Pith. sign in

REVIEW 5 minor 294 references

Heat-smoothed weighted measures on compact manifolds have an exact pairwise entropy profile and a scale-aware geometric effective sample size.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-02 08:13 UTC pith:MROX4JOZ

load-bearing objection A sound, honest repackaging of Leinster–Cobbold diversity with heat-kernel overlaps into a scale-dependent geometric ESS; the math holds, the experiments don't go beyond demonstrative.

arxiv 2607.06696 v3 pith:MROX4JOZ submitted 2026-07-07 stat.ML cs.LGstat.ME

Heat-Kernel Entropy Profiles and Geometric Effective Sample Size for Weighted Measures on Manifolds

Kisung You , Boram Cho This is my paper
classification stat.ML cs.LGstat.ME MSC 62R3058J35
keywords heat-kernel entropy profilegeometric effective sample sizeweighted empirical measuresorder-two entropyimportance samplingdirectional statisticsspherical harmonicsmerge invariance
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper tries to establish that a weighted collection of points on a smooth finite space without boundary carries a geometric degeneracy that ordinary effective sample size cannot see. After smoothing the weighted measure by heat diffusion for time t, the paper defines the profile as the log of the integral of the squared smoothed density; it proves this integral is exactly a weighted sum of pairwise heat-kernel overlaps, so the whole curve is computable from pairwise kernel values rather than manifold integrals. Normalizing each overlap by the self-overlap of the two points gives a geometric effective sample size gESS(t), a scale-dependent count of distinguishable atoms: it is between 1 and the usual ESS, unchanged if duplicate atoms are merged, and it moves from the ESS of the distinct support points to 1 as heat destroys structure. On a sphere, the unlogged profile separates into spherical-harmonic energies, with mean-direction and axial-anisotropy terms first, so the curve detects antipodal, girdle, and multimodal patterns that weights and first moments miss.

Core claim

The central claim is that diffuseness of a weighted empirical measure on a compact, connected, boundaryless Riemannian manifold is quantified exactly, at every scale, by the order-two entropy of its heat-smoothed density. The paper proves the pairwise identity ∫(Σ_i w_i k_t(y,x_i))² dν(y) = Σ_{i,j} w_i w_j k_{2t}(x_i,x_j), which turns a manifold integral into kernel-matrix arithmetic. From this it defines s_t(x,y) = k_{2t}(x,y)/√(k_{2t}(x,x)k_{2t}(y,y)) and gESS_w(t) = 1/Σ w_i w_j s_t(x_i,x_j), and proves gESS is merge-invariant, satisfies 1 ≤ gESS ≤ 1/Σ w_i², has small-time limit equal to the ESS of the distinct support masses, and large-time limit 1. The same identity gives a spectral expa

What carries the argument

The load-bearing object is the pairwise heat-kernel overlap identity and its normalized form. For a weighted measure, the heat-smoothed density is Σ w_i k_t(y,x_i), and the semigroup property of the heat kernel makes the order-two overlap equal to Σ_{i,j} w_i w_j k_{2t}(x_i,x_j); normalizing each pair by the geometric mean of the two self-overlaps gives s_t in (0,1]. That normalized overlap matrix defines gESS as the inverse of the quadratic form w^T S_t w, so the proof reduces to elementary bounds on a positive similarity matrix.

Load-bearing premise

The calibration of gESS — small-time limit equal to the ESS of distinct support masses and large-time limit 1 — rests on the assumption that the space is a compact, connected manifold without boundary with normalized total volume, so heat flow cannot leak out and must converge to the uniform state; on a bounded domain, a noncompact space, or with an unnormalized reference measure, these endpoints would fail.

