pith. sign in

arxiv: 2509.06392 · v2 · submitted 2025-09-08 · 🧮 math.OC

What are Capra-Convex Sets?

Jean-Philippe Chancelier (CERMICS) , Michel de Lara (CERMICS) , Adrien Le Franc (CERMICS) , Seta Rakotomandimby (CERMICS) This is my paper

Pith reviewed 2026-05-18 18:55 UTC · model grok-4.3

classification 🧮 math.OC
keywords Capra-convexityabstract convexityℓ0 pseudonormCapra couplingFenchel conjugacysparsityoptimizationmachine learning
0
0 comments X

The pith

The ℓ0 pseudonorm equals its Capra-biconjugate, proving it is Capra-convex and yielding a characterization of Capra-convex sets.

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

This paper defines Capra-convexity as an abstract convexity in which a coupling constant along primal rays replaces the usual scalar product when forming generalized Fenchel conjugacies. The central result is that the ℓ0 pseudonorm, which simply counts the nonzero entries of a vector, equals the value obtained by taking its Capra-biconjugate. This equality establishes that the ℓ0 pseudonorm itself is a Capra-convex function. The authors then derive a characterization of the sets that satisfy Capra-convexity under the same coupling. The construction matters because it supplies duality tools for functions and sets that enforce sparsity.

Core claim

Capra-convexity arises when a Capra coupling replaces the scalar product in the definition of generalized Fenchel conjugacies. A key motivating result is that the ℓ0 pseudonorm equals its Capra-biconjugate. This shows that the ℓ0 pseudonorm is a Capra-convex function. The paper then supplies a characterization of Capra-convex sets.

What carries the argument

The Capra coupling, a function that remains constant along primal rays and replaces the scalar product to define generalized conjugacies.

If this is right

  • The ℓ0 pseudonorm becomes treatable by the duality machinery of abstract convex analysis.
  • Capra-convex sets can be identified by checking properties induced by the Capra coupling.
  • Duality-based methods become available for optimization problems that require sparsity.

Where Pith is reading between the lines

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

  • The characterization may support construction of Capra-convex surrogates for other nonconvex sparsity penalties.
  • Subdifferential calculus under the Capra coupling could produce new first-order conditions for sparse recovery problems.
  • The same ray-constant coupling idea might apply to discrete selection functions beyond simple counting.

Load-bearing premise

The Capra coupling preserves the duality properties needed for the biconjugate of a function or indicator to recover the original object.

What would settle it

An explicit vector for which the numerical value of the ℓ0 pseudonorm differs from the value of its Capra-biconjugate would disprove the central equality.

Figures

Figures reproduced from arXiv: 2509.06392 by Adrien Le Franc (CERMICS), Jean-Philippe Chancelier (CERMICS), Michel de Lara (CERMICS), Seta Rakotomandimby (CERMICS).

Figure 1
Figure 1. Figure 1: Illustration of Theorem 2.3. Example of a set [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of Example 2.9, where cone(H) in Fig. 2a (left) is not Capra-convex, but cone(X) in Fig. 2b (right) is Capra-convex 2.2.3 Relationship with spherical convexity On the one hand, spherically-convex sets are defined using the unit sphere S and the associ￾ated radial projection ̺: R n → S (0) [7, 8]. Indeed, following [8, Definition 2.5]1 spherically￾convex sets are subsets X ⊂ S of the unit spher… view at source ↗
Figure 3
Figure 3. Figure 3: The cones {Ki}i∈{1,2,3} in (14) (left column) and the closed convex hull of their image by the radial projection ̺ in (4) defined with the source norm ||·|| = k·k2 (right column) 12 [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The cones {Ki}i∈{1,2,3} in (14) (left column) and the closed convex hull of their image by the radial projection ̺ in (4) defined with the source norm ||·|| = k·k∞ (right column) 14 [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
read the original abstract

This paper focuses on a specific form of abstract convexity known as Capra-convexity, where a constant along primal rays (Capra) coupling replaces the scalar product used in standard convex analysis to define generalized Fenchel conjugacies. A key motivating result is that the ${\ell}$0 pseudonorm - which counts the number of nonzero components in a vector - is equal to its Capra-biconjugate. This implies that ${\ell}$0 is a Capra-convex function, highlighting potential applications in statistics and machine learning, particularly for enforcing sparsity in models. Building on prior work characterizing the Capra-subdifferential of ${\ell}$0 and the role of source norms in defining the Capra-coupling, the paper provides a characterization of Capra-convex sets.

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

1 major / 2 minor

Summary. The paper introduces Capra-convexity as a form of abstract convexity in which a constant-along-primal-rays (Capra) coupling replaces the standard scalar product in the definition of generalized Fenchel conjugacies. A central motivating result is that the ℓ0 pseudonorm equals its Capra-biconjugate (hence is Capra-convex). Building on earlier characterizations of the Capra-subdifferential of ℓ0 and the role of source norms, the manuscript supplies a characterization of Capra-convex sets.

Significance. If the biconjugate equality and the set characterization are rigorously established, the work supplies a new abstract-convexity framework that directly accommodates the ℓ0 pseudonorm. This could be useful for sparsity-constrained problems in statistics and machine learning. The explicit link to source norms and prior subdifferential results is a concrete strength.

major comments (1)
  1. The manuscript asserts that ℓ0 equals its Capra-biconjugate, but the precise definition of the Capra coupling (via source norms) and the verification that the biconjugate recovers ℓ0 exactly are not visible in the supplied abstract; a self-contained proof or explicit reference to the relevant theorem in the body is required to confirm that no post-hoc adjustment of the coupling is used.
minor comments (2)
  1. Add explicit citations to the prior work on Capra-subdifferentials and source norms already in the introduction.
  2. Clarify the notation for the Capra coupling at its first appearance so that readers can follow the generalized conjugacy without consulting external references.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their positive evaluation of the significance of our work and for the constructive major comment. We address the point below and have revised the manuscript to improve clarity and visibility of the relevant definitions and proofs.

read point-by-point responses

  1. Referee: The manuscript asserts that ℓ0 equals its Capra-biconjugate, but the precise definition of the Capra coupling (via source norms) and the verification that the biconjugate recovers ℓ0 exactly are not visible in the supplied abstract; a self-contained proof or explicit reference to the relevant theorem in the body is required to confirm that no post-hoc adjustment of the coupling is used.

    Authors: We thank the referee for highlighting this point. The Capra coupling is defined explicitly via source norms in Section 2 of the manuscript, directly extending the framework from our prior characterizations of the Capra-subdifferential of ℓ0. The equality ℓ0 equals its Capra-biconjugate is established with a self-contained proof in Theorem 4.1 (main results section), which verifies recovery of ℓ0 exactly from the defined coupling without any post-hoc adjustments. To ensure this is immediately visible, we have added an explicit cross-reference in the abstract and introduction to Theorem 4.1, along with a brief outline of the proof's key steps. This revision confirms the result follows rigorously from the stated coupling. revision: yes

Circularity Check

0 steps flagged

No significant circularity; central biconjugate result and set characterization are independent of inputs

full rationale

The paper defines Capra-convexity via a Capra coupling that replaces the scalar product in generalized Fenchel conjugacies, then states as a key result that the ℓ0 pseudonorm equals its Capra-biconjugate (hence is Capra-convex) and provides a characterization of Capra-convex sets. This equality and characterization are presented as derived outcomes building on the coupling definition and prior background results on subdifferentials and source norms. No equation or step reduces the target claims to a self-definition, a fitted parameter renamed as prediction, or a load-bearing self-citation chain whose content is unverified within the paper. The cited prior work supplies context for the coupling but does not force the biconjugate equality or set characterization by construction. The derivation chain is therefore self-contained against external mathematical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on standard definitions of abstract convexity and Fenchel conjugacy together with the specific Capra coupling introduced in earlier papers by the same research group.

axioms (1)
  • domain assumption Generalized Fenchel conjugacy can be defined via an arbitrary coupling function that satisfies suitable monotonicity and constancy properties along rays.
    Invoked to replace the scalar product while preserving duality relations.

pith-pipeline@v0.9.0 · 5680 in / 1249 out tokens · 50054 ms · 2026-05-18T18:55:43.321675+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

18 extracted references · 18 canonical work pages

  1. [1]

    H. H. Bauschke and P. L. Combettes. Convex analysis and monotone operator theory in Hilbert spaces. CMS Books in Mathematics/Ouvrages de Mathématiques de la S MC. Springer-Verlag, New York, second edition, 2017

    work page 2017

  2. [2]

