A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations

http://aofa.cs.princeton.edu
Orientation
Second half of class
• Surveys fundamental combinatorial classes.
• Considers techniques from analytic combinatorics to study them .
AN INTRODUCTION

ANALYSIS
TO THE

ALGORITHMS
OF

• Includes applications to the analysis of algorithms.

S E C O N D E D I T I O N

ROBERT SEDGEWICK
PHILIPPE FLAJOLET

chapter combinatorial classes type of class type of GF

6 Trees unlabeled OGFs

7 Permutations labeled EGFs

8 Strings and Tries unlabeled OGFs

9 Words and Mappings labeled EGFs

Note: Many more examples in book than in lectures.

2
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations
• Basics
• Sets of cycles
• Left-right-minima
OF
• Other parameters
• BGFs and distributions
http://aofa.cs.princeton.edu

7a.Perms.Basics
Basics

Definition. A permutation is an ordering or the numbers 1 through N.

Ex. A group of N students who live in single rooms go to a party that leads to a state of inebriation.
When returning, they each end up in a random room.

student 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

room 9 12 11 10 5 15 1 3 7 6 13 8 2 16 4 14
4
Review: permutations

Def. A permutation is a sequence of labelled atoms.

1 2 3 4 1 2 4 3

2 1 3 4 2 1 4 3

3 1 2 4 3 1 4 2

1 2 3 4 1 2 3 4 1 3 2

2 1 3 1 3 2 4 1 3 4 2 counting sequence EGF

1 2 3 1 2 2 3 1 4 2 3 4 1
1 = !
2 1 1 3 2 3 2 1 4 3 2 4 1
P1 = 1
P2 = 2 2 3 1 4 2 1 3 4 2 3 1
!
1 4 2 3 1 4 3 2
= =
3 2 1 !
P3 = 6 2 4 1 3 2 4 3 1

3 4 1 2 3 4 2 1

4 3 1 2 4 3 2 1

P4 = 24
5
Inverse

Alternate def. A permutation is a mapping of the numbers 1 through N to itself.

student 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

room 9 12 11 10 5 15 1 3 7 6 13 8 2 16 4 14

Def. The inverse of a permutation is the inverse of that mapping.

student 7 13 8 15 5 10 9 12 1 4 3 2 11 16 6 14

room 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

room 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

student 7 13 8 15 5 10 9 12 1 4 3 2 11 16 6 14

6
Computing the inverse of a permutation

permutation
1 2 3 4 5 6 7 8 9
8 1 3 7 6 2 9 4 5

public static int[] inverse(int[] a)

inverse
{
1
int N = a.length;
int[] b = new int[N]; 2 1
for (int i = 0; i < N; i++)
2 3 1
b[a[i]-1] = i+1;
return b; 2 3 4 1
}
2 3 5 4 1

2 6 3 5 4 1
Java arrays are 0-based
2 6 3 5 4 1 7

2 6 3 8 5 4 1 7

2 6 3 8 9 5 4 1 7

7
Application: Substitution cipher

Algorithm (traditional)
• Generate random permutation of A-Z (stay tuned).
• Apply as a mapping to encrypt.
• Use inverse to decrypt.

Encryption A B C D E F G H I J K L M N O P Q R S T U V W X Y Z -
random permutation W V L Q I X J A B G - U N F K R Y C D P Z E O M H T S

plaintext A T T A C K - A T - D A W N
ciphertext W P P W L - S W P S Q W O F

Decryption A B C D E F G H I J K L M N O P Q R S T U V W X Y Z -
inverse H I R S V N J Y E G O C X M W T D P - Z L B A F Q U K

ciphertext W P P W L - S W P S Q W O F
plaintext A T T A C K - A T - D A W N

Caveat. Not useful in modern applications because of susceptibility to character frequency analysis.
8
Lattice representation of a permutation
marked
column
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1 7
2 13
3 8
4 15
5 5
6 10
7 9
8 12 inverse
9 1
10 4
11 3
12 2
13 11
14 16
15 6
16 14
marked
9 12 11 10 5 15 1 3 7 6 13 8 2 16 4 14
row
permutation

Implication. Representation of inverse is transpose of representation of permutation.

9
Review: A combinatorial bijection

Alternate def. A permutation is a set of cycles.

