Pith. sign in

REVIEW 3 major objections 7 minor 36 references

Hilbert geometry of the symmetric positive-definite bicone: Application to the geometry of the extended Gaussian family

T0 review · 3 major / 7 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper proves that the extended Gaussian parameter space, viewed as a symmetric positive-definite bicone, carries an explicit Hilbert metric whose distances are read from four extreme eigenvalues, and shows its isometries are exactly ort

desk verdict The paper gets the Hilbert distance formula right, but the isometry classification rests on an unproven affine-reduction step. read the letter →

arxiv 2508.14369 v1 pith:J77H5T27 submitted 2025-08-20 cs.CG cs.LGmath.PR

Jacek Karwowski , Frank Nielsen This is my paper · ORCID
classification cs.CGcs.LGmath.PR
keywords HilbertgeometryextendedGaussiandistributionsvariance-precisionmodelsymmetricpositivesemi-definitebiconeBirkhoffprojectivedistancematrixMöbiustransformationsisometryclassificationconvexcones
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 studies the open bounded convex set VPM(n) = {0 ≺ X ≺ I} of symmetric matrices, which parameterizes the extended Gaussian family: ordinary Gaussians with covariance Σ correspond to the interior point X = Σ(I+Σ)^{-1}, and degenerate covariance or precision matrices appear on the boundary. The central result is an explicit Hilbert distance formula: for A, B in VPM(n), dH(A,B) is the logarithm of the ratio of the larger of two maximal eigenvalues to the smaller of two minimal eigenvalues. Because the formula uses only extreme eigenvalues, distances are computable in closed form. The paper also proves that for n > 1 the only distance-preserving transformations are the complement map X ↦ I-X and conjugation by orthogonal matrices X ↦ UᵀXU.

What carries the argument

The load-bearing object is the cone C_n = {(tX, t) : X ∈ VPM(n), t > 0} over the bicone. Hilbert distance on VPM is Birkhoff distance between rays of this cone; the paper computes the Birkhoff constants by solving Loewner-order inequalities, which collapse to ratios of extreme eigenvalues. For isometries, the paper shows C_n is not symmetric for n > 1, invokes the result that Hilbert isometries coincide with collineations in that case, and uses projective geometry to classify those collineations as linear cone automorphisms; a classification of symmetric-positive-definite-cone automorphisms then forces the orthogonal form.

What would settle it

Compute the Hilbert distance in VPM(2) directly from boundary intersections for a pair of noncommuting matrices and compare with the four-eigenvalue formula; a mismatch would refute Theorem 1. For the isometry claim, test the one-dimensional analog: on (0,1), x ↦ 2x/(1+x) preserves the cone over (0,1) and is a Hilbert isometry but is not affine; seek an analogous non-affine projective map preserving VPM(n) for n = 2. Existence of such a map in Isom(VPM(2)) outside the group generated by orthogonal conjugation and X ↦ I−X would refute Theorem 5.

Watch

Extended reading notes

Core claim

On the open bicone VPM(n) = {X ∈ Sym(n) : 0 ≺ X ≺ I}, the Hilbert distance between A and B is log( max( λmax(B^{-1}A), λmax((I−B)^{-1}(I−A)) ) / min( λmin(B^{-1}A), λmin((I−B)^{-1}(I−A)) ) ). The authors derive this by viewing Hilbert distance as Birkhoff projective distance on rays of the cone over VPM, where the required bounds reduce to Loewner-order inequalities A ⪯ λB and I−A ⪯ λ(I−B). They further prove that for n > 1 the isometry group is generated by X ↦ UᵀXU with U orthogonal and X ↦ I−X. Degenerate covariance and precision Gaussians lie on the boundary; the paper introduces an enlarged bicone VPM_epsilon that makes distances between such boundary points finite.

Load-bearing premise

The proof that there are no other isometries rests on the unstated claim that every projective automorphism of the cone over VPM induces an affine isomorphism on each horizontal matrix slice; if non-affine fractional-linear maps occur for n ≥ 2, the isometry classification could be missing a family, as happens already in dimension one.

Editorial extensions

If this is right

  • The Hilbert VPM distance needs only four extreme eigenvalues, so it is cheaper to compute than the affine-invariant Riemannian metric, which requires the full spectrum.
  • Boundary points corresponding to degenerate covariance or precision matrices have infinite Hilbert distance, but the enlarged bicone VPM_epsilon assigns them finite distances and lower-bounds the original distance.
  • Because straight lines are geodesics in Hilbert geometry, computational geometry primitives such as smallest enclosing balls and Voronoi diagrams transfer directly to the extended Gaussian parameter space.
  • The isometry invariance under orthogonal conjugation matches the invariance used in diffusion-tensor style processing, while the complement map I−X mirrors the covariance-precision duality.
  • The paper shows the Hilbert VPM distance is invariant under the two named transformations and, for n > 1, that these generate all its isometries.

Reading between the lines

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

  • Because only extreme eigenvalues are needed, the Hilbert VPM distance may be evaluated with a single extreme-eigenvalue routine per ratio matrix; benchmarking it against AIRM in high-dimensional Gaussian inference would be a natural next test.
  • The complement map X ↦ I−X swaps the two boundary cones, so the metric treats degenerate covariance and degenerate precision symmetrically; this may yield a natural loss for parameter estimation when both kinds of singular Gaussians are present.
  • The paper's enlarged bicone VPM_epsilon could define a practical 'nearly degenerate' metric with a well-posed optimization landscape over the full closed extended Gaussian family, though the paper leaves that application open.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. The paper studies the Hilbert geometry of the open bounded convex set VPM(n) = {X ∈ Sym(n) : 0 ≺ X ≺ I}, the interior of the variance-precision bicone parameterizing extended centered Gaussian families. Theorem 1 gives a closed-form Hilbert distance: dH(A,B)=log( max(λmax(B^{-1}A),λmax((I−B)^{-1}(I−A))) / min(λmin(B^{-1}A),λmin((I−B)^{-1}(I−A))) ). Propositions 7 and 8 record the isometries X↦I−X and X↦U^T XU, and Theorem 5 asserts that these generate the full isometry group for n>1. The derivation of Theorem 1 via Birkhoff's characterization is essentially correct, and the invariance proofs are correct. The isometry classification, however, depends on Theorem 4, whose proof contains an unjustified and in general false affine-reduction step; the 1D analogue provides a concrete non-affine projective automorphism. Thus the 'only these isometries' contribution is not proven as written.

Significance. If the missing step can be supplied, the paper is a useful contribution: it provides an explicit Hilbert metric on the variance-precision manifold that depends only on extreme eigenvalues, exhibits two natural invariance groups, and gives a plausible classification of isometries. The Birkhoff-cone derivation is reproducible from cited standard results and involves no fitted parameters or numerical tuning. The weaknesses are confined to the classification part, but that part is advertised as a main result, so the manuscript requires a substantial repair before it can be accepted.

major comments (3)
  1. [§5.2.1, proof of Theorem 4] The paragraph beginning 'Because L maps the origin...' asserts that a linear automorphism L of Cn induces an affine isomorphism on each projective slice. This is false. In the 1D analogue, the linear map L(x,t)=(2x,x+t) preserves the cone C1={(x,t):0<x<t}, but the induced projectivization on the chart t=1 is x↦2x/(1+x), a non-affine fractional linear map that is nonetheless a Hilbert isometry of (0,1). Therefore the application of Lemma 3 to the slice maps is not justified, and the rest of the proof—in particular the conclusion C=0—does not follow. Since Theorem 4 is the bridge from Corollary 1 to Theorem 5, the claim that the only isometries are O(n)-conjugation and complement is unsupported. The authors need to exclude 'matrix Möbius' automorphisms for n≥2 by a genuinely different argument (e.g., by showing that a projective automorphism of VPM(n) must preserve the boundary hypersurfac
  2. [§5.2.1, proof of Theorem 4 (slice chart)] The slice restriction is also not set up correctly. The chart identifying VPM(n) with the slice t=t0 of Cn is X↦(t0X,t0), not X↦(X,t0). With the latter, π∘L(·,t0)∘ιt0 need not map VPM(n) into itself, so Lemma 3 cannot be applied even if that map happened to be affine. The proof should fix the chart t=1 and work with the homogeneous scaling consistently.
  3. [Lemma 3] The proof of Lemma 3 does not establish the claimed dichotomy. It shows that no interior point can be mapped to 0, but that does not imply L(0) and L(I) must be the two corners. The later step 'substituting Y=0 ... and Y=I' uses points in the closure, not in the open set VPM(n), and requires a limiting argument that is not supplied. This can likely be repaired by working with the closure, but as written the lemma is incomplete.
minor comments (7)
  1. [Theorem 1] In the statement, 'minimal and maximal eigenvalues of the A^{-1}B matrix' should read B^{-1}A (the proof uses B^{-1}A).
  2. [Theorem 1 proof] In the computation of m, the sentence 'taking the infimum' should be 'taking the supremum', since m is defined by a supremum.
  3. [Definition 7 / Remark 2] VPM(n) is used for both the open manifold (Definition 7) and the closed model from [16] (Remark 2). Please adopt a consistent notation such as VPM and overline.
  4. [Lemma 3 proof] 'L is onto VPM(n)' should read 'onto VPM(n)' (closure); the argument with Y=0 and Y=I is a limiting argument.
  5. [Proposition 10] The phrase 'the infimum is achieved by letting Z→...' is imprecise, since the extremal Z lies in the closure of VPM(n); write 'attained in the limit'.
  6. [§6] The final sentence 'We leave these extensions ... for the next revision of this manuscript' is informal; replace with a standard future-work statement.
  7. [Theorem 5] The map X↦I−X is called 'inversion'; consider using 'complement' to avoid confusion with matrix inversion.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 1 follows directly from Birkhoff's external characterization, self-citations are background only, and the identified flaw in Theorem 4 is a proof gap, not a circular dependency.