What would settle it

Take a manifold with boundary (for example, a flat disk) and two equal atoms; compute gESS(t) up to large t. The paper predicts the profile converges to 1 because the heat-smoothed density becomes uniform; with boundary heat loss the density need not become uniform, so a large-time limit other than 1 would falsify the claimed calibration. Alternatively, on a compact homogeneous manifold, verify the near-diagonal merging law s_t(x,y) = exp(-d(x,y)^2/(8t))(1+O(t+d^2)) at small t; any leading exponent other than -d^2/(8t) would falsify the two-atom merging expansion.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • At any fixed scale, gESS lies between 1 and ordinary ESS; a value near ordinary ESS means geometry adds no degeneracy, while a lower value means nearby or duplicate particles are reducing the effective count.
  • Splitting an atom into equal labels no longer changes the diagnostic: gESS is invariant under merging duplicates, whereas ordinary ESS changes.
  • The small-time limit is the ordinary ESS of the distinct support locations after merging duplicates, so exact duplicates are handled automatically; the large-time limit 1 says that after enough diffusion every measure looks uniform.
  • On spheres, the unlogged profile expands as a sum of exponentially damped spherical-harmonic energies, with squared mean resultant length and traceless second-moment anisotropy as the first two terms; this connects the profile to established directional summaries in one scale-dependent curve.
  • The profile is statistically stable in a positive scale window: it converges uniformly for deterministic weights, for self-normalized importance weights with bounded density ratio, and for measures converging in geodesic transport distance, so it can be estimated from samples rather than only computed on known atoms.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A consequence the authors leave implicit is that the gESS curve is a scale-space signature of the weighted measure: the times at which the curve drops mark the scales at which spatial structure becomes distinguishable, so an analyst could use the curve itself to choose a resolution, not just report one number.
  • The merge-invariance and location awareness suggest a practical use for particle filters and Markov-chain output: a gESS that stays near ordinary ESS across scales indicates weight degeneracy without spatial redundancy, while a drop with decreasing scale flags clusters of near-duplicate particles; the experiments illustrate this but the paper does not state it as a formal diagnostic rule.
  • Because gESS monotonicity in t is only guaranteed on homogeneous manifolds (where the self-overlap is constant), a natural extension is a location-aware normalization for general manifolds; any such variant would need a new proof of monotonicity, since the paper shows the current argument does not carry over.
  • The bounded-ratio importance-sampling result suggests the profile may be extendable to heavy-tailed weights via truncation or a renormalized overlap, but the paper proves rates only for bounded density ratios and states heavy-tailed weights as future work.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 5 minor

Summary. This paper introduces heat-kernel entropy profiles for weighted empirical measures on compact, connected, boundaryless Riemannian manifolds with normalized volume. For order-two Rényi entropy, it shows that the overlap integral of the heat-smoothed density equals the quadratic form wᵀK_{2t}w (Theorem 1). The inverse of this quantity is an effective occupied volume U₂(t); normalizing each heat kernel by its self-overlap yields a geometric effective sample size gESS(t). The paper proves profile monotonicity, bounds and scale limits for gESS, merge invariance, a two-atom merging law, a spectral expansion, a spherical-harmonic specialization with vMF/Bingham moment energies, and stability and consistency results including a self-normalized importance-sampling bound. Experiments on S² illustrate the profiles on unimodal, antipodal, girdle, multimodal, and duplicate configurations, and show that gESS reveals structure missed by ordinary ESS and first-moment summaries.

Significance. The central mathematical claims are clean and, as far as I can verify, correct. The pairwise identity makes the profile exactly computable from a Gram matrix, and the small- and large-time calibrations give the curve a clear operational meaning. The paper is careful about its domain assumptions (compact, boundaryless, connected, normalized volume) and explicitly states where gESS monotonicity is not proved. The availability of replication code and the self-contained appendix proofs are further strengths. If the results hold, the method is a useful geometry-aware complement to ordinary ESS and first-moment directional summaries. The principal weakness is empirical: the experiments are demonstrative, confined to S², and the main tables report single values without uncertainty bands, although the appendix provides replicated IQR curves and a truncation study.

minor comments (5)
  1. [Section 8, Tables 1 and 2] These tables report single values without any measure of variability. The replicated synthetic profiles in Figure 4 (Appendix B) show stable medians and IQRs, but the main-text tables do not refer to them. Please either state explicitly that the tabulated numbers are from a single representative construction or include uncertainty intervals (e.g., IQR over replications) so that the practical claims are not overstated.
  2. [Section 8, Figure 3] The 'Small-Time Guide' curve in panel (a) is not defined in the caption or text. Specify the formula (presumably the leading term V(8πt)^{-m/2}Sα from Theorem 4) so the comparison is reproducible.
  3. [References] The bibliography contains a formatting artifact: 'chung 1997 SpectralGraphTheory.' in the Chung reference. Also, García-Portugués and Verdebout (2018) lacks a publisher or arXiv identifier. Please clean up the reference list.
  4. [Author byline] The author line appears as 'Kisung Y ou Boram Cho' with an unintended line break. Please fix the byline formatting.
  5. [Section 4 / Algorithm 1] Since gESS monotonicity is not proved on nonhomogeneous manifolds (as noted in Section 9), a brief remark in Section 4 would help avoid over-generalizing the profile curves to nonhomogeneous settings.

