Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

On the Wasserstein Geodesic Principal Component Analysis of probability measures

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper shows that geodesic PCA in the Wasserstein space can be carried out exactly for centered Gaussians and approximated with neural-parameterized geodesics for general absolutely continuous measures.

desk verdict Gaussian GPCA is a solid, citable contribution; the general-a.c. method is promising but its 'exact' claim is not supported by the current geodesic certificate. read the letter →

arxiv 2506.04480 v2 pith:NJAUUBTL submitted 2025-06-04 stat.ML cs.LGstat.ME

Nina Vesseron , Elsa Cazelles , Alice Le Brigant , Thierry Klein This is my paper
classification stat.MLcs.LGstat.ME MSC 49Q2262R30
keywords geodesicprincipalcomponentanalysisWassersteinspaceOtto-WassersteingeometryBures-WassersteindistanceoptimaltransportneuralnetworkparameterizationtangentPCAprobabilitymeasures
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper attacks a step that has blocked PCA on collections of probability measures: finding the geodesic curve in Wasserstein space that best explains the data's variation, without first flattening the space through a tangent approximation. For centered Gaussians it lifts the problem into the flat space of invertible matrices via Otto's fiber bundle, so the Geodesic PCA objective is optimized over true Wasserstein geodesics rather than linearized ones. For general absolutely continuous measures it parameterizes the same kind of geodesics through a reference diffeomorphism and a scalar potential, both represented by multilayer perceptrons, and fits them by minimizing the sum of squared projection residuals. The method gives curves that are genuine Wasserstein geodesics whenever the parameterization constraint is met, and it makes sampling along components possible. A sympathetic reader would care because tangent PCA, while cheap, can distort data on curved parts of the Wasserstein manifold, and the paper shows where that distortion matters.

What carries the argument

The central object is Otto's fiber bundle over the Wasserstein space: the bottom space is probability measures (or SPD matrices for centered Gaussians), the top space is diffeomorphisms (or $\mathrm{GL}_d$), and the projection is pushforward by a fixed reference measure (or $A\mapsto AA^\top$). A horizontal line segment in the top space projects to a Wasserstein geodesic, so every geodesic writes as $\mu(t)=(\mathrm{id}+t\nabla f)_\#(\phi_\#\rho)$, with $f$ smooth and $\mathrm{id}+t\nabla f$ a diffeomorphism. This parametrization carries the whole argument: it replaces the curved optimization by a flat one, makes projection times explicit, distinguishes horizontal directions so that orthogonality of components has a concrete meaning in $L^2(\rho)$, and, because the map is explicit, lets the practitioner sample points along any component. In the Gaussian restriction, the same machinery specializes to $A+tX$ with $X^\top A = A^\top X$ and gives the Bures-Wasserstein distance as a minimal Frobenius distance between fibers.

What would settle it

Evaluate the learned maps on a dense grid: if $\mathrm{Id}+tH_{f_\psi}$ has a non-positive eigenvalue for any $t$ within the fitted $[t_{\min}, t_{\max}]$, then the GPCA GEN curve is not a Wasserstein geodesic and the loss was minimized over an invalid family; in the Gaussian case, run the same-eigenvalue circle experiment with $|a-b|/|a+b|$ near $0.8$ and check whether GPCA's residual cost beats TPCA by the tens of percent that the paper's Figure 4 and equation (17) imply.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that Otto's Riemannian submersion picture turns Wasserstein GPCA from a geodesic-fitting problem on a curved space into a horizontal-line-fitting problem in a flat total space. In the Gaussian case the total space is $\mathrm{GL}_d$ with the Frobenius metric, the geodesic is a line $A+tX$ with $X$ horizontal at $A$, and the cost becomes explicit in terms of rotations $Q_i$ representing each covariance in its fiber. The paper proves this reformulation minimizes the original residual objective, and shows that TPCA and GPCA generically agree except near the boundary of the SPD cone, where linearization distorts pairwise distances by a factor quantified in Proposition 5. For absolutely continuous measures, the analogous statement is that every Wasserstein geodesic has the form $(\mathrm{id}+t\nabla f)_\#(\phi_\#\rho)$, so fitting a geodesic component reduces to learning $\phi$ and $f$; the paper's contribution is a neural parameterization and an algorithm (GPCA GEN) that jointly learns the component, the projection times, and, for the second component, an orthogonal intersection enforced by regularization.

