LeastSquaresSolver Class

Use the LeastSquaresSolver class to compute the least squares solution of a system of equations. When a system of linear equations is over-determined, an exact solution may not exist. The least squares solution is the vector that minimizes the sum of the squares of the residuals. To make sure the solution is uniquely defined, the solution with the smallest norm is returned.

The coefficient matrix and the right-hand side must be specified in the constructor. Several methods are available to compute the solution. These can be selected through the SolutionMethod property. This is a LeastSquaresSolutionMethod value. The default is to use a QR decomposition. For ill-conditioned problems, it may be better to use a singular value decomposition. If speed is the dominant factor, then the method using the normal equations is the fastest, at the expense of some accuracy.

A non-negative least squares solution is restricted by having only positive or zero entries in the solution. To compute the non-negative least squares solution, select the value NonNegative for the solution method.