1: \begin{abstract}

2: We consider minimization of a smooth nonconvex objective function

3: %using an iterative algorithm based on Newton's method and linear

4: %conjugate gradient, with explicit detection and use of negative

5: %curvature directions for the Hessian of the objective function.

6: using an iterative algorithm \refer{based on Newton's method and the 

7: linear conjugate gradient algorithm}, with explicit detection and use 

8: of negative curvature directions for the Hessian of the objective function. 

9: The algorithm tracks Newton-conjugate gradient procedures developed in the

10: 1980s closely, but includes enhancements that allow worst-case

11: complexity results to be proved for convergence to points that satisfy

12: approximate first-order and second-order optimality conditions. The

13: complexity results match the best known results in the literature for

14: second-order methods.

15: \end{abstract}

16: