A Theme Park Simulation

1
A Theme Park Simulation

• COMP 114
• Tuesday April 16

2
Discrete Event Simulators

• Grocery store example we did in class
• Treat time as discrete
• Move in small increments
• Why? Much easier to program
• Simulate whats happening over time

3
Roller Coaster Tycoon

• Simulate events at an amusement park through the
  day.
• Amusement park has rides with
• Capacity (number of people that can go)
• Time to run the ride
• Lines have no size limit

4
Program 5 (last one!)

• Build an Amusement Park simulation
• Read park description from file
• Simplified ride and customer behavior
• No graphics!
• Due Tues 4/30

5
Goals

• Code linked list class
• Code queue class
• Use Java classes to deal with instances of rides,
  and perhaps with customers
• Document properly
• Use good Java style (naming, modularity,
  encapsulation)

6
Customers

• Customers arrive at park each time step
• with some probability
• When customers leave ride, they may leave the
  park or go to another ride
• Customers leave park
• with some probability
• If stay, go to another ride

7
Description File

• Description of rides
• Arrival probability
• Departure probability
• Simulation time

8
Comment Lines

• Lines that begin with '' are comment
  lines      all other lines have data.

9
Description of Rides

• ride      Rides begin with the word
  "ride", then a list of colon
  separated values        Name of ride       
  Time it takes to cycle the ride        Capacity
  - total number of people it holds.        Cost
  - pennies that it costs to run one cycle
  (not
  per-person)        Price - price (in pennies)
  per person to riderideFerris
  Wheel421225rideMerry-Go-Round610820

10
Ride Model

• Simplified
• Like Merry-Go-Round or Bumper Cars
• not like Ferris Wheel
• If ride is empty
• capacity customers get on at once
• Ride goes for cycle time
• All customers get off ride

11
Other Data

• arrival      Probability that a person
  will enter park at any given time step
  departure      Probability that a person will
  leave park when exiting ride
  closes      Time when park closes (0 is when
  park opens)arrival 0.7
• departure 0.2
• closes 100

12
Customer Movement

• There is no travel delay between rides
• When customer leaves ride, departure is
  probability he/she will leave park
• If stay, customer has even probability of going
  to any ride
• If there are four rides, chance of going to each
  is 25

13
When Park Closes

• At closing time no more customers admitted to
  park.
• After the park closes, you still need to empty
  it.
• All customers in line are given rides
• No more customers admitted to lines
• They leave as soon as ride over

14
A Simulation Step

• Empty any rides that are finished.
• Send customers that just left rides either to the
  line at a ride, or to leave the park.
• Check for a new arrival to the park.
• Max one person per time step
• If arrival, send to ride (doesnt leave)
• Start any rides that are ready to go.
• Was stopped and has customers

15
Random Numbers

• Really pseudo-random numbers
• Generated by a function
• Distribution is pretty good (usually)

16
The Random Class

• In java.util package
• Make an instance of Random
• Random rand new Random( 0 )
• Argument is the seed, the number that the
  generating function uses to start
• If you give no seed, one is chosen based on the
  current time of day

17
Getting a Random Number

• Several methods to get random
• integers, floats, doubles, bytes
• Most useful is
• double rand.nextDouble()
• The number is between 0.0 and 1.0

18
Using

• Roll dice (get random number)
• If number is lower than arrival, customer enters
• If number is lower than departure, customer
  leaves

19
Repeatable Sequences

• For debugging youd like to have the same thing
  happen every time you try program.
• Explicitly give a seed (like 0) to Random( ).

20
Park 1 Description

• ride Ferris Wheel, 4, 2, 12, 25
• ride Merry-Go-Round, 6, 10, 8, 20
• arrival 0.7
• closes 100
• departure0.2

21
My Results

• Time 0
• rand 0.730967787376657
• no arrival this time period
• Time 1
• rand 0.24053641567148587
• 0 arrives at park
• rand 0.6374174253501083
• 0 on line for Merry-Go-Round
• 0 riding Merry-Go-Round, out at 7
• ...

22
Park Closes

• PARK CLOSED AT 100
• Time 103
• 38 leaves park
• 39 leaves park
• 64 leaves park
• 65 leaves park
• 17 leaves park
• 66 leaves park
• 59 leaves park
• 57 leaves park
• 68 leaves park
• 23 42 riding Ferris Wheel, out at 107
• 33 riding Merry-Go-Round, out at 109

23
Park Empties

• Time 167
• 69 leaving Ferris Wheel
• 69 leaves park
• LAST CUSTOMER LEFT PARK AT 167
• Ride Ferris Wheel cost 4.92 and made 19.50
• Average wait was 31.3 and maximum was 65
• Ride Merry-Go-Round cost 1.44 and made 15.60
• Average wait was 1.4 and maximum was 5
• Total profit 28.74

24
Product

• Print
• Time park can close,
• average and max wait times for each of the rides,
  
• cost to run each ride and
• amount of money made for each ride.
• Finally, print the profit for the day.
• Look at class DecimalFormat for nice printout.
• In package java.text

25
Debugging

• I also printed the events as they occurred. This
  was very useful for debugging.
• DONT HAND IN ON PAPER.

26
Hand In

• Print the final results of running the test cases
  (Park1 and Park2) posted on the web site.
• I have also posted the results of my
  implementation.
• Let me know if you find bugs in mine!
• I seeded the random number generator with a zero
  (IMPORTANT).
• To get my results, must use same event order

27
Requirements

• Use Classes and encapsulation
• Implement a class Queue
• Implement a SinglyLinkedList class
• Implement the queues with a linked list
• Add at tail, remove from head
• Do not electronically copy Queue or Linked List
  code! I want you to understand code.
• Make your Queues and List general
• Should hold any Object
• Not be tied to this problem!

28
Suggestions

• Implement List class and test
• Make a simple implementation
• Just what you need for queue
• Simpler than my example
• Use a simple test program
• Add, remove Strings
• Implement Queue class and test
• Then implement park

29
Event-Driven Simulation

• Program 5 is simplistic
• Customer arrival very basic
• More accurate
• subdivide time into micro-steps
• Then impractical to visit each step
• Instead, keep list of events go to next
• Book Section 13.2