function [ f, g, H ] = opt05_fgh ( x, flag, a ) %% OPT05_FGH evaluates F, G and H for test case #5. % % Discussion: % % This example is known as the Rosenbrock "banana" function. % % This example is discussed in Dennis and Schnabel, page 157. % (There is a misprint in Dennis and Schabel. When they print the % rescaled function, they include an incorrect sign on the second term.) % % This example tests the scaling matrices. % % Suggested starting points are % % X(init) = ( -1.2/A, A) % or % X(init) = ( 6.39/A, -0.221*A ) % % The optimizing value is % % X* = (1/A, A). % % for which % % F(X*) = 0. % % A typical value of A is 1. For values of A that are greater or lesser % than 1, the optimization can take significantly longer. % % Modified: % % 09 January 2008 % % Author: % % Jeff Borggaard, % Gene Cliff, % Virginia Tech. % % Reference: % % John Dennis, Robert Schnabel, % Numerical Methods for Unconstrained Optimization % and Nonlinear Equations, % SIAM, 1996, % ISBN13: 978-0-898713-64-0, % LC: QA402.5.D44. % % Parameters: % % Input, real X(2), the evaluation point. % % Input, string FLAG, indicates what must be computed. % 'f' means only the value of F is needed, % 'g' means only the value of G is needed, % 'all' means F, G and H (if appropriate) are needed. % It is acceptable to behave as though FLAG was 'all' % on every call. % % Input, real A, the scale factor. A typical value is 1. % A should not be 0. % % Output, real F, the optimization function. % % Output, real G(2,1), the gradient column vector. % % Output, real H(2,2), the Hessian matrix. % n = length ( x ); if ( n ~= 2 ) fprintf ( '\n' ); fprintf ( 'OPT05_FGH - Fatal error!\n' ); fprintf ( ' The input vector X should have length 2.\n'), fprintf ( ' Instead, it has length = %d.\n', n ); keyboard end if ( a == 0 ) fprintf ( '\n' ); fprintf ( 'OPT05_FGH - Fatal error!\n' ); fprintf ( ' The scale factor A should be nonzero.\n'), keyboard end f = 100*( (a*x(1))^2 - x(2)/a )^2 + ( 1 - a*x(1) )^2; g(1,1) = 400*a^4*x(1)^3 - 400*a*x(1)*x(2) - 2*a + 2*a^2*x(1); g(2,1) =-200*a*x(1)^2 + 200*x(2)/a^2; H(1,1) = 1200*a^4*x(1)^2 - 400*a*x(2) + 2*a^2; H(1,2) =-400*a*x(1); H(2,1) = H(1,2); H(2,2) = 200/a^2;