Semi-automatic Differentiation (SD) Toolkit is a Matlab implementation

of the complex step derivative (CSD) technique for the differentiation of real-valued functions. The Toolkit consists of three core functions:



sdGrad.m - Returns CSD approximation of the gradient (g) of the

scalar-valued target function fun(p,Extra), according to Equation 3

of the paper.



sdJac.m - Returns CSD approximation of the Jacobian (J) of the

scalar-valued target function fun(p,Extra), according to Equation 3

of the paper.



sdHg.m - Rerurns CSD approximation of the Hessian (H) of the

scalar-valued target function fun(p,Extra), according to Equation 7 of the

paper. It also returns the centered-difference CSD approximation of the

gradient as a by-product.



For a brief describtion of the functions in the toolkit,

type ,help sdToolkit> at Matlab command prompt.-The sdToolkit demonstrates the complex step derivative method on a variety of functions and geophysically oriented examples