Transcript 01-linkedlist

TCSS 342, Winter 2005
Lecture Notes
Linked List Implementation
Weiss Ch. 6, pp. 183-205
Weiss Ch. 17, pp. 537-548
1
Collections

collection: an object that stores data inside it;
a.k.a. a "data structure"
 the objects stored are called elements
 some maintain an ordering, some don't
 some collections allow duplicates, some don't
 an array is like a very crude "collection"
 typical operations: add element, remove element,
clear all elements, contains or find element, get size
 most collections are built with particular kinds of
data, or particular operations on that data, in mind

examples:
java.util.ArrayList, java.util.HashMap,
java.util.TreeSet
2
Java's Collection interface

Java provides an interface java.util.Collection
to represent many kinds of collections. It has the
following methods:
public boolean add(Object o)
Appends the specified element to this collection.
public void clear()
Removes all of the elements of this collection.
public boolean contains(Object o)
Returns true if this collection contains the specified element.
public boolean containsAll(Collection coll)
Returns true if this collection contains all of the elements in the
specified collection.
3
Collection interface,cont'd.
public boolean isEmpty()
Returns true if this list contains no elements.
public Iterator iterator()
Returns a special object for examining the elements of the list in
order (seen later).
public boolean remove(Object o)
Removes the first occurrence in this list of the specified element.
public int size()
Returns the number of elements in this list.
public Object[] toArray()
Returns an array containing all of the elements from this list.
4
An example collection: List

list: an ordered sequence of elements, each
accessible by a 0-based index

one of the most basic collections
5
List features



ORDERING: maintains order elements were added
(new elements are added to the end by default)
DUPLICATES: yes (allowed)
OPERATIONS: add element to end of list, insert
element at given index, clear all elements, search for
element, get element at given index, remove element
at given index, get size


some of these operations are inefficient! (seen later)
list manages its own size; user of the list does not
need to worry about overfilling it
6
Java's List interface

Java also has an interface java.util.List
to represent a list of objects. It adds the
following methods to those in Collection:
(a partial list)
public void add(int index, Object o)
Inserts the specified element at the specified position in this list.
public Object get(int index)
Returns the element at the specified position in this list.
public int indexOf(Object o)
Returns the index in this list of the first occurrence of the
specified element, or -1 if the list does not contain it.
7
List interface, cont'd.
public int lastIndexOf(Object o)
Returns the index in this list of the last occurrence of the specified
element, or -1 if the list does not contain it.
public Object remove(int index)
Removes the object at the specified position in this list.
public Object set(int index, Object o)
Replaces the element at the specified position in this list with the
specified element.

Notice that the methods added to Collection by
List all deal with indexes; a list has indexes while a
general collection may not
8
Some list questions




all of the list operations on the previous slide
could be performed using an array instead!
open question: What are some reasons why we
might want to use a list class, rather than an
array, to store our data?
thought question: How might a List be
implemented, under the hood?
why do all the List methods use type Object?
9
Array list

array list: a list implemented using an array to
store the elements



encapsulates array and # of elements (size)
in Java: java.util.ArrayList
when you want to use ArrayList, remember to
import java.util.*;
10
ArrayList features




think of it as an auto-resizing array that can
hold any type of object, with many convenient
methods
maintains most of the benefits of arrays, such
as fast random access
frees us from some tedious operations on
arrays, such as sliding elements and resizing
can call toString on an ArrayList to print
its elements

[1, 2.65, Marty Stepp, Hello]
11
Collections class

Class java.util.Collections has many utility
methods that do useful things to collections
public
public
public
public
public
public
public
public

static
static
static
static
static
static
static
static
void copy(List dest, List src)
void fill(List list, Object value)
Object max(Collection coll)
Object min(Collection coll)
void reverse(List list)
void shuffle(List list)
void sort(List list)
void swap(List list, int i, int j)
Example:
Collections.sort(myArrayList);
12
Analysis of ArrayList runtime
OPERATION
add to start of list
add to end of list
add at given index
clear
get
find index of an object
remove first element
remove last element
remove at given index
set
size
toString
RUNTIME (Big-Oh)
O(n)
O(1)
O(n)
O(1)
O(1)
O(n)
O(n)
O(1)
O(n)
O(1)
O(1)
O(n)
13
Open questions