Circularity Check

0 steps flagged

No significant circularity: the derivation chain is self-contained and no fitted input is relabeled as a prediction.

full rationale

The paper's central claims are derived from standard, external mathematical facts about heat kernels on compact boundaryless manifolds rather than from its own conclusions. Theorem 1's pairwise identity follows directly from the heat semigroup identity ∫ k_t(y,x_i)k_t(y,x_j)dν(y)=k_{2t}(x_i,x_j); monotonicity follows from the Dirichlet energy identity d/dt∫bp_t^2 dν = -2∫‖∇bp_t‖²dν. Definition 2 defines gESS as the inverse of a weighted sum of normalized heat-kernel overlaps. Theorem 3's bounds, merge invariance, and small/large-time limits are direct consequences of s_t(x,y)≤1, s_t(x,x)=1, the spectral gap, and standard heat-kernel asymptotics; no parameter is fitted to data and then called a prediction. Theorem 4 rests on the Minakshisundaram-Pleijel diagonal expansion and Gaussian off-diagonal bounds from Rosenberg/Grigoryan; Theorem 6 uses the spectral representation of the heat kernel; Corollary 7 uses standard spherical-harmonic identities; Theorem 8 uses Kantorovich-Rubinstein duality and McDiarmid's inequality. There are no self-citations, no imported uniqueness theorem, and no ansatz smuggled in via citation. Some properties are definitional (gESS is explicitly an order-two Leinster-Cobbold diversity), but the paper acknowledges this and does not present definitional consequences as empirical discoveries. The experiments are explicitly demonstrative and their limitations are stated in Section 9 and Appendix B; they are not used as evidence for the theorems. Therefore no circular step exists.

Axiom & Free-Parameter Ledger

2 free parameters · 7 axioms · 0 invented entities

The method is an order-two Leinster-Cobbold diversity with heat-kernel similarity; the only user-chosen parameter is the diffusion scale (and numerical truncation in experiments). The paper introduces no fitted constants and no new physical entities.

free parameters (2)
  • Diffusion time t (or angular scale α=√(8t))
    User-selected scale parameter; the whole output is a profile over t. No data-driven or optimal choice is proposed.
  • Spherical harmonic truncation Lmax = 120-180 in main figures; 360 reference
    Numerical approximation in the S² heat-kernel series, not a scientific parameter; truncation error is shown to be small for reported scales.
axioms (7)
  • domain assumption M is compact, connected, boundaryless Riemannian manifold with normalized volume ν.
    Stated in §3; required for heat kernel to converge to uniform density and for limits in Theorems 1 and 3.
  • standard math Heat kernel semigroup identity and spectral representation k_t(x,y)=Σ e^{-λt} φ_r(x)φ_r(y).
    Used in Theorems 1, 6, Cor 7; standard results from Rosenberg/Grigoryan.
  • standard math Minakshisundaram-Pleijel diagonal expansion and Gaussian/exponential heat-kernel bounds.
    Used in Theorem 4 and Prop 5 to derive small-time laws and off-diagonal decay.
  • standard math Kantorovich-Rubinstein duality and McDiarmid bounded-difference inequality.
    Used in Theorem 8 proof for W1 stability and empirical convergence rates.
  • domain assumption Self-normalized IS: density ratio ρ=dP/dQ bounded with 0≤ρ≤M, E_Q ρ=1.
    Used in Theorem 8 to get O_p(√(log n/n)) rate; compactness gives boundedness for the S² example.
  • standard math On S^{p-1}, spherical harmonic identities: √p x_a orthonormal and fourth-moment identity ∫(x^T A x)(x^T B x)dν=2tr(AB)/(p(p+2)).
    Used in Corollary 7 to identify B1 and B2 as vMF- and Bingham-type energies.
  • domain assumption On homogeneous manifolds, k_{2t}(x,x) is constant, so gESS(t)=k_{2t}(x,x)bU2(t).
    Used in Definition 2 and Algorithm 1 for spheres and Lie groups; not required for definition.

pith-pipeline@v1.3.0-alltime-deepseek · 13374 in / 19034 out tokens · 178200 ms · 2026-08-02T08:13:23.347877+00:00 · methodology

