | 1.04e-3 | 2.16e-3 | 2.74e-4 | 4.31e-2 | 7.47e-4 | 6.14e-3 |
| 3D-Unet | 6.35e-6 | 5.71e-5 | 6.23e-5 | 5.84e-5 | 8.46e-5 | 6.69e-4 | 5.20e-4 | 5.58e-4 |
| M2NO | 8.61e-4 | 7.49e-3 | 7.68e-5 | 7.39e-5 | 6.94e-4 | 7.16e-3 | 1.85e-3 | 2.93e-3 |
| SAOT | 1.33e-5 | 1.91e-4 | 7.64e-5 | 6.83e-5 | 2.38e-5 | 3.27e-4 | 7.39e-4 | 6.67e-3 |
| ADE-Solver | 2.62e-5 | 3.44e-4 | 9.92e-6 | 9.20e-6 | 4.15e-5 | 3.99e-4 | 5.28e-4 | 1.73e-4 |
In the original analysis for Table 1, we used a two-sided Welch -test to compare ADE-Solver with the baseline with the lowest mean error for each dataset and metric. Each of the seven independent training runs per method contributed one test-set score for each metric. We treated the runs as unpaired and allowed the variances to differ.
| Dataset | RMSE baseline | RMSE -value | Rel. L2 baseline | Rel. L2 -value |
|---|---|---|---|---|
| DR | SAOT | 7.01e-9 | SAOT | 4.68e-9 |
| SW | M2NO | 1.50e-5 | M2NO | 1.33e-5 |
| NS | SAOT | 3.05e-8 | SAOT | 5.25e-8 |
| CF | M2NO | 1.81e-3 | 3D-Unet | 5.72e-10 |
The lower bound of 0.096 for the paired test, as calculated by you, is derived based on the standard deviation reported previously. With the corrected values, the two-sided Welch test gives for Incompressible NS RMSE. All eight comparisons meet Table 1’s threshold. The final manuscript will report the corrected standard deviations, test method, baselines, and -values.
- Thank you for examining the released implementation and pointing out the discrepancy. We will clarify the attention mechanisms in Eq. (4) and Appendix A.6. We also measured inference at the resolution you identified, where the bottleneck contains 1,024 spatial tokens.
(1) Attention used in ADE-Solver. In the released configuration, the U-Net encoder and decoder use linear spatial attention. Each block normalizes queries across channels within each head and keys across spatial tokens, aggregates keys and values, and then applies the queries. Its computation scales linearly with the spatial token count at fixed attention widths and temporal length. The middle U-Net block uses standard scaled dot-product spatial attention at the coarsest resolution. Its pairwise attention computation costs per sample, where is the temporal sequence length, is the number of bottleneck spatial tokens, and is the total width across attention heads. Temporal attention uses scaled dot-product attention along the time axis at each spatial location, with rotary positional embeddings and a relative position bias. In the final manuscript, we will clarify where Eq. (4) applies, provide the linear spatial attention formulation, and revise Appendix A.6 to assign the quadratic term in the spatial token count to the bottleneck block.
(2) Measured efficiency at 1,024 bottleneck tokens. The released configurations use two downsampling stages, each with stride 2 along both spatial axes. Increasing the input grid from to therefore increases the bottleneck token count from 256 to 1,024. We benchmarked both resolutions using the same trained ADE-Solver checkpoint and fixed downsampling.
| Input grid | Bottleneck grid | Bottleneck tokens | Median (ms/sample) | P90 (ms/sample) | Peak allocated GPU memory (MiB) |
|---|---|---|---|---|---|
| 64×64 | 16×16 | 256 | 57.45 | 66.70 | 488.33 |
| 128×128 | 32×32 | 1,024 | 227.16 | 264.63 | 1,661.89 |
With four times as many input grid points, the median inference latency increased by 3.95×, P90 latency by 3.97×, and peak allocated GPU memory by 3.40×. The bottleneck's pairwise spatial attention work grows by 16× at fixed sequence length and attention width. Over the two tested resolutions, the measured full-model latency grew approximately in proportion to the number of input grid points, so inference time per grid point remained nearly constant between the two tested resolutions. We will add these measurements and the benchmarking protocol to the final manuscript.
Thank you again for your careful review. Your questions helped us clarify the two residual ablations, evaluate 80-step rollouts on Diffusion-Reaction and Shallow Water, measure test runtimes under matched conditions, and report variability across seven runs on all four datasets. We will include all of these clarifications and results in the final manuscript.
Thank you for carefully reading our paper and raising these specific questions. We appreciate the opportunity to clarify the residual ablations and evaluate ADE-Solver beyond the original 20-step horizon.
1. Residual ablations (Weakness 1)
Both NS variants share the same backbone. Each prediction head is a single 1×1×1 Conv3d layer.
In w/o Spatial Transfer, the two velocity channels share a sigmoid gate, and the magnitude penalty is applied to Vk before gating. The registered parameter count includes 130 parameters in the displacement head. Its output is set to zero, so the head receives no loss gradient and both displacement penalties are zero. The w/o DRL variant uses an ungated residual and trains with prediction MSE alone. Relative to the full model, NS RMSE increases by 4.95% for w/o Spatial Transfer and by approximately 42% for w/o DRL.
2. Free-running rollout beyond 20 steps (Weakness 2 and Q1)
We evaluated 80-step free-running rollouts on Diffusion-Reaction (DR) and Shallow Water (SW) using the original 20→20 checkpoints without retraining. Each model starts with 20 observed frames and feeds each predicted 20-frame block back as input for the next block. Both tables report stepwise RMSE (×10⁻⁴) at the end of each block.
Diffusion-Reaction:
Shallow Water:
At step 80, ADE-Solver still achieves the lowest RMSE among the evaluated methods, with reductions of 50.1% on DR and 28.4% on SW relative to the strongest baseline on each dataset.
3. Measured runtime (Q2)
We timed on one hardware in FP32 with AMP disabled. Within each dataset, all methods used the same test windows, 64×64 grids, and prediction duration. Batch sizes were 4 for DR and 1 for Incompressible NS. We report end-to-end test time after checkpoint loading, including data loading, preprocessing, prediction, postprocessing, and metric computation.
4. Seven-run statistics (Q3)
We report the mean ± standard deviation across seven runs. To fit the rebuttal’s character limit, we round both values to two significant figures.
Across seven runs, ADE-Solver achieves the lowest mean RMSE and mean relative L2 error on all four datasets.
We commit to including all these experiments in the final version.