Load-bearing premise

In the general absolutely continuous case, the whole construction assumes the fitted maps $\phi_\theta$ and $\mathrm{id}+t\nabla f_\psi$ are diffeomorphisms on the chosen time interval; the implementation neither enforces invertibility of $\phi_\theta$ nor verifies the Hessian condition outside the minibatches used to estimate $t_{\min}$ and $t_{\max}$, so the output curve may not be a true Wasserstein geodesic.

Editorial extensions

If this is right

  • For Gaussian datasets, practitioners can compute GPCA components by optimizing over $\mathrm{GL}_d$ and $\mathrm{SO}_d$, with a certified global minimizer for the first component, instead of relying on tangent-space PCA.
  • Because the Gaussian formulation is exact, it provides a benchmark to measure how much accuracy tangent PCA loses; Proposition 5 predicts that loss grows near the SPD cone boundary.
  • For general distributions, GPCA GEN gives geodesic components that can be sampled at any time, enabling interpolations and visualizations that discrete TPCA methods cannot produce directly.
  • The Otto parameterization avoids convexity constraints, replacing them with Hessian-eigenvalue monitoring, which suggests a new route for parameterizing Wasserstein geodesics without input-convex neural networks.
  • The univariate result that the first principal geodesic component stays Gaussian (Proposition 4) shows that the known one-dimensional coincidence between GPCA and linearized PCA persists within Gaussian families.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the paper's Gaussian analysis supplies a practical diagnostic for when tangent PCA is acceptable: compare the fitted horizontal line's endpoints with the barycenter and with the SPD cone boundary, as in the paper's Figure 4, and similar diagnostics could guide users of linearized Wasserstein distances elsewhere.
  • Editorial inference: the same neural parameterization could be compared head-to-head against input-convex neural network geodesic solvers on identical datasets; since Otto's $f$ need not be convex, such a comparison would clarify when convexity is a restrictive assumption.
  • Editorial inference: a natural stress test would replace the single scalar potential $f$ by a time-dependent potential, which would approximate non-geodesic interpolation structures while still projecting from a flat total space.
  • Editorial inference: the ratio $|a-b|/|a+b|$ acts as a curvature warning in the Gaussian case, and the paper's Proposition 5 suggests that a similar curvature-based heuristic could be derived for other positively curved manifolds.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. This paper develops two geodesic PCA procedures in Wasserstein space. In the Gaussian case, it uses the Otto bundle over SPD matrices to lift the geodesic PCA problem to GL_d with horizontal lines, proves existence and equivalence, and compares with tangent PCA, including a second-order distortion formula. In the general absolutely continuous case, it parameterizes geodesics as t ↦ (id+t∇f_ψ)#(φ_θ#ρ) using MLPs, optimizes a Sinkhorn-based residual loss with soft orthogonality/intersection penalties, and demonstrates the method on synthetic MNIST geodesics, 3D point clouds, and image color distributions.

Significance. The Gaussian contribution is a valuable exact finite-dimensional formulation of Wasserstein GPCA; if the proofs are completed, it gives practitioners a principled alternative to tangent PCA and a quantitative account of when linearization fails. The neural parameterization of Otto geodesics is also a novel and promising idea, and the experiments are a useful sanity check. However, the paper's central claim of exactness for general a.c. measures is not supported by the implemented algorithm: the geodesic conditions are only checked on minibatches, the loss is a Sinkhorn approximation, and the constraints for later components are soft. With appropriate certification or reframing, the paper would be a solid contribution; as it stands, the general-case claims outrun the evidence.

