Solves unconstrained nonlinear least squares problems using Powell's dog-leg
 method. The user specifies a callback C function that returns the value and
 gradients of the model function at a particular operating point. This library
 takes a series of dog-leg steps to find a local minimum of the residual
 surface.
 .
 This library works with sparse matrices, which makes it suitable for
 efficiently solving very large problems.
 .
 Development files