On approximation of convex functionals with a convexity constraint and general Lagrangians

Basic Information

• Paper ID: 2504.07783
• Title: On approximation of convex functionals with a convexity constraint and general Lagrangians
• Author: Young Ho Kim
• Classification: math.AP (Analysis of PDEs)
• Publication Date: October 13, 2025 (arXiv version)
• Paper Link: https://arxiv.org/abs/2504.07783v2

Abstract

This paper proves that minimization problems for convex functionals with convexity constraints under the class of general Lagrangian functions can be approximated by solutions of Abreu-type fourth-order equations. This result generalizes the work of Le (2023), which only addressed the case of quadratic growth Lagrangian functions.

Research Background and Motivation

Core Problem

The research addresses variational problems with convexity constraints: $\inf_{u \in S[\phi,\Omega_0]} \int_{\Omega_0} F(x, u(x), Du(x)) dx$

where the constraint set is: $S[\phi,\Omega_0] = \{u : \Omega \to \mathbb{R} \text{ convex}, u = \phi \text{ on } \Omega \setminus \Omega_0\}$

Problem Significance

1. Economic Applications: Such variational problems arise in economics in monopolist problems (Rochet-Choné model)
2. Numerical Computation Difficulties: The presence of convexity constraints makes it difficult to derive tractable Euler-Lagrange equations
3. Numerical Scheme Challenges: Such variational problems are difficult to handle numerically

Limitations of Existing Methods

1. Dimensional Restrictions: Previous results were either limited to two-dimensional cases
2. Growth Condition Constraints: Required Lagrangian functions to satisfy quadratic growth conditions
3. Narrow Applicability: Could not handle general classes of Lagrangian functions

Research Motivation

Generalize the results to general Lagrangian functions in n ≥ 2 dimensional spaces by modifying the quadratic term in the approximation scheme.

Core Contributions

1. Extended Approximation Theory: Generalized results from Le et al. from quadratic growth Lagrangian functions to general Lagrangian function classes
2. Expanded Dimensional Range: Established approximation results in n ≥ 2 dimensional spaces
3. Improved Approximation Scheme: Introduced a new penalty term G(u-φ̃ε) replacing the original quadratic term
4. Established Existence Theory: Proved existence and convergence of solutions to the modified Abreu equation

Methodology Details

Problem Formulation

Consider the variational problem: $\min_{u \in S[\phi,\Omega_0]} J(u) = \min_{u \in S[\phi,\Omega_0]} \int_{\Omega_0} F(x, u(x), Du(x)) dx$

where F satisfies general growth conditions rather than quadratic growth conditions.

Core Technical Framework

1. Lagrangian Function Conditions

The paper assumes F satisfies:

• (F1) F is smooth and convex in variables z and p
• (F2) The derivatives of F satisfy growth estimates: $\left|\frac{\partial F}{\partial z}\right| + \left|\frac{\partial F}{\partial p_i}\right| \leq f_0(|z|)g_0(|p|)$

2. Modified Approximation Functional

Original approximation functional: $J(v) + \frac{1}{2\varepsilon}\int_{\Omega\setminus\Omega_0}(v-\phi)^2 dx - \varepsilon\int_\Omega \log \det D^2v dx$

Modified approximation functional: $J_\varepsilon(u) = \int_{\Omega_0} F(x,u,Du) dx + \frac{1}{\varepsilon}\int_{\Omega\setminus\Omega_0} G(u-\tilde{\phi}_\varepsilon) dx - \varepsilon\int_\Omega \log \det D^2u dx$

3. Key Modifications

• Boundary Function Correction: $\tilde{\phi}_\varepsilon(x) = \phi(x) + \varepsilon^{1/(3n^2)}(e^{\rho(x)} - 1)$
• Penalty Function: $G(x) = \int_0^{x^2} H(t) dt$, where $H(x) = x(1 + f_0(x)g_0(x) + f_2(x)g_2(x) + xf_3(x)g_3(x))$

Technical Innovations

1. Non-quadratic Penalty Term: Replacing the quadratic term with G(u-φ̃ε) enables control of Lagrangian functions with general growth
2. Boundary Correction Technique: Using φ̃ε to make the function "sufficiently" uniformly convex, handling non-uniformly convex Lagrangian functions
3. Growth Function Matching: The construction of G precisely matches the growth properties of F

Theoretical Analysis

Main Theorem

Theorem 1.1: Let Ω₀ and Ω be smooth convex domains in ℝⁿ, with Ω uniformly convex and Ω₀ ⋐ Ω. Let φ ∈ C⁵(Ω), ψ ∈ C³(Ω), φ convex, and min∂Ω ψ > 0. Let F satisfy (F1)-(F2). If 0 < ε < ε₀ < 1, then:

(i) The second boundary value problem (1.14) has a uniformly convex W⁴'ˢ(Ω) solution uε for all s ∈ (n,∞)

(ii) The solution sequence (uεₖ) converges uniformly to the minimizer of the variational problem on compact subsets of Ω

Proof Strategy

Existence Proof (Theorem 1.1(i))

1. L∞ Estimates: Establish uniform boundedness of uε
2. Gradient Estimates: Utilize convexity to obtain gradient bounds
3. Hessian Estimates: Obtain upper and lower bounds for det D²uε via the ABP maximum principle
4. Regularity Theory: Combine Hölder estimates and W⁴'ˢ estimates
5. Degree Theory: Use Leray-Schauder degree theory to establish existence

Convergence Proof (Theorem 1.1(ii))

1. Compactness: Obtain subsequence convergence using uniform estimates
2. Boundary Conditions: Prove the limit function satisfies boundary conditions
3. Minimality: Demonstrate the limit function is a minimizer via variational inequalities

Technical Difficulties and Solutions

Key Technical Challenges

1. Difficulty in L∞ Estimates

Challenge: Under general growth conditions, the integral $\int_{\Omega_0} F_{p_ix_i}(u_\varepsilon - \tilde{\phi}_\varepsilon) dx$ cannot be controlled by quadratic terms.

