pith. sign in

arxiv: 2604.27427 · v1 · submitted 2026-04-30 · 🧮 math.OC

A Geometric Perspective on Polynomially Solvable Convex Maximization

Shaoning Han , Liangju Li , Yongchun Li This is my paper

Pith reviewed 2026-05-07 08:18 UTC · model grok-4.3

classification 🧮 math.OC
keywords convex maximizationcomonotonicitypolynomial solvabilityfixed-rank optimizationmatroid maximizationsparse PCAgeometric optimization
0
0 comments X

The pith

Comonotonicity of the feasible region makes fixed-rank convex maximization polynomially solvable.

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

Convex maximization problems are generally NP-hard, even for objectives of fixed low rank. The paper introduces comonotonicity, a geometric structural property of the feasible region, and shows it is the key condition that changes the complexity. Mathematical characterizations of the property are given, followed by a unified enumerative framework that solves the problems in polynomial time under comonotonicity plus mild assumptions. The same framework recovers several earlier tractability results that had required separate proofs, such as fixed-rank convex matroid maximization and sparse principal component analysis. For the subclass of standard comonotone regions, a lifting technique improves the running-time bound by a square-root factor.

Core claim

Under the comonotonicity property of the feasible region together with mild additional assumptions, fixed-rank convex maximization admits a unified enumerative algorithm that runs in polynomial time. This framework recovers known results such as fixed-rank convex matroid maximization and sparse principal component analysis without needing separate proofs. For standard comonotone regions, a lifting technique yields a square-root improvement in the running time.

What carries the argument

comonotonicity, a structural property of the feasible region that enables an enumerative polynomial-time algorithm for fixed-rank convex objectives

If this is right

  • Fixed-rank convex maximization becomes polynomially solvable over any comonotone feasible region.
  • A single enumerative framework recovers and unifies prior separate proofs for matroid maximization and SPCA.
  • A lifting technique gives a square-root improvement in complexity for standard comonotone regions.
  • The framework applies directly to SPCA and its variants with demonstrated effectiveness.

Where Pith is reading between the lines

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

  • Comonotonicity may identify tractable cases in other families of nonconvex optimization beyond convex maximization.
  • Algorithms for real-world problems could first test for comonotonicity to decide whether the enumerative method is guaranteed to succeed.
  • Similar geometric properties on feasible sets might extend polynomial solvability results to objectives of higher rank.

Load-bearing premise

The feasible region must satisfy the comonotonicity property in addition to the mild assumptions needed for the enumerative framework to run in polynomial time.

What would settle it

A concrete feasible region that meets the comonotonicity definition but for which no polynomial-time algorithm exists for some fixed-rank convex maximization instance would disprove the central claim.

Figures

Figures reproduced from arXiv: 2604.27427 by Liangju Li, Shaoning Han, Yongchun Li.