full rationale

The paper's central distance formula (Theorem 1) is obtained by applying the Birkhoff characterization of the Hilbert metric (Proposition 2, citing Lemmens–Nussbaum [17]) to the explicitly constructed cone K(C) over the affine image of VPM. The required hypotheses—openness, convexity, boundedness—are proven in Proposition 3, and the eigenvalue computations in the proof of Theorem 1 are straightforward. No parameter is fitted to data, and no target quantity is used to define its own input. The invariance results (Propositions 7 and 8) are algebraic consequences of the formula. The isometry classification (Theorems 4 and 5) relies on Walsh's theorem [35] and Gowda–Sznajder–Tao [14], both external to the authors; the paper's own self-citations ([21]–[24]) appear only as background references for Hilbert geometry, not as load-bearing evidence. The proof of Theorem 4 does contain an unsupported assertion: that π∘L(·,t0)∘ιt0 is an affine isomorphism for each t0. This inference is not justified in the text and is false in the one-dimensional analog, as the projective map x ↦ 2x/(1+x) shows. However, this is a correctness gap in the classification argument, not circular reasoning: the claim is not derived from the conclusion it is meant to prove, and no self-citation or fitted input supplies the missing step. The distance formula and invariance claims remain unaffected. Therefore the circularity score is 0.

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

No fitted parameters and no invented physical entities. The bicone/boundary structure is James's classical variance-precision model (ref [16]); the VPM_ϵ enlargement is a plain homothety of the domain, not a new postulated entity. The central formula rests on the standard Birkhoff characterization; the isometry classification imports three external theorems plus one unproven affine-reduction step, which is the main fragility.

assumptions (5)
  • standard math Birkhoff's characterization: for an open bounded convex set C ⊂ V, the space P(K(C)) with Birkhoff metric is isometric to C with the Hilbert metric (Prop. 2, cited to Lemmens-Nussbaum [17]).
    Used to convert the Hilbert distance on VPM into a cone-order / Birkhoff computation, which yields Theorem 1.
  • domain assumption Walsh's theorem (Thm 2, cited [35], Cor. 1.4): for a non-symmetric (and non-Lorentzian) cone, the isometries of the projective Hilbert geometry coincide with collineations.
    External deep theorem. The quoted statement's symmetric-cone direction appears inconsistent with the inversion isometry on the projectivized PSD cone; the paper only uses the non-symmetric branch for C_n.
  • domain assumption The automorphism group of the SPD cone as a cone is the congruence group X → U^T X U, U ∈ GL(n) (Lemma 1, cited [14] Gowda-Sznajder-Tao).
    Used at the end of Theorem 4's proof to get X → P^T X P with P ∈ O(n).
  • standard math Fundamental theorem of projective geometry / Shiffman's extension (Thm 3, Prop 11, cited [29]): collineations of open regions in RP^n extend to projective linear maps.
    Bridges collineations of VPM to linear cone isomorphisms.
  • ad hoc to paper The projectivization π∘L(·,t0)∘ι of a linear cone automorphism is affine (asserted in the proof of Theorem 4).
    This is the load-bearing unproven step; false for n=1. If it fails for n ≥ 2, the classification of cone automorphisms (and hence the isometry group) may be incomplete.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hilbert geometry of the symmetric positive-definite bicone: Application to the geometry of the extended Gaussian family." pith.science (2026). https://pith.science/paper/J77H5T27