Standard representation

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

9 12 11 10 5 15 1 3 7 6 13 8 2 16 4 14

Set of cycles representation

12
15 2 8
7 16
6 4 5
9 1 14
10 13 3
11
10
Review: The symbolic method for labelled classes (transfer theorem)

Theorem. Let A and B be combinatorial classes of labelled objects with EGFs A(z) and B(z). Then

construction notation semantics EGF

disjoint union A +B disjoint copies of objects from A and B ( )+ ( )

ordered pairs of copies of objects,
labelled product A★B ( ) ( )
one from A and one from B

SEQk ( A ) k- sequences of objects from A ( )

sequence
SEQ ( A ) sequences of objects from A
( )
SETk ( A ) k-sets of objects from A ( ) / !
set
SET ( A ) sets of objects from A ( )

CYCk ( A ) k-cycles of objects from A ( )/

cycle
CYC ( A ) cycles of objects from A ln
( )

11
Review: symbolic method to count permutations

How many permutations of length N ?

Class P, the class of all permutations Atom type class size GF

labelled atom Z 1 z
Size |p |, the length of p

! | |
OGF ( )= =
| |! !
∈

“a permutation is empty or an
Construction = + ! atom and a permutation”

OGF equation ( )= + ( )

Solution ( )=
−

![ ] ( )= ! ✓
12
Application: Sorting algorithms
[hundreds of algorithms since 1950]
{
public static
private class Merge
int partition(Comparable[] a, int lo, int hi) input (maybe not in random order)
{{
private
public class
static
j =Quick
int partition(Comparable[] a, int lo, int hi)
int
{{
while
i = lo,
(true)
hi+1; T S R P O N M L I
{ int
private
i = lo,
static
j = int
hi+1;partition(Comparable[] a, int lo, int hi)
while
{
while (true)
(less(a[++i], a[lo])) if (i == hi) break;
{
while int i = lo, j =
(less(a[lo], hi+1; if (j == lo) break;
a[--j]))
if (iwhile
while
>= j)(less(a[++i],
(true)
break; a[lo])) if (i == hi) break; random permutation of the input
while
{ i,(less(a[lo],
exch(a, j); a[--j])) if (j == lo) break;
if (iwhile
>= j)(less(a[++i],
break; a[lo])) if (i == hi) break;
} N L T R M O I P S
exch(a,exch(a,
lo,while
j);i,(less(a[lo],
j); a[--j])) if (j == lo) break;
return} j; if (i >= j) break;
} exch(a,exch(a,
lo, j);i, j);
return
} j;
}
private exch(a,
static voidlo, j);
sort(Comparable[] a, int lo, int hi) sorted output
{ return j;
ifprivate
} <=static
(hi void sort(Comparable[] a, int lo, int hi)
lo) return;
{ j = partition(a, lo, hi);
I L M N O P R S T
int
if
sort(a,private
(hi
lo,<= static
lo) return;
j-1); void sort(Comparable[] a, int lo, int hi)
int
sort(a,{ j+1,
j = partition(a,
hi); lo, hi);
} sort(a,
if (hi
lo,<=j-1);
lo) return;
} sort(a,
int jj+1,
= partition(a,
hi); lo, hi);
} sort(a, lo, j-1);
} sort(a, j+1, hi);
}
}

Q. Model for input?

A. Random permutation.

Q. Realistic?

Q. Absolutely, if we put entries in random order before the sort! Chapter 2

13
Application: Randomly permuting an array/generate a random permutation

input (maybe not in random order)

Algorithm (Knuth)
T S R P O N M L I
• Move from left to right.
• Exch each entry with a random entry to its right. N S R P O T M L I

N L R P O T M S I
for (int i = 0; i < N; i++)
{ N L T P O R M S I
int r = i + StdRandom.uniform(N-i);
int t = a[r]; a[r] = a[i]; a[i] = t; N L T R O P M S I
}
N L T R M P O S I

N L T R M O P S I
All permutations are equally likely:
N L T R M O I S P
• 1st entry equally likely to be any of the N entries.
• 2nd equally likely to be any of the N −1 remaining entries. N L T R M O I P S
• 3rd equally likely to be any of the N −2 remaining entries.
random permutation of the input
• ...
N L T R M O I P S

use 1 2 3 4 5 6 7 8 9 as input to get a random permutation 6 8 1 3 7 5 9 4 2

14
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations
• Basics
• Sets of cycles
• Left-right-minima
OF
• Other parameters
• BGFs and distributions
http://aofa.cs.princeton.edu

7b.Perms.Cycles
Review: Permutations and derangements

How many sets of cycles of length N ?

∗ “A permutation is a set of cycles"

Construction =( ( ))
∗
! "
EGF equation ( ) = exp ln =
− −
∗ ∗
Counting sequence = ![ ] ( )= !

How many derangements of length N ?