major comments (3)
  1. [Section 4, Eq. (14), Algorithm 1, Appendix E] The abstract and Section 1 state that in the a.c. case the components are 'true geodesics that minimize (1)' and that the method is 'exact'. The implemented procedure does not support this claim. First, the objective in Algorithm 1 line 7 is the Sinkhorn divergence S_ε, not W_2^2, so (14) is minimized only approximately; no convergence or bias result is provided for ε→0 or as the batch size grows. Second, the interval [tmin,tmax] that should guarantee that id+t∇f_ψ is a diffeomorphism is estimated from Hessian eigenvalues on a finite minibatch (line 5), which cannot certify global positive-definiteness for an MLP Hessian on all of R^d. Third, φ_θ is an unconstrained MLP, whereas the geodesic representation requires φ to be a diffeomorphism and φ_θ#ρ to be absolutely continuous; Appendix E explicitly notes that a non-diffeomorphic φ can make the computed curve degenerate. Consequently, the optimized curves are not certified to be Wasserstein geodesics, and the minimization is effectively over a superset of the true geodesics. The exactness claims should either be accompanied by certificates (for example, invertibility-by-construction architectures and global Hessian bounds) or be restated as an approximate algorithm.
  2. [Section 3 / Proposition 4 (Appendix D.1, Proposition 13)] The proof of Proposition 4 shows only that the Gaussian projection of any candidate geodesic has cost no larger than the candidate, and hence that the projected geodesic is also optimal; it does not imply that a given optimal geodesic coincides with its projection. Equality of optimal costs does not force equality of minimizers without a uniqueness or strict-convexity argument. As written, the statement that 'the first principal geodesic component ... remains in the space of Gaussian distributions' is not established. Please provide a uniqueness argument or weaken the proposition to the existence of a Gaussian optimal component.
  3. [Section 4, second and higher-order components] The definitions of successive GPCA components require exact intersection and orthogonality, but the implementation uses soft penalties λ_I I and λ_O O with λ_I=λ_O=1.0 (Appendix E, Table 1). No constraint-satisfaction guarantee or post-hoc verification is given; the text only states that the regularization terms are monitored to decrease. Thus even if each estimated curve were a true geodesic, the output pair need not intersect orthogonally, and the result need not satisfy the recursive definition of GPCA components used in Section 3 and in the narrative of Section 4. This further weakens the claim that the method computes exact GPCA in the a.c. case.
minor comments (5)
  1. [Appendix D.1, Lemma 1] The coercivity argument in the proof of Lemma 1 is difficult to follow: the 'diagonal' A=Σ_i^{1/2}Q_i is discussed before the constraint Q_i∈SO_d is exploited, and the conclusion is stated over (R^{d×d})^n. Coercivity on GL_d × B × SO_d^n follows directly from the compactness of the sets {Σ_i^{1/2}Q_i}; the proof should be reorganized accordingly.
  2. [Section 3, Eq. (11)] The projection p_{A,X}(t_i) and the interval [tmin,tmax] are used in Eq. (11) before they are defined; please introduce both before Proposition 3.
  3. [Algorithm 1, line 7] The Sinkhorn divergence S_ε is used without definition in the main text; either define it and state its relation to W_2^2 in Section 4, or explicitly defer to Appendix E at first use.
  4. [Section 5.1] The claim that GPCA reduces the objective by less than 1% relative to TPCA 'on average for 100 trials' is reported without standard deviations or a precise description of the 'uniform distribution on the spectral parameter space'; please specify the sampling procedure and report dispersion.
  5. [Figures 9 and 10] The captions contain the typo 'GPCAG EN' for 'GPCA GEN', and Section 3 contains 'Propositon'; a proofreading pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Gaussian and general-case objectives are exact reformulations of the stated GPCA problems via standard Otto-Wasserstein geometry, not reductions to their own outputs.

full rationale

The paper's central derivation is self-contained against external mathematics. In the Gaussian case, problem (10) is transformed into (11) by substituting the horizontal-line geodesic representation (6) and the fiber-distance formula (7); Proposition 3 proves the equivalence directly, so the lifted objective is the original GPCA criterion by a proven identity, not by construction. In the general a.c. case, the objective (14) follows from substituting Otto's geodesic parametrization (8) (Propositions 2 and 8), which is cited to Otto [28] and McCann [24], into the GPCA cost (13); no fitted parameter enters the definition of the geodesic family itself. The MLP parameterization of phi and f is an approximation ansatz, but it is not smuggled in through the paper's own claims, and the MNIST experiment is a recovery sanity check on known geodesics rather than a prediction forced by a fit. The estimation of tmin and tmax from minibatch Hessian eigenvalues (Algorithm 1, line 5, and Appendix E) means the implemented algorithm does not certify that its output curves are true Wasserstein geodesics; this is a correctness/verification gap in the 'exact' claim, not a circular reduction, because the paper never uses the target components to define the optimization. Self-citations such as [5] and [10] are used for background or standard quantile-isometry facts and are not load-bearing; the univariate Gaussian result of Proposition 4 is proved in the text from an external inequality [32]. Appendix E's admission that a non-diffeomorphic phi can make the curve degenerate further supports a caution about guarantees, but does not indicate that any result is defined in terms of its own conclusion.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rely on standard Otto submersion geometry and McCann interpolation, plus three practical assumptions introduced by the algorithm: the MLP phi approximates a diffeomorphism, Hessian positivity is validated on finite samples, and Sinkhorn divergence stands in for W2. No new physical or mathematical entities are postulated.

