Pith. sign in

REVIEW 3 major objections 5 minor 80 references

Distance-profile embedding yields analytic independence tests on any metric space

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

2026-08-03 16:01 UTC pith:FCU6E57U

load-bearing objection A serious, mostly sound kernel-based framework for independence and conditional independence on metric spaces, but the recommended empirical reference measure falls outside the theory, so the default implementation lacks a proven validity guarantee. the 3 major comments →

arxiv 2607.28981 v1 pith:FCU6E57U submitted 2026-07-31 stat.ME econ.EMmath.STstat.APstat.MLstat.TH

Distance Profile Embedding for Independence and Conditional Independence Testing of Random Objects

Wenxi Tan , Bing Li , Lingzhou Xue This is my paper
classification stat.ME econ.EMmath.STstat.APstat.MLstat.TH MSC 62H1562H2062G1062G20
keywords distance profile embeddingindependence testconditional independence testrandom objectsmetric spacesreproducing kernel Hilbert spacecharacteristic kernelobject-valued conditioning
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.

The paper introduces the Distance Profile Embedding (DPE), which maps every point of a metric space to the function of its distances to all other points, viewed as an element of a Hilbert space. The central claim is that this map is injective and measurable under mild conditions, so that independence and conditional independence of random objects—such as microbiome compositions, brain networks, or probability distributions—are exactly equivalent to independence and conditional independence of the embedded Hilbert elements. Because the embedded space is Hilbertian, the paper applies reproducing-kernel machinery to build tests whose statistics have explicit weighted-chi-square limits, giving analytic p-values without permutation. If correct, this provides the first conditional independence test in which the conditioning variable itself may be an object in a general metric space, with no need for negative-type metrics, isometric embeddings, or one-to-one correspondence assumptions.

Core claim

DPE maps each random object X to the distance function u ↦ d(u, X) in L2(ΩX, λX). Under a reference measure with full support, the map is injective (Theorem 1), so X⊥Y iff Φ(X)⊥Φ(Y) (Corollary 2) and X⊥Y|Z iff Φ(X)⊥Φ(Y)|Φ(Z) (Corollary 3). With characteristic Gaussian kernels on the embedded space, the cross-covariance operator vanishes exactly under the null (Theorems 4 and 10), and the test statistics Tₙ and Sₙ have explicit weighted-chi-square null limits (Theorems 7 and 13).

What carries the argument

The distance profile embedding Φ(x) = dX(·, x), landing in the Hilbert space L2(ΩX, λX). Its injectivity and measurability let the authors transfer independence and conditional independence questions from any Polish metric space to a Hilbert space, where RKHS cross-covariance operators and their Moore–Penrose inverses provide exact characterizations and tractable null distributions.

Load-bearing premise

The reference measure used to build the distance embedding must put mass on every region of the metric space; the paper's recommended default, an empirical measure on the observed sample, does not, so the central equivalence is unproven for the default implementation.

What would settle it

Construct a metric space and two distinct distributions that agree on distance profiles to all sample points but differ elsewhere; with the empirical reference measure, DPE would fail to distinguish dependence. Concretely, simulate X and Y as independent while Y depends on a rare-but-influential region absent from the sample, then check whether the analytic test keeps its nominal size.

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

If this is right

  • Independence and conditional independence tests now apply to metric spaces that admit no isometric Hilbert embedding, including spheres with geodesic distance, SPD matrices with affine-invariant Riemannian metric, and Wasserstein spaces of distributions.
  • Conditioning variables can be object-valued; for example, testing whether female mortality is conditionally independent of fertility given the male mortality distribution.
  • Analytic p-values replace permutation tests, enabling fast inference for large samples.
  • The distance profile representation itself becomes a bridge between metric-space statistics and Hilbert-space operator theory, independent of the testing application.

