- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper studies transport-based multivariate prediction regions . Main negative result (Thm. 2/18): with a standard Gaussian reference, the Brenier (OT) map yields HDR-volume regions at every level only for Gaussian or spherical elliptical targets; for all other finite-second-moment elliptical targets the Brenier region is strictly larger on an open interval of levels. For uniform-ball and spherical-uniform references the gap holds for every non-spherical elliptical target (Thm. 17). Constructively, any transport is factored as with measure-preserving; smooth -preserving flows are characterized by skew-symmetric band matrix fields with at most entries (Prop. 5/Cor. 6), the HDR volume is shown attainable over measurable rearrangements (Prop. 8), and is trained with a log-domain surrogate; split conformal calibration on gives marginal coverage. Experiments: Student-t sweep (Table 1), synthetic unit-Jacobian targets (Table 2), and five real multi-output datasets (Fig. 2, Table 4).
Critical strengths
- A clean, rigorous and somewhat surprising theoretical point: Brenier-induced quantile regions, widely treated as canonical, are provably volume-suboptimal except in exactly characterized cases (full dichotomy in Thm. 18, reference-independent rigidity lemma). Notably, Thm. 17 covers the uniform-ball reference used by center-outward-rank methods, so the result also gives a principled justification for the post-hoc volume corrections that recent OT-conformal methods add heuristically, e.g. the rank-space cutoff optimization in FACTOR (Gao & Han, 2026) and references therein. Assumptions and limits (second moment, "open interval" of levels) are stated honestly.
- A principled, post-hoc correction: coverage is fixed by and volume controlled by ; the sparse skew-symmetric parametrization gives exact measure preservation and a closed-form Jacobian , and it applies to any invertible transport (also demonstrated on RealNVP). Because only reshapes the latent set, it is in principle composable with monotone rank post-processing used elsewhere (e.g. subgroup synchronization of ranks).
C. Gao and L. Han. FACTOR: Fairness-Aligned Conformal Transport for Optimal Regions. STAI-X 2026.
Critical weaknesses
- Theory/method gap and modest real-data gains. The suboptimality theory covers only elliptical targets, where the fix is a trivial affine whitening; nothing quantifies the gap or achievable gain for the non-elliptical targets the flow is designed for. On real data the rearrangement improves normalized log-volume by only ~0.01–0.05 per dimension (NOT: 4.53→4.50 Bio, 5.82→5.80 SCM20D), is neutral for NF on QM9 and worse on ONP (1.19→1.26). The claim that rearranged transports "dominate all baselines when " is largely inherited from the base NOT model (plain NOT already beats every baseline on SCM20D); CP-Split wins on both 2-D datasets.
- The surrogate and the parametrized class do not target the stated objective. The log-domain loss is only a lower bound on log-volume with a non-vanishing Jensen gap (Remark 10); flow images of a ball are diffeomorphic to a ball, so disconnected HDRs are unattainable (Remark 11); the oracle inequality (App. H) assumes a known and excludes approximation error. No guarantee links the trained to the volume it is claimed to minimize.
- Missing comparison with simpler latent-set families and related volume-reduction work. The paper fixes the latent set to a ball and only reshapes it via , but it does not discuss or compare against cheaper alternatives that keep the Brenier map and change the latent set instead: (a) affine/whitened balls (sufficient for every target covered by Sec. 3), and (b) radial bands/annuli in rank space, which minimizes the same -weighted volume estimator (Prop. 7's identity) over . The theory actually implies that radial-only corrections cannot reach the HDR for anisotropic targets (the obstruction is angular, Lemma 16), which would be a nice additional result to state and verify empirically.
Questions for authors
- What is the training/inference overhead of the ODE-based and relative to the base NOT/NF model?
- Sec. 4.2 allows as an input to (non-nested regions), but App. I.1 writes without . Which was used, and how often does region crossing occur?
- How much of the gain over baselines is due to the rearrangement vs. the base model? Please add (i) an affine/whitening correction of and (ii) a radial-band / cutoff-optimized baseline (keeping ) so that the value of reshaping, rather than rescaling, the latent set is isolated. Also explain why NF-R degrades on ONP.
- For Student-t in , can you provide a high-precision value of the Brenier–HDR gap to validate Table 1 beyond the control?
- Is there a route to a measure-preserving correction for the uniform-ball reference (e.g. velocity fields tangent at the boundary), so the construction applies to center-outward ranks directly?
7 comments
Follow-up to bB5R on the Student-t gap and uniform-ball reference
Student- reference check. The exact gap is zero for the spherical Student- control () and for the anisotropic Gaussian controls. Both give zero at the precision of Table 1. For , , and scatter , the HDR area at mass is . We independently quantized the Gaussian reference with 65,536 equally weighted scrambled Sobol atoms and integrated the Student- density in the resulting mass-balanced Laguerre cells. At matched mass 0.90002, the NVF was approximately −0.0008, −0.0026, −0.0052, −0.0082, and −0.0111 for . The spherical control gave −0.00013 (exact value zero). At 16,384 atoms, the two scrambles gave −0.0116 and −0.0122 for . The check corroborates the scale and trend of the gap. Finite quantization limits the precision.
Uniform-ball reference. A sufficient construction is a smooth skew-symmetric field supported strictly inside the ball. Its row-wise divergence is divergence-free by skew symmetry and vanishes near the boundary. Its flow remains inside the ball and preserves uniform volume. The spherical-uniform reference with a different radial density requires a weighted construction. Our experiments use the Gaussian reference.
