0% found this document useful (0 votes)
32 views2 pages

Midterm 20

This document is a 75 minute, closed book midterm exam for a convex optimization course. It contains 4 questions testing concepts of convexity, curvature of functions, properties of positive semidefinite matrices, and expressing a function in disciplined convex programming form.

Uploaded by

Hamdi M. Saif
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
0% found this document useful (0 votes)
32 views2 pages

Midterm 20

This document is a 75 minute, closed book midterm exam for a convex optimization course. It contains 4 questions testing concepts of convexity, curvature of functions, properties of positive semidefinite matrices, and expressing a function in disciplined convex programming form.

Uploaded by

Hamdi M. Saif
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
You are on page 1/ 2

EE 364a: Convex Optimization I J.

Duchi
January 23, 2020

Name: SUID:

Midterm Quiz

This is a 75 minute, closed notes, closed book midterm. Each question is worth 20 points. In
questions 1–3, we will award 5 points per part for correct responses, and 2 points for parts left
blank. For question 4, we’ll award some credit to partially correct responses. No justification is
required for any of these questions, however.
By taking this quiz you’re agreeing to respect the honor code. Good luck!

1. Convexity of some sets. Determine if each set below is convex.


   
3 x y
(a) (x, y, z) ∈ R | 0
y z
 convex  not convex
(b) {(x, y, z) ∈ R3 | xz − y 2 ≥ 0, x ≥ 0, z ≥ 0}
 convex  not convex

(c) {(x, y, z) ∈ R3 | xz − y 2 ≥ 0}
 convex  not convex
2
(d) {(x, y, z) ∈ R3 | yz ≥ x, z < 0, x ≤ 0}
 convex  not convex

2. Curvature of some functions. Determine the curvature of the functions below. For affine func-
tions (which are both convex and concave), select only the ‘affine’ box. If a function is neither
convex nor concave, select ‘neither’.

(a) f (x) = max{2, x, 1/ x, x3 }, with dom f = R+
 convex  concave  affine  neither
14
(b) f (x, t) = kxk
t13
with dom f = {x ∈ Rn , t > 0}
 convex  concave  affine  neither

(c) f (x) = (1/2)x2 − (1/12)x4 , with dom f = R


 convex  concave  affine  neither

(d) f (x, y, z) = log(y log yz − x) + log(zy), with dom f = {(x, y, z) ∈ R × R2++ | yex/y < z}
 convex  concave  affine  neither

1
3. Convexity of some sets of positive semidefinite matrices. In each part of the question, n, k are
fixed numbers with k < n. Determine if each set below is convex.
(a) {A ∈ Sn+ | Rank(A) ≥ k}, where k < n.
 convex  not convex
(b) {A ∈ Sn+ | Rank(A) ≤ k}, where k < n.
 convex  not convex
(c) {A ∈ Sn+ | Rank(A) = n}.
 convex  not convex
(d) {C ∈ Sn++ | A − B T C −1 B  0} where A, B are fixed matrices of appropriate size.
 convex  not convex

4. DCP rules. The function


(aT x)2
    
T −1/2
f (x) = log exp + exp (c x − d)
cT x − d

is convex in x over {x ∈ Rn | cT x − d > 0}. Express f using disciplined convex programming


(DCP), limited to the following atoms:

inv_pos(u), which is 1/u, with domain R++


square(u), which is u2 , with domain R

sqrt(u), which is u, with domain R+
geo_mean(u), which is ( ni=1 ui )1/n , with domain Rn+
Q

quad_over_lin(u,v), which is u2 /v, with domain R × R++


log_sum_exp(u), which is log( ni=1 exp(ui )), with domain Rn .
P

log(u), which is log u, with domain R++


exp(u), which is eu , with domain R

You may also use addition, subtraction, scalar multiplication, and any constant functions.
Assume that DCP is sign-sensitive, e.g., square(u) increasing in u when u ≥ 0. Please only
write down your composition. No justification is required.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy