ISE 530 Optimization Methods for Analytics Homework 3
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ISE 530 Optimization Methods for Analytics
Homework 3
1. Do exercises 2.5, 2.7, 2.9, 2.10, 2.13 from the book “Optimization methods in finance” .
2. Starting at the point (0, 0), do three iterations of gradient descent for the problem
min f(x) = (x1 + x2 )2 + (3 − 2x1 + x2 )2 + (1 + 2x2 )2 . x=R2
In all cases, compute step sizes by solving the optimization problem
minf(xk + λdk )
λ
for current point xk and descent direction dk .
3. Solve the problem
min f(x) = (x1 + x2 )2 + (3 − 2x1 + x2 )2 + (1 + 2x2 )2
x=R2
using the Newton method.
4. You are a production manager at a factory that produces three different products: A, B, and C. Each product needs to go through two machines, Machine 1 and Machine 2, for processing. The time required on each machine and the profit obtained from each product are as follows:
Product |
Time on Machine 1 (hours) |
Time on Machine 2 (hours) |
Profit per Unit (✩) |
A |
1 |
1 |
30 |
B |
2 |
1 |
50 |
C |
3 |
2 |
70 |
Table 1: Product processing time and profit
The factory operates for 16 hours a day. Your goal is to maximize the daily profit.
Formulate this problem as a linear programming problem. Solve it using any solver (e.g., AMPL).
5. Consider the support vector machine problem seen in class. Given data {(ai , yi)}mi=1, where each ai ∈ Rn represents the features of individual i and yi ∈ {−1, 1} in- dicates its class, support vector machines ask for the solution of the optimization problem
w=Rnz=Rm (1 − β) wj(2) + β zi
s.t. yi(ai(⊤)w − b) ≥ 1 − Mzi i = 1,...,m.
In the formulation, 0 ≤ β ≤ 1 is a parameter to be tuned, (w, b) gives the equation of the hyperplane that separates the data, and zi = 1 if point i is missclassified. Modify the formulation to account for the following situations.
❼ To ensure interpretability, it is desired the vector w is sparse, that is, at most k of its entries can be nonzeros.
❼ To ensure fairness, it is desired that the proportion of missclassified points from groups with class 1 and −1 is (approximately)the same.
❼ Point #1 and #2 must be on opposite sides of the hyperplane.
2023-12-19