“Derangements are permutations

Construction = ( > ( )) with no singleton cycles"

! " −
/ + / + / +...
EGF equation ( )= = exp ln − =
− −
( )
Expansion [ ] ( ) =
! !

16
Review: generalized derangements

How many permutations of length N have no cycles of length ≤ M ?

Construction = ( > ( ))

+ + ! "
+ +... = exp ln − − / − ... − /
OGF equation ( )= + +
−

− − − −...
=
−

!
Asymptotics [ ] ( )

17
Involutions
are permutations composed of cycles of length 1 or 2.

I1 = 1
I2 = 2

I3 = 4
I4 = 10

18
Review: Inverse

Alternate def. A permutation is a mapping of the numbers 1 through N to itself.

index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

permutation 9 12 11 10 5 15 1 3 7 6 13 8 2 16 4 14

Def. The inverse of a permutation is the inverse of that mapping.

7 13 8 15 5 10 9 12 1 4 3 2 11 16 6 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

inverse 7 13 8 15 5 10 9 12 1 4 3 2 11 16 6 14

Q. What is the inverse of an involution?

19
Inverse of an involution

An involution is a mapping of the numbers 1 through N to itself with all 1- or 2-cycles

index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

involution 9 12 11 4 5 15 7 13 1 10 3 2 8 16 6 14

Def. The inverse of an involution is the inverse of that mapping.

9 12 11 4 5 15 7 13 1 10 3 2 8 16 6 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

inverse 9 12 11 4 5 15 7 13 1 10 3 2 8 16 6 14

Q. What is the inverse of an involution? A. ITSELF!

20
Lattice representation of an involution
marked
column
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1 9
2 12
3 11
4 4
5 5
6 15
7 7
8 13 inverse
9 1
10 10
11 3
12 2
13 8
14 16
15 6
16 14
marked
9 12 11 4 5 15 7 13 1 10 3 2 8 16 6 14
row
involution

Representation of involution is symmetric about the main diagonal.

21
Application: Reciprocal cipher

An involution is a permutation that is its own inverse.

Implication: Can encrypt and decrypt with the same machine.

Enigma (WW II)

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z -
involution D K R A Z F U J N J B L X I O S V C P - G Q Y M W E T

plaintext A T T A C K - A T - D A W N
Encryption
ciphertext D K R A Z F U J N J B L X I

ciphertext D K R A Z F U J N J B L X I
Decryption
plaintext A T T A C K - A T - D A W N

Caveat. Still susceptible to character frequency analysis but can be useful as a component.
22
Application: How many different Enigma settings?

23
Warmup

How many perms are comprised entirely of 2-cycles?

Example: ROT-13 (world's weakest cryptosystem)

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

N O P Q R S T U V W X Y Z A B C D E F G H I J K L M

A T T A C K A T D A W N
R2 = 1 Encryption
N G G N P X N G Q N J A
R4 = 3
N G G N P X N G Q N J A
Decryption
A T T A C K A T D A W N

Construction = ( ( )
/
OGF equation ( )= Stirling's approximation
√
!∼( / )
/ ! √ ! " /
Coefficients ≡ ![ ] /
∼
( / )!
24
Involutions

How many involutions of size N ?

2 6
1 7 5
5 3

“Involutions are permutations with

Construction = ( ( )) ! ( ( )) all cycles of length 1 or 2"

OGF equation ( )= + /

+ / !
Coefficients ![ ] =
! ( )!
Laplace method

/
Asymptotics
Complex asymptotics
(stay tuned for Part 2)

25
Generalized involutions

How many permutations of length N have no cycles of length > M ?

Construction = ( ( )) ! ( ( )) ! . . . ! ( ( ))

( )= + / + ... + /
OGF equation

+ / +...+ /
Coefficient asymptotics
Complex asymptotics
(stay tuned for Part 2)

26
In-class exercise

Find [ ] + / + / + / + /

ln − / − / − / − / − / − ...
=[ ] −

− / − / − / − / − /
=[ ] ...
−

=[ ] ( − )( − )( − )( − )( − )...
−

! "
=[ ]( + + + ... + ) − − − − −

.
= =

27
100 prisoners

Problem. 100 prisoners, each uniquely identified by a numbered ID card (1 to 100), have been
sentenced to death, but are given a last chance.
• The ID cards are collected and put in the drawers of a cabinet with 100 numbered drawers
(1 to 100) in random order, one card per drawer
• One at a time, the prisoners are allowed to enter the room containing the cabinet and open,
then close again, at most half the drawers.
• If all prisoners find their own number, they will all be spared.
• If one prisoner fails, they will all be executed.