@misc{pith2026250814369,
  author       = {Pith},
  title        = {Pith review of: Hilbert geometry of the symmetric positive-definite bicone: Application to the geometry of the extended Gaussian family},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J77H5T27}},
  note         = {Machine review of arXiv:2508.14369}
}
read the original abstract

The extended Gaussian family is the closure of the Gaussian family obtained by completing the Gaussian family with the counterpart elements induced by degenerate covariance or degenerate precision matrices, or a mix of both degeneracies. The parameter space of the extended Gaussian family forms a symmetric positive semi-definite matrix bicone, i.e. two partial symmetric positive semi-definite matrix cones joined at their bases. In this paper, we study the Hilbert geometry of such an open bounded convex symmetric positive-definite bicone. We report the closed-form formula for the corresponding Hilbert metric distance and study exhaustively its invariance properties. We also touch upon potential applications of this geometry for dealing with extended Gaussian distributions.

Figures

Figures reproduced from arXiv: 2508.14369 by the authors.

Figure 1
Figure 1. The parameter space of the extended Gaussian family forms a closed double cone. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The parameter space of VPM(2) can be viewed as a double Lorentz cone in R 3 : Three views of the double Lorentz cone. uniqueness of geodesics depending on the smoothness of the boundary ∂C of the domain [22]. When clear from context, we shall write dH instead of d C H for sake of brevity. Birkhoff geometry [6, 18] is a geometry defined on any open regular cone K (i.e., convex pointed cone). The cone defines a partia… view at source ↗
Figure 3
Figure 3. The parameter space of VPMϵ(2) (viewed as a Lorentz bicone in R 3 ): Three views of the boundary of the Lorentz bicone with VPMϵ (in blue) encapsulating VPM (in black). Next, we may consider the full Gaussian family {N(µ, Σ) : µ ∈ R n , Σ ∈ PD(n)} instead of zero￾centered Gaussians by embedding full Gaussian distributions into the SPD cone of higher dimension following the Calvo-Oller embedding [7]: (µ, Σ) 7→∈ Σ + µ… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Fine approximation of the smallest enclosing ball (SEB, in red) with respect to the Hilbert [PITH_FULL_IMAGE:figures/full_fig_p019_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

  1. [1]

    On approximating the Riemannian 1-center

    Marc Arnaudon and Frank Nielsen. On approximating the Riemannian 1-center. Computa- tional Geometry, 46(1):93–104, 2013

    work page 2013

  2. [2]

    Log-Euclidean metrics for fast and simple calculus on diffusion tensors

    Vincent Arsigny, Pierre Fillard, Xavier Pennec, and Nicholas Ayache. Log-Euclidean metrics for fast and simple calculus on diffusion tensors. Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine , 56(2):411–421, 2006

    work page 2006

  3. [3]

    Smaller core-sets for balls

    Mihai Bˆ adoiu and Kenneth L Clarkson. Smaller core-sets for balls. In Proceedings of the fourteenth annual ACM-SIAM symposium on Discrete algorithms , pages 801–802, 2003

    work page 2003

  4. [4]

    On The Heine-Borel Property and Minimum Enclosing Balls

    Hridhaan Banerjee, Carmen Isabel Day, Megan Hunleth, Sarah Hwang, Auguste H Gezalyan, Olya Golovatskaia, Nithin Parepally, Lucy Wang, and David M Mount. On the Heine-Borel property and minimum enclosing balls. arXiv preprint arXiv:2412.17138 , 2024. 18 Figure 4: Fine approximation of the smallest enclosing ball (SEB, in red) with respect to the Hilbert VP...

    work page Pith review arXiv 2024

  5. [5]

    Alan F. Beardon. The Klein, Hilbert and Poincar´ e metrics of a domain. Journal of computa- tional and applied mathematics , 105(1-2):155–162, 1999

    work page 1999

  6. [6]

    Extensions of Jentzsch’s theorem

    Garrett Birkhoff. Extensions of Jentzsch’s theorem. Transactions of the American Mathemat- ical Society, 85(1):219–227, 1957

    work page 1957

  7. [7]

