Abstract:We investigate the Weighted Energy-Dissipation variational approach to semilinear gradient flows with state-dependent dissipation. A family of parameter-dependent functionals defined over entire trajectories is introduced and proved to admit global minimizers. These global minimizers correspond to solutions of elliptic-in-time regularizations of the limiting causal problem. By passing to the limit in the parameter we prove that such global minimizers converge, up to subsequences, to a solution of the gradient flow.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to apply the Weighted Energy - Dissipation (WED) variational method in semilinear gradient flows with state - dependent dissipation. Specifically, the authors study how to approximate these gradient flow problems by minimizing a family of parameter - dependent functional functionals. These functional functionals are defined over the entire trajectory, and it is proven that they have global minima. These global minima correspond to time - elliptic regularized solutions of the limit - causal problem. By taking the limit under the parameter, the authors prove that these global minima converge to a solution of the gradient flow.
### Main contributions of the paper:
1. **Introduction of WED functional functionals**: The authors introduce a family of parameter - dependent functional functionals \( W_\varepsilon \), which are defined over the entire trajectory, and prove that they have global minima.
2. **Elliptic regularization**: These global minima correspond to solutions of the limit - causal problem with time - elliptic regularization.
3. **Limit process**: By taking the limit under the parameter \(\varepsilon\), the authors prove that these global minima converge to a solution of the gradient flow.
4. **Application of the variational method**: The WED method provides a method for variational approximation of the limit - differential problem, which paves the way for the general method of applying variational principles.
### Mathematical description:
Consider the abstract gradient flow problem in the following form:
\[ d_2 \psi(u, \dot{u}) + A u(t) + \partial \phi_2(u) \ni 0 \quad \text{in } H, \text{ almost everywhere in } (0, T). \]
where:
- \( u: [0, T] \to H \) is a trajectory from the finite - time interval \([0, T]\) to the Hilbert space \( H \).
- \( \psi(u, v) \) is a smooth function, convex in the rate \( v \).
- \( A: X \to X^* \) is a linear, continuous and self - adjoint operator, mapping the second Hilbert space \( X \) to its dual \( X^* \).
- \( \phi_2: H \to [0, \infty]\) is a convex function, and \(\partial \phi_2\) represents its sub - differential.
### WED functional functionals:
\[ W_\varepsilon(u) = \begin{cases}
\int_0^T e^{-t/\varepsilon} \left( \varepsilon \psi(u(t), \dot{u}(t)) + \phi(u(t)) \right) dt & \text{if } u \in K(u_0), \\
\infty & \text{otherwise},
\end{cases} \]
where:
- \( K(u_0) = \{ u \in L^2(0, T; H) : u(0) = u_0, \phi(u), \psi(u, \dot{u}) \in L^1(0, T) \} \).
- \( \varepsilon > 0 \) is a relaxation time parameter.
### Main results:
1. For all \(\varepsilon > 0\), there exists a global minimum \( u_\varepsilon \) of \( W_\varepsilon \).
2. \( u_\varepsilon \) solves the strong form of the Euler - Lagrange equation.
3. For any sequence \(\varepsilon_k \to 0\), there exists a sub - sequence such that \( u_{\varepsilon_k} \) converges to a solution of the gradient flow (1.1).
### Conclusion:
This paper provides a new variational approximation method to solve the semilinear gradient flow problem with state - dependent dissipation by introducing the WED method. This method not only proves the existence of global minima, but also through the limit process proves...