- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
They propose a time-series generator that does flow matching on the coefficients of a fixed cubic B-spline basis instead of on the sampled values (GalerkinFlow). The coefficients are fitted using ridge regression objective. They use a DiT-style Transformer with one token per basis coefficient and learn the flow with standard rectified flow; for doing the decoding they use a fixed linear map to generate new coefficient matrices. For incomplete data, missing entries are linearly interpolated before fitting, and the loss is only applied on observations after decoding. They also show that because the decoder is linear the Wasserstein error divides into a spline error and a coefficient learning error. Their results show the method has the best Context-FID on all six regular benchmarksnd they argue the their model processes significantly less tokens than a comparable model that is based on sample value generation.
Critical strengths
1- The idea of using a fixed linear encoder or decoder around a standard flow-matching transformer looks simple and practical . It is well motivated in the paper with a clean error decomposition, so it should be easy to reproduce and train. Also, the decomposition together with the synthetic study gives a useful rule to check before training and thinking about reducing token reduction for horizon time series is an exciting open question to be investigated.
2- They have competitive results on all benchmarks, with arguably context FID gains compared to SMF and Diffusion-TS. If these results hold with a proper seed variance and statistical significance, it is an interesting finding although in current state is overstated by the paper claims in the abstract and also see Weakness .
Critical weaknesses
1- They made a strong SOTA claim, but the results doesnt statistically significant to support them. Most margins are within the reported uncertainty of the baselines. In Table 1, their method is clearly outside the error bound of all baselines only for FID on ETTh. On MuJoCo and Energy it is better than SMF and Diffusion-TS but overlaps with FFM-OT (here also +- values are much larger than its means, which is also strange). Everywhere else it overlaps with at least one baseline. Also, GalerkinFlow’s uncertainty doesn’t include generator variance, which makes the bounds less reliable. I can see similar patterns in Tables 2 and 3. This needs a discussion or narrowing down the claims accordingly.
2- in the motivation section the authors claim token reduction when K << N and a per-step cost that doesn’t depend on the grid. However on six of the benchmarks here K = 32 > N = 24, so there is no reduction. Therefore, how the 2.5–4.6× speedup over SMF? is it probably coming from the model size or architecture differences and not the spline usage.
3- The abstract and the theory talk about the method in function space, but the generated samples are never evaluated as functions. All metrics are on the training grid, and there is no off-grid or resolution-transfer experiment.
Questions for authors
1- +- values for GalerkinFlow are 95% CI? if so it shows the evaluator variance and not generator variance. Can you report the results in Tables 1- 3 with mean +- std over independently trained generators (with different training seeds) while you keep a single uncertainty rule across all of them?
2- In Table 1, GalerkinFlow’s advantage falls outside the reported uncertainty only for FID on ETTh, MuJoCo, and Energy. In other places its intervals overlap those of at least one baseline. Also similarly in Table 2 and 3. Given this, on which dataset–metric you consider the improvement statistically significant to claim SOTA?
3- the paper claimed token reduction when K<<N, but on all six regular benchmarks K = 32 and N = 24 which shows the experiments aren't selected carefully selected. As I see here the model processes more tokens than a raw-value model then what exactly explains the 2.5–4.6× training speedup over SMF in Table 10? Also can you report parameter counts and flops too?
4- The abstract and Section 4 you frame the method in the function space but all metrics are computed on the training grid and no offgrid or resolution transfer evaluation is being done here, can you provide such an experiment?
9 comments
Statistical significance
2. Statistical significance and the scope of the performance claim GalerkinFlow’s strongest result in the submitted Table 1 is its consistently low mean Context-FID across datasets. It has the lowest reported mean on all six regular benchmarks. The differences from SMF are particularly large on ETTh1, MuJoCo and Energy, as the following comparison shows.
| Dataset | GF Context-FID ↓ | SMF Context-FID ↓ | Diffusion-TS Context-FID ↓ |
|---|---|---|---|
| Stocks | 0.005 | 0.008 | 0.169 |
| ETTh1 | 0.014 | 0.058 | 0.126 |
| Energy | 0.029 | 0.051 | 0.113 |
| fMRI | 0.108 | 0.116 | 0.118 |
| MuJoCo | 0.008 | 0.018 | 0.015 |
| Sines | 0.003 | 0.004 | 0.013 |
These are reported means from the submitted Table 1, in the original metric units. They summarize the observed differences and are not significance-test results. The original intervals measure evaluator variability and do not establish statistical superiority across independently trained generators. We will therefore replace the broad SOTA wording with the specific dataset and metric findings supported by the evidence.
The close predictive scores also deserve a task-specific interpretation. This metric measures how well a forecaster trained on generated sequences predicts real observations. On Stocks, GF, SMF and Diffusion-TS all have a reported predictive score of 0.037, while their mean Context-FIDs are 0.005, 0.008 and 0.169. The predictive evaluator gives similar scores at the displayed precision, while Context-FID separates their reported means. Similar predictive scores can therefore coexist with differences in measured distributional fit. We will describe predictive near ties as similar reported scores and discuss the differences measured by the other metrics.
The submitted Table 2 also reports better mean results than SMF on all three metrics for both long-horizon datasets. Table 3 contains both favorable means and close comparisons. For example, at 70% missingness in irregular-to-regular generation, GF and SMF score 0.016 and 0.015. We will describe this as a close comparison and retain the setting-specific conclusions.
