Recursion

22c:19 Chapter 8 Hantao Zhang

Recursion
Recursion means defining something, such as a function, in terms of itself
For example, let f(x) = x! We can define f(x) as f(x) = x * f(x-1)

Sequences are functions from natural numbers to reals.

Recursion example

a)

Find f(1), f(2), f(3), and f(4), where f(0) = 1 Let f(n+1) = f(n) + 2
f(1) = f(0) + 2 = 1 + 2 = 3 f(2) = f(1) + 2 = 3 + 2 = 5 f(3) = f(2) + 2 = 5 + 2 = 7 f(4) = f(3) + 2 = 7 + 2 = 9

b)

Let f(n+1) = 3f(n)

f(1) = 3 * f(0) = 3*1 = 3 f(2) = 3 * f(1) = 3*3 = 9 f(3) = 3 * f(2) = 3*9 = 27 f(4) = 3 * f(3) = 3*27 = 81

Recursion example

c)

Find f(1), f(2), f(3), and f(4), where f(0) = 1 Let f(n+1) = 2f(n)
f(1) = 2f(0) = 21 = 2 f(2) = 2f(1) = 22 = 4 f(3) = 2f(2) = 24 = 16 f(4) = 2f(3) = 216 = 65536 f(n) = 2n + 1

d)

Let f(n+1) = f(n)2 + f(n) + 1

f(1) = f(0)2 + f(0) + 1 = 12 + 1 + 1 = 3 f(2) = f(1)2 + f(0) + 1 = 32 + 3 + 1 = 13 f(3) = f(2)2 + f(0) + 1 = 132 + 13 + 1 = 183 f(4) = f(3)2 + f(0) + 1 = 1832 + 183 + 1 = 33673 f(n) = 3n
4

Start here - Instructions

1. Transfer all the disks from pole A to pole B. 2. You may move only ONE disk at a time. 3. A large disk may not rest on top of a smaller one at anytime.

1 2 3 A B C

Try this one!

1 2 3 4 A B C

Shortest number of moves??

And this one

1 2 3 4 5 A B C

Shortest number of moves??

Now try this one!

1 2 3 4 5 6

B Shortest number of moves??

How to solve Tower of Hanoi of n disks?

If n = 1, move disk 1 from A to B, done. If n > 1, 1. Solve the Tower of Hanoi of n-1 disks, from A to C; 2. move disk n from A to B , 3. Solve the Tower of Hanoi of n-1 disks, from C to B.
Hanoi ( int n, char A, char B, char C ) { if (n==1) cout << move disk from << A << to << B; else { Hanoi(n-1, A, C, B); cout << move disk << n << A << to << B; Hanoi(n-1, C, B, A); } }

Counting the moves: Let f(n) be the number of moves for n disks. f(1) = 1; f(n) = 2f(n-1) + 1.
9

Let f(n) be the number of moves for n disks. f(1) = 1; f(n) = 2f(n-1) + 1.
Number of Disks 1 2 3 4 5 6 Number of Moves f(1) = 1 f(2) = 2*1 + 1 = 3 f(3) = 2*3 + 1 = 7 f(4) = 2*7 + 1 = 15 f(5) = 2*15 + 1 = 31 f(6) = 2*31 + 1 = 63

Let f(n) be the number of moves for n disks. f(1) = 1; f(n) = 2f(n-1) + 1. Prove: f(n) = 2n - 1. By induction.
Base step: n = 1 1. Left = f(1) = 1; Right = 21 - 1 = 1 Induction hypothesis: f(n-1) = 2n-1 - 1. Inductive step: Left = f(n) = 2f(n-1) + 1 = 2(2n-1 - 1) + 1 = 2n - 1. Right = 2n - 1.
11

Fascinating fact
So the formula for finding the number of steps it takes to transfer n disks from post A to post C is:

2 n- 1
If n = 64, the number of moves of single disks is 2 to the 64th minus 1, or 18,446,744,073,709,551,615 moves! If one worked day and night, making one move every second it would take slightly more than 580 billion years to accomplish the job! - far, far longer than some scientists estimate the solar system will last.

Fibonacci sequence
Definition of the Fibonacci sequence
Non-recursive:

F ( n) =

(1 + 5 ) (1 5 )
n

5 2

Recursive: or:

F(n) = F(n-1) + F(n-2) F(n+1) = F(n) + F(n-1)

Must always specify base case(s)!

F(1) = 1, F(2) = 1 Note that some will use F(0) = 1, F(1) = 1
13

Fibonacci sequence in Java

long Fibonacci if ( (n == return else return } (int n) { 1) || (n == 2) ) 1; Fibonacci (n-1) + Fibonacci (n-2);

long Fibonacci2 (int n) { return (long) ((Math.pow((1.0+Math.sqrt(5.0)),n)Math.pow((1.0-Math.sqrt(5.0)),n)) / (Math.sqrt(5) * Math.pow(2,n))); }

14

Recursion pros
Easy to program Easy to understand

15

Recursion cons
Consider the recursive Fibonacci generator How many recursive calls does it make?
F(1): 1 F(2): 1 F(3): 3 F(4): 5 F(5): 9 F(10): 109 F(20): 13,529 F(30): 1,664,079 F(40): 204,668,309 F(50): 25,172,538,049 F(100): 708,449,696,358,523,830,149 7 * 1020
At 1 billion recursive calls per second (generous), this would take over 22,000 years But that would also take well over 1012 Gb of memory!
16

Bad recursive definitions

Consider:
f(0) = 1 f(n) = 1 + f(n-2) What is f(1)?

Consider:
f(0) = 1 f(n) = 1+f(-n) What is f(1)?
17

Defining sets via recursion

Same as mathematical induction:
Base case (or basis step) Recursive step

Example: the set of positive integers

Basis step: 1 S Recursive step: if x S, then x+1 S

18

Defining sets via recursion

Give recursive definitions for:
a)

The set of odd positive integers

1S If x S, then x+2 S

b)

The t f Th set of positive i t iti integer powers of 3 f

3S If x S, then 3*x S

c)

The set of polynomials with integer coefficients

0S If p(x) S, then p(x) + cxn S

c Z, n Z and n 0
19

Defining strings via recursion

Terminology
is the empty string: is the set of all letters: { a, b, c, , z }
The set of letters can change depending on the problem

We can define a set of strings * as follows

Base step: * If w * and x , then wx * Thus, * s the set of all the possible strings that can be generated with the alphabet

20

Defining strings via recursion

Let = { 0, 1 } Thus, * is the set of all binary all binary strings
Or all possible computer files

21

String length via recursion

How to define string length recursively?
Basis step: len() = 0 Recursive step: len(wx) = len(w) + 1 if w * and x

Example: len(aaa) len( aaa )

len(aaa) = len(aa) + 1 len(aa) = len(a) + 1 len(a) = len() + 1 len() = 0 Result: 3
22

Strings via recursion example

Given a string x = a1a2an xR stands for its reversal: xR = anan-1a1. Eg. x = abc, xR = cba. A string x is a palindrome if x = xR. Eg. x = aba. Give a recursive definition for the set of string that are palindromes
We will define set P, which is the set of all palindromes

Basis step: P
Second basis step: x P when x

Recursive step: xpx P if x and p P

23

How many binary strings of length n that do not contain the pattern 11?
n = 0: 1 string (, the empty string) n = 1: 2 strings (0 and 1) n = 2: 3 strings (00, 01, 10) n = 3: 5 strings (000 001 010 100 101) (000, 001, 010, 100, n = 4: 8 strings (0000, 0001, 0010, 0100, 1000, 0101, 1001, 1010)

Any pattern? A Fibonacci sequence!

24

How many binary strings of length n that do not contain the pattern 11?
n = 2: 3 strings (00, 01, 10) n = 3: 5 strings (000, 001, 010, 100, 101) n = 4: 8 strings (0000, 0001, 0010, 0100, 1000, 0101, 1001, 1010) The strings of n=4 can be divided into two classes: X = { 0000, 0001, 0010, 0100, 0101 } and Y = { 1000, 1001, 1010 } X can be obtained from n = 3: adding a leading 0 Y can be obtained from n = 2: adding leading 10.
25

How many binary strings of length n that do not contain the pattern 11?
Let Sn be the set of binary strings of length n that do not contain the pattern 11. For any string x in Sn-1, y = 0x is a string of Sn. For any string x in Sn-2, z = 10x is a string of Sn. A string of Sn i either y or z. Any t i f is ith Hence |Sn| = |Sn-1| + |Sn-2| From |S0| = 1, |S2| = 2, we can compute any |Sn|.

26

Stirling Numbers of the Second Kind

You have 3 students in a class, and you want to divide them up into 2 groups (perhaps groups of 1) to work on a project. In how many ways can you do this? Answer: 3 ({a, b}{c}, {a, c}{b}, {a}{b, c}). You have 4 students and want to divide them up into 2 groups? Answer: 7 ({a b c}{d} {a b d}{c} {a c d}{b} {a}{b c d} {a b}{c d} ({a,b,c}{d}, {a,b,d}{c}, {a,c,d}{b}, {a}{b,c,d}, {a,b}{c,d}, {a,c}{b,d}, {a,d}{b,c}). You have 5 students and want to divide them up into 2 groups? Answer: S2(5,2)=15 You have 5 students and want to divide them up into 3 groups? Answer: S2(5,3)=25 You have 8 students and want to divide them up into 5 groups? Answer: S2(8,5)=1050.
27

Stirling Numbers of the Second Kind

The numbers S2(n, k) are called the Stirling Numbers of the Second Kind. S2(n, k) = the number of ways of dividing n objects into k non-empty subsets. You can find a nice page about them at http://mathworld.wolfram.com/StirlingNumberoftheSecondKi nd.html.

28

Stirling Numbers of the Second Kind

Now evaluate S2(4,3). Let X = {x, y, z, w} and Y = {x, y, z}. How many ways to divide Y into two non-empty subsets?
Answer: S2(3,2) = 3 ({x,y}{z}, {x,z}{y}, {x}{y,z}) Adding {w} into either solution of S2(3 2) will produce a solution of (3,2) S2(4,3).

How many ways to divide Y into three non-empty subsets?

Answer: S2(3,3) = 1 ({x}{y}{z}) Adding w into either set of ({x}{y}{z}) will produce a solution of S2(4,3).

Hence, S2(4,3) = S2(3,2) + 3S2(3,3) = 3 + 3*1 = 6.

29

Stirling Numbers of the Second Kind

In general, let X1 = {x1} and Xn = Xn-1 U {xn}. (A) For any solution of dividing Xn-1 into k-1 non-empty sets, adding {xn} will produce a solution of dividing Xn into k nonempty sets . (B) For any solution of dividing Xn-1 into k non-empty sets sets, 1 adding xn into one of k sets will produce a solution of dividing Xn into k non-empty sets. For any solution of dividing Xn into k non-empty sets, if xn is in a singleton set, it is a solution from (A); otherwise, its solution from (B).

Hence, S2(n,k) = S2(n-1,k-1) + kS2(n-1,k).

30

10

Stirling Numbers of the Second Kind

Theorem: For n>k>1, S2(n,k)=S2(n1,k1)+kS2(n 1,k). Proof: Count the partitions of Xn into k non-empty subsets according to whether xn is in a subset by itself. If so, then there are S2(n1,k1) ways to partition the remaining elements of Xn.. If not, then there are S2(n1,k) ways to partition the remaining elements of Xn and then k ways to choose the subset to place xn in.
31

Stirling Numbers of the Second Kind

S2(n,k) = S2(n-1,k-1) + kS2(n-1,k). Intuitively S2(0,0)=1. If n>0, then S2(n,0)=0 ( ) S2(n,1)=1 S2(n,n)=1 If n<k, then S2(n,k)=0. S2(n,2)= 2n-1 1
Note: There are 2n subsets of Xn. A subset and its complement consist of a solution of S2(n,2), except the 32 empty set.

Stirling Numbers of the Second Kind

There is also a formula for S2(n,k), namely

S2 (n, k ) =

k 1 k (1)k j j j n k ! j =0

but for many purposes the recurrence relation is easier to use.

33

11

Stirling Numbers of the Second Kind

k S2(n,k) 0 1 2 3 4 5 6 7 8 9 10 0 1 1 1 2 1 1 3 1 3 1 4 1 7 6 1 n 5 1 15 25 10 1 6 1 31 90 65 15 1 7 1 63 301 350 140 21 1 8 1 127 966 1701 1050 266 28 1 9 1 255 3025 7770 6951 2646 462 36 1 10 1 511 9330 34105 42525 22827 5880 750 45 1

34

Stirling Numbers of the Second Kind

How many onto functions are there from a 5-set to a 4-set? Answer: There are S2(5,4) ways to divide 5 elements into four non-empty subsets. For each division, there are 4! ways to map the four subsets to the codomain codomain. The total number of onto functions is 4!S2(5,4)=2410 = 240. In general, the number of onto functions from a m-set to a n-set is n!S2(m, n).

35

12