no. 17 · Wednesday 18 November 2026
A Divergence-Free Finite Element Discretization of a Convection-Cooling Optimal Control Problem
Abstract
In this paper, we develop and analyze a finite element discretization for a bilinear optimal control problem governed by a stationary convection-diffusion equation, in which an incompressible velocity field acts as the control through the convective term. Continuous piecewise linear finite elements are employed to approximate the state and adjoint variables, while the Stokes-type optimality equation for the velocity control is discretized by a low-order, pointwise divergence-free ($P_1^c\oplus RT_0-P_0$) finite element. To accommodate the nonconforming velocity space, we construct a discrete cost functional by replacing the continuous $H^1$-seminorm regularization with a suitable discrete velocity bilinear form and derive the corresponding discrete first-order optimality system. Since the bilinear coupling between the velocity control and the state variable renders the underlying optimization problem nonconvex, we establish local strong convexity of the discrete reduced cost functional under a second-order sufficient optimality condition at an isolated strict local minimizer of the continuous problem. Furthermore, by exploiting a divergence-preserving interpolation operator, auxiliary state and adjoint error estimates, and a stability relation between the discrete reduced-gradient residual and the control error, we derive a priori error estimates for the state, adjoint state and velocity control. Numerical experiments are presented to demonstrate the effectiveness of the optimized incompressible velocity in redistributing the temperature field and to verify the predicted convergence rates.
- comments
- 31 pages, 6 figures, 5 tables
- fields
- Numerical Analysis, Optimization and Control