The University of Texas at Austin
Department of Electrical and Computer Engineering
EE381K: Convex Optimization Fall 2015
Problem Set One Solutions
Constantine Caramanis
Due: Tuesday, September 8, 2015.
Matlab and Computational Assignments
1.
Algorithm 1
EE 381V: Large Scale Optimization
Fall 2015
Lecture 2 September 1
Lecturer: Constantine Caramanis
2.1
Scribe: 2012 and 2014 Class
Overview of the last Lecture
In the last lecture, we gave the basic properties and denitions of convex sets and convex
functi
The University of Texas at Austin
Department of Electrical and Computer Engineering
EE381K: Large Scale Optimization Fall 2016
Problem Set Eight
Constantine Caramanis
Due: Tuesday, October 4, 2016.
Related Reading. Read Chapter 3 in S. Bubecks notes, espe
2.
least square
dataset 1
a. succeed
b. average time 0.72s
c.
regression error: 2.4922e-15
testing error: 21.2254
dataset 2
a. succeed
b. average time 58.471s
c.
regression error: 2.6772e-09
testing error: 22.7721
dataset 3
a. unsucceed
LASSO
dataset 1
a.
The University of Texas at Austin
Department of Electrical and Computer Engineering
EE381K: Large Scale Optimization Fall 2016
Problem Set Two
Constantine Caramanis
Due: Wednesday, September 21, 2016.
Reading Assignments
1. (?) Reading: Boyd & Vandenbergh
Chance Constrained Optimization of Process Systems
under Uncertainty
vorgelegt von
Diplom-Ingenieur
Harvey Arellano-Garcia
aus Piura-Per
von der Fakultt III- Prozesswissenschaften
der Technischen Universitt Berlin
zur Erlangung des akademischen Grades
Dok
\I‘iVi’T TA M mm W1 (MAM): WWW Pdw VtUl’BV‘MJ’
“WA/“419d
WTV? to A cal—Al
A; e R
We 03" 1N;u =1 ~41 v; I vd:
+7 JHLJ’ ‘3‘}
meml M I; Ma'va
M: 016.31 ")0 OJ + o<1f1)2("‘)('~'):}
M ‘: i[l i]
U “L 5M5”‘
k)th .5: mm ﬁmwmum bums L93me
Pﬁszm‘ﬂ/jmﬁmmgn _
TWLeuur
The University of Texas at Austin
Department of Electrical and Computer Engineering
EE381K: Large Scale Optimization Fall 2015
Problem Set Three Solutions
Constantine Caramanis
.
Matlab and Computational Assignments Please provide a printout of the Matlab
EE 381K: Large Scale Optimization
Fall 2015
Lecture 11 October 6
Lecturer: Constantine Caramanis
Scribe: 2012 and 2014 Class
Recall the constrained convex optimization problem:
min
f (x)
x
(11.1)
subject to x X
where f is a convex function and X is a conv
The University of Texas at Austin
Department of Electrical and Computer Engineering
EE381K: Large Scale Optimization Fall 2015
Problem Set Two Solutions
Constantine Caramanis
.
Matlab and Computational Assignments.
1. This problem illustrates how the grad
EE 381K: Large Scale Optimization
Fall 2015
Lecture 9 September 24
Lecturer: Constantine Caramanis
Scribe: 2012 and 2014 Class
(Note that the main reference for this lecture is Nocedal & Wright (Chapter 5 and 6)
9.1
Topics Covered Last Time
Newtons Metho
EE 381K: Large Scale Optimization
Fall 2015
Lecture 8 September 22
Lecturer: Constantine Caramanis
8.1
8.1.1
Scribe: 2012 and 2014 Classes
Newtons method
Convergence
In the last lecture we gave a proof of the convergence of Newtons method under the assump
EE 381K: Large Scale Optimization
Fall 2015
Lecture 10 October 1
Lecturer: Constantine Caramanis
Scribe: 2012 and 2014 Class
Reference: Nocedal & Wright - Numerical Optimization, Ch 6. in the new edition, Ch.
8 in the original.
10.1
Last time
In the last
EE 381K: Large Scale Optimization
Fall 2015
Lecture 7 September 17
Lecturer: Constantine Caramanis
7.1
Scribe: 2012 and 2014 Classes
Motivation
We have seen that for strongly convex, smooth functions, gradient descent (with xed step
size 1/M , or with exa
EE 381K: Large Scale Optimization
Fall 2015
Lecture 4 September 8
Lecturer: Constantine Caramanis
4.1
4.1.1
Scribe: 2012 and 2014 Classes
Recall from last lecture
Gradient descent
x(k+1) = x(k) (k) f (x(k) )
(4.1)
f (x(k) ) is the descent direction
is t
EE 381K: Large Scale Optimization
Fall 2015
Lecture 3 September 3
Lecturer: Constantine Caramanis
3.1
Scribe: 2012 and 2014 Class
Recap
The previous lecture began with the rst order conditions of optimality which gives a characterization of the optimal so
EE 381K: Large Scale Optimization
Fall 2015
Lecture 1 August 27, 2015
Lecturer: Constantine Caramanis
Scribe: 2012 and 2014 class
In this lecture we give the basic denition of a convex set, a convex function, and convex
optimization. As an application of
EE 381K: Large Scale Optimization
Fall 2015
Lecture 5 September 10
Lecturer: Constantine Caramanis
5.1
Scribe: 2012 and 2014 Classes
Recap: Gradient Descent using Exact Line Search
In the last class we discussed the gradient descent for solving an unconst