    A distance between elliptical distributions based in an embed- ding into the Siegel group

    Miquel Calvo and Josep M Oller. A distance between elliptical distributions based in an embed- ding into the Siegel group. Journal of Computational and Applied Mathematics , 145(2):319– 334, 2002

    work page 2002

  8. [8]

    Stochastic control liaisons: Richard Sinkhorn meets Gaspard Monge on a Schr¨ odinger bridge.Siam Review, 63(2):249–313, 2021

    Yongxin Chen, Tryphon T Georgiou, and Michele Pavon. Stochastic control liaisons: Richard Sinkhorn meets Gaspard Monge on a Schr¨ odinger bridge.Siam Review, 63(2):249–313, 2021

    work page 2021

Show all 36 references
  1. [9]

    On Hilbert’s Metric for Simplices

    Pierre de la Harpe. On Hilbert’s Metric for Simplices. In Graham A. Niblo and Martin A. Roller, editors, Geometric Group Theory , volume 1 of London Mathematical Society Lecture Note Series , pages 97–119. Cambridge University Press, Cambridge, 1993

    1993

  2. [10]

    On representing the positive semidefinite cone using the second-order cone

    Hamza Fawzi. On representing the positive semidefinite cone using the second-order cone. Mathematical Programming, 175(1):109–118, 2019

    2019

  3. [11]

    Delaunay Triangulations in the Hilbert Metric

    Auguste H Gezalyan, Soo H Kim, Carlos Lopez, Daniel Skora, Zofia Stefankovic, and David M Mount. Delaunay Triangulations in the Hilbert Metric. In 19th Scandinavian Symposium on Algorithm Theory, 2024

    2024

  4. [12]

    Gezalyan and David M

    Auguste H. Gezalyan and David M. Mount. Voronoi Diagrams in the Hilbert Metric. In Erin W. Chambers and Joachim Gudmundsson, editors, 39th International Symposium on 19 Computational Geometry, SoCG 2023, June 12-15, 2023, Dallas, Texas, USA , volume 258 of LIPIcs, pages 35:1–3...

    2023

  5. [13]

    Geometric structures on manifolds, volume 227

    William M Goldman. Geometric structures on manifolds, volume 227. American Mathematical Society, 2022

    2022

  6. [14]

    Seetharama Gowda, Roman Sznajder, and Jiyuan Tao

    M. Seetharama Gowda, Roman Sznajder, and Jiyuan Tao. The automorphism group of a completely positive cone and its Lie algebra.Linear Algebra and its Applications, 438(10):3862– 3871, May 2013

    2013

  7. [15]

    ¨Uber die gerade linie als k¨ urzeste verbindung zweier punkte: Aus einem an herrn f

    David Hilbert. ¨Uber die gerade linie als k¨ urzeste verbindung zweier punkte: Aus einem an herrn f. klein gerichteten briefe. Mathematische Annalen, 46(1):91–96, 1895

  8. [16]

    The variance information manifold and the functions on it

    Alan Treleven James. The variance information manifold and the functions on it. In Multi- variate Analysis–III, pages 157–169. Elsevier, 1973

    1973

  9. [17]

    Birkhoff’s version of Hilbert’s metric and its applications in analysis

    Bas Lemmens and Roger Nussbaum. Birkhoff’s version of Hilbert’s metric and its applications in analysis. Handbook of Hilbert Geometry , pages 275–303, 2014

    2014

  10. [18]

    Nussbaum

    Bas Lemmens and Roger D. Nussbaum. Birkhoff’s version of Hilbert’s metric and its applica- tions in analysis. In Athanase Papadopoulos and Marc Troyanov, editors, Handbook of Hilbert geometry, pages 275–303. EMS press, 2014

    2014

  11. [19]

    Isometries of polyhedral Hilbert geometries

    Bas Lemmens and Cormac Walsh. Isometries of polyhedral Hilbert geometries. Technical Report arXiv:0904.3306, April 2009

    2009 arXiv

  12. [20]

    Differ- ential geometry with extreme eigenvalues in the positive semidefinite cone

    Cyrus Mostajeran, Natha¨ el Da Costa, Graham Van Goffrier, and Rodolphe Sepulchre. Differ- ential geometry with extreme eigenvalues in the positive semidefinite cone. SIAM Journal on Matrix Analysis and Applications , 45(2):1089–1113, 2024

    2024

  13. [21]