Where Pith is reading between the lines

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

  • The recommended empirical reference measure (λ = average of Dirac masses at the observed sample) has finite support and does not satisfy the full-support assumption behind the injectivity theorems; the paper's validity guarantees may not cover the default implementation, and no asymptotic argument connects the empirical DPE to the population DPE.
  • The equivalence results suggest a general recipe: any injective, measurable distance-based representation of metric-space objects could inherit RKHS testing machinery, making DPE one instance of a broader 'reference-measure embedding' class.
  • The conditional test's regularity assumptions (range inclusions, kernel eigenvalue decay, representability of conditional expectations) are substantial; applying the method to non-smooth distributions on metric spaces will require checking these conditions case by case.
  • One could test the empirical-reference shortcut directly: simulate independent X and Y but with a dependence driven by a region not covered by the reference sample; if the analytic null distribution fails to control size, the shortcut is invalid.

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

3 major / 5 minor

Summary. The paper introduces the Distance Profile Embedding (DPE), which maps a random object in a general metric space to a distance function in L2 of a reference measure. Under Assumptions 2–3 the map is claimed to be injective and measurable, so that independence of X and Y is equivalent to independence of their DPEs (Corollary 2), and similarly for conditional independence (Corollary 3). The paper then builds Hilbert–Schmidt cross-covariance operators on these embedded objects and, using characteristic Gaussian kernels, claims that vanishing of the operators characterizes (conditional) independence (Theorems 4 and 10). Test statistics T_n and S_n are given explicit trace forms, and asymptotic null distributions are stated as weighted chi-square limits (Theorems 7 and 13), with fixed and local alternative results in Theorems 8–9 and 14–15. Numerical experiments on spheres, SPD matrices, and Wasserstein Gaussian distributions, plus microbiome and mortality applications, are reported. All proofs are deferred to a supplement that is not included in the submitted manuscript.

Significance. If the central theoretical claims are correct, DPE would be a substantial contribution: it would provide independence and conditional-independence tests for random objects in general metric spaces without negative-type, isometric-embeddability, or one-to-one-correspondence assumptions, and it would be the first framework to allow an object-valued conditioning variable. The analytic asymptotic null distributions are a practical improvement over permutation-based procedures. The paper also gives explicit trace-based statistics that are easy to implement. However, the validity of the recommended default implementation is currently not covered by the stated theorems, and the missing supplement prevents verification of the proof of every central result. The theoretical framework is plausible and follows the standard kernel-operator template, but the gap between the full-support Assumption 3 and the recommended empirical reference measure is load-bearing.

