MA3236 NONLINEAR PROGRAMMING
Semester 1, 2018/2019
Assignment 2
Deadline: 9 November. Late submission will not be accepted.
Submit a report including solutions you obtained and your observa-
tions and comments (softcopy only), and les of matlab codes ( .m), to
Lam Xin Yee (email: ).
1. Program Frank-Wolfe algorithm. You can use any line search method, e.g.,
exact, bisection or backtracking line search. Use it to solve the problem in
Tutorial 8 question 2 and the following problem
min 12xTQx + cTx
s:t: Ax b
where Q;c;A;b are given in the le ‘Q1data.mat’. Choose the initial point
x(0) = 0 and set the tolerance = 10 2 and maximum iteration number
= 10000.
2. Program the quadratic penalty and augmented Lagrangian algorithms using
the backtracking line search. Use them to solve the following problem
min 12xTQx + cTx
s:t: Ax b = 0
where Q;c;A;b are given in the le ‘Q2data.mat’. Choose the initial point
x(0) = 0 and set the tolerance = 10 6 and maximum iteration number
= 1000.