Prisoner A, a mathematician, bemoans their fate, claiming

the probability of success is on the order of 2−100 ≈ 8⋅10−31.

Prisoner B, who knows analytic combinatorics, claims to know

a strategy that gives them better than 30% chance of success.

What is Prisoner B’s strategy?

28
100 prisoners solution

Problem. 100 prisoners, each uniquely identified by a numbered ID card

(1 to 100), have been sentenced to death, but are given a last chance.
• The ID cards are collected and put in the drawers of a cabinet with 100
numbered drawers (1 to 100) in random order, one card per drawer.
• One at a time, the prisoners are allowed to enter the room containing
the cabinet and open, then close again, at most half the drawers.
• If all prisoners find their own number, they will all be spared.
• If one prisoner fails, they will all be executed.

Prisoner B’s strategy: Each prisoner "follows the cycle"

• Opens the drawer corresponding to his ID.
• Uses the number in that drawer to decide which drawer to open next.
• Continues until finding the drawer containing his ID.

Q. When does Prisoner's B strategy succeed?

A. When the random permutation has no cycles of length greater than 50.

.
Probability of success: [ ] exp + + ... + = ( )= .
29
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations
• Basics
• Sets of cycles
• Left-right-minima
OF
• Other parameters
• BGFs and distributions
http://aofa.cs.princeton.edu

7c.Perms.LRM
General approach for analyzing parameters

Review: Cumulated cost approach for parameters

• Define GF for counting sequence and CGF.
• Identify construction to give CGF equation.
• Solve to get explicit formula for CGF.
• Extract coefficients from GF to get counting seq.
• Extract coefficients from CGF to get cumulated cost.
• Divide to compute expected value

Small trick for permutations:

! | | !
• Use exponential CGF. ( )= ( ) =
| |! ≥
!
∈P
• Treat as OGF to extract expected value directly.
Why does it work? ![ ] ( )
cumulated cost
= [ ] ( )=
• N ! is the normalizing factor for ECGF. counting sequence ! !
• N ! is also the counting sequence.
31
Application: Selection sort

public static void sort(Comparable[] a)

{
S O R T I N G E X A M P L E
int N = a.length;
for (int i = 0; i < N; i++)
{ min
int min = i; min
(1st update) min
for (int j = i+1; j < N; j++) (2nd update) min min
(3rd update)
if (less(a[j], a[min])) min = j; (4th update) min
exch(a, i, min); (5th update)
}
}

Q. How many times is min updated in the first pass (assuming keys distinct)?

Section 2.1
A. The number of left-right minima in the permutation.

Q. How many left-right minima in a random permutation?

Caveat. Cost for whole sort is complicated, but not significant relative to the number of compares.

32
Left-right minima

Def. A left-right minimum (lrm) in a permutation is a smaller than any item to its left.

Q. How many lrm in a random permutation of size N ? 1 1 2 3 4 1 2 4 3 1

2 2 1 3 4 2 1 4 3 2
2 3 1 2 4 3 1 4 2 2
2 4 1 2 3 4 1 3 2 2
1 2 3 1 1 1
1 3 2 4 1 3 4 2
2 1 3 2 2 2 3 1 4 2 3 4 1 2
1 2 3 1 2 2 3 3 2 1 4 3 2 4 1 3
1
2 1 1 3 2 1 3 4 2 1 3 4 2 3 1 3
P1 = 1 2 3 1
P2 = 2 2 1 1 4 2 3 1 4 3 2 1
B1 = 1
B2 = 1 + 2 = 3 3 2 1 3 2 4 1 3 2 4 3 1
B1/P1 = 1 2 2
B2/P2 = 3/2 = 1.5 P3 = 6 2 3 4 1 2 3 4 2 1 3
B3 = 2･1 + 3･2 + 1･3 = 11 3 4 3 1 2 4 3 2 1 4
B3/P3 = 11/6 ≐ 1.833 P4 = 24
B4 = 6･1 + 11･2 + 6･3 + 1･4 = 50
B4/P4 = 50/24 ≐ 2.083 33
Construction for left-right minima

Create |p|+1 perms from a perm p by star product construction.

4 7 3 6 2 5 1

4 7 3 6 1 5 2

4 7 2 6 1 5 3

3 6 2 5 1 4 ★ 1 = 3 7 2 6 1 5 4

3 7 2 6 1 4 5

3 7 2 5 1 4 6

3 6 2 5 1 4 7

Original perm has lrm(p) left-right minima.

