---
title: Capra Conjugacy in Convex Analysis
url: https://www.emergentmind.com/topics/capra-conjugacy
type: topic
---

# Capra Conjugacy in Convex Analysis

Capra conjugacy is a generalized convex duality framework, introduced to address the limitations of classical Fenchel duality for functions exhibiting invariance along primal rays in normed vector spaces. The Capra conjugacy, defined via a coupling that is constant along rays (hence "Constant Along Primal Rays," or Capra), yields nontrivial duality results and reveals hidden convexity structures for intrinsically nonconvex, 0-homogeneous, or support-based functions such as the $\ell_0$ pseudonorm and rank function. This concept has led to new theoretical and algorithmic approaches for sparse and low-rank optimization in convex analysis and applied mathematics.

## 1. Definition of Capra Coupling and Capra Conjugacy

Let $X = \mathbb{R}^d$ be equipped with a source norm $\|\cdot\|$ and dual norm $\|\cdot\|_*$. The Capra coupling is defined for $x, y \in \mathbb{R}^d$ as
\[
C(x, y) = 
\begin{cases}
  \dfrac{\langle x, y \rangle}{\|x\|}, & x \neq 0, \\
  0 & x = 0.
\end{cases}
\]
This coupling is homogeneous of degree zero in $x$: $C(\lambda x, y) = C(x, y)$ for all $\lambda > 0$, making it constant along rays. The associated Capra conjugate and biconjugate of a function $f: \mathbb{R}^d \to \overline{\mathbb{R}}$ are
\[
f^{*_{\!C}}(y) = \sup_{x \in \mathbb{R}^d} \{C(x, y) - f(x)\}, \quad
f^{*_{\!C}*_{\!C}}(x) = \sup_{y \in \mathbb{R}^d} \{C(x, y) - f^{*_{\!C}}(y)\}.
\]
A function $f$ is called **Capra-convex** if $f^{*_{\!C}*_{\!C}} = f$ [2010.13323, 2002.01314, 2509.06392].

## 2. Motivation: Failure of Fenchel Conjugacy for 0-Homogeneous and Support-Based Functions

For functions $f$ that are 0-homogeneous or constant along rays, such as the $\ell_0$ pseudonorm or indicator functions of level sets of support, the Fenchel conjugacy is degenerate. For instance, the Fenchel conjugate of the indicator of $\{x: \mathrm{supp}(x) \subseteq K\}$ is the zero function, identifying all sparse-structure functions as zero. Similarly, the Fenchel biconjugate of $\ell_0$ is identically zero, which precludes classical convex-analytic tools for analyzing sparsity and support-based structures [2001.11828, 1906.04038]. The Capra conjugacy overcomes this by factoring out the ray-invariance intrinsic to 0-homogeneous functions via normalization.

## 3. Fundamental Results: Capra-Convexity, Biconjugacy, and Hidden Convexity

The key theorem asserts that if both the source norm $\|\cdot\|$ and its dual $\|\cdot\|_*$ are orthant-strictly monotonic (e.g., $\ell_p$ for $1 < p < \infty$), then any nondecreasing, finite-valued function of the support, $f(x) = F(\mathrm{supp}(x))$, is Capra-convex:
\[
f(x) = f^{*_{\!C}*_{\!C}}(x), \qquad \forall x \in \mathbb{R}^d.
\]
For the $\ell_0$ pseudonorm, this gives
\[
\ell_0(x) = (\ell_0)^{*_{\!C}*_{\!C}}(x),
\]
with the Capra conjugate
\[
(\ell_0)^{*_{\!C}}(y) = \max_{0 \leq k \leq d} \{\|y\|_{*,k} - k\},
\]
where $\|y\|_{*,k}$ is a "coordinate-$k$" dual norm specific to the support size [2010.13323, 2002.01314, 2001.11828, 2509.06392, 2112.15335].

Capra-convexity implies that functions which are highly nonconvex in the classical sense (e.g., cardinality) become the exact representative of their own Capra biconjugate.

A striking consequence is **hidden convexity**: every such $f$ coincides with a proper convex lower semicontinuous function $g$ composed with the normalization mapping,
\[
f(x) = g\left(\frac{x}{\|x\|}\right), \quad x \neq 0,
\]
where $g = (f^{*_{\!C}})^*$ is convex and lsc. On the unit sphere, $f$ and $g$ coincide [2010.13323, 2002.01314, 1906.04038].

## 4. Capra-Subdifferential and Variational Representations

The Capra-subdifferential of a function $f$ at $x$ is
\[
\partial_{C}\,f(x) = \{ y : f^{*_{\!C}}(y) = C(x, y) - f(x) \}.
\]
For $f = \ell_0$ and source norm $\ell_p$ with $p > 1$, the Capra-subdifferential at $x\neq 0$ consists of $y$ such that:
- $y_L \in N_{B_p}(x/\|x\|_p)$, where $L = \mathrm{supp}(x)$,
- certain monotonicity and ordering conditions on the dual coordinates,
resulting in a convex, nonempty, but not generally linear set [2112.15335].

Capra-convexity allows for variational formulations:
\[
\frac{f(x)}{\|x\|} = \inf_{\sum_{K} z_K = x} \left\{ \sum_{K} F(K) \|z_K\|_{*,K}^{\mathrm{supp}} : \sum_{K} \|z_K\| \leq \|x\| \right\},
\]
where $z_K$ are supported on $K$, and $\|\cdot\|_{*,K}^{\mathrm{supp}}$ are generalized local $K$-support dual norms. This representation is convex and finite-dimensional, making it suitable for optimization [2010.13323, 2002.01314, 2001.11828, 1906.04038].

## 5. Capra-Convex Sets and Geometric Characterization

A set $S \subseteq \mathbb{R}^n$ is Capra-convex if its indicator is Capra-convex: $\delta_S^{CC} = \delta_S$. Any Capra-convex set is necessarily a cone (closed under positive scaling), though not necessarily classically convex or closed.

The main theorem is that $S$ is Capra-convex if and only if:
- $S$ is a cone,
- $S^\circ = \overline{\mathrm{conv}(S \cap S^{n-1})}$,
where $S^\circ = \{ x/\|x\| : x \in S, x \neq 0 \} \cup \{0\}$. The directions of $S$ (its intersection with the unit sphere) form a closed convex subset. Closed convex cones are always Capra-convex; so are conical hulls of spherically-convex subsets of the unit sphere [2509.06392].

These properties offer a sharp geometric criterion for Capra-convexity, distinct from classical convex sets.

## 6. Extensions: Matrix Functions and Rank-Based Capra Conjugacy

Capra conjugacy extends to matrix spaces, replacing the scalar product with the trace. For a matrix norm $\|\cdot\|$, the Capra coupling is
\[
\langle M, N \rangle_{\mathrm{Capra}} = 
\begin{cases}
  \frac{\mathrm{Tr}(M^T N)}{\|M\|}, & M \neq 0, \\
  0, & M = 0.
\end{cases}
\]
For the rank function, the Capra conjugate is
\[
\operatorname{rank}^{\capra}(N) = \max_{r = 0, \ldots, d} \{ \|N\|_{(r)*} - r \},
\]
with generalized $r$-rank dual norms. The Capra biconjugate gives a variational lower bound (tight for the Frobenius norm), providing a convex formulation for the rank:
\[
\operatorname{rank}(M) = \frac{1}{\|M\|} \min \left\{ \sum_{r=1}^{d} r\, \|M^{(r)}\|_{(r)} \mid M = \sum_{r=1}^{d} M^{(r)},\, \sum_{r=1}^{d} \|M^{(r)}\|_{(r)} \leq \|M\| \right\}
\]
when $\|\cdot\|$ is the Frobenius norm [2105.14982].

## 7. Applications and Implications for Sparse and Low-Rank Optimization

The Capra framework supports exact convex-analytic reformulations for cardinality-constrained or penalized problems common in statistics, machine learning, and signal processing. For example, minimizing $\|x\|_0$ over an affine set or the $k$-sparse least-squares problem can be recast as minimizing a proper convex function over the unit sphere using hidden convexity:
\[
\min_{x \in X} \|x\|_0 = \min_{u \in S^{d-1}} G(u),
\]
where $G$ is convex, enabling sphere-based or conic optimization techniques. Feature selection, compressive sensing, and low-rank matrix recovery now admit new algorithmic approaches based on convex minorants and variational formulations unlocked by Capra conjugacy [2509.06392, 2002.01314, 2010.13323].

---

**References**:  
- [2010.13323]  
- [2001.11828]  
- [2002.01314]  
- [2112.15335]  
- [2509.06392]  
- [2105.14982]  
- [1906.04038]

Source: https://www.emergentmind.com/topics/capra-conjugacy