C/C++ implementation of the Levenberg-Marquardt non-linear
least squares algorithm. levmar includes double and single precision LM versions, both
with analytic and finite difference approximated jacobians
cordic methods describe essentially the same algorithm that with suitably chosen inputs can be used to calculate a whole range of scientific functions including sin, cos, tan, arctan, arcsin, arccos, sinh, cosh, tanh, arctanh, log, exp, square root and even multiply and divide.
the method dates back to volder [1959], and due to its versatility and compactness, it made possible the microcoding of the hp35 pocket scientific calculator in 1972.
here is some code to illustrate the techniques. ive split the methods into three parts linear, circular and hyperbolic. in the hp35 microcode these would be unified into one function (for space reasons). because the linear mode can perform multiply and divide, you only need add/subtract and shift to complete the implementation.
you can select in the code whether to do the multiples and divides also by cordic means. other multiplies and divides are all powers of 2 (these dont count). to eliminate these too, would involve ieee hackery.