0 comments
read the original abstract

Weighted empirical measures on compact manifolds appear in importance sampling, particle approximations, posterior summaries, quadrature, and representation learning. Ordinary effective sample size and related weight summaries ignore the geometry of the support. We introduce heat-kernel entropy profiles to measure nonuniformity after intrinsic diffusion at a range of scales. For order-two R\'enyi entropy, pairwise heat-kernel overlaps give an exact profile and a geometric effective sample size. This effective sample size discounts nearby or duplicate particles. It approaches ordinary effective sample size as overlaps between distinct particles vanish. On compact boundaryless manifolds, we establish profile monotonicity, gESS scale limits, deterministic-weight consistency, and a bounded-ratio result for self-normalized importance sampling. On spheres, the unlogged profile decomposes into spherical-harmonic energies. The first terms are squared mean-resultant and traceless-second-moment energies, which give vMF- and Bingham-type scalar summaries. Experiments identify antipodal, girdle, multimodal, and duplicate-particle structures that weight-only and first-moment summaries miss.

Figures

Figures reproduced from arXiv: 2607.06696 by Boram Cho, Kisung You.

Figure 1
Figure 1. Figure 1: Preview of the proposed profiles on represen [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The reported synthetic values correspond to the con￾structions described in Appendix B. All quantities are computed by the routine in Section 4. The only sphere-specific step is evaluating the heat Gram ma￾trix. The heat kernel is kt(x, y) = X∞ ℓ=0 (2ℓ + 1)e −ℓ(ℓ+1)tPℓ(x ⊤y), (14) where Pℓ is the Legendre polynomial. For a fixed weighted cloud, we compute the harmonic energies Bℓ = (2ℓ + 1)X i,j wiwjPℓ(x ⊤… view at source ↗
Figure 2
Figure 2. Figure 2: Mollweide projections of the five synthetic [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Two-regime validation for a unimodal vMF cloud on [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Replicated synthetic profiles on S 2 . Each curve is a median over 30 replications, with an interquartile band. The left panel shows gESS. The right panel shows effective occupied volume. Profile shapes are stable across replications. four-mode cloud. Fine angular scales require more terms, consistent with Lmax of order t −1/2 . The main￾paper cutoff is adequate for the reported scales. The finest scales a… view at source ↗
Figure 4
Figure 4. Figure 4: Replicated synthetic profiles on S 2 . Each curve is the median over 30 replications, and the shaded region is the interquartile range. Left: geometry-aware effective sample size, which counts distinguishable atoms or clusters. Right: effective occupied volume fraction, which increases as heat flow removes nonuniformity. The plot checks that the qualitative profile shapes are stable across replications. sa… view at source ↗
Figure 5
Figure 5. Figure 5: Sensitivity of the S 2 heat profile to harmonic truncation. The plot compares Abw(t) with a longer￾series reference. Smaller angular scales need more harmonics because heat flow has not yet damped the high frequencies. 101 102 Angular Scale α (Degrees) 0 10 20 30 40 Effective Number Self-Normalized IS for an Antipodal Target gESS profile median ordinary ESS [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 5
Figure 5. Figure 5: Sensitivity of the S 2 heat-profile computation to harmonic truncation. The plot shows relative error in Abw(t) against a longer-series reference as a function of angular resolution. Smaller angular scales require more spherical-harmonic terms because high frequencies have not yet been damped by heat flow. Errors below 10−16 are clipped for log-scale visualization. 101 102 Angular Scale α (Degrees) 0 10 20… view at source ↗
Figure 6
Figure 6. Figure 6: Self-normalized importance sampling for an antipodal two-mode target on [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 6
Figure 6. Figure 6: Self-normalized importance-sampling stress test. Particles are drawn from the uniform proposal on [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Full profiles for the normalized embeddings in Table 2. The left panel shows gESS, and the right shows [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 7
Figure 7. Figure 7: Full profiles for the normalized-embedding example from Table 2. Left: geometry-aware effective [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

294 extracted references · 11 linked inside Pith

  1. [1]

    The Annals of Statistics , author =

    Invariant. The Annals of Statistics , author =. doi:10.1214/aos/1176343283 , number =

    arXiv

  2. [2]

    and Gretton, Arthur and Fukumizu, Kenji and Schölkopf, Bernhard and Lanckriet, Gert R.G