Figure 1
Figure 1. Figure 1: Examples of comonotone sets in R 2 2.1. Properties of standard comonotone sets In this subsection, we study properties of standard comonotone sets. We begin with the planar case to build geometric intuition. While restricted to 𝑛 = 2, it provides a clear lens for understanding the restrictions imposed by standard comonotonicity, or equivalently, the structural information it carries. As shown below, standa… view at source ↗
Figure 2
Figure 2. Figure 2: Examples of standard comonotone sets in R 2 In R 2 , ordering information is encoded by only two directions (𝑥1 +𝑥2 or −𝑥1 −𝑥2) as indicated by Proposition 1. Unfortunately, the simplification does not directly extend to high dimensions. To see this, consider the distributed lat￾tice X = {𝑥 ∈ {0,1} 3 : 𝑥1 ≥ 𝑥2 ≥ 𝑥3}. On the one hand, 0 ∈ argmax{−𝑒 ⊤𝑥 : 𝑥 ∈ X} and 𝑒 ∈ argmax{𝑒 ⊤𝑥 : 𝑥 ∈ X}, and moreover {0, … view at source ↗
Figure 3
Figure 3. Figure 3: A theoretical framework of complexity analysis. Beyond its broad applicability, the proposed framework is also highly adaptable. When additional problem structure is available (e.g., standard comonotonicity), each step can be tailored accordingly to tighten the complexity bounds. To avoid redundancy, we do not repeat the full framework for special families of X in the subsequent sections; instead, we highl… view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of hyperplane arrangements (𝑟 = 3, 𝑛 = 4, slice 𝑐3 = 1) DEFINITION 2. For any vector 𝑐 ∈ R 𝑟 and parameters 𝜆 ≤𝜆, define the set Q  𝑐,𝜆, 𝜆 ≜ {𝑥 ∈ R 𝑛 : (a) sign constraints, (b) ordering constraints} , where (a) Sign constraints: For all 𝑖 ∈ [𝑛], 𝑥𝑖 > 0 if (𝐴 ⊤𝑐)𝑖 >𝜆, 𝑥𝑖 = 0 if 𝜆 < (𝐴 ⊤𝑐)𝑖 <𝜆, 𝑥𝑖 < 0 if (𝐴 ⊤𝑐)𝑖 < 𝜆, 𝑥𝑖 ≥ 0 if (𝐴 ⊤𝑐)𝑖 =𝜆 and 𝜆 > 𝜆, 𝑥𝑖 ≤ 0 if (𝐴 ⊤𝑐)𝑖 =𝜆 and 𝜆 > 𝜆. (b) Ordering c… view at source ↗
Figure 5
Figure 5. Figure 5: The optimal solution cannot be exposed in X. 5.2. Affine restriction of binary comonotone sets In this subsection, we relax the assumption that X is a comonotone set. Instead, we study the case where X =S ∩ P satisfying Assumption 3 below. ASSUMPTION 3. conv(S∩P) = conv(S)∩P, whereS ⊆ {0,1} 𝑛 is comonotone and P = {𝑥 ∈ R 𝑛 : 𝑀𝑥 ≤ 𝑏} for some 𝑀 ∈ R 𝑚×𝑛 and 𝑏 ∈ R 𝑚 view at source ↗
read the original abstract

Convex maximization encompasses a broad class of optimization problems and is generally NP-hard, even for low-rank objectives. This paper investigates structural conditions under which convex maximization becomes polynomially solvable. From a geometric perspective, we introduce comonotonicity, a structural property of the feasible region crucial for problem tractability, and establish mathematical characterizations of this property. Under comonotonicity and mild additional assumptions, we develop a unified enumerative framework showing that fixed-rank convex maximization is polynomially solvable. This viewpoint recovers several known tractability results that previously required separate analyses, such as fixed-rank convex matroid maximization and sparse principal component analysis (SPCA). Furthermore, for the more structured class of standard comonotone feasible regions, we refine the analysis via a lifting technique to achieve a square-root improvement in the complexity bound. Finally, applications to SPCA and its variants illustrate the broad applicability and effectiveness of the proposed framework.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 4 minor

Summary. The paper claims that convex maximization, generally NP-hard even for low-rank objectives, becomes polynomially solvable under a newly introduced geometric property called comonotonicity of the feasible region, along with mild additional assumptions. It provides mathematical characterizations of comonotonicity and develops a unified enumerative framework for fixed-rank convex maximization. This framework recovers known tractable cases such as fixed-rank convex matroid maximization and sparse principal component analysis (SPCA). For standard comonotone feasible regions, a lifting technique achieves a square-root improvement in the complexity bound. Applications to SPCA and its variants are presented to illustrate the framework.

Significance. If the characterizations and enumerative framework hold, this work offers a significant unified geometric perspective on polynomially solvable convex maximization problems. The recovery of disparate known results (matroid maximization, SPCA) as special instances within a single framework is a notable strength, as is the explicit complexity analysis of the enumeration and the lifting technique for improved bounds. The paper provides mathematical characterizations, a complexity analysis, and a falsifiable structural property (comonotonicity) that could guide further research in optimization.

