Data Structures for Java

William H. Ford
William R. Topp


Chapter 6
Recursion

2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Bret Ford
2005, Prentice Hall
The Concept of Recursion
An algorithm is recursive if it can be broken
into smaller problems of the same type.
Continue this decomposition until one or
more of the smaller problems has a solution.
Backing up a directory and its subdirectories is
recursive. At each step, the backup copies files
in the current directory and then repeats the
process in a subdirectory. Eventually, the
backup reaches a directory with no
subdirectories.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Iterative factorial()
n-factorial is the product of the first n
integers.
n! = n * (n-1) * (n-2) * ... * 2 * 1
n! can be computed iteratively.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
// n is a non-negative integer
public static int factorial(int n)
{
int factValue = 1;

while (n >= 1)
{
factValue *= n;
n--;
}

return factValue;
}
Recursive Computation of n!
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Define 0! = 1

4! = 4 * 3! // multiplication awaits evaluation of 3!
3! = 3 * 2! // need value of 2!
2! = 2 * 1! // need value of 1!
1! = 1 * 0! // need value of 0!
0! = 1 // know 0! = 1 (stopping condition)

1! = 1 * 0! // compute 1! with value of 0!
= 1
2! = 2 * 1! // compute 2! with value of 1!
= 2 * 1 = 2
3! = 3 * 2! // compute 3! with value of 2!
= 3 * 2 = 6
4! = 4 * 3! // compute 4! with value of 3!
= 4 * 6 = 24

2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Recursive Computation of n!
(continued)
The computation of n! can be viewed
recursively as follows:
n! = n * (n-1)!
(n-1)! = (n-1) * (n-2)!
...
2! = 2 * 1
1! = 1 * 0!
0! = 1 (stopping condition)

After reaching the stopping condition, revisit the
multiplication steps in reverse order.

>
==
=
1 )! 1 *(
0 1
!
n n n
n
n
Describing a Recursive Algorithm
The design of a recursive method consists of
the following elements:
One or more stopping conditions that can be
directly evaluated for certain arguments.
One or more recursive steps, in which a current
value of the method can be computed by
repeated calling of the method with arguments
that will eventually arrive at a stopping condition.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Implementing Recursive Methods
Use an if-else statement to distinguish
between a stopping condition and a recursive
step.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
// n is a non-negative integer
public static double factorial(int n)
{
if (n == 0)
// stopping condition
return 1;
else
// recursive step
return n * factorial(n-1);
}
Recursive Computation of
1 + 2 + ... + n

> +
=
=
step recursive 1 n n 1) S(n
condition stopping 1 n 1
S(n)
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Example: S(5) = 5 + (4 + 3 + 2 + 1) = 5 + S(4)
public static int sumToN(int n)
{
if (n == 1)
return 1;
else
return sumToN(n-1) + n;
}
Infinite Recursion
A recursive algorithm must lead to a
stopping condition. Infinite recursion
occurs when a stopping condition is never
reached.
Example:

> + +
=
=
1 , ) 1 4 / (
0 , 0
) (
n n n f
n
n f
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
f(5) = f(5/4 + 1) + 5; // recursive call to f(5/4 + 1) = f(2)
f(2) = f(2/4 + 1) + 2; // recursive call to f(2/4 + 1) = f(1)
f(1) = f(1/4 + 1) + 1; // recursive call f(1/4 + 1) = f(1)
. . . // infinite recursion!
How Recursion Works
Think of factorial(n) as a machine called n-Fact
that computes n! by carrying out the
multiplication n*(n-1)!. The machine must be
networked with a series of other n-Fact
machines that pass information back and
forth.
The 0-Fact machine is an exception and can
work independently and produce the result 1
without assistance from another machine.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
The n-Fact Machine
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
4-Fact Starts up the 3-Fact machine
Waits for return value (3!) from 3-Fact.
Computes 4 * 3! = 24 and returns to main
3-Fact Starts up the 2-Fact machine
Waits for return value (2!) from 2-Fact.
Computes 3 * 2! = 6 and returns to 4-Fact
2-Fact Starts up the 1-Fact machine
Waits for return value (1!) from 1-Fact.
Computes 2 * 1! = 2 and returns to 3-Fact
1-Fact Starts up the 0-Fact machine
Waits for return value (1!) from 1-Fact.
Computes 1 * 0! = 1 and returns to 2-Fact
0-Fact Determines that 0! = 1
Returns 1 to 1-Fact
The n-Fact Machine
(concluded)
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Multibase Representations
Decimal is only one representation for
numbers. Other bases include 2 (binary), 8
(octal), and 16 (hexadecimal).
Hexadecimal uses the digits 0-9 and a=10, b=11,
c=12, d=13, e=14, f=16.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
95 = 1011111
2
// 95 = 1(2
6
)+0(2
5
)+0(2
4
)+0(2
3
)+1(2
2
)+1(2
1
)+1(2
0
)
= 1(64)+0(32)+1(16)+1(8)+1(4)+1(2)+1
95 = 340
5
// 95 = 3(52) + 4(51) + 0(5
0
)
= 3(25) + 4(5) + 0
95 = 137
8
// 95 = 1(82) + 3(81) + 7(8
0
)
= 1(64) + 3(8) + 7
748 = 2ec
16
// 748 = 2(16
2
) + 14(16
1
) + 12(16
0
)
= 2(256) + 14(16) + 12
= 512 + 224 + 12