free parameters (4)
  • Sinkhorn regularization epsilon = 0.01 * E_{x,x' ~ nu_i} ||x - x'||^2, estimated per minibatch
    Controls the bias of the Sinkhorn divergence surrogate for W2 in the GPCA GEN objective. Chosen per dataset rather than derived from first principles.
  • Intersection and orthogonality weights lambda_I, lambda_O = 1.0
    Selected by grid search over 0.1, 1.0, 10.0, 100.0 in Appendix E. Balance the intersection and orthogonality constraints for second and higher components.
  • Projection times t_i = One optimized scalar per data point
    Latent variables in the objective (14) that locate each measure relative to the learned geodesic. They are fitted as part of the optimization.
  • Neural network weights of phi_theta and f_psi = Trained by Adam
    Approximate the diffeomorphism and scalar potential in Otto's geodesic parameterization. No convergence guarantee to the true geodesic PCA optimum is provided.
assumptions (5)
  • standard math The projection pi from GL_d to SPD matrices is a Riemannian submersion with isometric horizontal spaces, so horizontal line segments project to Bures-Wasserstein geodesics.
    Used in Propositions 1 and 3 and throughout the Gaussian GPCA derivation. Standard result cited to Refs. [36, 23, 4].
  • standard math Every Wasserstein geodesic of absolutely continuous measures can be written as (id + t grad f)#(phi#rho) for some diffeomorphism phi and smooth f on a time interval.
    Proposition 2 and Appendix B.3. This is Otto's geodesic representation, equivalent to McCann interpolation when the time interval is reparameterized.
  • domain assumption The input measures are absolutely continuous with finite second moments, while in practice they are represented by finite samples and approximated by empirical measures in the Sinkhorn loss.
    Assumed in Section 2 and Appendix B.1 for the theoretical framework. Algorithm 1 operates on batches of samples, so the theoretical regularity conditions are not exactly satisfied in experiments.
  • ad hoc to paper The MLP phi_theta behaves as a diffeomorphism and produces an absolutely continuous pushforward measure.
    Appendix E explicitly states that phi is a standard MLP with no invertibility constraint, and that a non-diffeomorphic phi may yield degenerate geodesics. This assumption is load-bearing for the theoretical validity of GPCA GEN.
  • ad hoc to paper The interval [tmin, tmax] on which id + t grad f remains a diffeomorphism can be determined from Hessian eigenvalues evaluated on a finite minibatch.
    Algorithm 1 line 5 estimates extrema of Hessian eigenvalues over a finite sample set, whereas the theoretical condition requires positivity of the Hessian on all of R^d.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the Wasserstein Geodesic Principal Component Analysis of probability measures." pith.science (2026). https://pith.science/paper/NJAUUBTL

@misc{pith2026250604480,
  author       = {Pith},
  title        = {Pith review of: On the Wasserstein Geodesic Principal Component Analysis of probability measures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NJAUUBTL}},
  note         = {Machine review of arXiv:2506.04480}
}
read the original abstract

This paper focuses on Geodesic Principal Component Analysis (GPCA) on a collection of probability distributions using the Otto-Wasserstein geometry. The goal is to identify geodesic curves in the space of probability measures that best capture the modes of variation of the underlying dataset. We first address the case of a collection of Gaussian distributions, and show how to lift the computations in the space of invertible linear maps. For the more general setting of absolutely continuous probability measures, we leverage a novel approach to parameterizing geodesics in Wasserstein space with neural networks. Finally, we compare to classical tangent PCA through various examples and provide illustrations on real-world datasets.

Figures

Figures reproduced from arXiv: 2506.04480 by the authors.