Long-sequence computation provides a separate practical advantage. GalerkinFlow evaluates its velocity network on coefficient tokens, independently of the number of output positions. A controlled comparison with the same Transformer architecture measures the benefit when and the output grid is much longer.
| Output grid points | Coefficient-network evaluation ms | Raw-grid network evaluation ms | Speedup per network evaluation | One-time spline decoding ms |
|---|---|---|---|---|
| 728 | 2.149 | 25.879 | 12.045× | 0.136 |
| 791 | 2.178 | 29.632 | 13.606× | 0.137 |
Both networks use width 96, four Transformer blocks, four heads, six channels and batches of 64 on one single GPU. Times are medians of 30 repetitions after ten warmup iterations, measured on synthetic inputs.
At 791 positions, the coefficient network takes 2.178 ms per evaluation compared with 29.632 ms for the raw-grid network. This reduces the cost of the operation repeated throughout sampling, while the final decoding takes 0.137 ms per batch. The experiment demonstrates the computational benefit of the fixed coefficient state as the output grid grows.
We will revise the performance claim to emphasize low reported Context-FID, competitive predictive scores and reduced network computation on long grids. These findings support the practical value of a fixed coefficient-space generator and complement the paper’s continuous representation and error analysis.
Explain the Speed Ups
3. Short-window speedup, parameter counts and FLOPs GalerkinFlow’s computational advantage comes from processing K coefficient tokens at each sampling step, independently of the number of output positions. This benefits long sequences when is much smaller than . You are correct that it cannot explain the short-window result with . We will separate the timing of the original implementations from the benefit demonstrated by the controlled experiment below.
To isolate that benefit, we compare two versions of the same Transformer. One processes 32 spline coefficients per channel, and the other processes the values at all grid points. Both use width 96, four Transformer blocks, four attention heads and feed-forward width 256. We measure them on one GPU with six channels and batches of 64.
Cost with the same Transformer architecture
Timings are in milliseconds per batch and report the median of 30 repetitions after ten warmup iterations. A training step includes the forward pass, backward pass and optimizer update. MFLOPs are approximate forward matrix-product operations per sample, counting two operations per multiply-add. The final column measures the separate spline decoding operation.
| Representation | Grid size | Network tokens | Parameters | MFLOPs per sample | Forward ms | Training-step ms | Decode ms |
|---|---|---|---|---|---|---|---|
| Spline coefficients | 24 | 32 | 518,950 | 23.998 | 2.168 | 14.089 | 0.174 |
| Grid values | 24 | 24 | 518,182 | 17.787 | 2.157 | 13.976 | Not needed |
| Spline coefficients | 96 | 32 | 518,950 | 23.998 | 2.159 | 14.102 | 0.135 |
| Grid values | 96 | 96 | 525,094 | 80.769 | 2.426 | 14.066 | Not needed |
| Spline coefficients | 192 | 32 | 518,950 | 23.998 | 2.157 | 14.071 | 0.137 |
| Grid values | 192 | 192 | 534,310 | 189.518 | 4.130 | 14.118 | Not needed |
| Spline coefficients | 384 | 32 | 518,950 | 23.998 | 2.147 | 14.038 | 0.137 |
| Grid values | 384 | 384 | 552,742 | 491.950 | 9.526 | 26.840 | Not needed |
| Spline coefficients | 728 | 32 | 518,950 | 23.998 | 2.149 | 14.069 | 0.136 |
| Grid values | 728 | 728 | 585,766 | 1317.022 | 25.879 | 71.536 | Not needed |
| Spline coefficients | 791 | 32 | 518,950 | 23.998 | 2.178 | 14.169 | 0.137 |
| Grid values | 791 | 791 | 591,814 | 1507.510 | 29.632 | 81.622 | Not needed |
The Transformer blocks have the same architecture in both versions. Parameter counts differ because the learned positional embeddings contain one vector for each network token. The FLOP estimates include the linear layers and attention matrix products. Normalization, softmax, nonlinear functions and spline decoding are outside this estimate.
The benefit becomes clear as grows. At , the coefficient model requires 2.178 ms for a forward pass, compared with 29.632 ms for the model operating on grid values. Training steps take 14.169 and 81.622 ms, respectively. These correspond to approximately 13.6 times faster forward passes and 5.8 times faster training steps in this controlled benchmark. The coefficient model's network cost stays nearly constant because it continues to process 32 tokens.
Generator parameter counts in the regular comparison
In the following, we record the number of parameters of each model. FA denotes Functional Attention used within GF's velocity network.
| Dataset | GF | OFM adapted | FMF adapted | GF + FA direct | GF + FA matched | SMF storage-adapted |
|---|---|---|---|---|---|---|
| Stocks | 518,950 | 419,526 | 1,657,478 | 433,150 | 518,842 | 356,800 |
| ETTh1 | 519,143 | 1,657,607 | 3,714,887 | 433,343 | 519,035 | 356,864 |
| Energy | 523,196 | 3,770,780 | 1,663,132 | 437,396 | 523,088 | 1,086,432 |
| fMRI | 527,442 | 1,668,658 | 1,668,786 | 441,642 | 527,334 | 1,237,152 |
| MuJoCo | 520,494 | 3,766,286 | 3,717,134 | 434,694 | 520,386 | 1,233,696 |
| Sines | 518,757 | 3,714,053 | 3,714,245 | 432,957 | 518,649 | 356,736 |
We will revise the efficiency claim around these results. The fixed coefficient representation reduces repeated network computation when the output grid is much larger than the chosen basis.
Error Bars
We want to express our sincere thankness for reviewer's valuable opinions and problems for our work. We will answer each problem one by on in the following.
1. Error bars and variation across trained generators
Our original Context-FID uncertainty convention followed SMF’s released regular-benchmark workflow. Its evaluation code repeats metric evaluation on a fixed generated dataset, and its reporting function calculates 95% Student-t confidence-interval half-widths. Our Appendix C.3 states the same convention. These intervals describe evaluator variability.
We agree that including variation from generator training provides a more complete assessment, so we provide the required table in the following: