You are on page 1of 1

CE 235: Optimization Methods

Jan-April 2024
Homework 2

Due Date: April 12 (Friday)

Important note: Please write down the appropriate equations/expressions in relevant


places (instead of giving only verbal description). Assume non-trivial numerical values
for all parameters that are not specied. Submission should be done only by hard copy
(include a print-out of your computer code), please drop it in the instructors mailbox.

1. A Single degree of freedom (SDOF) model as shown in the Figure (1) is excited by random
force. Minimize the mass of the system such that the maximum steady state displacement does not
exceed δ . The mass should lie between mmin and mmax and mean stiness between kmin and kmax .
The following quantities are probabilistic and follow normal distribution with mean and standard
deviations as indicated:
Stiness, k = (k̄, σk )
Forcing frequency, λ = (λ̄, σλ ).
Forcing amplitude, P = (P̄ , σP ). The constraints have to be satised with a probability of at least
0.95. Formulate the problem. Solution is not required.
The steady state response of the system is given

Figure 1: llustration of the problem

2. Solve using Genetic Algorithm:


Maximize the function

f (x) = sin(x), 0≤x≤π


Consider initial population size of 6. Show crossover and mutation operation. Do hand calculation,
though for binary conversion you can use any tool. Justify all your choices.

3. The aim of this question is to understand the geometric properties of gradient descent for
the simple two-dimensional quadratic form

f (x) = 0.1x21 + 2x22 .


Use x0 = [4, 6] as the initial point.
a) Perform gradient descent with a xed step size of η = 0.4 and display the rst 20 iterations on
a contour plot.
b) Now, perform gradient descent with a step size of η = 0.5 and compare the results.
Compose code in your preferred programming language. Subsequently, present the outcomes/plots
of your code. Comment on the results. Do not use built-in function.

You might also like