Scribd
Upload
4 views2 pages

Tutorial 6

The document outlines a tutorial for Numerical Optimization at the Birla Institute of Technology and Science, Pilani, focusing on various methods such as Newton's method, steepest descent, and Bisection method for function minimization. It includes specific problems and MATLAB programming tasks aimed at applying these optimization techniques to different functions, including Rosenbrock's function. The tutorial also emphasizes the convergence properties of Newton's method and the effects of initial guesses on optimization outcomes.

Uploaded by

eklavyaadwait
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views2 pages

Tutorial 6

The document outlines a tutorial for Numerical Optimization at the Birla Institute of Technology and Science, Pilani, focusing on various methods such as Newton's method, steepest descent, and Bisection method for function minimization. It includes specific problems and MATLAB programming tasks aimed at applying these optimization techniques to different functions, including Rosenbrock's function. The tutorial also emphasizes the convergence properties of Newton's method and the effects of initial guesses on optimization outcomes.

Uploaded by

eklavyaadwait
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE, PILANI

K. K. BIRLA GOA CAMPUS


SECOND SEMESTER
2025-2026

Numerical Optimization
Tutorial–6

1. Find the minimum of the function


 
0.75 1
f (x) = 0.65 − − 0.65 tan−1
1 + x2 x

using Newton and Bisection method.

2. If {x(k) }∞ n (k)
k=0 is the steepest descent sequence for f : R → R and if ∇f (x ) ̸= 0, then show that
f (x(k+1) ) < f (x(k) ).
1 1
3. Perform two iterations leading to the minimization of f (x1 , x2 ) = x1 + x2 + x21 + x22 + 3 using
2 2
the steepest descent method with the starting point x(0) = (0, 0)T . Also determine an optimal
solution analytically.

4. Write a simple MATLAB program for implementing the steepest descent algorithm using the
Bisection method for the line search. For the stopping criterion, use the condition ∥g (k) ∥ ≤ ϵ,
where ϵ = 10−6 . Test your program for the function f (x1 , x2 , x3 ) = (x1 , x2 , x3 ) = (x1 − 4)4 +
(x2 − 3)3 + 4(x3 + 5)4 , x(0) = (4, 2, −1)T . Also test your program using an initial condition of
(−4, 5, 1)T , and determine the number of iterations required to satisfy the stopping criterion.
Evaluate the objective function at the final point to see how close it is to 0.

5. Apply the MATLAB program from Question 4 to Rosenbrock’s function

f (x) = 100(x2 − x21 )2 + (1 − x1 )2 .

Use an initial condition of x(0) = (−2, 2)T . Terminate the algorithm when the norm of the
gradient of f is less than 10−4 .

6. Suppose that f ∈ C 3 and x∗ ∈ Rn is a point such that ∇f (x∗ ) = 0 and H(x∗ ) is invertible.
Then, for all x(0) sufficiently close to x∗ , Newton’s method is well-defined for all k and converges
to x∗ with an order of convergence at least 2.

7. Consider Rosenbrock’s Function: f (x) = 100(x2 − x21 )2 + (1 − x1 )2 , where x = [x1 , x2 ]T


(also known to be a “nasty” function – often used as a benchmark for testing algorithms). This
function is also known as the banana function because of the shape of its level sets.

a. Prove that [1, 1]T is the unique global minimizer of f over R2 .

1
b. With a starting point of [0, 0]T , apply two iterations of Newton’s method (with unit step
size).
c. Repeat part b using a gradient algorithm with a fixed step size of αk = 0.05 at each
iteration.

8. Compute the minimizer of the following functions by using all of the following methods by start-
ing with the given initial guess. Also indicate what happens when we change the initial guess.

Methods: Steepest Descent, Conjugate Gradient and Newton method.

a. (x1 − x2 )2 + (x2 − x3 )4 , Initial guess: (−2, 6, 2).


b. 9 − 8x1 − 6x2 − 4x3 + 2x21 + 2x22 + x23 + 2x1 x2 + 2x1 x3 , Initial guess: (0.5, 0.5, 0.5).
c. (x1 + x2 )2 + (x2 + x3 )2 , Initial guess: (−4, 1, 1).
d. (x1 − 1)2 + (x1 − x2 )2 + (x2 − x3 )4 , Initial guess: (2, 2, 2).
e. 4x21 + 2x22 + 2x23 − 33x1 + 16x2 − 24x3 , Initial guess: (0, 0, 0).
f. 1000 − x21 − 2x22 − x23 − x1 x2 − x1 x3 , Initial guess: (2, 2, 2).

All the Best

You might also like