4 Numerical Computation

A

4.1 Overflow and Underflow

4.2 Poor conditioning

4.3 Gradient Based Optimization

A image image image image

4.4 Constrained Optimization

A The KKT approach generalizes the method of Lagrange multipliers, which allows equality constraints but not inequality constraints. image

4.5 Example of Linear Least Square