    Sriperumbudur, Bharath K. and Gretton, Arthur and Fukumizu, Kenji and Schölkopf, Bernhard and Lanckriet, Gert R.G. , month = aug, year =. Hilbert. Journal of Machine Learning Research , publisher =

  3. [3]

    R. On. Proceedings of the. 1961 , pages =

    1961

  4. [4]

    Journal of the American Statistical Association , author =

    Sequential. Journal of the American Statistical Association , author =. 1998 , pages =. doi:10.1080/01621459.1998.10473765 , language =

    arXiv 1998

  5. [6]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Sequential. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2006 , pages =. doi:10.1111/j.1467-9868.2006.00553.x , language =

    arXiv 2006

  6. [7]

    Ecology , author =

    Diversity and. Ecology , author =. 1973 , pages =. doi:10.2307/1934352 , abstract =

    doi:10.2307/1934352 1973

  7. [8]

    Measuring

    Gorham, Jackson and Mackey, Lester , year =. Measuring. Proceedings of the 34th

  8. [9]

    Measuring

    Gorham, Jackson and Mackey, Lester , year =. Measuring. Proceedings of the 29th

  9. [10]

    and Giné, Evarist , editor =

    De La Peña, Víctor H. and Giné, Evarist , editor =. Decoupling , isbn =. doi:10.1007/978-1-4612-0537-1 , urldate =

    doi:10.1007/978-1-4612-0537-1

  10. [11]

    Journal of Multivariate Analysis , author =

    Modifications of the. Journal of Multivariate Analysis , author =. 2001 , pages =. doi:10.1006/jmva.2000.1922 , language =

    arXiv 2001

  11. [12]

    Causality:

    Pearl, Judea , year =. Causality:

  12. [13]

    and Ivanov, Sergej and Ivanov, Sergei V

    Burago, Dmitrij Jurʹevič and Burago, Jurij D. and Ivanov, Sergej and Ivanov, Sergei V. , year =. A

  13. [14]

    Methods of

    Amari, Shun'ichi and Nagaoka, Hiroshi and Amari, Shun'ichi and Amari, Shun'ichi , translator =. Methods of

  14. [15]

    Lange, Richard D and Kwok, Devin and Matelsky, Jordan Kyle and Wang, Xinyue and Rolnick, David and Kording, Konrad , editor =. Deep. Proceedings of 2nd. 2023 , pages =

    2023

  15. [16]

    Nature Communications , author =

    Single-cell analysis reveals inflammatory interactions driving macular degeneration , volume =. Nature Communications , author =. 2023 , pages =. doi:10.1038/s41467-023-37025-7 , language =

    doi:10.1038/s41467-023-37025-7 2023

  16. [17]

    The Annals of Statistics , author =

    Data-driven. The Annals of Statistics , author =. doi:10.1214/009053607000000541 , number =

    doi:10.1214/009053607000000541

  17. [18]

    , month = jan, year =

    Brown, Lawrence D. , month = jan, year =. Fundamentals of. doi:10.1214/lnms/1215466757 , urldate =

    arXiv

  18. [19]

    Transactions of the American Mathematical Society , author =

    Strong laws for \. Transactions of the American Mathematical Society , author =. 1996 , pages =. doi:10.1090/S0002-9947-96-01681-9 , language =

    doi:10.1090/s0002-9947-96-01681-9 1996

  19. [20]

    Journal of the American Statistical Association , author =

    Asymptotic. Journal of the American Statistical Association , author =. 1987 , pages =. doi:10.1080/01621459.1987.10478472 , language =

    arXiv 1987

  20. [21]

    Journal of Machine Learning Research , author =

    Scikit-learn:. Journal of Machine Learning Research , author =. 2011 , pages =

    2011

  21. [22]

    Biometrika , author =

    Saddlepoint approximations for the. Biometrika , author =. 2005 , pages =. doi:10.1093/biomet/92.2.465 , language =

    doi:10.1093/biomet/92.2.465 2005

  22. [23]

    Computational Statistics , author =

    On maximum likelihood estimation of the concentration parameter of von. Computational Statistics , author =. 2014 , pages =. doi:10.1007/s00180-013-0471-0 , language =

    doi:10.1007/s00180-013-0471-0 2014

  23. [24]