Multibase Representations (continued)
An integer n > 0 can be represented in
different bases using repeated division.
Generate the digits of n from right to left using
operators '%' and '/'. The remainder is the next
digit and the quotient identifies the remaining
digits.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Multibase Representations
(continued)
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Multibase Representations
(continued)
Convert n to base b by converting the smaller
number n/b to base b (recursive step) and
adding the digit n%b.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Multibase Representations
(continued)
// returns string representation
// of n as a base b number
public static String baseString(int n, int b)
{
String str = "", digitChar = "0123456789abcdef";

// if n is 0, return empty string
if (n == 0)
return "";
else
{
// get string for digits in n/b
str = baseString(n/b, b); // recursive step

// return str with next digit appended
return str + digitChar.charAt(n % b);
}
}
Multibase Representations
(concluded)
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.1
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.1 (continued)
<import statements>

public class Program6_1 extends JFrame
{
JTextField decimalField, baseField;
JButton binButton, octButton, hexButton;
JLabel baseLabel;

public static void main(String[] args)
{
Program6_1 app = new Program6_1();
app.setVisible(true);
}

public Program6_1()
{
<set up text fields and labels>
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.1 (continued)
// create binButton and add the action listener
binButton = new JButton("Binary");
binButton.addActionListener(new ConvertNumber());
. . .
}
private class ConvertNumber implements ActionListener
{
public void actionPerformed(ActionEvent ae)
{
// getSource() method of ae returns reference
// to the pressed button
JButton buttonPressed =
(JButton)ae.getSource();
int n = Integer.parseInt(decimalField.getText());
String str;
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.1 (concluded)
if (buttonPressed == binButton)
{ baseLabel.setText(" Binary");
str = baseString(n, 2);
}
else if (buttonPressed == octButton)
{
baseLabel.setText(" Octal");
str = baseString(n, 8);
}
else
{ baseLabel.setText(" Hex");
str = baseString(n, 16);
}
baseField.setText(str);
}
}
}
Building a Ruler
Assume that the method drawMark() takes a
point x and an integer value h as parameters
and draws a vertical line at point x with size
proportional to h.
The recursive method drawRuler() takes a
range low x high and an integer h and
draws a ruler.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Building a Ruler (continued)
drawRuler() draws a vertical line at
mid = (low + high)/2 whose height is
proportional to h. Using the midpoint to
separate the interval into half-lines, make two
recursive calls to drawRuler() that draw
vertical lines at the middle of each half-line
with height proportional to h-1. Continue in
this fashion until lines with height
proportional to h = 1 are drawn (stopping
condition).
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Building a Ruler (continued)
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Method drawRuler()
public static void drawRuler(double low,
double high, int h)
{
double midpt;
if (h >= 1)
{
// find the midpoint of interval (low,high)
midpt = (high + low)/2;
// draw a mark at the midpoint proportional to h
drawMark(midpt, h);
// draw all marks on the left half-interval with h-1
drawRuler(low, midpt, h - 1);
// draw all marks on the right half-interval with h-1
drawRuler(midpt, high, h - 1);
}
}
Run of Application
Ruler.java
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Towers of Hanoi
You are given a stack of n graduated disks and
a set of three needles called A, B, and C. The
initial setup places the n disks on needle A.
The task is to move the disks one at a time
from needle to needle until the process
rebuilds the original stack, but on needle C. In
moving a disk, a larger disk may never be
placed on top of a smaller disk.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Towers of Hanoi (continued)
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Towers of Hanoi Example
Stage 1
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Towers of Hanoi Example
Stage 2
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Towers of Hanoi Example
Stage 3
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Method hanoi()
The method hanoi() has arguments n, the
number of disks, and three string arguments
that denote the name of the starting needle
(initNeedle), the destination needle
(endNeedle), and the intermediate needle
(tempNeedle) that temporarily holds disks
during the moves.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
// move n disks from initNeedle to endNeedle
// using tempNeedle for temporary storage
public static void hanoi(int n, String initNeedle,
String endNeedle, String tempNeedle)
Method hanoi() (continued)
Move n-1 disks from initNeedle to
tempNeedle using endNeedle for temporary
storage.
hanoi(n-1, initNeedle, tempNeedle, endNeedle);
Move largest disk to endNeedle.
System.out.println("Move " + initNeedle + " to " +
endNeedle);
Move n-1 disks from tempNeedle to
endNeedle using initNeedle for temporary
storage.
hanoi(n-1, tempNeedle, endNeedle, initNeedle);
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Method hanoi() (continued)
// move n disks from initNeedle to endNeedle, using tempNeedle
// for intermediate storage of the disks
public static void hanoi(int n, String initNeedle,
String endNeedle,
String tempNeedle)
{
// stopping condition: move one disk
if (n == 1)
System.out.println("move " + initNeedle +
" to " + endNeedle);
else
{
// move n-1 disks from initNeedle to
// tempNeedle using endNeedle
// for temporary storage
hanoi(n-1,initNeedle,tempNeedle,endNeedle);
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Method hanoi() (concluded)
// move largest disk to endNeedle
System.out.println("move " + initNeedle +
" to " + endNeedle);