major comments (3)
  1. [§6, 'The Empirical Reference Measure'; Assumption 3; Theorem 1(i); Corollary 2] The recommended default reference measure λ_X = n^{-1} Σ δ_{X_i} has finite support, so Assumption 3 (supp λ_X = Ω_X) is violated. Consequently Theorem 1(i) does not apply, and the injectivity-based equivalences Corollaries 2–3, Theorem 4, and Theorem 10 are not established for the implemented statistic. The claim that the empirical measure 'preserves the characteristic properties' is informal, and no asymptotic argument (e.g., uniform convergence of the empirical embedding to a population embedding, or a separate proof of injectivity up to P_X-null sets with data-dependent λ_n) is supplied. Since §8.1 uses exactly this empirical reference measure, the validity guarantee of the default implementation is unproven. This is the weakest load-bearing link between theory and practice and must be addressed—either by restricting the implementation to fixed full-support reference measures, or by
  2. [§5.3, Theorem 12; §7.3, 'The tuning parameter ϵ_n … is fixed at 0.005'] Theorem 12 requires ϵ_n ≍ n^{-η(β∧1)/(2η(β∧1)+η+1)} for its CLT, but the simulations fix ϵ_n = 0.005 independent of n. The asymptotic null distribution of S_n in Theorem 13 is therefore not directly applicable to the simulated procedure. A fixed ϵ_n may be viewed as a finite-sample approximation, but the paper does not provide any result showing that the test with fixed ϵ_n is asymptotically valid (e.g., that the effect of regularization vanishes or that the distribution of S_n with fixed ϵ_n is stochastically bounded by the theoretical null). This gap should be closed or explicitly discussed.
  3. [General; 'All technical proofs are presented in the supplement'] Every central theorem—Theorem 1 (injectivity/measurability), Theorems 6–7 (CLT and null distribution), Theorems 10 and 12–13 (conditional independence characterization and CLT)—is deferred to a supplement that is not included in this submission. As a referee, I cannot verify the correctness of the proofs, and several claims (e.g., the Gaussian-product-kernel condition in Theorem 10, the spectral decomposition arguments in Theorem 9(ii), and the matrix trace representation for S_n) are nontrivial. The authors must provide the supplement as part of the submission for review. This is a blocking issue for acceptance, though it is fixable by supplying the missing supplement.
minor comments (5)
  1. [§6, Eq. (13)] The notation in (13) is slightly ambiguous: the integral over Ω_X with respect to λ_X is written as ∫_{u∈Ω_X} ... dλ_X(u); if λ_X is a probability measure, this is fine, but for general finite measures the normalization should be explicit (e.g., λ_X(Ω_X) factor).
  2. [References] Reference 'Dubey, P., , Y. & Müller' has a typo; the author list appears corrupted. Also, 'Hoffmann-Jorgensen' is spelled in multiple ways; use the standard 'Hoffmann-Jørgensen'.
  3. [§7, Figures] The figures are referenced but not included in the text; the captions are informative but actual plots are needed to assess the empirical claims.
  4. [§4.4, Eq. (6)] The identification of T_n as hSIC is useful; however, the authors should note that the Gaussian kernel on the DPE space uses the L2(λ) distance, so the Gram matrix entries are exp(-γ times squared DPE distances), which should be stated explicitly to avoid confusion with a kernel on the original space.
  5. [§8.1, Table 3] The p-values for 'Ball' and 'dCov' are reported to four decimals, which is fine, but the number of Monte Carlo or permutation replications used for those methods is not stated; please give the computational details for reproducibility.

Circularity Check

0 steps flagged

No significant circularity: the DPE independence/conditional-independence equivalences are derived from injectivity and characteristic-kernel theory, not from fitted inputs or self-referential definitions. The empirical-reference-measure gap is a support/validity concern, not a circular reduction.

full rationale

The paper's central chain is not circular. The DPE is defined as Φ_X(x)=d_X(·,x), and Theorem 1 proves injectivity from full support of the reference measure together with continuity of distance functions; Corollaries 2–3 then convert independence/conditional independence of the original objects to the embedded Hilbert-space objects via measurability and injectivity, a standard σ-field argument, not by assuming the target equivalence. The independence characterization (Theorem 4) and conditional characterization (Theorem 10) rely on external characteristic-kernel results (Gretton et al. 2007; Fukumizu et al. 2004, 2007; Ziegel et al. 2024), and the Gaussian-kernel special cases are stated as consequences of those external results. No parameter is fitted to the response variable: the reference measures are either fixed surface measures, Wishart distributions, or Monte Carlo samples independent of the data; the bandwidths use the usual median heuristic; and the regularization constant ϵ_n is fixed at 0.005. Thus no fitted input is renamed as a prediction. The paper's self-citations (Li 2018a,b; Li & Song 2017; Bhattacharjee et al. 2025a; Tang & Li 2026; Sang & Li 2026) are used for technical conventions, coordinate representations, or examples of smoothness conditions; the theorems explicitly assume Assumptions 6–8 rather than deriving them from those self-citations, so the self-citations are not load-bearing in a circular sense. One genuine gap, but not a circularity, is that Section 6 recommends the empirical reference measure λ_X = n^{-1}Σδ_{X_i} as 'always available' and claims it 'preserves the characteristic properties,' while Assumption 3 requires supp(λ_X)=Ω_X. The finite-support empirical measure does not satisfy this condition, so the theoretical injectivity and the resulting Corollaries 2–3 are not directly available for the default implementation, and no asymptotic bridging to a fixed full-support λ is supplied. This is a correctness/robustness concern about the implemented test, not an equivalence-by-construction of the claimed prediction to its input.

Axiom & Free-Parameter Ledger

4 free parameters · 7 axioms · 0 invented entities

The test does not fit any parameter to the outcome, so the circularity burden is low; but the method's validity depends on several domain assumptions (5–8) inherited from KCI theory, the full-support reference measure, and hand-chosen hyperparameters (γ, ϵ_n, n_MC). The total burden is typical for kernel-method papers, with the notable exception that the recommended empirical reference measure is outside the stated theory.

free parameters (4)
  • Kernel bandwidth γ (per variable) = inverse median of pairwise squared DPE distances
    Standard median heuristic for Gaussian kernels; affects power and the null approximation but is not fitted to the outcome.
  • Tikhonov regularization ϵ_n = 0.005
    Hand-chosen and fixed across all conditional-independence simulations; the theory only prescribes a rate, not a constant.
  • Reference measure λ_X, λ_Y, λ_Z = per-space choice (surface measure, Wishart, Monte Carlo draws, empirical)
    Determines the L² geometry of the embedding; the empirical variant recommended in §6 has finite support and violates the full-support hypothesis of Theorem 1.
  • Monte Carlo sample size n_MC for reference integrals = 500
    Used to approximate the λ-integrals in (13); arbitrary choice, no sensitivity analysis reported.
axioms (7)
  • domain assumption Assumption 3: Ω_Xi Polish, λ_Xi finite with supp(λ_Xi)=Ω_Xi, Borel σ-fields
    Needed for Theorem 1 injectivity/continuity/measurability; the empirical reference measure recommended in §6 does not satisfy the full-support condition.
  • domain assumption Assumption 4: E[κ(Ẋ,Ẋ)]<∞ and E[κ(Ỹ,Ỹ)]<∞
    Required for Bochner integrals and the operator CLT in Theorem 6.
  • domain assumption Assumption 5: range inclusions ran(Σ_{Ẑ(ẊẐ)}) ⊆ ran(Σ_{ẐẐ}) and ran(Σ_{ẐỸ}) ⊆ ran(Σ_{ẐẐ})
    Required for the Moore–Penrose-based conditional operator (7) to be well defined; not verifiable from data.
  • domain assumption Assumption 6: E[f(Ẋ,Ẑ)|Ẑ] ∈ H_Ẑ and E[g(Ỹ)|Ẑ] ∈ H_Ẑ for all RKHS f, g
    Standard in KCI-type theory (Fukumizu et al. 2004); if false, Theorem 10's equivalence between the vanishing operator and conditional independence breaks.
  • domain assumption Assumptions 7–8: Σ_{ỸẐ}=S Σ^{1+β}, Λ = S Σ^{1+β}, eigenvalue decay λ_j(Σ_{ẐẐ}) ⪯ j^{-η}, η>1
    Smoothness and spectral-rate conditions for the regularized inverse CLT (Theorem 12); justified mainly by reference to the authors' own prior work; not checked on microbiome or mortality data.
  • standard math Lusin–Suslin theorem: injective Borel maps between Polish spaces preserve Borel structure
    Implicit in the assertion 'measurable and injective mappings do not change σ-fields' (Subsection 3.4, Corollaries 2–3); holds because Ω and L² are Polish.
  • standard math Gaussian RBF kernels on separable Hilbert spaces are characteristic (Ziegel et al. 2024, Theorem 3.1)
    Cited to make Theorem 4 and Corollary 5 operational; this is the bridge that lets HSIC/KCI machinery work on DPE features.

pith-pipeline@v1.3.0-daily-deepseek · 24912 in / 20941 out tokens · 186279 ms · 2026-08-03T16:01:22.629724+00:00 · methodology

0 comments
read the original abstract

Testing independence or conditional independence is fundamental to statistical inference, yet existing methods for non-Euclidean random objects often face a difficult trade-off between geometric flexibility and theoretical tractability. We introduce the Distance Profile Embedding (DPE), a novel representation that maps random objects from general metric spaces into a Hilbert space of square-integrable functions. We prove that this mapping is injective and preserves full distributional information without requiring isometric Hilbert embeddings or one-to-one correspondence conditions. Leveraging the DPE, we develop a unified framework for marginal and conditional independence testing of random objects that enjoys a rigorous asymptotic theory for both size and power. Notably, our framework is the first in the literature to accommodate object-valued conditioning variables when testing conditional independence, overcoming the Euclidean or Hilbertian constraints of existing methodologies. We facilitate the calculation of analytic $p$-values using closed-form asymptotic null distributions, which avoids the computational burden of permutation tests common in existing metric-based methods. The numerical properties of our methods are demonstrated through both simulations and two real-world applications involving gut microbiome compositions and global human mortality distributions, respectively.

Figures

Figures reproduced from arXiv: 2607.28981 by Bing Li, Lingzhou Xue, Wenxi Tan.

Figure 1
Figure 1. Figure 1: Empirical power for testing independence on [PITH_FULL_IMAGE:figures/full_fig_p023_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Empirical power for testing independence on SPD matrices (Scenario 2) and bivariate [PITH_FULL_IMAGE:figures/full_fig_p024_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Empirical rejection rates for the proposed DPE-based conditional independence test. [PITH_FULL_IMAGE:figures/full_fig_p026_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Empirical p-values for conditional independence testing between female mortality and fertility distributions given male mortality distribution. The curves show the proposed DPE-based test evaluated across five-year intervals under different Wasserstein−p metrics. The dashed horizontal line denotes the significance level 0.05. We apply the proposed DPE-based conditional independence test to these distributi… view at source ↗

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

80 extracted references · 3 linked inside Pith

  1. [1]

    Association and independence test for random objects , author=. Ann. Statist., to appear , year=

  2. [2]

    Li, Bing and Song, Jun , title =. Ann. Statist. , year =

  3. [3]

    1993 , isbn =

    Serge Lang , title =. 1993 , isbn =

    1993

  4. [4]

    Canadian Journal of Statistics , year =

    Bing Li , title =. Canadian Journal of Statistics , year =

  5. [5]

    Journal of Nonparametric Statistics , volume=

    Sparse kernel sufficient dimension reduction , author=. Journal of Nonparametric Statistics , volume=. 2025 , publisher=

    2025

  6. [6]

    Distance-based and RKHS-based dependence metrics in high dimension , author=. Ann. Statist. , volume=. 2020 , publisher=

    2020

  7. [7]

    Journal of the American Statistical Association , volume=

    Martingale difference correlation and its use in high-dimensional variable screening , author=. Journal of the American Statistical Association , volume=. 2014 , publisher=

    2014

  8. [8]

    Nonparametric statistical inference via metric distribution function in metric spaces , author=. J. Am. Stat. Assoc. , volume=. 2024 , publisher=

    2024

  9. [9]

    Demography , volume=

    Coherent mortality forecasts for a group of populations: An extension of the Lee-Carter method , author=. Demography , volume=. 2005 , publisher=

    2005

  10. [10]

    Metric statistics: Exploration and inference for random objects with distance profiles , author=. Ann. Statist. , volume=. 2024 , publisher=

    2024

  11. [11]

    arXiv preprint arXiv:2506.22754 , year=

    Doubly robust estimation of causal effects for random object outcomes with continuous treatments , author=. arXiv preprint arXiv:2506.22754 , year=

    Pith/arXiv arXiv

  12. [12]

    Testing mutual independence in metric spaces using distance profiles , author=. Ann. Statist., to appear , year=

  13. [13]

    Equivalence of distance-based and RKHS-based statistics in hypothesis testing , author=. Ann. Statist. , volume=

  14. [14]

    Lecture Notes, Columbia University , volume=

    A gentle introduction to empirical process theory and applications , author=. Lecture Notes, Columbia University , volume=

  15. [15]

    2015 , publisher =

    Theoretical Foundations of Functional Data Analysis, with an Introduction to Linear Operators , author =. 2015 , publisher =

    2015

  16. [16]

    , author=

    Statistical consistency of kernel canonical correlation analysis. , author=. Journal of Machine Learning Research , volume=

  17. [17]

    A nonparametric test for elliptical distribution based on kernel embedding of probabilities , author=. Ann. Statist. , volume=. 2024 , publisher=

    2024

  18. [18]

    Advances in Neural Information Processing Systems , volume=

    Testing for homogeneity with kernel Fisher discriminant analysis , author=. Advances in Neural Information Processing Systems , volume=

  19. [19]

    2020 , publisher=

    Gaussian measures in Hilbert space: construction and properties , author=. 2020 , publisher=

    2020

  20. [20]

    A new form of the spherical expansion of zonal functions and Fourier transforms of SO (d)-finite functions , author=. SIGMA. Symmetry, Integrability and Geometry: Methods and Applications , volume=. 2006 , publisher=

    2006

  21. [21]

    2012 , publisher=

    Spherical Harmonics and Approximations on the Unit Sphere: An Introduction , author=. 2012 , publisher=

    2012

  22. [22]

    arXiv preprint arXiv:0710.2063 , year=

    Distance matrices and isometric embeddings , author=. arXiv preprint arXiv:0710.2063 , year=

    Pith/arXiv arXiv 2063

  23. [23]

    IEEE Trans

    Riemannian Gaussian distributions on the space of symmetric positive definite matrices , author=. IEEE Trans. Inf. Theory , volume=. 2017 , publisher=

    2017

  24. [24]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    Kernel methods on the Riemannian manifold of symmetric positive definite matrices , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

  25. [25]

    Journal of Machine Learning Research , year =

    Nina Miolane and Nicolas Guigui and Alice Le Brigant and Johan Mathe and Benjamin Hou and Yann Thanwerdas and Stefan Heyder and Olivier Peltre and Niklas Koep and Hadi Zaatiti and Hatem Hajri and Yann Cabanes and Thomas Gerald and Paul Chauchat and Christian Shewmake and Daniel Brooks and Bernhard Kainz and Claire Donnat and Susan Holmes and Xavier Pennec...

  26. [26]

    Advances in Neural Information Processing Systems , volume=

    A fast, consistent kernel two-sample test , author=. Advances in Neural Information Processing Systems , volume=

  27. [27]

    Russian Mathematical Surveys , volume=

    The eigen-and singular values of the sum and product of linear operators , author=. Russian Mathematical Surveys , volume=. 1964 , publisher=

    1964

  28. [28]

    Advances in Neural Information Processing Systems , pages =

    Zaid Harchaoui and Francis Bach and Eric Moulines , title =. Advances in Neural Information Processing Systems , pages =

  29. [29]

    Bing Li and Jun Song , title =. Ann. Statist. , number =

  30. [30]

    2018 , publisher =

    Sufficient Dimension Reduction: Methods and Applications with R , author =. 2018 , publisher =

    2018

  31. [31]

    Nonlinear global Fr

    Bhattacharjee, Satarupa and Li, Bing and Xue, Lingzhou , journal=. Nonlinear global Fr

  32. [32]

    Journal of Machine Learning Research , volume=

    Nonlinear function-on-function regression by RKHS , author=. Journal of Machine Learning Research , volume=

  33. [33]

    arXiv preprint arXiv:2603.13704 , year=

    A kernel-based nonparametric test for conditional independence of functional data , author=. arXiv preprint arXiv:2603.13704 , year=

    Pith/arXiv arXiv

  34. [34]

    Advances in Neural Information Processing Systems , volume=

    A kernel statistical test of independence , author=. Advances in Neural Information Processing Systems , volume=

  35. [35]

    Testing in microbiome-profiling studies with MiRKAT, the microbiome regression-based kernel association test , author=. Am. J. Hum. Genet. , volume=. 2015 , publisher=

