Stacks

Dr. Hakim Mellah

Department of Computer Science & Software Engineering
Concordia University, Montreal, Canada

These slides has been extracted, modified and updated from original slides of :
• Data Structures and Algorithms in Java, 5th edition. John Wiley& Sons, 2010. ISBN 978-0-470-38326-1.
• Dr. Hanna’s slides (http://aimanhanna.com/concordia/comp352/index.htm)

Copyright © 2010 Michael T. Goodrich, Roberto Tamassia

All rights reserved
Abstract Data Types (ADTs)
❑ An abstract data type (ADT) is an
abstraction/model of a data structure.

❑ An abstract data type is defined indirectly, only

by the operations that may be performed on
it. An ADT specifies:
 Data stored
 Operations on the data
 Error conditions associated with operations

Stacks 2
Abstract Data Types (ADTs)
❑ Example: ADT modeling a simple stock trading
system
◼ The data stored are buy/sell orders
◼ The operations supported are
 order buy(stock, shares, price)
 order sell(stock, shares, price)
 void cancel(order)

◼ Error conditions:
 Buy/sell a nonexistent stock
 Cancel a nonexistent order

Stacks 3
The Stack ADT
❑ The Stack ADT stores arbitrary objects.

❑ Insertions and deletions follow the last-in first-out

(LIFO) scheme. Think of a spring-loaded plate
dispenser.

❑ Formally, a stack is an ADT that supports the following

main operations:
◼ push(object): inserts an element
◼ object pop(): removes and returns the last inserted element

❑ Examples: operations of “Back” button on a browser

or “undo” on text editors.
Stacks 4
The Stack ADT
❑ Secondary stack operations include:
◼ object top(): returns the last inserted element
without removing it

◼ integer size(): returns the number of elements

stored

◼ boolean isEmpty(): indicates whether no elements

are stored

Stacks 5
The Stack ADT
❑ The following table shows a series of stack operations
and their effects on an initially empty stack of integers:
Operation Output Stack Contents
push(5) -- [5]
push(2) -- [5, 2]
push(8) -- [5, 2, 8]
pop() 8 [5, 2]
isEmpty() false [5, 2]
top() 2 [5, 2]
pop() 2 [5]
pop() 5 []
pop() ”error” []
Stacks 6
The Java Built-in Stack Class
❑ Because of its importance, Java has a built-in class for
the stack (java.util.Stack).

❑ The class has various methods, including:

◼ push(), pop(), peek(), empty(), and size().

❑ pop() and peek() throw EmptyStackException if operations

are attempted on an empty stack. .

❑ While this class is convenient, it is very important to

know how to design and implement a Stack class from
scratch.
Stacks 7
Exceptions
❑ Attempting the ❑ In the Stack ADT,
execution of an operations pop and
operation of ADT may top cannot be
sometimes cause an performed if the stack
error condition, called is empty.
an exception. ❑ Attempting the
❑ Exceptions are said to execution of pop or
be “thrown” by an top on an empty
operation that cannot stack throws an
be executed. EmptyStackException.
Stacks 8
Applications of Stacks
❑ Direct applications
◼ Page-visited history in a Web browser
◼ Undo sequence in a text editor
◼ Chain of method calls in the Java Virtual
Machine
❑ Indirect applications
◼ Auxiliary data structure for algorithms
◼ Component of other data structures

Stacks 9
Method Stack in the JVM
❑ The Java Virtual Machine (JVM) main() {
keeps track of the chain of int i = 5;
active methods with a stack. bar
foo(i); PC = 1
❑ When a method is called, the } m=6
JVM pushes on the stack a
frame containing foo(int j) {
foo
◼ Local variables and return value int k; PC = 3
◼ Program counter, keeping track of k = j+1; j=5
the statement being executed bar(k); k=6
❑ When a method ends, its frame }
is popped from the stack and main
control is passed to the method bar(int m) { PC = 2
on top of the stack. … i=5
❑ Allows for recursion. }
Stacks 10
Array-based Stack
❑ A simple way of Algorithm size()
implementing the return t + 1
Stack ADT uses an
array. Algorithm pop()
❑ We add elements if isEmpty() then
from left to right. throw EmptyStackException
❑ A variable keeps else
track of the index of tt−1
the top element. return S[t + 1]
◼ Initialized to -1 when stack
is created.
…
S
0 1 2 t
Stacks 11
Array-based Stack (cont.)
❑ The array storing the
stack elements may
Algorithm push(o)
become full.
if t = S.length − 1 then
❑ A push operation will
throw FullStackException
then throw a
FullStackException else
◼ Limitation of the array- tt+1
based implementation S[t]  o
◼ We need to define this
class; it is not intrinsic
to the Stack ADT

…
S
0 1 2 t
Stacks 12
Performance and Limitations
❑ Performance
◼ Let n be the number of elements in the stack
◼ The space used is O(n)
◼ Each operation runs in time O(1)
❑ Limitations
◼ The maximum size of the stack must be defined a
priori and cannot be changed
◼ Trying to push a new element into a full stack
causes an implementation-specific exception