minor comments (4)
  1. [Abstract and Introduction] The abstract and introduction refer to 'mild additional assumptions' without a concise summary or pointer to their precise statement in the main theorems; adding this would improve accessibility for readers.
  2. [Section on characterizations of comonotonicity] The characterizations of comonotonicity would benefit from an illustrative low-dimensional example or diagram showing how the geometric property manifests in a simple feasible region.
  3. [Applications section] In the applications to SPCA, the discussion of effectiveness would be strengthened by including a brief complexity comparison between the proposed enumerative framework and existing specialized SPCA solvers.
  4. [Throughout the manuscript] Notation for the rank parameter and the enumeration size should be consistently defined early and used uniformly across theorems and complexity statements.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive and constructive report, including the favorable assessment of the significance of the comonotonicity framework and its ability to unify results on matroid maximization and SPCA. The recommendation for minor revision is noted. No specific major comments were provided in the report, so we have no points requiring rebuttal or substantive revision at this time. Any minor editorial or typographical issues will be addressed in the revised manuscript.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper defines a new geometric property (comonotonicity) of the feasible region, supplies explicit mathematical characterizations of it, and then proves that fixed-rank convex maximization is polynomially solvable via an enumerative procedure under this property together with mild additional assumptions. The framework is shown to subsume known tractable cases (fixed-rank convex matroid maximization, SPCA) as special instances rather than using those cases to justify the general claim. No parameter fitting, self-definitional loops, load-bearing self-citations, or renaming of prior results as new derivations appear; the derivation chain consists of independent definitions followed by complexity analysis and recovery of special cases.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Only the abstract is available; the paper relies on standard convex optimization axioms and the newly defined comonotonicity property whose independent justification is not visible.

axioms (1)
  • domain assumption Standard assumptions of convex optimization (convex objective, convex feasible set) hold.
    Invoked implicitly as background for the maximization problem class.
invented entities (1)
  • comonotonicity no independent evidence
    purpose: Structural property of the feasible region that enables polynomial-time enumeration for fixed-rank convex maximization.
    Newly introduced geometric property whose definition and characterizations form the core of the tractability result.

pith-pipeline@v0.9.0 · 5455 in / 1284 out tokens · 34971 ms · 2026-05-07T08:18:55.397131+00:00 · methodology

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

84 extracted references · 84 canonical work pages

  1. [1]

    Bioinformatics27(21):3029–3035

    Allen GI, Maleti´c-Savati´c M (2011) Sparse non-negative generalized PCA with applications to metabolomics. Bioinformatics27(21):3029–3035

    work page 2011

  2. [2]

    Asteris M, Papailiopoulos D, Dimakis A (2014) Nonnegative sparse PCA with provable guarantees.International Conference on Machine Learning, 1728–1736 (PMLR)

    work page 2014

  3. [3]

    Asteris M, Papailiopoulos D, Kyrillidis A, Dimakis AG (2015) Sparse PCA via bipartite matchings.Advances in Neural Information Processing Systems28

    work page 2015

  4. [4]

    Asteris M, Papailiopoulos DS, Karystinos GN (2014) The sparse principal component of a constant-rank matrix.IEEE Transactions on Information Theory60(4):2281–2290

    work page 2014

  5. [5]

    Atamt¨ urk A, Narayanan V (2009) The submodular knapsack polytope.Discrete Optimization6(4):333–344

    work page 2009

  6. [6]

    Balas E (2018)Disjunctive programming(Springer)

    work page 2018

  7. [7]

    Bector C (1970) Some aspects of quasi-convex programming.ZAMM-Journal of Applied Mathematics and Mechanics/Zeitschrift f¨ur Angewandte Mathematik und Mechanik50(7-9):495–497

    work page 1970

  8. [8]

    Benson HP (1985) A finite algorithm for concave minimization over a polyhedron.Naval Research Logistics Quarterly 32(1):165–177

    work page 1985

  9. [9]

    Benson HP (1995) Concave minimization: theory, applications and algorithms.Handbook of global optimization, 43–148 (Springer)

    work page 1995

  10. [10]

    Revue franc ¸aise d’automatique informatique recherche op´erationnelle

    Bereanu B (1972) Quasi-convexity, strictly quasi-convexity and pseudo-convexity of composite objective functions. Revue franc ¸aise d’automatique informatique recherche op´erationnelle. Math´ematique6(R1):15–26

    work page 1972

  11. [11]

    Calinescu G, Chekuri C, Pal M, Vondr´ak J (2011) Maximizing a monotone submodular function subject to a matroid constraint.SIAM Journal on Computing40(6):1740–1766

    work page 2011

  12. [12]

