function [a, b, N, nn] = min_fibonacci(fun_expression, alpha, beta, epsilon, lambda) % Fibonacci method for finding the local minimum of a function. % % fun_expr: (symbolic expression over x) The symbolic expression of the % objective function % alpha: (number) The starting point of the interval in which we seek % for minimum % beta: (number) The ending point of the interval in which we seek % for minimum % epsilon: (number) The epsilon value (the interval of the last step) % lambda: (number) The lambda value (accuracy) % % return: % a: (vector) Starting points of the interval for each iteration % b: (vector) Ending points of the interval for each iteration % N: (number) The number of iterations needed. % nn: (number) The calls of objective function fun_expr % % Error checking if alpha > beta || lambda <= 0 || epsilon <= 0 error ('Input criteria not met') end % Use Binet's formula instead of matlab's recursive fibonacci % implementation fibonacci = @(n) ( ((1 + sqrt(5))^n - (1 - sqrt(5))^n) / (2^n * sqrt(5)) ); % Init variables a = alpha; b = beta; fun = matlabFunction(fun_expression); % calculate number of iterations N = 0; while fibonacci(N) < (b(1) - a(1)) / lambda N = N + 1; end % calculate x1, x2 of the first iteration, since the following iteration % will not require to calculate both x_1 = a(1) + (fibonacci(N-2) / fibonacci(N)) * (b(1) - a(1)); x_2 = a(1) + (fibonacci(N-1) / fibonacci(N)) * (b(1) - a(1)); % All but the last calculation for k = 1:N-2 % set new search interval if fun(x_1) < fun(x_2) a(k+1) = a(k); b(k+1) = x_2; x_2 = x_1; x_1 = a(k+1) + (fibonacci(N-k-2) / fibonacci(N-k)) * (b(k+1) - a(k+1)); else a(k+1) = x_1; b(k+1) = b(k); x_1 = x_2; x_2 = a(k+1) + (fibonacci(N-k-1) / fibonacci(N-k)) * (b(k+1) - a(k+1)); end end % Last calculation x_2 = x_1 + epsilon; if fun(x_1) < fun(x_2) a(N) = a(N-1); b(N) = x_1; else a(N) = x_1; b(N) = b(N-1); end % Set objective function calls nn = 2*N -2;