Stacks 13
Array-based Stack in Java *

public class ArrayStack<E> public E pop()

implements Stack<E> { throws EmptyStackException {
if isEmpty()
// holds the stack elements
throw new EmptyStackException
private E S[ ];
(“Empty stack: cannot pop”);
// index to top element E temp = S[top];
private int top = -1; // facilitate garbage collection:
S[top] = null;
// constructor
top = top – 1;
public ArrayStack(int capacity) {
return temp;
S = (E[]) new Object[capacity]);
}
}
… (other methods of Stack interface)
* Notice that this is not a built-in Java implementation

Stacks 14
Example use in Java
/** A non-recursive generic method for reversing an array */
public static <E> void reverse(E[] a){
Stack<E> s = new ArrayStack<E>(a.length);
for(int i = 0; i < a.length; i++)
s.push(a[i]);
for(int i = 0; i < a.length; i++)
a[i] = s.pop();
}
Time complexity is: O(n)
Space complexity is: O(n )
Stacks 15
Example use in Java
public class Tester {
// … other methods public floatReverse(Float f[]) {
public intReverse(Integer a[]) { Stack<Float> s;
Stack<Integer> s; s = new ArrayStack<Float>();
s = new ArrayStack<Integer>(); … (code to reverse array f) …
… (code to reverse array a) … }
}

Stacks 16
Better Stack Implementation
❑ Use linked lists instead of arrays.

❑ No need to define a maximum size.

❑ When push(), add new element/node at the

tail of the list.

❑ pop() removes the node at the tail of the list.

What is the complexity?
Stacks 17
Parentheses Matching
❑ Each “(”, “{”, or “[” must be paired with
a matching “)”, “}”, or “[”
◼ correct: ( )(( )){([( )])}
◼ correct: ((( )(( )))){([( )])}
◼ incorrect: )(( )){([( )])}
◼ incorrect: ({[ ])}
◼ incorrect: (

Stacks 18
 Parentheses Matching Algorithm
Algorithm ParenMatch(X,n):
Input: An array X of n tokens, each of which is either a grouping symbol, a
variable, an arithmetic operator, or a number
Output: true if and only if all the grouping symbols in X match
Let S be an empty stack
for i=0 to n-1 do
if X[i] is an opening grouping symbol then
S.push(X[i])
else if X[i] is a closing grouping symbol then
Notice that item is if S.isEmpty() then
removed here
return false {nothing to match with}
if S.pop() does not match the type of X[i] then
return false {wrong type}
if S.isEmpty() then
return true {every symbol matched}
else return false {some symbols were never matched}
Stacks 19
HTML Tag Matching
For fully-correct HTML, each <name> should pair with a matching </name>

<body>
<center> The Little Boat
<h1> The Little Boat </h1>
</center> The storm tossed the little boat
<p> The storm tossed the little
like a cheap sneaker in an old
boat like a cheap sneaker in an
old washing machine. The three washing machine. The three
drunken fishermen were used to drunken fishermen were used to
such treatment, of course, but such treatment, of course, but not
not the tree salesman, who even as the tree salesman, who even as
a stowaway now felt that he a stowaway now felt that he had
had overpaid for the voyage. </p> overpaid for the voyage.
<ol>
<li> Will the salesman die? </li>
<li> What color is the boat? </li> 1. Will the salesman die?
<li> And what about Naomi? </li> 2. What color is the boat?
</ol> 3. And what about Naomi?
</body>

Stacks 20
Evaluating Arithmetic Slide by Matt Stallmann
included with permission.

Expressions
14 – 3 * 2 + 7 = (14 – (3 * 2) ) + 7
Operator precedence
* has precedence over + and –
Example: x + y * z is:
x + (y * z) rather than (x + y) * z
Associativity
operators of the same precedence group
evaluated from left to right
Example: x – y + z is:
(x – y) + z rather than x – (y + z)
Idea: push each operator on the stack, but first pop and
perform higher and equal precedence operations.
Stacks 21
Algorithm for
Evaluating Expressions
Two stacks:
❑ opStk holds operators Algorithm EvalExp()
Input: a stream of tokens representing an
arithmetic expression (with numbers)
❑ Use $ to hold a special “end of input”
Output: the value of the expression
token with lowest precedence
Algorithm doOp()
x  valStk.pop();
while there’s another token z
y  valStk.pop(); if isNumber(z) then
op  opStk.pop(); valStk.push(z)
Force the execution of
valStk.push( y op x ) else all remaining operators
Algorithm repeatOps( refOp ) repeatOps(z);
while ( valStk.size() > 1  opStk.push(z)
prec(refOp) ≤ repeatOps($);
prec(opStk.top()) return valStk.top()
doOp()

Stacks 22
Algorithm on an Slide by Matt Stallmann
included with permission.