    2015

  36. [36]

    Proceedings of the Twenty-Seventh Conference on Uncertainty in Artificial Intelligence , pages=

    Kernel-based conditional independence test and application in causal discovery , author=. Proceedings of the Twenty-Seventh Conference on Uncertainty in Artificial Intelligence , pages=

  37. [37]

    Advances in Neural Information Processing Systems , volume=

    Universal kernels on non-standard input spaces , author=. Advances in Neural Information Processing Systems , volume=

  38. [38]

    Dimension reduction for Fr

    Zhang, Qi and Xue, Lingzhou and Li, Bing , journal=. Dimension reduction for Fr. 2024 , publisher=

    2024

  39. [39]

    Journal of Multivariate Analysis , volume=

    Nonlinear sufficient dimension reduction for distribution-on-distribution regression , author=. Journal of Multivariate Analysis , volume=. 2024b , publisher=

  40. [40]

    Gates Open Research , volume=

    Mortality, fertility, and economic development: An analysis of 201 countries from 1960 to 2015 , author=. Gates Open Research , volume=

    1960

  41. [41]

    2013 , publisher=

    Measure Theory , author=. 2013 , publisher=

    2013

  42. [42]

    The Journal of Machine Learning Research , volume=

    Hilbert space embeddings and metrics on probability measures , author=. The Journal of Machine Learning Research , volume=. 2010 , publisher=

    2010

  43. [43]

    Advances in Neural Information Processing Systems , volume=

    Kernel measures of conditional dependence , author=. Advances in Neural Information Processing Systems , volume=

  44. [44]

    and Jordan, Michael I

    Fukumizu, Kenji and Bach, Francis R. and Jordan, Michael I. , title =. Ann. Statist. , year =. doi:10.1214/08-AOS637 , url =

    doi:10.1214/08-aos637

  45. [45]

    On some measures analogous to Haar measure

    Christensen, Jens Peter Reus , journal =. On some measures analogous to Haar measure. , volume =

  46. [46]

    Mathematika , author=

    Measures not approximable or not specifiable by means of balls , volume=. Mathematika , author=. 1971 , pages=

    1971

  47. [47]

    , journal =

    Hoffmann-Jorgensen, J. , journal =. Measures which agree on balls. , volume =

  48. [48]

    International conference on algorithmic learning theory , pages=

    Measuring statistical dependence with Hilbert-Schmidt norms , author=. International conference on algorithmic learning theory , pages=. 2005 , organization=

    2005

  49. [49]

    Journal of Machine Learning Research , volume=

    Dimensionality reduction for supervised learning with reproducing kernel Hilbert spaces , author=. Journal of Machine Learning Research , volume=

  50. [50]

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

    Riemann manifold langevin and hamiltonian monte carlo methods , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2011 , publisher=

    2011

  51. [51]

    Wasserstein F-tests for Fr

    Xu, Haoshu and Li, Hongzhe , journal=. Wasserstein F-tests for Fr

  52. [52]

    Székely and Maria L

    Gábor J. Székely and Maria L. Rizzo and Nail K. Bakirov , journal =. Measuring and testing dependence by correlation of distances , urldate =

  53. [53]

    Journal of Statistical Software , volume=

    Ball: An R package for detecting distribution difference and association in metric spaces , author=. Journal of Statistical Software , volume=

  54. [54]

    Wu and Jun Chen and Christian Hoffmann and Kyle Bittinger and others , title =

    Gary D. Wu and Jun Chen and Christian Hoffmann and Kyle Bittinger and others , title =. Science , volume =

  55. [55]

    Bioinformatics , volume=

    A general framework for association analysis of microbial communities on a taxonomic tree , author=. Bioinformatics , volume=. 2017 , publisher=

    2017

  56. [56]

    Annual Review of Statistics and Its Application , author =

    Microbiome, metagenomics, and high-dimensional compositional data analysis , volume =. Annual Review of Statistics and Its Application , author =. 2015 , pages =