Based on the preceding analysis, when is an
ArrayList a good collection to use? When is
it a poor performer?
Is there a way that we could fix some of the
problems with the ArrayList?
Should we represent our list in a different way?
14
some ArrayList problems





an insertion into a full list causes a large
reallocation
an insertion to the front "slides" down the
subsequent items (slow!)
a removal also "slides" down subsequent items
still need to use indexes/subscripts a lot
somewhat ugly syntax to use it
15
The underlying issue


the elements of an ArrayList are too tightly
attached; can't easily rearrange them
can we break the element storage apart into a
more dynamic and flexible structure?
16
Nodes: objects to store elements



let's make a special "node" type of object that
represents a storage slot to hold one element
of a list
each node will keep a reference to the node
after it (the "next" node)
the last node will have next == null
(drawn as / ), signifying the end of the list
17
Node implementation
/* Stores one element of a linked list. */
public class Node {
public Object element;
public Node next;
public Node(Object element) {
this(element, null);
}
public Node(Object element, Node next) {
this.element = element;
this.next = next;
}
}
18
Linked node problems (a)

Let's examine sample chains of nodes together,
and try to write the correct code for each

each Node stores an Integer object
1.
 before:

after:
19
Linked node problems (b)
2.
 before:

after:
3.
 before:

after:
20
Linked node problems (c)
4.

before:

after:
5.
 before:

after:
21
Linked node problems (d)
6.
 before:

after:
7.

before:

after:
22
Linked list

linked list: a list implemented using a linked
sequence of nodes


the list only needs to keep a reference to the first
node (we might name it myFront)
in Java: java.util.LinkedList
(but we'll write our own)
23
Linked list implementation
/* Models an entire linked list. */
public class MyLinkedList {
private Node myFront;
public MyLinkedList() {
myFront = null;
}
/* Methods go here */
}
24
Some list states of interest



empty list
(myFront == null)
list with one element
list with many
elements
25
Let's draw them together...

an add operation





at the front, back, and middle
a remove operation
a get operation
a set operation
an index of (searching) operation
26
Analysis of LinkedList runtime
OPERATION
add to start of list
add to end of list
add at given index
clear
get
find index of an object
remove first element
remove last element
remove at given index
set
size
toString
RUNTIME (Big-Oh)
O(1)
O(n)
O(n)
O(1)
O(n)
O(n)
O(1)
O(n)
O(n)
O(n)
O(n)
O(n)
27
An optimization: mySize


problem: array list has a O(1) size method,
but the linked list needs O(n) time
solution: add a mySize field to our linked list

what changes must be made to the implementation
of the methods of the linked list?
28
A variation: dummy header

dummy header: a front node intentionally left
blank



myFront always refers to dummy header
(myFront will never be null)
requires minor modification to many methods
surprisingly, makes implementation much easier
29
An optimization: myBack


problem: array list has O(1) get/remove of last
element, but the linked list needs O(n)
solution: add a myBack pointer to the last node


which methods' Big-Oh runtime improve to O(1)?
what complications does this add to the
implementation of the methods of the list?
30
Doubly-linked lists

add a prev pointer to our Node class
allows backward iteration (for ListIterator)

some methods need to be modified



when adding or removing a node, we must fix the
prev and next pointers to have the correct value!
can make it easier to implement some methods
such as remove
31
Combining the approaches


Most actual linked list implementations are
doubly-linked and use a dummy header and
dummy tail
this actually makes a very clean implementation
for all linked list methods and provides good
efficiency for as many operations as possible
32
Improved LinkedList runtime
OPERATION
add to start of list
add to end of list
add at given index
clear
get
find index of an object
remove first element
remove last element
remove at given index
set
size
toString
RUNTIME (Big-Oh)
O(1)
O(1)
O(n)
O(1)
O(n)
O(n)
O(1)
O(1)
O(n)
O(n)
O(1)
O(n)
33
A particularly slow idiom
// print every element of linked list
for (int i = 0; i < list.size(); i++) {
Object element = list.get(i);
System.out.println(i + ": " + element);
}

This code executes an O(n) operation (get)
every time through a loop that runs n times!