Figure 1
Figure 1. The Otto-Wasserstein geometry of centered non-degenerate Gaussian distributions. Figure inspired by [17]. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. First (red) and second (blue) geodesic components of Gaussian GPCA, where [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. GPCA on a set of diagonal covariance matrices [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Comparison between tangent and geodesic PCA on a set of [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Densities of probability distributions uniformly sampled along the first and second principal geodesics [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Empirical distributions uniformly sampled along the geodesics corresponding to the first ( [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Each lamp point cloud (left) and each image (right) is embedded in the plane according to its projection times onto the first and second principal components computed by GPCAGEN. 6 Discussion We have proposed two methods for computing exact GPCA : one tailored for Gaus…
Figure 4
Figure 4. Figure 4: In the general case of a.c. probability measures, a key advantage of our approach is that it operates directly on [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 8
Figure 8. Figure 8: Principal geodesic components of a set of Gaussian distributions whose covariance matrices have same [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Densities of probability distributions uniformly sampled along the geodesics corresponding to the first and [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Each point cloud, corresponding to a distribu [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 12
Figure 12. Figure 12: Each MNIST digit is embedded in the plane (the arrows indicate the exact position of each digit) according [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: The Otto-Wasserstein geometry of (left) centered Gaussian distributions and (right) a.c. probability distributions. Figures inspired by [17]. Proposition 10 (Propositon 1 in main). Any geodesic t 7→ Σ(t) in S ++ d for the Bures-Wasserstein metric (3) is the π-projecti…

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. PCA of probability measures: Sparse and Dense sampling regimes

    stat.ML 2026-02 conditional novelty 6.0 of 10

    PCA of random probability measures converges at rate n^{-1/2}+m^{-α} in the double asymptotic regime, with the dense-regime covariance rate minimax optimal.

Reference graph

Works this paper leans on

44 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [1]

    Barycenters in the Wasserstein space

    Martial Agueh and Guillaume Carlier. Barycenters in the Wasserstein space. SIAM Journal on Mathematical Analysis , 43(2):904–924, 2011

    work page 2011

  2. [2]

    A user’s guide to optimal transport

    Luigi Ambrosio, Alberto Bressan, Dirk Helbing, Axel Klar, Enrique Zuazua, Luigi Ambrosio, and Nicola Gigli. A user’s guide to optimal transport. Modelling and Optimisation of Flows on Networks: Cetraro, Italy 2009, Editors: Benedetto Piccoli, Michel Rascle, pages 1–155, 2013

    work page 2009

  3. [3]

    Gradient flows: in metric spaces and in the space of probability measures

    Luigi Ambrosio, Nicola Gigli, and Giuseppe Savaré. Gradient flows: in metric spaces and in the space of probability measures. Springer Science & Business Media, 2008

    work page 2008

  4. [4]

    On the Bures–Wasserstein distance between positive definite matrices

    Rajendra Bhatia, Tanvi Jain, and Yongdo Lim. On the Bures–Wasserstein distance between positive definite matrices. Expositiones Mathematicae, 37(2):165–191, 2019

    work page 2019

  5. [5]

    Geodesic PCA in the Wasserstein space by convex PCA

    Jérémie Bigot, Raúl Gouet, Thierry Klein, and Alfredo López. Geodesic PCA in the Wasserstein space by convex PCA. Annales de l’Institut Henri Poincaré - Probabilités et Statistiques, 53(1):1–26, 2017

    work page 2017

  6. [6]

    Distribution’s template estimate with Wasserstein metrics

    Emmanuel Boissard, Thibaut Le Gouic, and Jean-Michel Loubes. Distribution’s template estimate with Wasserstein metrics. Bernoulli, 21(2):740–759, 2015

    work page 2015

  7. [7]

    An introduction to optimization on smooth manifolds

    Nicolas Boumal. An introduction to optimization on smooth manifolds. Cambridge University Press, 2023

    work page 2023

  8. [8]

    Polar factorization and monotone rearrangement of vector-valued functions

    Yann Brenier. Polar factorization and monotone rearrangement of vector-valued functions. Communications on pure and applied mathematics, 44(4):375–417, 1991

    work page 1991

Show all 44 references
  1. [9]

    Populations of unlabelled networks: Graph space geometry and generalized geodesic principal components

    Anna Calissano, Aasa Feragen, and Simone Vantini. Populations of unlabelled networks: Graph space geometry and generalized geodesic principal components. Biometrika, pages 147–170, 2024

    2024

  2. [10]

    Geodesic PCA versus log-PCA of histograms in the Wasserstein space

    Elsa Cazelles, Vivien Seguy, Jérémie Bigot, Marco Cuturi, and Nicolas Papadakis. Geodesic PCA versus log-PCA of histograms in the Wasserstein space. SIAM Journal on Scientific Computing, 40(2):B429–B456, 2018

