Review and Catchup
Nonlinear Programming
IE426: Optimization Models and Applications:
Lecture 21
Jeff Linderoth
Department of Industrial and Systems Engineering
Lehigh University
November 21, 2006
Jeff Linderoth
IE426:Lecture 21
Review and Catchup
Nonlinear Programming
Stochastic Programming
Review
Quiz #2
HW #4
Questions?
No Class
on 11/30
Homework Due 11/30 – In my mailbox by close of business!
TV People – Please have proctor verify that you turned in the
homework by 11/30.
You will need to know a little about nonlinear programming,
AMPL, and NEOS, so we’ll do that today.
Jeff Linderoth
IE426:Lecture 21
Review and Catchup
Nonlinear Programming
Stochastic Programming
Final Discussion
Final is on 12/7 from 2:45PM–5:45PM
Location:
410 Packard Lab
TV People – You can take it
anytime
on 12/7
Contact me if you cannot take it from 2:45–5:45 on 12/7.
You will take it in the morning of 12/7
Jeff Linderoth
IE426:Lecture 21
Review and Catchup
Nonlinear Programming
Stochastic Programming
Stochastic Programming Review
For stochastic programming, you will need to know how to do
the following:
1
Create the “supermodel”: The full stochastic model to
optimize the expected value of the decisions
2
Compute Value of Stochastic Solution
3
Compute the Expected Value of Perfect Information
Jeff Linderoth
IE426:Lecture 21
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Review and Catchup
Nonlinear Programming
Stochastic Programming
VSS and EVPI: Definitions
Let
S
: Set of scenarios
Let
s
: Be the
average
scenario. A scenario in which each
element is the average over all the elements in
S
Be sure to weight the average by the
probability
of each
scenario
Let
SP
: Stochastic program
Let
LP
(ˆ
s
)
: Linear program that gives optimal (first and
second stage) actions if scenario
ˆ
s
happens
Let
LP
(ˆ
x, s
)
: Linear program that gives optimal (second
stage) actions if scenario
ˆ
s
happens,
and
first stage decisions
ˆ
x
are taken
Jeff Linderoth
IE426:Lecture 21
Review and Catchup
Nonlinear Programming
Stochastic Programming
More Definitions
Let
v
(
·
)
: Be the
value
of the corresponding problem.
Let
x
(
·
)
: Be an
optimal solution
to the named problem.
v
(
SP
)
: Value of optimal solution to SP
v
(
LP
(
s
))
: Optimal solution value of
LP
(
s
)
x
(
LP
(
s
))
: Optimal solution to
LP
(
s
)
Jeff Linderoth
IE426:Lecture 21
Review and Catchup
Nonlinear Programming
Stochastic Programming
VSS
These definitions assume
maximization
This is the end of the preview.
Sign up
to
access the rest of the document.
 Spring '08
 Linderoth
 Operations Research, Linear Programming, Optimization, Systems Engineering, Jeff Linderoth, NEOS Portfolio Optimization, Catchup Nonlinear Programming

Click to edit the document details