    The Siegel–Klein disk: Hilbert geometry of the Siegel disk domain

    Frank Nielsen. The Siegel–Klein disk: Hilbert geometry of the Siegel disk domain. Entropy, 22(9):1019, 2020

    2020

  14. [22]

    On balls in a Hilbert polygonal geometry

    Frank Nielsen and Laetitia Shao. On balls in a Hilbert polygonal geometry. In 33rd Interna- tional Symposium on Computational Geometry (SoCG), pages 67–1. Schloss Dagstuhl–Leibniz- Zentrum f¨ ur Informatik, 2017

    2017

  15. [23]

    Clustering in Hilbert’s projective geometry: The case studies of the probability simplex and the elliptope of correlation matrices

    Frank Nielsen and Ke Sun. Clustering in Hilbert’s projective geometry: The case studies of the probability simplex and the elliptope of correlation matrices. In Geometric Structures of Information, pages 297–331. Springer, 2018

    2018

  16. [24]

    Non-linear embeddings in Hilbert simplex geometry

    Frank Nielsen and Ke Sun. Non-linear embeddings in Hilbert simplex geometry. In Topological, Algebraic and Geometric Learning Workshops 2023 , pages 254–266. PMLR, 2023

    2023

  17. [25]

    Hilbert’s projective metric and iterated nonlinear maps , volume 1

    Roger D Nussbaum. Hilbert’s projective metric and iterated nonlinear maps , volume 1. Amer- ican Mathematical Soc., 1988

    1988

  18. [26]

    Doubly autoparallel structure on positive definite matrices and its applica- tions

    Atsumi Ohara. Doubly autoparallel structure on positive definite matrices and its applica- tions. In International Conference on Geometric Science of Information (GSI), pages 251–260. Springer, 2019. 20

    2019

  19. [27]

    Dualistic differential geometry of positive definite matrices and its applications to related problems

    Atsumi Ohara, Nobuhide Suda, and Shun-ichi Amari. Dualistic differential geometry of positive definite matrices and its applications to related problems. Linear Algebra and Its Applications, 247:31–53, 1996

    1996

  20. [28]

    A Riemannian framework for tensor computing

    Xavier Pennec, Pierre Fillard, and Nicholas Ayache. A Riemannian framework for tensor computing. International Journal of computer vision , 66(1):41–66, 2006

    2006

  21. [29]

    Synthetic projective geometry and Poincar´ e’s theorem on automorphisms of the ball

    Bernard Shiffman. Synthetic projective geometry and Poincar´ e’s theorem on automorphisms of the ball. 1995

    1995

  22. [30]

    A Category for Unifying Gaussian Probability and Non- determinism

    Dario Stein and Richard Samuelson. A Category for Unifying Gaussian Probability and Non- determinism. In 10th Conference on Algebra and Coalgebra in Computer Science (CALCO 2023), pages 13–1. Schloss Dagstuhl–Leibniz-Zentrum f¨ ur Informatik, 2023

    2023

  23. [31]

    Loewner order in terms of eigenvalues

    Viktor Stein. Loewner order in terms of eigenvalues. Mathematics Stack Exchange. URL:https://math.stackexchange.com/q/4606814 (version: 2022-12-28)

    2022

  24. [32]

    O(n)-invariant Riemannian metrics on SPD matrices

    Yann Thanwerdas and Xavier Pennec. O(n)-invariant Riemannian metrics on SPD matrices. Linear Algebra and its Applications , 661:163–201, 2023

    2023

  25. [33]

    Simplicial faces of the set of correlation matrices

    Joel A Tropp. Simplicial faces of the set of correlation matrices. Discrete & Computational Geometry, 60(2):512–529, 2018

    2018

  26. [34]

    Support Vector Machines in the Hilbert Geometry

    Julian Vanecek, Auguste H Gezalyan, and David M Mount. Support Vector Machines in the Hilbert Geometry. In 31st Annual Fall Workshop onComputational Geometry , 2024

    2024

  27. [35]

    Gauge-reversing maps on cones, and Hilbert and Thompson isometries

    Cormac Walsh. Gauge-reversing maps on cones, and Hilbert and Thompson isometries. Ge- ometry & Topology, 22(1):55–104, October 2017

    2017

  28. [36]

    Open stochastic systems

    Jan C Willems. Open stochastic systems. IEEE Transactions on Automatic Control, 58(2):406– 421, 2012. 21

    2012

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.