Solution: Construct a growth-matching penalty function G such that: $G'(u_\varepsilon - \tilde{\phi}_\varepsilon)(u_\varepsilon - \tilde{\phi}_\varepsilon) \geq H(C_8(1 + |u_\varepsilon|))$

2. Non-uniform Ellipticity

Challenge: Non-uniform convexity of the Lagrangian function causes ellipticity to degenerate.

Solution: Correct the boundary function φ̃ε to introduce sufficient uniform convexity.

3. Singular Abreu Equation

Challenge: The term $\frac{\partial}{\partial x_i}\left(\frac{\partial F}{\partial p_i}\right)$ in equation (1.14) depends on D²uε, which is only a measure under the convexity assumption.

Solution: Use transformation techniques and twisted Harnack inequalities to handle singularities.

Experimental Verification

Theoretical Verification

The paper is primarily theoretical work, verified through rigorous mathematical proofs of the following key results:

1. Uniform Estimates: Proved L∞, W¹'∞, W⁴'ˢ estimates for the solution sequence
2. Convergence: Established uniform convergence on compact subsets
3. Minimality: Proved the limit function is indeed a minimizer of the original variational problem

Application Examples

A specific example mentioned in the paper is the Lagrangian function: $F(x,z,p) = e^{|p|^2}$

This function does not satisfy quadratic growth conditions but satisfies the general growth conditions (F1)-(F2) of the paper.

Related Work

Historical Development

1. Carlier-Radice (2019): First introduced Abreu equation approximation, but only for F = F(x,z)
2. Le (2020-2023): Extended to F(x,z,p) = F⁰(x,z) + F¹(x,p), but required quadratic growth
3. Le-Zhou (2021): General Lagrangian functions in two dimensions
4. This Paper: General Lagrangian functions in n ≥ 2 dimensions

Technical Connections

• Monge-Ampère Equation Theory: Utilized boundary regularity theory from Savin, Trudinger-Wang, and others
• Linearized Monge-Ampère Equation: Based on boundary regularity results of Le-Savin
• Twisted Harnack Inequality: Used recent results from Le

Conclusions and Discussion

Main Conclusions

1. Successfully generalized convex functional approximation theory to general Lagrangian function classes
2. Established complete existence and convergence theory in n ≥ 2 dimensional spaces
3. Provided effective methods for handling non-quadratic growth Lagrangian functions

Limitations

1. Smoothness Requirements: Requires high regularity φ ∈ C⁵, ψ ∈ C³
2. Geometric Restrictions: Requires domain Ω to be uniformly convex
3. Growth Conditions: While generalizing quadratic growth, still requires specific growth estimates

Future Directions

1. Numerical Implementation: Convert theoretical results into practically computable numerical schemes
2. Reduced Regularity: Investigate results under lower regularity conditions
3. Non-convex Domains: Extend to non-convex domain cases

In-Depth Evaluation

Strengths

1. Theoretical Breakthrough: Significantly generalizes existing theory, removing quadratic growth restrictions
2. Technical Innovation: The construction of penalty function G is clever and precisely matches the growth of the Lagrangian function
3. Rigorous Proofs: Employs advanced techniques from modern partial differential equations
4. Clear Exposition: Mathematically precise and logically coherent

Weaknesses

1. Limited Application Verification: Lacks concrete numerical experiments validating theoretical results
2. Strong Conditions: High requirements on function regularity and domain geometry
3. Complex Construction: The construction of correction function φ̃ε and G is quite technical

Impact

1. Theoretical Contribution: Provides important generalization to variational problem approximation theory
2. Methodological Value: Techniques employed may apply to similar problems
3. Application Prospects: Provides theoretical foundation for economics, optimization, and related fields

Applicable Scenarios

1. Theoretical Research: Partial differential equations, calculus of variations, convex analysis
2. Numerical Computation: Provides theoretical guidance for developing new numerical methods
3. Applied Mathematics: Optimal transport, monopoly theory, and other economic problems

References

The paper cites 16 important references, primarily including:

• Abreu's pioneering work on Kähler metrics on toric manifolds
• Carlier-Radice's foundational work on Abreu equation approximation
• Le's important series of results on singular Abreu equations
• Regularity theory for Monge-Ampère equations by Savin, Trudinger-Wang, and others

Overall Assessment: This is a high-quality theoretical mathematics paper that achieves significant progress in variational problem approximation theory. While primarily theoretical, it provides a solid mathematical foundation for related application fields. The paper has high technical content, rigorous proofs, and represents an important contribution to the field.