    Chancelier and M

    J.-P. Chancelier and M. De Lara. Hidden convexity in the ℓ0 pseudonorm. Journal of Convex Analysis , 28(1):203–236, 2021

    work page 2021

  3. [3]

    Chancelier and M

    J.-P. Chancelier and M. De Lara. Constant along primal ra ys conjugacies and the ℓ0 pseudonorm. Optimization, 71(2):355–386, 2022. doi: 10.1080/02331934.2020.18228 36

    work page doi:10.1080/02331934.2020.18228 2022

  4. [4]

    Chancelier and M

    J.-P. Chancelier and M. De Lara. Capra-convexity, conve x factorization and variational formulations for the ℓ0 pseudonorm. Set-Valued and Variational Analysis , 30:597–619, 2022

    work page 2022

  5. [5]

    Chancelier and M

    J.-P. Chancelier and M. De Lara. Orthant-strictly monot onic norms, generalized top- k and k-support norms and the ℓ0 pseudonorm. Journal of Convex Analysis , 30(3): 743–769, 2023

    work page 2023

  6. [6]

    W. Fenchel. Convexity through the ages. In Convexity and its applications , pages 120–130. Springer, 1983

    work page 1983

  7. [7]

    O. P. Ferreira and S. Z. Iusem, A. N.and Németh. Concepts a nd techniques of op- timization on the sphere. TOP, 22(3):1148–1170, Oct 2014. ISSN 1863-8279. doi: 10.1007/s11750-014-0322-3

    work page doi:10.1007/s11750-014-0322-3 2014

  8. [8]

    Guo and Y

    Q. Guo and Y. Peng. Spherically convex sets and spherical ly convex functions. J. Convex Anal , 28(1):103–122, 2021

    work page 2021

  9. [9]

    Hiriart-Urruty and C

    J.-B. Hiriart-Urruty and C. Lemaréchal. Fundamentals of convex analysis . Springer, 2004

    work page 2004

  10. [10]

    J. L. W. V. Jensen. Sur les fonctions convexes et les inég alités entre les valeurs moyennes. Acta mathematica, 30(1):175–193, 1906

    work page 1906

  11. [11]

    Le Franc, J.-P

    A. Le Franc, J.-P. Chancelier, and M. De Lara. The capra- subdifferential of the ℓ0 pseudonorm. Optimization, 73(4):1229–1251, 2024. doi: 10.1080/02331934.2022. 2145172

    work page doi:10.1080/02331934.2022 2024

  12. [12]

    J. E. Martínez-Legaz. Generalized convex duality and i ts economic applications. In N. Hadjisavvas, S. Komlósi, and S. Schaible, editors, Handbook of Generalized Con- vexity and Generalized Monotonicity. Nonconvex Optimizat ion and Its Applications , volume 76, pages 237–292. Springer-Verlag, New York, 2005. 20

    work page 2005

  13. [13]

    Rakotomandimby, J.-P

    S. Rakotomandimby, J.-P. Chancelier, M. De Lara, and A. Le Franc. Subgradient selec- tor in the generalized cutting plane method with an applicat ion to sparse optimization. Optimization Letters, 2025. Accepted for publication

    work page 2025

  14. [14]

    R. T. Rockafellar. Conjugate Duality and Optimization . CBMS-NSF Regional Confer- ence Series in Applied Mathematics. Society for Industrial and Applied Mathematics, 1974

    work page 1974

  15. [15]

    R. T. Rockafellar and R. J.-B. Wets. Variational Analysis . Springer-Verlag, Berlin, 1998

    work page 1998

  16. [16]

    Simon and H

    F. Simon and H. Rauhut. A Mathematical Introduction to Compressive Sens- ing. Birkhäuser New York, NY, 2013. ISBN 978-0-8176-4948-7. do i: 10.1007/ 978-0-8176-4948-7

    work page 2013

  17. [17]

    I. Singer. Abstract Convex Analysis . Canadian Mathematical Society Series of Mono- graphs and Advanced Texts. John Wiley & Sons, Inc., New York, 1997. ISBN 0-471- 16015-6

    work page 1997

  18. [18]

    A. M. Tillmann, D. Bienstock, A. Lodi, and A. Schwartz. C ardinality minimization, constraints, and regularization: a survey. SIAM Review , 66(3):403–477, 2024. 21

    work page 2024