The Optimization Lab provides a state-of-the art suite of COMSOL Script functions for setting up and solving optimization problems.
Based on the SNOPT and SQOPT codes developed by Philip E. Gill, UC San Diego, and Walter Murray and Michael A. Saunders, Stanford University, it contains solvers for the optimization of constrained linear, quadratic, and nonlinear objective functions as well as constrained linear and nonlinear least-squares problems. There may be a mixture of linear and nonlinear constraints, and sparsity is exploited. A separate Nelder-Mead solver adds functionality for unconstrained optimization of nonsmooth functions of relatively few variables.
The command syntax allows you to state the components of your problem easily. An algorithm then analyzes the problem and chooses the most appropriate optimization function.
An essential part of the COMSOL Script product line, the Optimization Lab is fully integrated with the other Labs. It is also available to optimize COMSOL Multiphysics models.
Application Examples
- Accessed from the COMSOL Script command line
- Automatic choice of solver
- Based on the highly respected and widely used SNOPT and SQOPT codes
- Integration with COMSOL Script and its products
-
Large-scale, constrained and unconstrained problems:
- Linear
- Nonlinear
- Quadratic
- Linear least squares
- Nonlinear least squares
- Nelder-mead search algorithm for non-smooth unconstrained problems
- Optimize FEA and multiphysics problems by including COMSOL Multiphysics models
- User-friendly problem definition in flexible data structure