Skip to content

Moreau Envelope

Smooth a proper lower-semicontinuous convex function by infimizing its value plus a quadratic distance penalty, linking nonsmooth optimization to the proximal map.

Version
v2 · 2026-09-06 · History
Domain-specific #
2312
Origin domain
mathematics
Subdomain
convex analysis
Aliases
Moreau-Yosida regularization, Moreau regularization, Moreau approximation

Core Idea

For a proper lower-semicontinuous convex function \(f\) on a Hilbert space and \(\lambda>0\), the Moreau envelope is

\[ e_\lambda f(x)=\inf_y\left\{f(y)+\frac{1}{2\lambda}\lVert x-y\rVert^2\right\}. \]

It replaces the value at \(x\) by the best tradeoff between a low value of \(f\) and staying near \(x\). This is the infimal convolution of \(f\) with a scaled squared norm. Under the standard convex assumptions, the minimizer is unique and is the proximal point \(\operatorname{prox}_{\lambda f}(x)\); the envelope is continuously differentiable even when \(f\) is not.[1]

The key identity

\[ \nabla e_\lambda f(x)=\frac{x-\operatorname{prox}_{\lambda f}(x)}{\lambda} \]

turns a nonsmooth problem into a smooth gradient field without changing the minimum value or minimizer set. The recognition invariant is function + positive scale + quadratic proximity penalty + infimum over a latent point + proximal minimizer + smooth envelope/gradient relation.

Structural Signature

  • Input function: proper, lower-semicontinuous, convex extended-real function in the classical theory.
  • Hilbert geometry: norm from the inner product.
  • Positive parameter: \(\lambda\) controls smoothing and displacement.
  • Quadratic kernel: \(\|x-y\|^2/(2\lambda)\).
  • Infimal convolution: minimize function value plus proximity cost.
  • Proximal point: unique minimizing \(y\).
  • Envelope value: finite smooth under standard hypotheses.
  • Gradient identity: scaled proximal residual.
  • Minimizer preservation: \(\arg\min e_\lambda f=\arg\min f\).
  • Dual relation: conjugacy connects the envelope to quadratic regularization of \(f^*\).
  • Approximation limit: envelope rises to \(f\) as \(\lambda\downarrow0\) under standard conditions.

What It Is Not

It is not mollification by averaging and not convolution with a Gaussian; it uses an infimum. It is not the proximal operator, though its inner minimizer and gradient are determined by that operator. It is not an optimization algorithm by itself, although proximal-point and envelope-based algorithms exploit it.

It is not automatically globally smooth for every nonconvex or arbitrary extended-real function. Nonconvex Moreau envelopes require prox-boundedness and parameter restrictions, and may be nondifferentiable or set-valued at the proximal step.

Scope of Application

The envelope is foundational in convex analysis, monotone operators, variational inequalities, nonsmooth optimization, splitting methods, stochastic optimization, and Hamilton–Jacobi theory. Indicator functions recover squared distance to a closed convex set, and absolute-value or norm penalties yield explicit piecewise-smooth envelopes.[2]

Generalizations replace the squared Hilbert norm with Bregman divergences, variable metrics, or nonconvex prox-regular functions. These are related envelopes, but their guarantees must be restated rather than inherited silently.

Clarity

Parameter conventions vary. Some write the envelope of \(\lambda f\) with penalty \(\|x-y\|^2/2\); others write the envelope of \(f\) with \(1/(2\lambda)\). Formulas for gradients and conjugates must match the declared convention.

The envelope lies below \(f\), because \(y=x\) is feasible. Larger \(\lambda\) weakens the proximity penalty and increases smoothing; smaller \(\lambda\) keeps the proximal point near \(x\) and makes the envelope closer to \(f\).

Manages Complexity

The construction packages a nonsmooth subdifferential into a Lipschitz gradient and replaces local kink handling with a proximal subproblem. Resolvent and conjugacy identities allow the analyst to switch among primal, dual, and monotone-operator views.[3]

The cost is displaced: evaluating the gradient requires computing a proximal point, which may itself be hard. Smoothing can also worsen conditioning as \(\lambda\) becomes small and can blur features relevant outside minimization.

Abstract Reasoning

  1. Verify properness, lower semicontinuity, convexity, and Hilbert-space setting.
  2. Fix and declare the parameter convention.
  3. Form the quadratic-penalized inner objective.
  4. Prove existence and uniqueness of its minimizer.
  5. Identify the proximal map and calculate the envelope.
  6. Derive the gradient from the proximal residual.
  7. Check minimizer and minimum-value preservation.
  8. Use conjugacy or Moreau decomposition for dual analysis.
  9. For nonconvex/generalized variants, re-establish finiteness, single-valuedness, and smoothness.
  10. Include proximal-evaluation cost in algorithmic claims.

Knowledge Transfer

The portable idea is to regularize a rough objective by allowing a nearby surrogate point and charging for displacement. The proposed parent is Regularization; Moreau’s particular penalty-and-infimum construction is an optimization-specific specialization.

Examples

Absolute value. The envelope of \(|x|\) is quadratic near zero and linear outside, the Huber-type smoothing generated by soft thresholding.

Indicator function. For the indicator of a closed convex set \(C\), the proximal point is projection onto \(C\) and the envelope is \(d(x,C)^2/(2\lambda)\).

Non-example. Replacing \(f\) by an arbitrary moving average is smoothing, but not the Moreau envelope.

Structural Tensions

  • Smoothness versus faithfulness to sharp features.
  • Small approximation error versus gradient conditioning.
  • Cheap outer gradients versus expensive proximal subproblems.
  • Convex guarantees versus nonconvex extensions.
  • Primal geometry versus dual simplification.
  • Shared name versus incompatible parameter conventions.

Structural–Framed Character

Infimum, quadratic penalty, proximal minimizer, and gradient relation are structural. Function class, norm, scale, generalized kernel, numerical solver, and tolerance are framed.

Structural Core vs. Domain Accent

The portable core is optimize-over-nearby-surrogates to smooth a rough object. Extended-real convex functions, Hilbert norms, infimal convolution, proximal maps, conjugates, and nonsmooth algorithms are constitutive domain accent.

Regularization is the proposed immediate parent. Optimization, Infimal Convolution, Proximal Operator, Convex Conjugacy, and Gradient Descent are related. L-BFGS is an algorithmic neighbor, not coverage.

The prospective queue contains one strict edge to prime:regularization. No live DAG mutation is authorized.

Relationships to Other Abstractions

Local relationship map for Moreau EnvelopeParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Moreau EnvelopeDOMAINPrime abstraction: Regularization — is a kind ofRegularizationPRIME

Current abstraction Moreau Envelope Domain-specific

Parents (1) — more general patterns this builds on

  • Moreau Envelope is a kind of Regularization Prime

    Regularization is the proposed immediate parent.

Hierarchy path (1) — routes to 1 parentless root

Neighborhood in Abstraction Space

Moreau Envelope sits in a sparse region of the domain-specific corpus (81st percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With

  • Ordinary convolution or mollification.
  • The proximal operator itself.
  • Moreau decomposition as the whole identity.
  • An optimization algorithm rather than a transform.
  • Unqualified smoothness in the nonconvex case.
  • A formula using an undeclared \(\lambda\) convention.

References

[1] Jean-Jacques Moreau, “Proximité et dualité dans un espace hilbertien,” Bulletin de la Société Mathématique de France 93, 1965, 273–299. DOI 10.24033/bsmf.1625. registry ↩

[2] Heinz H. Bauschke and Patrick L. Combettes, Convex Analysis and Monotone Operator Theory in Hilbert Spaces, 2nd ed., Springer, 2017. DOI 10.1007/978-3-319-48311-5. registry ↩

[3] R. Tyrrell Rockafellar and Roger J-B Wets, Variational Analysis, Springer, 1998. DOI 10.1007/978-3-642-02431-3. registry ↩

[4] Warren L. Hare and R. Tyrrell Rockafellar, “Prox-Regularity and Stability of the Proximal Mapping,” Journal of Convex Analysis 11, 2004, 589–606. registry ↩