Simulating by Excel
Download
Report
Transcript Simulating by Excel
Outline
terminating
and non-terminating systems
analysis of terminating systems
generation of random numbers
simulation by Excel
a
terminating system
a non-terminating system
basic
operations in Arena
1
Two Types of Systems
Terminating and Non-Terminating
2
Two Types of Systems
chess piece
starts at vertex F
moves equally likely to
adjacent vertices
to estimate E(# of moves)
to reach the upper
boundary
B
A
GI/G/ 1 queue
infinite buffer
service times ~ unif[6, 10]
interarrival times ~ unif[8, 12]
to estimate the
E[# of customers in system]
C
…
N(t)
D
E
t, time
F
3
Two Types of Systems
chess piece
initial condition defined
by problem
termination of a
simulation run defined
by the system
estimation of the mean
or probability of a
random variable
run length defined by
number of replications
4
GI/G/ 1 queue
initial condition unclear
termination of a
simulation run defined
by ourselves
estimation of the mean
or probability of the
limit of a sequence of
random variables
run length defined by
run time
Two Types of Systems
Terminating and Non-Terminating
chess piece: a
terminating systems
analysis: Strong Law of
Large Numbers (SLLN)
and Central Limit
Theorem (CLT)
GI/G/ 1 queue: a nonterminating system
analysis: probability
theory and statistics
related to but not
exactly SLLN, nor CLT
5
Analysis of Terminating Systems
6
Strong Law of Large Numbers
- Basis to Analyze Terminating Systems
i.i.d. random variables X1, X2, …
finite mean and variance 2
define X n
X1 ... X n
n
P {} | lim X n () 1
n
7
Strong Law of Large Numbers
- Basis to Analyze Terminating Systems
a fair die thrown continuously
Xi = the number shown on the ith throw
lim X n ?
n
n
Y
1, if X n {3,4},
What
Yn
Why should lim i 1
n
n
0, otherwise.
8
i
1
be??
3
Strong Law of Large Numbers
- Basis to Analyze Terminating Systems
in terminating systems, each replication is an
independent draw of X
Xi
E(X)
are i.i.d.
(X1 + … + Xn)/n
9
Central Limit Theorem
- Basis to Analyze Terminating Systems
interval estimate & hypothesis testing of normal random
variables
t, 2, and F
i.i.d. random variables X1, X2, … of finite mean and
variance 2
Xn d
standard normal
/ n
CLT: approximately normal for “large enough” n
can use t, 2, and F for
10
Generation of
Random Numbers & Random Variates
11
To Generate
Random Variates in Excel
for
uniform [0, 1]: rand() function
for other distributions: use Random
Number Generator in Data Analysis Tools
uniform,
discrete, Poisson, Bernoulli,
Binomial, Normal
tricks
to transform
uniform
[-3.5, 7.6]?
normal (4, 9) (where 4 is the mean and 9 is the variance)?
12
To Generate the
Random Mechanism
general overview, with details discussed later this semester
everything based on random variates from uniform (0, 1)
each stream of uniform (0, 1) random variates being a
deterministic sequence of numbers on a round robin
“first” number in the robin to use: SEED
many simple, handy generators
13
Simulation by Excel
for Terminating Systems
14
Examples
Example 1: Generate 1000 samples of X ~ uniform(0,1)
Example 2: Generate 1000 samples of Y ~ normal(5,1)
Example 3: Generate 1000 samples of Z ~
z: 5
10
15
20
25
30
p:
0.1 0.15 0.3
0.2
0.14 0.11
Example 4. Use simulation to estimate
(a) P(X > 0.5)
(b) P(2 < Y < 8)
(c) E(Z)
Using 10 replications, 50 replications, 500 replications,
5000 replications. Which is more accurate?
15
Examples: Probability and Expectation
of Functions of Random Variables
X
~
x:
p(x):
Y
=
Find
100 150 200 250 300
0.1 0.3 0.3 0.2 0.1
2 X 2 50
E(Y) and P(Y 30)
16
Examples: Probability and Expectation
of Functions of Random Variables
X
~ N(10, 4), Y ~ N(9,1), independent
estimate
P(X
< Y)
Cov(X, Y) = E(XY) - E(X)E(Y)
17
Example: Newsboy Problem
Pieces of “Newspapers” to Order
order 2012 calendars in Sept 2011
cost: $2 each; selling price: $4.50 each
salvage value of unsold items at Jan 1 2012: $0.75 each
from historical data: demand for new calendars
Demand:
Prob.
:
100
150
0.3
0.2
objective: profit maximization
questions
200
250
0.3 0.15
how many calendars to order
with the optimal order quantity, P(profit 400)
18
300
0.05
Example: Newsboy Problem
Pieces of “Newspapers” to Order
D
= the demand of the 2012 calendar
D
follows the given distribution
Q
= the order quantity {100, 150, 200, 250, 300}
V
= the profit in ordering Q pieces
=
4.5 min (Q, D) + 0.75 max (0, Q - D) - 2Q
objective:
find Q* to maximize E(V)
19
Example: Newsboy Problem
Pieces of “Newspapers” to Order
two-step
1
solution procedure
estimate E(profit) for a given Q
generate demands
find the profit for each demand sample
find the (sample) mean profit of all demand samples
look for Q*, which gives the largest
mean profit
2
20
Example: Newsboy Problem
Pieces of “Newspapers” to Order
our
simulation of 1000 samples,
Q
= 100: E(V) = 250
Q
= 150: E(V) = 316.31
Q
= 200: E(V) = 348.31
Q
= 250: E(V) = 328.75
Q
= 300: E(V) = 277.17
Q* = 200 is optimal
remarks: many papers on this issue
21
Simulation by Excel
for a Non-Terminating System
22
Simulation a GI/G/1 Queue
by its Special Properties
Dn = delay time of the nth customer; D1 = 0
Sn = service time of the nth customer
Tn = inter-arrival time between the nst and the
(n+1)st customer
Dn+1 = [Dn + Sn - Tn]+, where []+ = max(, 0)
N
average delay = Dn / N
n 1
23
Arena Model 03-1,
Model 03-02, Model 03-03
24
Model 03-01
a drill press processing one type of product
interarrival times ~ i.i.d. exp(5)
service times ~ i.i.d. triangular (1,3,6)
all random quantities are independent
one type of parts; parts come in
and are processed one by one
25
a drill press
Model 03-02 and Model 03-03
Model 03-02: sequential servers
Alfie
checks credit
Betty prepares covenant
Chuck prices loan
Doris disburses funds
Model 03-03: parallel servers
Each
employee can do any tasks
26