// move n-1 disks from tempNeedle to
// endNeedle using initNeedle
// for temporary storage
hanoi(n-1,tempNeedle,endNeedle,initNeedle);
}
}
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.2
import java.util.Scanner;

public class Program6_2
{
public static void main(String[] args)
{
// number of disks and the needle names
int n;
String beginNeedle = "A",
middleNeedle = "B",
endNeedle = "C";

// the keyboard input stream
Scanner keyIn = new Scanner(System.in);

// prompt for n and solve
// the puzzle for n disks
System.out.print("Enter the number of disks: ");
n = keyIn.nextInt();
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.2 (concluded)
System.out.println("The solution for n = " + n);
hanoi(n, beginNeedle, endNeedle, middleNeedle);
}

< method hanoi() listed in the program discussion >
}
Run:

Enter the number of disks: 3
The solution for n = 3
Move A to C
Move A to B
Move C to B
Move A to C
Move B to A
Move B to C
Move A to C
Graphical Demonstration
for Towers of Hanoi
In Chapter 6 of the software supplement, the
program HanoiGraphic.java provides a
graphical version of the Towers of Hanoi for 1
to 6 disks. The figure shows two frames in a
run with four disks.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Evaluating Recursion
Sometimes recursion simplifies algorithm
design, but it is sometimes not efficient and
an iterative algorithm is preferable.
For the Towers of Hanoi, a recursive solution is
elegant and easier to code than the corresponding
iterative solution.
For some problems, a recursive solution is
inefficient.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Fibonacci Numbers
Fibonacci numbers are the sequence of
integers beginning with position n = 0. The
first two terms are 0 and 1 by definition.
Each subsequent term, beginning at n = 2, is
the sum of the two previous terms.
n Value Sum
2 1 0+1
3 2 1+1
4 3 1+2
5 5 2+3
6 8 3+5
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Recursive method fib()
// compute Fibonacci number n using recursion
public static int fib(int n)
{
if (n <= 1) // stopping conditions
return n;
else
return fib(n-1) + fib(n-2); // recursive step
}

>
=
=
+
=
2 n
1 n
0 n
2), - fib(n 1) - fib(n
1,
0,
fib(n)
Recursive method fib()
(continued)
The recursive method fib() makes multiple
calls to itself with the same argument. This
creates enounous redundancy. If numCall(n) is
the number of recursive calls required to
evaluate fib(n), then it can be shown that
numCall(n) = 2 * fib(n+1) - 1
For example, numCall(35) = 29,860,703
Because fib(n) gets large quickly, recursion is
not an efficient way to compute the Fibonacci
numbers.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Iterative method fibIter()
// compute Fibonacci number n iteratively
public static int fibIter(int n)
{
// integers to store previous two Fibonacci value
int oneback = 1, twoback = 0, current = 0;
int i;

// return is immediate for first two numbers
if (n <= 1)
current = n;
else
// compute successive terms beginning at 2
for (i = 2; i <= n; i++)
{
current = oneback + twoback;
// update for next calculation
twoback = oneback;
oneback = current;
}
return current;
}
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.3
import ds.time.Timing;

public class Program6_3
{
public static void main(String[] args)
{
int fib_45;
Timing timer = new Timing();
double fibTime;

// evaluate fibIter(45) using iterative method
System.out.println(Value of fibIter(45) by +
iteration is + fibIter(45));
// evaluate fib(45) using recursive method
System.out.print("Value of fib(45) by recursion is ");
// start/stop timing the recursive method
timer.start();
fib_45 = fib(45);
fibTime = timer.stop();
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.
Program 6.3 (concluded)
// output the value for fib(45) and time of computation
System.out.println(fib_45);
System.out.println(
" Time required by the recursive version is " +
fibTime + " sec");
}
< recursive method fib() defined in the
previous discussion >
< iterative method fibIter() defined in the
previous discussion >
}
Run:

Value of fibIter(45) by iteration is 1134903170
Value of fib(45) by recursion is 1134903170
Time required by the recursive version is 34.719 sec
Criteria of Using Recursion
With the overhead of method calls, a simple
recursive method could significantly
deteriorate runtime performance. In the case
of the Fibonacci numbers, use the O(n)
iterative solution in preference to the
recursive version.
Use recursion when it enhances the algorithm
design and provides a method
implementation that runs with reasonable
space and time efficiency.
2005 Pearson Education, Inc., Upper
Saddle River, NJ. All rights reserved.