    IEEE Transactions on Information Theory , author =

    Information projections revisited , volume =. IEEE Transactions on Information Theory , author =. 2003 , pages =. doi:10.1109/TIT.2003.810633 , language =

    arXiv 2003

  24. [26]

    IEEE Transactions on Information Theory , author =

    Information geometry on hierarchy of probability distributions , volume =. IEEE Transactions on Information Theory , author =. 2001 , pages =. doi:10.1109/18.930911 , number =

    doi:10.1109/18.930911 2001

  25. [27]

    Nature Reviews Neuroscience , author =

    Neural tuning and representational geometry , volume =. Nature Reviews Neuroscience , author =. 2021 , pages =. doi:10.1038/s41583-021-00502-3 , language =

    doi:10.1038/s41583-021-00502-3 2021

  26. [28]

    New insights and perspectives on the natural gradient method , url =

    Martens, James , month = sep, year =. New insights and perspectives on the natural gradient method , url =

  27. [29]

    International Statistical Review , author =

    Directions. International Statistical Review , author =. 2022 , pages =. doi:10.1111/insr.12481 , abstract =

    doi:10.1111/insr.12481 2022

  28. [30]

    doi:10.1007/978-981-19-1044-9 , language =

    Directional. doi:10.1007/978-981-19-1044-9 , language =

    doi:10.1007/978-981-19-1044-9

  29. [31]

    On the spherical

    You, Kisung and Shung, Dennis , year =. On the spherical. doi:10.48550/ARXIV.2208.11929 , abstract =

    Pith/arXiv arXiv doi:10.48550/arxiv.2208.11929

  30. [32]

    Statistics and Computing , author =

    Learning over von. Statistics and Computing , author =. 2026 , pages =. doi:10.1007/s11222-026-10844-6 , language =

    doi:10.1007/s11222-026-10844-6 2026

  31. [33]

    Diethe, Tom , month = feb, year =. A. doi:10.48550/arXiv.1502.07104 , abstract =

    Pith/arXiv arXiv doi:10.48550/arxiv.1502.07104

  32. [34]

    doi:10.18637/jss.v058.i10 , language =

    Journal of Statistical Software , author =. doi:10.18637/jss.v058.i10 , language =

    doi:10.18637/jss.v058.i10

  33. [35]

    Communications in Statistics - Theory and Methods , author =

    A solution to the multivariate behrens-fisher problem , volume =. Communications in Statistics - Theory and Methods , author =. 1986 , pages =. doi:10.1080/03610928608829342 , language =

    doi:10.1080/03610928608829342 1986

  34. [36]

    Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences , author =

    Dispersion on a. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences , author =. 1953 , pages =. doi:10.1098/rspa.1953.0064 , language =

    arXiv 1953

  35. [37]

    Annals of Eugenics , author =

    The. Annals of Eugenics , author =. 1936 , pages =. doi:10.1111/j.1469-1809.1936.tb02137.x , language =

    arXiv 1936

  36. [38]

    Translational Psychiatry , author =

    Brain network dysfunctions in addiction: a meta-analysis of resting-state functional connectivity , volume =. Translational Psychiatry , author =. 2022 , pages =. doi:10.1038/s41398-022-01792-6 , number =

    doi:10.1038/s41398-022-01792-6 2022

  37. [39]

    SIAM Journal on Matrix Analysis and Applications , author =

    Permutation-. SIAM Journal on Matrix Analysis and Applications , author =. 2024 , pages =. doi:10.1137/22M1538144 , language =

    doi:10.1137/22m1538144 2024

  38. [40]

    NeuroImage , author =

    Resting state functional connectivity in addiction:. NeuroImage , author =. 2012 , pages =. doi:10.1016/j.neuroimage.2012.01.117 , language =

    doi:10.1016/j.neuroimage.2012.01.117 2012

  39. [41]

    Frontiers in Human Neuroscience , author =

    Sex. Frontiers in Human Neuroscience , author =. 2019 , pages =. doi:10.3389/fnhum.2019.00104 , urldate =

    arXiv 2019

  40. [42]

    Trends in Cognitive Sciences , author =

    Functional connectomics from resting-state. Trends in Cognitive Sciences , author =. 2013 , pages =. doi:10.1016/j.tics.2013.09.016 , language =

    doi:10.1016/j.tics.2013.09.016 2013

  41. [43]

    Cerebral Cortex , author =

    Linked. Cerebral Cortex , author =. 2015 , pages =. doi:10.1093/cercor/bhu036 , language =

    doi:10.1093/cercor/bhu036 2015

  42. [44]