    2015

  57. [57]

    The Lancet , volume=

    Global, regional, and national levels and causes of maternal mortality during 1990--2013: a systematic analysis for the global burden of disease study 2013 , author=. The Lancet , volume=. 2014 , publisher=

    1990

  58. [58]

    2019 , institution =

    Maternal mortality: levels and trends 2000 to 2017 , author =. 2019 , institution =

    2000

  59. [59]

    Conditional distance correlation , author=. J. Am. Stat. Assoc. , volume=. 2015 , publisher=

    2015

  60. [60]

    A non-parametric test of independence , author =. Ann. Math. Statist. , volume =

  61. [61]

    Biometrika , volume=

    A new measure of rank correlation , author=. Biometrika , volume=. 1938 , publisher=

    1938

  62. [62]

    Biometrika , volume=

    A consistent multivariate test of association based on ranks of distances , author=. Biometrika , volume=. 2013 , publisher=

    2013

  63. [63]

    Biometrika , volume=

    Nonparametric independence testing via mutual information , author=. Biometrika , volume=. 2019 , publisher=

    2019

  64. [64]

    Biometrika , volume=

    Distribution-free tests of independence in high dimensions , author=. Biometrika , volume=. 2017 , publisher=

    2017

  65. [65]

    Asymptotic distribution-free independence test for high-dimension data , author=. J. Am. Stat. Assoc. , volume=. 2024 , publisher=

    2024

  66. [66]

    The American Statistician , volume=

    On conditional and partial correlation , author=. The American Statistician , volume=. 1976 , publisher=

    1976

  67. [67]

    Biometrika , volume=

    Testing independence for sparse longitudinal data , author=. Biometrika , volume=. 2024 , publisher=

    2024

  68. [68]

    Biometrical Journal , volume=

    Overview of object oriented data analysis , author=. Biometrical Journal , volume=. 2014 , publisher=

    2014

  69. [69]

    Science Advances , volume=

    Air pollution and COVID-19 mortality in the United States: Strengths and limitations of an ecological regression analysis , author=. Science Advances , volume=. 2020 , publisher=

    2020

  70. [70]

    Probability measures , booktitle =

    Da Prato, Giuseppe and Zabczyk, Jerzy , year =. Probability measures , booktitle =

  71. [71]

    Mathematical foundations of beta diversity: why common metrics fail in microbiome analysis , author=. J. Stat. Theory Appl. , volume=. 2026 , publisher=

    2026

  72. [72]

    arXiv preprint arXiv:2509.13685 , year=

    Variable Selection for Additive Global Fr 'echet Regression , author=. arXiv preprint arXiv:2509.13685 , year=

    arXiv

  73. [73]

    Alexander Petersen and Hans-Georg Müller , journal =. Fr

  74. [74]

    Bernoulli , volume=

    Characteristic kernels on Hilbert spaces, Banach spaces, and on sets of measures , author=. Bernoulli , volume=. 2024 , publisher=

    2024

  75. [75]

    Statistica Sinica, in press , year=

    High-dimensional log contrast models with measurement errors , author=. Statistica Sinica, in press , year=

  76. [76]

    arXiv preprint arXiv:2603.29058 , year=

    A unified framework for nonlinear mediation analysis of random objects , author=. arXiv preprint arXiv:2603.29058 , year=

    arXiv

  77. [77]

    Journal of the Royal Statistical Society Series B , pages=

    A copula graphical model for multi-attribute data using optimal transport , author=. Journal of the Royal Statistical Society Series B , pages=. 2026 , publisher=

    2026

  78. [78]

    Test of independence using generalized distance correlation , author=. Ann. Statist. , volume=

  79. [79]

    Linear algebra and its applications , volume=

    An observation on the Hadamard product of Hermitian matrices , author=. Linear algebra and its applications , volume=. 1995 , publisher=

    1995

  80. [80]

    , title =

    Fiedler, Miroslav and Markham, Thomas L. , title =. Linear Algebra and its Applications , year =