Its runtime is O(n2), which is much worse than O(n)
this code will take prohibitively long to run for large
data sizes
34
The problem of position

The code on the previous slide is wasteful
because it throws away the position each time



every call to get has to re-traverse the list!
it would be much better if we could somehow
keep the list in place at each index as we
looped through it
Java uses special objects to represent a
position of a collection as it's being examined...

these objects are called "iterators"
35
Iterators in Java

interface java.util.Iterator



public boolean hasNext()
Returns true if there are more elements to see
public Object next()
Returns the next object
in this collection, then
advances the iterator;
throws an exception
if no more elements
remain
public void remove()
Deletes the element that was
last returned by next (not always supported)
36
Iterators on array lists


An iterator on an array list is simple; it
maintains an index of its current node
iterators are not as useful on array lists




Why?
What is the Big-Oh of get and set?
What is the Big-Oh of add and remove?
Does the array list iterator improve this?
37
Iterators on linked lists


an iterator on a linked list maintains (at least)
its current index and a reference to that node
when iterator() is called on a linked list, the
iterator initially refers to the first node (index 0)
38
Linked list iterator iteration

when next() is called, the iterator:




grabs the current myNode's element value (32)
follows the next pointer on its node and increments its
index
returns the element it grabbed (32)
hasNext is determined by whether myNode is null

(Why?)
39
How does remove work?


remove is supposed to remove the last value
that was returned by next
to do this, we need to delete the node before
myNode, which may require modification
40
Fixing the slow LL idiom
// print every element of the list
for (int i = 0; i < list.size(); i++) {
Object element = list.get(i);
System.out.println(i + ": " + element);
}
Iterator itr = list.iterator();
for (int i = 0; itr.hasNext(); i++) {
Object element = itr.next();
System.out.println(i + ": " + element);
}
41
Iterator usage example
MyLinkedList names = new MyLinkedList();
// ... fill the list with some data ...
// print every name in the list, in upper case
Iterator itr = myList.iterator();
while (itr.hasNext()) {
String element = (String)itr.next();
System.out.println(element.toUpperCase());
}
// remove strings from list that start with "m"
itr = myList.iterator();
while (itr.hasNext()) {
String element = (String)itr.next();
if (element.startsWith("m"))
itr.remove(); // remove element we just saw
42
}
Benefits of iterators

speed up loops over linked lists' elements


What is the Big-Oh of each iterator method?
provide a unified way to examine all elements
of a collection

every collection in Java has an iterator method



in fact, that's the only guaranteed way to examine the
elements of any Collection (see Slide 4)
don't need to look up different collections' method
names to see how to examine their elements
don't have to use indexes as much on lists
43
Iterator is still not perfect
// print odd-valued elements, with their indexes
Iterator itr = list.iterator();
for (int i = 0; itr.hasNext(); i++) {
int element = ((Integer)itr.next()).intValue();
if (element % 2 == 1)
System.out.println(i + ": " + element);
i++;
}


We still had to maintain the index variable i so that
we could print the index of each element
Wouldn't it be nice if the iterator could just tell us the
index?
44
More iterator problems
// add a 0 after any odd element
Iterator itr = list.iterator();
int i = 0;
while (itr.hasNext()) {
int element = ((Integer)itr.next()).intValue();
if (element % 2 == 1)
list.add(i, new Integer(0));
// fails
}



can't use the iterator to add, set element values
(iterator is programmed to crash if list is modified)
the iterator speeds up get and remove loops only
the iterator really should be able to help us speed up
loops that add elements or set elements' values!
45
ListIterator interface

Java interface java.util.ListIterator extends
Iterator to add these methods:






public
public
public
public
void add(Object o)
int nextIndex()
int previousIndex()
void set(Object o)
ListIterator is also able to iterate in reverse:
 public boolean hasPrevious()
 public Object previous()
get a ListIterator by calling these methods on the
List: (both ArrayList and LinkedList have them)


public ListIterator listIterator()
public ListIterator listIterator(int index)
46
ListIterator usage
// print odd-valued elements, with their indexes
for (ListIterator itr = list.listIterator();
itr.hasNext(); ) {
int element = ((Integer)itr.next()).intValue();
if (element % 2 == 1)
System.out.println(itr.previousIndex() + ": " +
element);
}
// add a 0 after any odd element
for (ListIterator itr = list.listIterator();
itr.hasNext(); ) {
int element = ((Integer)itr.next()).intValue();
if (element % 2 == 1)
itr.add(0);
}
47
ListIterator benefits

using the ListIterator internally to
implement the methods of the linked list can
make the code cleaner and simpler:
public Object get(int index) {
ListIterator itr = listIterator(index);
return itr.next();
}

should the code check the index for validity?
48
Summary




lists are ordered, integer-indexed collections
that allow duplicates
lists are good for storing elements in order of
insertion and traversing them in that order
linked lists are faster for add / remove
operations at the front and back, but slower
than array lists for arbitrary get / set operations
lists are bad for searching for elements and for
lots of arbitrary add / remove operations
49