    Chandrasekaran R, Aneja Y, Nair K (1981) Minimal cost-reliability ratio spanning tree.North-Holland Mathematics Studies, volume 59, 53–60 (Elsevier)

    work page 1981

  13. [13]

    Conforti M, Cornu´ejols G, Zambelli G (2010) Extended formulations in combinatorial optimization.4OR8(1):1–48

    work page 2010

  14. [14]

    Journal of Pure and Applied Algebra213(8):1569–1577

    De Loera JA, Hemmecke R, Onn S, Rothblum U, Weismantel R (2009) Convex integer maximization via graver bases. Journal of Pure and Applied Algebra213(8):1569–1577

    work page 2009

  15. [15]

    De Loera JA, Hemmecke R, Onn S, Weismantel R (2008) N-fold integer programming.Discrete Optimization 5(2):231–241

    work page 2008

  16. [16]

    Del Pia A (2023) Sparse PCA on fixed-rank matrices.Mathematical Programming198(1):139–157. 32

    work page 2023

  17. [17]

    Dey SS, Molinaro M, Wang G (2023) Solving sparse principal component analysis with global support.Mathematical Programming199(1):421–459

    work page 2023

  18. [18]

    Deza A, Manoussakis G, Onn S (2018) Primitive zonotopes.Discrete & Computational Geometry60(1):27–39

    work page 2018

  19. [19]

    Deza A, Pournin L, Rakotonarivo R (2021) The vertices of primitive zonotopes.Contemporary Mathematics764:71–81

    work page 2021

  20. [20]

    Edelman PH, Jamison RE (1985) The theory of convex geometries.Geometriae Dedicata19(3):247–270

    work page 1985

  21. [21]

    Edelsbrunner H (1987)Algorithms in combinatorial geometry, volume 10 (Springer Science & Business Media)

    work page 1987

  22. [22]

    SIAM Journal on Computing15(2):341–363

    Edelsbrunner H, O’Rourke J, Seidel R (1986) Constructing arrangements of lines and hyperplanes with applications. SIAM Journal on Computing15(2):341–363

    work page 1986

  23. [23]

    Edelsbrunner H, Seidel R, Sharir M (1993) On the zone theorem for hyperplane arrangements.SIAM Journal on Computing22(2):418–429

    work page 1993

  24. [24]

    Edmonds J (1970) Submodular functions, matroids, and certain polyhedra.Combinatorial Structures and Their Applications, 69–87 (Gordon and Breach)

    work page 1970

  25. [25]

    Edmonds J (1971) Matroids and the greedy algorithm.Mathematical programming1(1):127–136

    work page 1971

  26. [26]

    Giannessi F, Niccolucci F (1976) Connections between nonlinear and integer programming problems.Symposia Mathematica, volume 19, 161–176 (Academic Press New Y ork)

    work page 1976

  27. [27]

    Goemans MX (2015) Smallest compact formulation for the permutahedron.Mathematical Programming153(1):5–11

    work page 2015

  28. [28]

    G´omez A, Neto J (2025) Outlier detection in regression: conic quadratic formulations.INFORMS Journal on Computing

    work page 2025

  29. [29]

    Operations Research Letters41(2):191–196

    Goyal V , Ravi R (2013) An FPTAS for minimizing a class of low-rank quasi-concave functions over a convex set. Operations Research Letters41(2):191–196

    work page 2013

  30. [30]

    Gretton A, Borgwardt KM, Rasch MJ, Sch¨olkopf B, Smola A (2012) A kernel two-sample test.Journal of Machine Learning Research13(1):723–773

    work page 2012

  31. [31]

    Han S, G ´omez A (2025) Compact extended formulations for low-rank functions with indicator variables.Mathematics of Operations Research50(3):1992–2016

    work page 2025

  32. [32]

    Mathematical Programming202:95–134

    Han S, G´omez A, Atamt¨ urk A (2023) 2×2-convexifications for convex quadratic optimization with indicator variables. Mathematical Programming202:95–134

    work page 2023

  33. [33]

    (1952)Inequalities(Cambridge University Press)

    Hardy GH, Littlewood JE, P´olya G, P´olya G, et al. (1952)Inequalities(Cambridge University Press)

    work page 1952

  34. [34]

