---
title: Weak Evolutionary Variational Inequality
url: https://www.emergentmind.com/topics/weak-evolutionary-variational-inequality
type: topic
---

# Weak Evolutionary Variational Inequality

A weak evolutionary variational inequality (Weak EVI) is a fundamental framework for describing evolutions subject to time-dependent convex constraints in Banach or Hilbert spaces, playing a central role in modern nonlinear analysis, PDEs, and continuum mechanics. This concept encompasses parabolic variational inequalities, evolutionary inclusions, and monotone operator flows, and admits both classical and abstract metric-space formulations. The term "weak" refers to the solution concept in appropriate Sobolev–Bochner spaces, in which constraints and evolution are realized in a variational or distributional sense.

## 1. Mathematical Formulation and Problem Setting

Let $\Omega \subset \mathbb{R}^n$ be a bounded domain with smooth (or Lipschitz) boundary, $T>0$, and $Q_T = \Omega \times (0,T)$. In the concrete vector-valued setting of [2504.19156], the prototypical weak evolutionary variational inequality is formulated for an unknown
$$
u: Q_T \to \mathbb{R}^N
$$
subject to the pointwise convex constraint $|u(x,t)| \leq 1$ a.e. $(x,t)\in Q_T$, and homogeneous Dirichlet boundary and initial data on the convex set
$$
K := \{ v \in H^1_0(\Omega; \mathbb{R}^N): |v(x)| \leq 1 \ \text{a.e. in} \ \Omega \}.
$$
Given data $f \in L^2(Q_T; \mathbb{R}^N)$ and $u_0 \in K$, one seeks $u \in L^2(0,T;H^1_0) \cap H^1(0,T;L^2) \cap L^p(Q_T)$ for every $1 < p < \infty$, such that for almost every $t\in(0,T)$ and all $v\in K$:
$$
\int_\Omega \partial_t u(t) \cdot (v-u(t)) + \int_\Omega \nabla u(t) : \nabla (v-u(t)) + \delta \int_\Omega u(t)\cdot(v-u(t))
\geq \int_\Omega f(t)\cdot(v-u(t)).
\tag{1}
$$
This is the weak (variational) formulation of the evolutionary problem, which naturally extends to more general Banach-space problems with linear operators, time-dependent or solution-dependent convex sets, and gradient-type or higher-order constraints [1805.06190].

## 2. Lagrange Multiplier and Strong Formulation

A crucial structural property is the equivalence between the variational inequality (1) and a system involving a Lagrange multiplier enforcing the constraint. One shows that $u$ solves (1) if and only if there is $\lambda\in L^p(Q_T)$, $1<p<\infty$, such that
$$
\begin{aligned}
\partial_t u - \Delta u + \delta u + \lambda u &= f,  && \text{in } Q_T \\
u &= 0, && \text{on } \partial\Omega\times(0,T) \\
u(0) &= u_0, && \text{in } \Omega \\
\lambda \geq \delta, \quad (\lambda-\delta)(|u|-1) &= 0, && \text{a.e. in } Q_T
\end{aligned}
\tag{2}
$$
with $|u| \leq 1$ a.e. The multiplier $\lambda$ enforces the constraint, and admits an explicit characterization: $\lambda(x,t)=\delta$ wherever $|u(x,t)|<1$, and $\lambda(x,t)\geq\delta$ on the contact set $|u(x,t)|=1$ [2504.19156]. In Banach-space settings with constraints on $|L u|$, similar multiplier structures arise [1805.06190].

## 3. Existence, Uniqueness, and Stability Theory

The fundamental result is the existence and uniqueness of solutions to both the weak EVI and the associated Lagrange system:

- If $f\in L^2(Q_T; \mathbb{R}^N)$ and $u_0\in K$, there exists a unique
  $$
  u \in L^2(0,T; H^1_0(\Omega; \mathbb{R}^N)) \cap H^1(0,T; L^2(\Omega; \mathbb{R}^N)) \cap \bigcap_{1<p<\infty} L^p(Q_T; \mathbb{R}^N)
  $$
  solving (1) with $|u|\leq 1$ a.e. and $u(0)=u_0$, $u=0$ on $\partial\Omega\times(0,T)$.

- There is a unique pair
  $$
  (\lambda, u) \in \bigcap_{1<p<\infty} L^p(Q_T)\times \Big(L^2(0,T;H^1_0)\cap H^1(0,T;L^2)\cap L^p(Q_T)\Big)
  $$
  satisfying (2) [2504.19156].

The proof employs penalized approximations, replacing the subdifferential of the indicator of the constraint with smooth or monotone-graph approximations (e.g., $k_\epsilon$), establishing uniform a priori bounds, and compactness arguments for convergence as $\epsilon \to 0$. Key monotonicity properties yield uniqueness, while strong-weak stability results guarantee continuous dependence: for convergent data $(f_n, u_{n0})\to(f, u_0)$ in $L^\infty(Q_T)\times H^1_0$, the solutions $(\lambda_n,u_n)$ of the Lagrange system converge to $(\lambda,u)$ [2504.19156].

The general evolutionary framework allows for nonlinear monotone operators and non-coercivity, and covers constraints on arbitrary linear combinations of derivatives, e.g., $\|L\,u\|_{L^p}$ for linear $L$ [1805.06190].