    ETS Research Bulletin Series , author =

    Estimating causal effects of treatments in randomized and nonrandomized studies , volume =. ETS Research Bulletin Series , author =. doi:10.1002/j.2333-8504.1972.tb00631.x , abstract =

    arXiv 1972

  43. [45]

    Biometrika , author =

    The. Biometrika , author =. 1983 , pages =. doi:10.1093/biomet/70.1.41 , language =

    doi:10.1093/biomet/70.1.41 1983

  44. [46]

    Journal of the American Statistical Association , author =

    Estimation of. Journal of the American Statistical Association , author =. 1994 , pages =. doi:10.1080/01621459.1994.10476818 , language =

    arXiv 1994

  45. [47]

    Biometrika , author =

    Quasi-oracle estimation of heterogeneous treatment effects , volume =. Biometrika , author =. 2021 , pages =. doi:10.1093/biomet/asaa076 , abstract =

    doi:10.1093/biomet/asaa076 2021

  46. [48]

    The Annals of Applied Statistics , author =

    Propensity score weighting for causal inference with multiple treatments , volume =. The Annals of Applied Statistics , author =. doi:10.1214/19-AOAS1282 , number =

    doi:10.1214/19-aoas1282

  47. [49]

    Journal of the American Statistical Association , author =

    Balancing. Journal of the American Statistical Association , author =. 2018 , pages =. doi:10.1080/01621459.2016.1260466 , language =

    arXiv 2018

  48. [50]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Causal inference on distribution functions , volume =. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2023 , pages =. doi:10.1093/jrsssb/qkad008 , language =

    doi:10.1093/jrsssb/qkad008 2023

  49. [51]

    Geodesic

    Kurisu, Daisuke and Zhou, Yidong and Otsu, Taisuke and Müller, Hans-Georg , year =. Geodesic. doi:10.48550/ARXIV.2406.19604 , abstract =

    Pith/arXiv arXiv doi:10.48550/arxiv.2406.19604

  50. [52]

    Proceedings of the National Academy of Sciences , author =

    Sex beyond the genitalia:. Proceedings of the National Academy of Sciences , author =. 2015 , pages =. doi:10.1073/pnas.1509654112 , abstract =

    doi:10.1073/pnas.1509654112 2015

  51. [53]

    and Robins, James M

    Hernan, Miguel A. and Robins, James M. , year =. Causal

  52. [54]

    The Econometrics Journal , author =

    Double/debiased machine learning for treatment and structural parameters , volume =. The Econometrics Journal , author =. 2018 , pages =. doi:10.1111/ectj.12097 , language =

    doi:10.1111/ectj.12097 2018

  53. [55]

    The Annals of Probability , author =

    Central limit theorems and bootstrap in high dimensions , volume =. The Annals of Probability , author =. doi:10.1214/16-AOP1113 , number =

    doi:10.1214/16-aop1113

  54. [56]

    The Annals of Statistics , author =

    Gaussian approximations and multiplier bootstrap for maxima of sums of high-dimensional random vectors , volume =. The Annals of Statistics , author =. doi:10.1214/13-AOS1161 , number =

    doi:10.1214/13-aos1161

  55. [57]

    Trends in Cognitive Sciences , author =

    Large-scale brain systems in. Trends in Cognitive Sciences , author =. 2012 , pages =. doi:10.1016/j.tics.2011.11.007 , language =

    doi:10.1016/j.tics.2011.11.007 2012

  56. [58]

    Biological Psychiatry: Cognitive Neuroscience and Neuroimaging , author =

    Intrinsic. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging , author =. 2016 , pages =. doi:10.1016/j.bpsc.2016.03.004 , language =

    doi:10.1016/j.bpsc.2016.03.004 2016

  57. [59]

    Riemannian

    Collas, Antoine and Ju, Ce and Salvy, Nicolas and Thirion, Bertrand , year =. Riemannian. doi:10.48550/ARXIV.2505.18193 , abstract =

    doi:10.48550/arxiv.2505.18193

  58. [60]

    Biometrics , author =

    Doubly. Biometrics , author =. 2005 , pages =. doi:10.1111/j.1541-0420.2005.00377.x , language =

    arXiv 2005

  59. [61]

