In this work, we present a gradient descent akin method for inequality constrained optimization. At each iteration, we compute a search direction using a linear combination of the negative and normalized objective and constraint gradient. The design of the method is inspired by the singular value decomposition. Using a dynamical systems approach, we show asymptotic global and local convergence of the method. We demonstrate the method using both common test cases and applications to large-scale shape optimizations.
«
In this work, we present a gradient descent akin method for inequality constrained optimization. At each iteration, we compute a search direction using a linear combination of the negative and normalized objective and constraint gradient. The design of the method is inspired by the singular value decomposition. Using a dynamical systems approach, we show asymptotic global and local convergence of the method. We demonstrate the method using both common test cases and applications to large-scale s...
»