## 4. Abstract and Metric-Space Evolution Variational Inequalities

A far-reaching abstraction is the metric-space Evolution Variational Inequality (EVI) formulation for gradient flows in $(X, d)$:
$$
\frac12\,\frac{d^+}{dt}\,d^2(u(t), v)\,+\,\frac\lambda2\,d^2(u(t), v)\,\leq\,\phi(v)-\phi(u(t)),
$$
for all $v\in{\rm Dom}(\phi)$ and $t>0$ [1810.03939]. This generalizes the Hilbert-space theory to arbitrary geodesic spaces and encodes well-posedness, contractivity, regularity, and semigroup properties. The EVI-characterization implies:

- Contraction semigroup: $d(u(t), v(t))\leq e^{-\lambda t}d(u(0), v(0))$.
- Absolute continuity in $t\mapsto \phi(u(t))$, energy-dissipation identities, and quantitative regularization effects.
- If the functional $\phi$ is $\lambda$-geodesically convex, existence and uniqueness of the EVI-trajectory follows, and the limiting $t\to\infty$ behavior exhibits exponential convergence to minimizers for $\lambda>0$.

The EVI paradigm is fully equivalent to the De Giorgi maximal-slope concept and underpins the convergence of time-discretization schemes such as Jordan–Kinderlehrer–Otto (JKO) [1810.03939].

## 5. Structural Properties and Applications

The weak EVI framework is applicable to a range of physical, geometric, and analytical models:

- **Vector-valued and system constraints:** The key result in [2504.19156] delivers existence/uniqueness for evolutionary VIs under pointwise norm constraints, relevant for multicomponent flows and phase-field models.
  
- **Constraints on derivatives:** The theory in [1805.06190] generalizes to constraints on $|L u|$, covering gradient, Laplacian, or curl constraints, and applies to plasticity, yield-stress fluids, and subelliptic PDEs.

- **Viscoelastoplastic fluid models:** In [2112.07480], the stress variable $S$ in a viscoplastic system satisfies an evolutionary variational inequality with a convex potential $P(S)$ (possibly nonsmooth), ensuring robust well-posedness and structural stability, including the weak–strong uniqueness principle.

- **Energy-variational and relative-entropy solutions:** When diffusion is degenerate or constraints become nonsmooth, energy-variational solution concepts (relative energy inequalities) extend the weak EVI notions and retain stability, convexity, and uniqueness properties [2112.07480].

A summary of principal results for weak EVIs in relevant contexts:

| Reference     | Setting                                  | Main Structural Results                                    |
|---------------|------------------------------------------|------------------------------------------------------------|
| [2504.19156]  | $|u|\leq 1$ constraint, vector-valued     | Existence, uniqueness, multiplier system, stability         |
| [1805.06190]  | $|L u|\leq g(x,t)$ constraints           | Double approximation, monotonicity, non-coercive operators  |
| [2112.07480]  | Tensorial VIs for viscoelastoplasticity  | EVI, energy-variational solutions, weak–strong uniqueness   |
| [1810.03939]  | Abstract EVI in metric spaces            | Equivalence to maximal slope, contraction, semigroup        |

## 6. Analytical Techniques and Proof Strategies

The central analytical tools are:

- **Penalty and monotone operator approximation:** Approximate indicators for convex constraints (e.g., via exponentials or maximal-monotone graphs) regularize the problem and enable the use of monotonicity theory or pseudomonotone operator arguments.
- **A priori and uniform estimates:** Energy testing yields uniform estimates in appropriate Sobolev–Bochner norms, independent of penalty parameters, ensuring compactness for passage to limits.
- **Convexity and monotonicity:** Strict convexity of the constraint set implies uniqueness, while monotonicity enables the derivation of Grönwall-type estimates for stability and continuous dependence on data.
- **Compactness and limiting arguments:** Aubin–Lions compactness lemmas and weak convergence principles are used to extract solutions in the limit of vanishing penalty and regularization.
- **Metric space and variational techniques:** In abstract settings, lower semi-continuity, geodesic convexity, and the energy-dissipation equality drive the analysis [1810.03939].

## 7. Stability, Continuous Dependence, and Extensions

Weak evolutionary variational inequalities are robust under perturbations of data:

- **Continuous dependence:** Solutions depend continuously on $f$, $u_0$, and constraint parameters, with quantitative estimates (Gronwall bounds) [2504.19156, 1805.06190].
- **Approximation and convergence:** Penalized and regularized solutions converge strongly in $L^\infty(0,T;L^2)$ and $L^2(0,T;H^1)$, and weakly for the multipliers in $L^p$.
- **Extensions:** The general theory encompasses quasi-variational inequalities (where the constraint set depends on the unknown), time-dependent or moving convex sets, and nonlinear monotone operators without coercivity.

A plausible implication is that the EVI methodology unifies classical Hilbert/Banach-space parabolic VI theory, PDE flows with constraints, and the modern metric-space theory of gradient flows and inclusions, ensuring analytic, geometric, and computational robustness.

---

**References**:  
- "An evolutionary vector-valued variational inequality and Lagrange multiplier" [2504.19156]  
- "Evolutionary quasi-variational and variational inequalities with constraints on the derivatives" [1805.06190]  
- "Weak-strong uniqueness and energy-variational solutions for a class of viscoelastoplastic fluid models" [2112.07480]  
- "Gradient flows and Evolution Variational Inequalities in metric spaces. I: structural properties" [1810.03939]

Source: https://www.emergentmind.com/topics/weak-evolutionary-variational-inequality