    The Annals of Statistics , author =

    Generalized random forests , volume =. The Annals of Statistics , author =. doi:10.1214/18-AOS1709 , number =

    doi:10.1214/18-aos1709

  60. [62]

    Magnetic Resonance in Medicine , author =

    Log‐. Magnetic Resonance in Medicine , author =. 2006 , pages =. doi:10.1002/mrm.20965 , abstract =

    doi:10.1002/mrm.20965 2006

  61. [63]

    1998 , keywords =

    Efficient and. 1998 , keywords =

    1998

  62. [64]

    Chwialkowski, Kacper and Strathmann, Heiko and Gretton, Arthur , editor =. A. Proceedings of. 2016 , pages =

    2016

  63. [65]

    Physical Review , author =

    On the. Physical Review , author =. 1930 , pages =. doi:10.1103/PhysRev.36.823 , language =

    doi:10.1103/physrev.36.823 1930

  64. [66]

    Orthogonal polynomials , isbn =

    Szegő, Gábor , year =. Orthogonal polynomials , isbn =

  65. [67]

    Journal of Multivariate Analysis , author =

    A new test for multivariate normality , volume =. Journal of Multivariate Analysis , author =. 2005 , pages =. doi:10.1016/j.jmva.2003.12.002 , language =

    doi:10.1016/j.jmva.2003.12.002 2005

  66. [68]

    Rayner, J. C. W. and Thas, O. and Best, D. J. , month = apr, year =. Smooth. doi:10.1002/9780470824443 , language =

    doi:10.1002/9780470824443

  67. [69]

    The Annals of Statistics , author =

    \. The Annals of Statistics , author =. doi:10.1214/aos/1176350374 , number =

    arXiv

  68. [70]

    Scandinavian Actuarial Journal , author =

    Smooth test for goodness of fit , volume =. Scandinavian Actuarial Journal , author =. 1937 , pages =. doi:10.1080/03461238.1937.10404821 , language =

    arXiv 1937

  69. [71]

    Journal für die reine und angewandte Mathematik , author =

    Ueber die. Journal für die reine und angewandte Mathematik , author =. 1866 , pages =

  70. [72]

    Biometrika , author =

    Measures of multivariate skewness and kurtosis with applications , volume =. Biometrika , author =. 1970 , pages =. doi:10.1093/biomet/57.3.519 , language =

    doi:10.1093/biomet/57.3.519 1970

  71. [73]

    Liu, Qiang and Lee, Jason and Jordan, Michael , editor =. A. Proceedings of. 2016 , pages =

    2016

  72. [74]

    Gaussian

    Janson, Svante , month = jun, year =. Gaussian. doi:10.1017/CBO9780511526169 , abstract =

    doi:10.1017/cbo9780511526169

  73. [75]

    IEEE Transactions on Neural Networks , author =

    Iterative. IEEE Transactions on Neural Networks , author =. 2011 , pages =. doi:10.1109/TNN.2011.2106511 , number =

    arXiv 2011

  74. [76]

    Journal of Machine Learning Research , author =

    Composite. Journal of Machine Learning Research , author =. 2025 , pages =

    2025

  75. [77]

    Bernoulli , author =

    A one-sample test for normality with kernel methods , volume =. Bernoulli , author =. doi:10.3150/18-BEJ1037 , number =

    doi:10.3150/18-bej1037

  76. [78]

    Communications in Statistics - Theory and Methods , author =

    A class of invariant consistent tests for multivariate normality , volume =. Communications in Statistics - Theory and Methods , author =. 1990 , pages =. doi:10.1080/03610929008830400 , language =

    doi:10.1080/03610929008830400 1990

  77. [79]

    Journal of Machine Learning Research , author =

    Spectral. Journal of Machine Learning Research , author =. 2024 , pages =

    2024

  78. [80]

    Biometrika , author =

    A test for normality based on the empirical characteristic function , volume =. Biometrika , author =. 1983 , pages =. doi:10.1093/biomet/70.3.723 , language =

    doi:10.1093/biomet/70.3.723 1983

  79. [81]

    Gaussianization , volume =

    Chen, Scott and Gopinath, Ramesh , editor =. Gaussianization , volume =. Advances in

  80. [82]

    Russian Mathematical Surveys , author =

    Ornstein–. Russian Mathematical Surveys , author =. 2018 , pages =. doi:10.1070/RM9812 , number =

    doi:10.1070/rm9812 2018

Showing first 80 references.