Response to bB5R (1/3): two statements added in the revision
Thank you for the detailed review. This follow-up addresses requests left open in our earlier reply. Two statements prompted by your review are proved in the revision, in a new Appendix J; the numbers of all earlier statements are unchanged.
W3(b). Radial corrections cannot reach the HDR (Corollary 35). Work in the setting of Lemma 16 (, elliptical target with a continuous, strictly decreasing generator): Gaussian reference and a non-Gaussian, non-spherical target with finite second moment, or the uniform-ball or spherical-uniform reference and a non-spherical target with finite first moment. Then is not almost everywhere a radial function. Consequently, no family of radial latent sets (balls, annuli, unions of shells) has the HDRs as images, even for a dense set of levels, and there is a nonempty open interval of levels on which every radial latent set of the given mass has an image strictly larger than the HDR. The proof needs no regularity of . If the preimages of the HDR ellipsoids were radial for a dense set of levels, then the Mahalanobis radius would satisfy almost everywhere for a Borel function . Monotonicity of a convex gradient and convexity of the ellipsoids force to be nondecreasing up to a null set of radii, hence equal almost everywhere to the radial map of Lemma 16. Every Brenier region would then be an HDR, which that lemma excludes.
Among radial sets of a given reference mass, the smallest image is obtained by thresholding , with partial selection at a tied threshold if needed. The ball is the best radial set at every level if and only if coincides almost everywhere with a nondecreasing function. We do not prove that is monotone for elliptical targets. In our two-dimensional Student- check (16,384 atoms, level 0.9), the best contiguous band was the ball for every tested , so a band correction would have gained nothing there. Radial sets do suffice exactly when is radial, for example for a spherical target with a ring-shaped density. In that example, an annular HDR can be attained by a single band at levels where the corresponding radial preimage is an annulus.
W1. Beyond elliptical targets (Proposition 28). For any target with density and any almost-everywhere invertible transport from a radial reference, write and let be its radially nonincreasing rearrangement under . Then, with the minimum volume at the same mass,
The gap vanishes at every level if and only if is, almost everywhere, a nonincreasing function of ; for a diffeomorphism and a reference density , if and only if with nonincreasing (Theorem 3 of Kondratyev et al., ICLR 2026, gives sufficiency for Brenier maps under strict decrease). This is a criterion and a computable quantity, not a closed-form gap for a family of targets, and it is elementary. The right-hand side can be estimated from the fitted base density and reference draws. Under the fitted law, it bounds what any rearrangement can gain. For the three non-elliptical synthetic targets of Table 2, where the HDR volume is known, the HDR-to-region volume ratio goes from 0.43 to 0.96 (banana), 0.89 to 0.98 (sinusoidal) and 0.94 to 0.98 (star-shaped) in dimension 2.
Response to bB5R on the effect and cost of rearrangement
Thank you for the detailed review. We agree that the paper should distinguish the theoretical result about exact Brenier maps from the empirical gains attributable to rearranging a fitted map.
What changes when the base map is fixed? Comparing each backbone with its own rearrangement, the reported median normalized log-volume decreases for NOT on all five datasets and for RealNVP on three. RealNVP ties on QM9 at the reported precision (−0.05 for both) and worsens on ONP (1.19 to 1.26, with overlapping interquartile ranges). CP-Split has the smallest median volume on both 2D datasets, and plain NOT already beats the other families on SCM20D. There, the additional reduction from 5.82 to 5.80 corresponds to roughly 27% on the volume scale, calculated as 1−exp(16(5.80−5.82)) from the rounded medians. For the non-elliptical 2D synthetic examples, where the HDR volumes are known, the median HDR-to-prediction volume ratios go from 0.43 to 0.96 (banana), 0.89 to 0.98 (sinusoidal), and 0.94 to 0.98 (star-shaped).
Surrogate, topology, and ONP. At a fixed level and under the assumptions of Lemma 9, the volume objective and its log surrogate identify the same optimal region over unrestricted measurable rearrangements. The Jensen gap can persist, so a decrease in the surrogate away from the optimum can leave volume unchanged or increase it. A smooth flow image of a ball cannot exactly represent a disconnected HDR, and Appendix H treats a known base map. An exact measure-preserving rearrangement leaves the fitted pushforward law unchanged. Errors in the base transport persist. After fitting and calibration, true conditional coverage can vary and the volume can increase, as it does on ONP. The current results do not distinguish the effects of base-map error, the surrogate, and the calibrated radius.
Simpler comparisons. With the true center and scatter of an elliptical target, unwhitening the radial map gives the HDR ellipsoids at every level. The composed map need not be Brenier for the original quadratic cost. The Elliptic comparator uses random forest residuals. A same-backbone whitening comparison would isolate the effect of whitening. In the discretized 2D Student-t check, we compared all contiguous radial bands with the same reference mass (16,384 atoms). For every tested k, the minimum-area band had zero inner radius, giving the original ball. Both same-backbone comparisons on real data remain open.
Overhead and nesting. With the base map frozen, the added training cost is fitting the flow, including forward ODE solves in the objective. The added scoring cost is applying Sθ−1 to the latent point obtained from T−1. The closed-form determinant simplifies volume estimation. Comparable timings for the reported setup still need to be verified. The fitted Sθ(u,x) takes no coverage level as input. Levels are sampled during training to choose the latent balls. At prediction time all levels threshold the same score, so the regions are nested. We corrected the conflicting description in Section 4.2 and Appendix H. Appendix I already gave the level-independent model.