    Hare WL, Lewis AS (2004) Identifying active constraints via partial smoothness and prox-regularity.Journal of Convex Analysis11(2):251–266

    work page 2004

  35. [35]

    Hashizume S, Fukushima M, Katoh N, Ibaraki T (1987) Approximation algorithms for combinatorial fractional programming problems.Mathematical programming37(3):255–267

    work page 1987

  36. [36]

    Hassin R, Tamir A (1989) Maximizing classes of two-parameter objectives over matroids.Mathematics of Operations Research14(2):362–375

    work page 1989

  37. [37]

    Helman P, Moret BM, Shapiro HD (1993) An exact characterization of greedy structures.SIAM Journal on Discrete Mathematics6(2):274–283

    work page 1993

  38. [38]

    Horst R, T uy H (2013)Global optimization: Deterministic approaches(Springer Science & Business Media)

    work page 2013

  39. [39]

    H¨ossjer O (1995) Exact computation of the least trimmed squares estimate in simple linear regression.Computational Statistics & Data Analysis19(3):265–282

    work page 1995

  40. [40]

    Hotelling H (1933) Analysis of a complex of statistical variables into principal components.Journal of educational psychology24(6):417. 33

    work page 1933

  41. [41]

    Jeffers JN (1967) T wo case studies in the application of principal component analysis.Journal of the Royal Statistical Society: Series C (Applied Statistics)16(3):225–236

    work page 1967

  42. [42]

    Kalra S, Adnan M, Taylor G, Tizhoosh HR (2020) Learning permutation invariant representations using memory networks.European conference on computer vision, 677–693 (Springer)

    work page 2020

  43. [43]

    Kim J, Tawarmalani M, Richard JPP (2022) Convexification of permutation-invariant sets and an application to sparse principal component analysis.Mathematics of Operations Research47(4):2547–2584

    work page 2022

  44. [44]

    Klouda K (2015) An exact polynomial time algorithm for computing the least trimmed squares estimate.Computational Statistics & Data Analysis84:27–40

    work page 2015

  45. [45]

    Korte B, Lov´asz L (1984) Greedoids-a structural framework for the greedy algorithm.Progress in combinatorial optimization, 221–243 (Elsevier)

    work page 1984

  46. [46]

    Lee J, Lee Y, Kim J, Kosiorek A, Choi S, Teh YW (2019) Set transformer: A framework for attention-based permutation-invariant neural networks.International conference on machine learning, 3744–3753 (PMLR)

    work page 2019

  47. [47]

    Li Y , Xie W (2024) Beyond symmetry: Best submatrix selection for the sparse truncated svd.Mathematical Programming 208(1):1–50

    work page 2024

  48. [48]

    Li Y , Xie W (2025) Exact and approximation algorithms for sparse principal component analysis.INFORMS Journal on Computing37(3):582–602

    work page 2025

  49. [49]

    Mathematical Programming1–58

    Li Y, Xie W (2025) On the partial convexification for low-rank spectral optimization: rank bounds and algorithms. Mathematical Programming1–58

    work page 2025

  50. [50]

    Lov ´asz L (1980) Matroid matching and some applications.Journal of Combinatorial Theory, Series B28(2):208–236

    work page 1980

  51. [51]

    Magdon-Ismail M (2017) NP-hardness and inapproximability of sparse PCA.Information Processing Letters126:35–38

    work page 2017

  52. [52]

    Mangasarian OL (1994)Nonlinear programming(SIAM)

    work page 1994

  53. [53]

    Martin RK, Rardin RL, Campbell BA (1990) Polyhedral characterization of discrete dynamic programming.Operations research38(1):127–138

    work page 1990

  54. [54]

    Megiddo N (1978) Combinatorial optimization with rational objective functions.Proceedings of the tenth annual ACM symposium on Theory of computing, 1–12

    work page 1978

  55. [55]

    Mittal S, Schulz AS (2013) An FPTAS for optimizing a class of low-rank functions over a polytope.Mathematical Programming141(1):103–120

    work page 2013

  56. [56]

    Moghaddam B, Weiss Y , Avidan S (2005) Spectral bounds for sparse PCA: Exact and greedy algorithms.Advances in neural information processing systems18

    work page 2005

  57. [57]