    2018

  3. [11]

    Groups of diffeomorphisms and the motion of an incompressible fluid

    David G Ebin and Jerrold Marsden. Groups of diffeomorphisms and the motion of an incompressible fluid. Annals of Mathematics, 92(1):102–163, 1970

    1970

  4. [12]

    Thomas Fletcher, Conglin Lu, and Sarang Joshi

    P. Thomas Fletcher, Conglin Lu, and Sarang Joshi. Statistics of shape via principal geodesic analysis on Lie groups. In 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings., volume 1, pages I–I. IEEE, 2003

    2003

  5. [13]

    Approximation of Riemannian distances and applications to distance- based learning on manifolds

    Philipp Harms, Elodie Maignant, and Stefan Schlager. Approximation of Riemannian distances and applications to distance- based learning on manifolds. arXiv preprint :1904.11860, 2019

    1904 arXiv

  6. [14]

    Riemannian proximal gradient methods

    Wen Huang and Ke Wei. Riemannian proximal gradient methods. Mathematical Programming, 194(1):371–413, 2022

    2022

  7. [15]

    Intrinsic shape analysis: Geodesic PCA for Riemannian manifolds modulo isometric Lie group actions

    Stephan Huckemann, Thomas Hotz, and Axel Munk. Intrinsic shape analysis: Geodesic PCA for Riemannian manifolds modulo isometric Lie group actions. Statistica Sinica, pages 1–58, 2010

    2010

  8. [16]

    Principal component analysis for Riemannian manifolds, with an application to triangular shape spaces

    Stephan Huckemann and Herbert Ziezold. Principal component analysis for Riemannian manifolds, with an application to triangular shape spaces. Advances in Applied Probability, 38(2):299–319, 2006

    2006

  9. [17]

    Geometric hydrodynamics and infinite-dimensional Newton’s equations

    Boris Khesin, Gerard Misiołek, and Klas Modin. Geometric hydrodynamics and infinite-dimensional Newton’s equations. Bulletin of the American Mathematical Society, 58(3):377–442, 2021

    2021

  10. [18]

    A geometric study of Wasserstein spaces: Euclidean spaces

    Benoit Kloeckner. A geometric study of Wasserstein spaces: Euclidean spaces. Annali della Scuola Normale Superiore di Pisa-Classe di Scienze, 9(2):297–323, 2010

    2010

  11. [19]

    MNIST handwritten digit database

    Yann LeCun, Corinna Cortes, and CJ Burges. MNIST handwritten digit database. ATT Labs [Online]. Available: http://yann.lecun.com/exdb/mnist, 2, 2010

    2010

  12. [20]

    Gluing methods for quantitative stability of optimal transport maps

    Cyril Letrouit and Quentin Mérigot. Gluing methods for quantitative stability of optimal transport maps. arXiv preprint arXiv:2411.04908, 2024

    2024 arXiv

  13. [21]

    Eigenvalue continuity and Gersgorin’s theorem

    Chi-Kwong Li and Fuzhen Zhang. Eigenvalue continuity and Gersgorin’s theorem. Electronic Journal of Linear Algebra, 2019

    2019

  14. [22]

    Improved SQP and SLSQP algorithms for feasible path-based process optimisation

    Yingjie Ma, Xi Gao, Chao Liu, and Jie Li. Improved SQP and SLSQP algorithms for feasible path-based process optimisation. Computers & Chemical Engineering, page 108751, 2024. 10

    2024

  15. [23]

    Wasserstein Riemannian geometry of Gaussian densities.Information Geometry, 1:137–179, 2018

    Luigi Malagò, Luigi Montrucchio, and Giovanni Pistone. Wasserstein Riemannian geometry of Gaussian densities.Information Geometry, 1:137–179, 2018

    2018

  16. [24]

    A convexity principle for interacting gases

    Robert J McCann. A convexity principle for interacting gases. Advances in mathematics, 128(1):153–179, 1997

    1997

  17. [25]

    Geomstats: A Python package for Riemannian geometry in machine learning

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

    2020

  18. [26]

    Geometry of matrix decompositions seen through optimal transport and information geometry

    Klas Modin. Geometry of matrix decompositions seen through optimal transport and information geometry. Journal of Geometric Mechanics, 9(3):335–390, 2017

    2017

  19. [27]

