Recursion
Alexander S. Kulikov
Steklov Mathematical Institute at St. Petersburg, Russian Academy of Sciences
and
University of California, San Diego
Outline
Recursion
Coin Problem
Hanoi Towers
Line to Louvre
By Edal Anton Lefterov, [Link]
Computing Queue Length
• Alice gets in line
Computing Queue Length
• Alice gets in line
• She wants to know the number of people
before her in the line
Computing Queue Length
• Alice gets in line
• She wants to know the number of people
before her in the line
• She asks a person (call him Bob) standing
before her in the line: Could you please
tell me the number of people before you?
Computing Queue Length
• Alice gets in line
• She wants to know the number of people
before her in the line
• She asks a person (call him Bob) standing
before her in the line: Could you please
tell me the number of people before you?
• Now, Bob faces exactly the same problem
Computing Queue Length
• Alice gets in line
• She wants to know the number of people
before her in the line
• She asks a person (call him Bob) standing
before her in the line: Could you please
tell me the number of people before you?
• Now, Bob faces exactly the same problem
• Bob somehow computes that the number
of people before him is 239
Computing Queue Length
• Alice gets in line
• She wants to know the number of people
before her in the line
• She asks a person (call him Bob) standing
before her in the line: Could you please
tell me the number of people before you?
• Now, Bob faces exactly the same problem
• Bob somehow computes that the number
of people before him is 239
• Now, Alice knows that the number of
people before her is 240
Recursive Program
numberOfPeopleBefore(A):
if there is nobody before A:
return 0
B ← person right before A
return numberOfPeopleBefore(B) + 1
Factorial Function
Definition
For a positive integer n, the factorial of n is
the product of integers from 1 to n.
Factorial Function
Definition
For a positive integer n, the factorial of n is
the product of integers from 1 to n.
5! = 1 × 2 × 3 × 4 × 5 = 120 = 4! × 5
Factorial Function
Definition
For a positive integer n, the factorial of n is
the product of integers from 1 to n.
5! = 1 × 2 × 3 × 4 × 5 = 120 = 4! × 5
Recursive definition
{
1 if n = 1
n! =
n × (n − 1)! if n > 1
Iterative Program
def f a c t o r i a l ( n ) :
assert ( n > 0)
result = 1
for i i n range ( 1 , n + 1 ) :
r e s u l t *= i
return r e s u l t
Recursive Program
def f a c t o r i a l ( n ) :
assert ( n > 0)
i f n == 1 :
return 1
else :
return n * f a c t o r i a l ( n − 1 )
Termination
• One must ensure that a recursive program
(or definition) terminates after a finite
number of steps
Termination
• One must ensure that a recursive program
(or definition) terminates after a finite
number of steps
• This is usually achieved by decreasing
some parameter until it reaches a base
case
Termination
• One must ensure that a recursive program
(or definition) terminates after a finite
number of steps
• This is usually achieved by decreasing
some parameter until it reaches a base
case
• Line length: the length of the line decreases
by 1 with each recursive call, until it
becomes 1
Termination
• One must ensure that a recursive program
(or definition) terminates after a finite
number of steps
• This is usually achieved by decreasing
some parameter until it reaches a base
case
• Line length: the length of the line decreases
by 1 with each recursive call, until it
becomes 1
• Factorial: n decreases to 1
Example of Infinite Recursion
def i n f i n i t e ( n ) :
i f n == 1 :
return 0
return n * i n f i n i t e ( n + 1 )
Example of Infinite Recursion
def i n f i n i t e ( n ) :
i f n == 1 :
return 0
return n * i n f i n i t e ( n + 1 )
• In theory, will never stop: the parameter
increases to infinity
Example of Infinite Recursion
def i n f i n i t e ( n ) :
i f n == 1 :
return 0
return n * i n f i n i t e ( n + 1 )
• In theory, will never stop: the parameter
increases to infinity
• In practice, will cause an error message
like stack overflow or maximum
recursion depth exceeded
More Examples of Infinite Recursion
• No base case:
def f a c t o r i a l ( n ) :
return n * f a c t o r i a l ( n − 1 )
More Examples of Infinite Recursion
• No base case:
def f a c t o r i a l ( n ) :
return n * f a c t o r i a l ( n − 1 )
• Parameter does not change:
def i n f i n i t e ( n ) :
i f n == 1 :
return 0
return 1 + i n f i n i t e ( n )
More (Jokey) Examples
• To understand recursion, you must first
understand recursion
More (Jokey) Examples
• To understand recursion, you must first
understand recursion
• By Google:
Outline
Recursion
Coin Problem
Hanoi Towers
Coin Problem
Problem
Prove that any monetary amount starting
from 8 can be paid using coins of
denominations 3 and 5.