Example Expression
Operator ≤ has lower
14 ≤ 4 – 3 * 2 + 7 precedence than + and –

4 –
14 ≤

3 *
$ $
4 –
7 $
14 ≤ F
+ -2 + 5
+ 14 ≤ 14 ≤
2 2
3 * 3 * 6
4 – 4 – 4 – -2 +
14 ≤ 14 ≤ 14 ≤ 14 ≤

Stacks 23
Computing Spans (not in book)
7
❑ Using a stack as an auxiliary
6
data structure in an algorithm
5
❑ Given an array X, the span
4
S[i] of X[i] is the maximum
number of consecutive
3
elements X[j] immediately 2
preceding X[i] and such that 1
X[j]  X[i] 0
❑ Spans have applications to 0 1 2 3 4
financial analysis
◼ E.g., stock at 52-week high X 6 3 4 5 2
S 1 1 2 3 1
Stacks 24
Quadratic Algorithm
Algorithm spans1(X, n)
Input array X of n integers
Output array S of spans of X #
S  new array of n integers n
for i  0 to n − 1 do n
s1 n
while s  i  X[i − s]  X[i] 1 + 2 + …+ (n − 1)
ss+1 1 + 2 + …+ (n − 1)
S[i]  s n
return S 1

Algorithm spans1 runs in O(n2) time

Stacks 25
 Computing Spans with a Stack
❑ We keep in a stack the indices of the 7
elements visible when “looking back”
❑ We scan the array from left to right
6
◼ Let i be the current index
5
◼ We push the index as long as the 4
one prior to it has a smaller value 3
◼ We pop all elements otherwise
2
Each index of the array
1
◼ Is pushed into the stack exactly
once 0
◼ Is popped from the stack at most 0 1 2 3 4
once
❑ Stack height for the pushed index
represents the needed value
Stacks 26
Linear Algorithm
We iterate on the array n times Algorithm spans2(X, n) #
S  new array of n integers n
Each index is then pushed and A  new empty stack 1
popped at most once, which for i  0 to n − 1 do n
totals to n + n
while (A.isEmpty() 
X[A.top()]  X[i] ) do n
We record the value at each
index of the resulted Span A.pop() n
array, which totals to n times if A.isEmpty() then n
S[i]  i + 1 n
Consequently algorithm spans2 else
has a complexity of O(n) S[i]  i − A.top() n
A.push(i) n
return S 1

Stacks 27
 Computing Spans with a Stack
X
index 0 1 2 3 4 5 6 7 8
❑ Example
Value 60 30 40 50 20 70 30 80 20

for i  0 to n − 1 do

while (A.isEmpty() 
X[A.top()]  X[i] ) do
A.pop()

if A.isEmpty() then

S[i]  i + 1

4 else
S[i]  i − A.top()
1 2 3 3 6 8
A.push(i)
0 0 0 0 0 5 5 7 7
return S

S 1 1 2 3 1 6 1 8 1
28
Stacks
Growable Array-based Stack
❑ In a push() operation, if the
stack is full (no more empty Algorithm push(o)
locations in the array), we can if t = S.length − 1 then
throw an exception and
abort/reject the operation. A  new array of
size …
❑ Alternatively, we can extend the for i  0 to t do
array; which is actually A[i]  S[i]
replacing it with a larger one. SA
tt+1
❑ How large should the new array S[t]  o
be?
◼ Incremental strategy: increase the
size by a constant c
◼ Doubling strategy: double the size
Stacks 29
Comparison of the Strategies
❑ We compare the incremental strategy and the
doubling strategy by analyzing the total time T(n)
needed to perform a series of n push() operations.

❑ We assume that we start with an empty stack

represented by an array of size 1.

❑ We refer to the average time taken by a push()

operations over the series of operations, i.e., T(n)/n,
the amortized time of a push() operation.

Stacks 30
Incremental Strategy Analysis
❑ We need to find the amortized time to perform
one push() operation.
◼ That is the total time to perform n push()
operations / n.

❑ In general, we need to replace the array k =

n/c times for all n push() to take place.
◼ For instance if n = 100, and c = 4, we need to go
through 25 (100/4) replacements for all push()
operations to take place.
◼ Notice also that each replacement is larger than the
previous one by c. Stacks 31
Incremental Strategy Analysis
❑ Consequently, the total time T(n) of a series of
n push() operations is proportional to:
n + c + 2c + 3c + 4c + … + kc =
n + c(1 + 2 + 3 + … + k) =
n + ck(k + 1)/2
❑ Since c is a constant, T(n) is O(n + k2), i.e.,
O(n2). That is
❑ T(n) is the complexity to perform n push()
operations. Hence, the amortized time of one
single push() operation is O(n).
Stacks 32
Doubling Strategy Analysis
❑ We replace the array k = log2 n
times.
◼ For instance, to perform 1000 push() geometric series
operations, we need to expand the
array 10 times (1 -> 2 -> 4 -> 8 ->
16 -> 32 -> 64 -> 128 -> 256 -> 2
512 -> 1024). 4
1 1
❑ The total time T(n) of a series of n
add operations is proportional to 8
n + 1 + 2 + 4 + 8 + …+ 2k =
n + 2k + 1 − 1 = n + 2n -1 =
3n − 1.

Stacks 33
Doubling Strategy Analysis
❑ Consequently, T(n) (which is
needed to perform n push()
geometric series
operations) is O(n).
2
4
❑ Hence, the amortized time of a 1 1
single push() operation is O(1).
8

Stacks 34