    Montanari A, Richard E (2015) Non-negative principal component analysis: Message passing algorithms and sharp asymptotics.IEEE Transactions on Information Theory62(3):1458–1484

    work page 2015

  58. [58]

    Mor´e JJ, Sorensen DC (1983) Computing a trust region step.SIAM Journal on scientific and statistical computing 4(3):553–572

    work page 1983

  59. [59]

    Algorithmica69(1):148–183

    Mount DM, Netanyahu NS, Piatko CD, Silverman R, Wu AY (2014) On the least trimmed squares estimator. Algorithmica69(1):148–183

    work page 2014

  60. [60]

    Murota K (2003)Discrete Convex Analysis(SIAM)

    work page 2003

  61. [61]

    Murty KG, Kabadi SN (1987) Some NP-complete problems in quadratic and nonlinear programming.Mathematical programming39(2):117–129

    work page 1987

  62. [62]

    Nocedal J, Wright SJ (2006)Numerical Optimization(New Y ork: Springer), 2nd edition. 34

    work page 2006

  63. [63]

    Oberlin C, Wright SJ (2006) Active set identification in nonlinear programming.SIAM Journal on Optimization 17(2):577–605

    work page 2006

  64. [64]

    Onn S (2003) Convex matroid optimization.SIAM Journal on Discrete Mathematics17(2):249–253

    work page 2003

  65. [65]

    Onn S (2010) Nonlinear discrete optimization.Zurich Lectures in Advanced Mathematics, European Mathematical Society75

    work page 2010

  66. [66]

    Onn S, Rothblum UG (2004) Convex combinatorial optimization.Discrete & Computational Geometry32:549–566

    work page 2004

  67. [67]

    Onn S, Schulman LJ (2001) The vector partition problem for convex objective functions.Mathematics of Operations Research26(3):583–590

    work page 2001

  68. [68]

    Pardalos PM, Vavasis SA (1991) Quadratic programming with one negative eigenvalue is NP-hard.Journal of Global optimization1(1):15–22

    work page 1991

  69. [69]

    Raghavachari M (1969) On connections between zero-one integer programming and concave programming under linear constraints.Operations Research17(4):680–684

    work page 1969

  70. [70]

    Rockafellar RT (1970)Convex analysis(Princeton university press)

    work page 1970

  71. [71]

    Rousseeuw PJ (1985) Multivariate estimation with high breakdown point.Mathematical statistics and applications 8(283-297):37

    work page 1985

  72. [72]

    (2003)Combinatorial optimization: polyhedra and efficiency, volume 24 (Springer)

    Schrijver A, et al. (2003)Combinatorial optimization: polyhedra and efficiency, volume 24 (Springer)

    work page 2003

  73. [73]

    Scott JR, Geunes J (2025) A normal fan projection algorithm for low-rank optimization.Mathematical Programming 209(1):681–702

    work page 2025

  74. [74]

    Shigeno M, Saruwatari Y, Matsui T (1995) An algorithm for fractional assignment problems.Discrete Applied Mathematics56(2-3):333–343

    work page 1995

  75. [75]

    Shishkin SL, Shalaginov A, Bopardikar SD (2019) Fast approximate truncated SVD.Numerical Linear Algebra with Applications26(4)

    work page 2019

  76. [76]

    Sigg CD, Buhmann JM (2008) Expectation-maximization for sparse and non-negative PCA.Proceedings of the 25th international conference on Machine learning, 960–967

    work page 2008

  77. [77]

    Stein C (1945) A two-sample test for a linear hypothesis whose power is independent of the variance.The Annals of Mathematical Statistics16(3):243–258

    work page 1945

  78. [78]

    Vu V , Lei J (2012) Minimax rates of estimation for sparse PCA in high dimensions.Artificial intelligence and statistics, 1278–1286 (PMLR)

    work page 2012

  79. [79]

    Wang J, Dey SS, Xie Y (2023) Variable selection for kernel two-sample tests.arXiv preprint arXiv:2302.07415

    work page arXiv 2023

  80. [80]

    White N (1992)Matroid applications(Cambridge University Press)

    work page 1992

Showing first 80 references.