Declaring objective functions

Nonlinear programming problems are classified according to the availability of derivative information. The three categories

• NLP0 - no available derivative information,
• NLP1 - analytic first derivatives available, and
• NLP2 - analytic first and second derivatives available

The NLPX classes are abstract base classes for NLFX.

Listed below are the constructors for the objective function.

• typedef void (*USERFCN0)(int, const ColumnVector&, real&, int&);
  

  In the following code fragment, we show the calling sequence for a trigonometric function of dimension n.

     trig(n, x, f, result);
  

  where x is the current point, f is the value of the objective function, and result is an output argument.

• typedef void (*USERFCN1)(int, int, const ColumnVector&, real&, 
                           ColumnVector&, int&);
  

  The function call for the Rosenbrock problem with analytic derivatives is

     rosen(mode, n, x, f, g, result);
  

  where g is the gradient of the objective function.

• typedef void (*USERFCN2)(int, int, const ColumnVector&, real&, 
                           ColumnVector&, SymmetricMatrix&, int&);
  

  Similary, the function call for the illumination problem with analytic derivatives and Hessian is

     illum2(mode, n, x, f, g, H, result);
  

  where H is the Hessian of the objective function.

Next Section: Bound-constrained minimization | Back to Unconstrained minimization

Last revised June 30, 2006

---

Bug Reports    OPT++ Developers    Copyright Information    GNU Lesser General Public License Documentation, generated by , last revised August 30, 2006.