Coin Problem
Problem
Prove that any monetary amount starting
from 8 can be paid using coins of
denominations 3 and 5.
• Looks plausibly: 8 = 3 + 5, 9 = 3 + 3 + 3,
10 = 5 + 5, 11 = 3 + 3 + 5
Coin Problem
Problem
Prove that any monetary amount starting
from 8 can be paid using coins of
denominations 3 and 5.
• Looks plausibly: 8 = 3 + 5, 9 = 3 + 3 + 3,
10 = 5 + 5, 11 = 3 + 3 + 5
• But how can we be sure that it is always
possible?
Speculating
• OK, we can definitely pay 8
Speculating
• OK, we can definitely pay 8
• But then we can also pay 11 = 8 + 3
(adding one 3-coin)
Speculating
• OK, we can definitely pay 8
• But then we can also pay 11 = 8 + 3
(adding one 3-coin)
• Then, 14, 17, 20, etc
Speculating
• OK, we can definitely pay 8
• But then we can also pay 11 = 8 + 3
(adding one 3-coin)
• Then, 14, 17, 20, etc
• Similarly, 9 gives 12, 15, 18, 21, etc
Speculating
• OK, we can definitely pay 8
• But then we can also pay 11 = 8 + 3
(adding one 3-coin)
• Then, 14, 17, 20, etc
• Similarly, 9 gives 12, 15, 18, 21, etc
• While 10 gives 13, 16, 19, 22, etc
Recursive Program
def change ( amount ) :
a s s e r t ( amount >= 8 )
i f amount == 8 :
return [ 3 , 5 ]
i f amount == 9 :
return [ 3 , 3 , 3 ]
i f amount == 1 0 :
return [ 5 , 5 ]
c o i n s = change ( amount − 3 )
c o i n s . append ( 3 )
return c o i n s
Outline
Recursion
Coin Problem
Hanoi Towers
Hanoi Towers
Puzzle
There are three sticks with n discs sorted by size on
one of the sticks. The goal is to move all n discs to
another stick subject to two constraints: move one disc
at a time and don’t place a larger disc on a smaller one.
[Link]
Is It Even Possible?
• For what n this is possible?
Is It Even Possible?
• For what n this is possible?
• For all!
Is It Even Possible?
• For what n this is possible?
• For all!
• But how can we be sure?
Is It Even Possible?
• For what n this is possible?
• For all!
• But how can we be sure?
• We will design a recursive program that
will solve the puzzle for every n
Simplest Possible Scenario
Simplest Possible Scenario
Simplest Possible Scenario
Two Discs
Two Discs
Two Discs
Two Discs
Two Discs
Two Discs
Two Discs
Speculating
Speculating
consider the largest disc
Speculating
when can we move it?
Speculating
when there are no other discs on this stick!
Speculating
where can we move it?
Speculating
only to an empty stick!
Speculating
all other discs should be at another stick
Speculating
all other discs should be at another stick
Speculating
need to move n − 1 discs
Speculating
this is the same problem!
Recursive Solution
Recursive Solution
move n − 1 discs recursively
Recursive Solution
move n − 1 discs recursively
Recursive Solution
move the largest disk
Recursive Solution
move the largest disk
Recursive Solution
move n − 1 discs recursively
Recursive Solution
move n − 1 discs recursively
Summary
• We’ve constructed a solution for all n:
• Base case: it is possible for n = 1
• Hence, it is possible for n = 2
• Hence, it is possible for n = 3
• ...
• We’ll later learn a related notion of
induction