    Mémoire sur la théorie des déblais et des remblais

    Gaspard Monge. Mémoire sur la théorie des déblais et des remblais. Mem. Math. Phys. Acad. Royale Sci., pages 666–704, 1781

  20. [28]

    The geometry of dissipative evolution equations: the porous medium equation

    Felix Otto. The geometry of dissipative evolution equations: the porous medium equation. Communications in Partial Differential Equations, 2001

    2001

  21. [29]

    Michael J.D. Powell. On search directions for minimization algorithms. Mathematical programming, 4:193–201, 1973

    1973

  22. [30]

    Ramsay and Bernard W

    James O. Ramsay and Bernard W. Silverman. Applied functional data analysis: methods and case studies. Springer, 2002

    2002

  23. [31]

    Landscape pictures dataset

    Arnaud Rougetet. Landscape pictures dataset. https://www.kaggle.com/datasets/arnaud58/landscape-pictures, 2020

    2020

  24. [32]

    On inequalities for moments and the covariance of monotone functions

    Klaus D Schmidt. On inequalities for moments and the covariance of monotone functions. Insurance: Mathematics and Economics, 55:91–95, 2014

    2014

  25. [33]

    Principal geodesic analysis for probability measures under the optimal transport metric

    Vivien Seguy and Marco Cuturi. Principal geodesic analysis for probability measures under the optimal transport metric. Advances in Neural Information Processing Systems, 28, 2015

    2015

  26. [34]

    The differential of the exponential map, Jacobi fields and exact principal geodesic analysis

    Stefan Sommer, François Lauze, and Mads Nielsen. The differential of the exponential map, Jacobi fields and exact principal geodesic analysis. CoRR, abs/1008.1902, 2010

    1902 arXiv

  27. [35]

    Optimization over geodesics for exact principal geodesic analysis

    Stefan Sommer, François Lauze, and Mads Nielsen. Optimization over geodesics for exact principal geodesic analysis. Advances in Computational Mathematics, 40:283–313, 2014

    2014

  28. [36]

    Wasserstein geometry of Gaussian measures

    Asuka Takatsu. Wasserstein geometry of Gaussian measures. Osaka J. Math., 48:1005–1026, 2011

    2011

  29. [37]

    Riemannian and stratified geometries on covariance and correlation matrices

    Yann Thanwerdas. Riemannian and stratified geometries on covariance and correlation matrices. PhD thesis, Université Côte d’Azur, 2022

    2022

  30. [38]

    Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J

    Pauli Virtanen, Ralf Gommers, Travis E. Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J. van der Walt, Matthew Brett, Joshua Wilson, K. Jarrod Millman, Nikolay Mayorov, Andrew R. J. Nelson, E...

    2020

  31. [39]

    A linear optimal transportation framework for quantifying and visualizing variations in sets of images

    Wei Wang, Dejan Slepˇcev, Saurav Basu, John A Ozolek, and Gustavo K Rohde. A linear optimal transportation framework for quantifying and visualizing variations in sets of images. International journal of computer vision, 101:254–269, 2013

    2013

  32. [40]

    3d shapenets: A deep representation for volumetric shapes, 2015

    Zhirong Wu, Shuran Song, Aditya Khosla, Fisher Yu, Linguang Zhang, Xiaoou Tang, and Jianxiong Xiao. 3d shapenets: A deep representation for volumetric shapes, 2015

    2015

  33. [41]

    1", 20 blue digits

    Juan Zhou, Kangkang Deng, Hongxia Wang, and Zheng Peng. Inexact Riemannian gradient descent method for nonconvex optimization. arXiv preprint arXiv:2409.11181, 2024. 11 A Additional experiments and figures In this section, we present additional figures to further explain the e...

    2024 arXiv

  34. [42]

    A2 − A1 = (T − I)A1 ∈ HorA1

  35. [43]

    LogΣ1 (Σ2) := dπA1 ((T − I)A1) = (T − I)Σ1 + Σ1(T − I)

  36. [44]

    BW2(Σ1, Σ2) = ∥LogΣ1 Σ2∥BW Σ1 = ∥(T − I)A1∥ where Log is the Riemannian logarithm map, ∥ · ∥BW Σ = q gBW Σ (·, ·) and ∥ · ∥is the Frobenius norm. This means that to compute the Bures-Wasserstein distance between two covariance matrices Σ1 and Σ2, one can consider any represent...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.