Q. How many left-right minima in the set of constructed perms?
A. (|p| + 1) lrm(p) +1

|p| + 1 copies of the only the one ending

original perm in 1 adds a lrm
34
Average number of left-right minima in a random permutation

! | | !
CGF. ( )= ( ) =
| |! ≥
!
∈P
| |+
Apply construction. = (| | + ) ( )+
(| | + )!
P

| |+ | |+
Simplify. = ( ) +
(| |)! (| | + )!
P P

! +
Substitute. = ( )+ = ( ) + ln
( + ) −
≥

Solve. ( )= ln OGF for the Harmonic numbers

cumulated cost
=

= + = .
Expand. [ ] ( )= =
! average # lrm in a = + +
.
= . ✓
random permutation .
= + + + = .
35
Cycles

Q. How many cycles in a random permutation of size N ?

all 2

4 1
3

2 3
1

2 3
2 1
1 2 3
1
B1 = 1 1
B1/P1 = 1 3
1
B2 = 1 + 2 = 3 1
B2/P2 = 3/2 = 1.5 3
1

3 1

B3 = 2･1 + 3･2 + 1･3 = 11

B3/P3 = 11/6 ≐ 1.833 B4 = 6･1 + 11･2 + 6･3 + 1･4 = 50
B4/P4 = 50/24 ≐ 2.083 36
Construction for cycles

Create |p|+1 perms from a perm p by inserting |p|+1 into 4 3

every position in every cycle (including the null cycle) 5 1
2
6 7

4 3
4 3 2
7 2 5 1 7 6
4 3 5 1 6
2
5 1 6

4 4
2 3 3
5 7 2
5 1 7 6 6
1

4
Original perm has cycles(p) cycles. 5 1 2
3

6
Q. How many cycles in the set of constructed perms? 7

A. (|p| + 1) cycles(p) +1 same as for lrm (!) 4

3
7 5 2
6
|p| + 1 copies of the from the 1
original perm null cycle
37
Average number of cycles in a random permutation (same derivation as for lrm)

! | | !
CGF. ( )= ( ) =
| |! !
∈P ≥

| |+
Decompose. = (| | + ) ( )+
(| | + )!
P

| |+ | |+
Simplify. = ( ) +
(| |)! (| | + )!
P P

! +
Substitute. = ( )+ = ( ) + ln
( + ) −
≥

Solve. ( )= ln OGF for the Harmonic numbers

cumulated cost
=

= + = .
Expand. [ ] ( )= =
! average # cycles in a = + +
.
= . ✓
random permutation .
= + + + = .
38
Left-right minima and cycles

Q. Is there a 1:1 correspondence ?

A. Yes!
12
15 2
7 16 8
To build a permutation from a set of cycles: 6 4 5
9 1 14
• Identify smallest as the leader in each cycle. 10 13 3
11
• Write cycles in decreasing order of leader.
14 16 5 4 10 6 15 2 12 8 3 11 13 1 7 9

14 16 5 4 10 6 15 2 12 8 3 11 13 1 7 9
To build a set of cycles from a permutation:
• Identify left-right minima. 4
2

14 13 12 1
• Build cycles with entries delimited by lrms 5 15 10
(start a new cycle with each lrm). 16 7 9
6 11 8
3
39
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations
• Basics
• Sets of cycles
• Left-right-minima
OF
• Other parameters
• BGFs and distributions
http://aofa.cs.princeton.edu

7d.Perms.Others
1-Cycles

Q. How many 1-cycles in a random permutation of size N ?

4 0 0
3

1 2
0 0 0

1 2
2 0
1 1 2
0
B1 = 1 0
B1/P1 = 1 2
0
B2 = 0 + 2 = 2 0
B2/P2 = 2/2 = 1 2
0

2 0

B3 = 2･0 + 3･1 + 1･3 = 6

B3/P3 = 1 all 1
B4 = 9･0 + 8･1 + 6･2 + 1･4 = 24
B4/P4 = 24/24 = 1
41
Construction for 1-cycles

Create |p|+1 perms from a perm p by inserting |p|+1 into 4 3

every position in every cycle (including the null cycle) 5 1
2
6 7

4 3
4 3 2
7 2 5 1 7 6
4 3 5 1 6
2
5 1 6

4 4
2 3 3
5 7 2
5 1 7 6 6
1

4
Original perm has cyc1(p) 1-cycles. 5 1 2
3

6
Q. How many 1-cycles in the set of constructed perms? 7

A. (|p| + 1) cyc1(p) + 1 − cyc1(p) = |p| cyc1(p) + 1 4

3
7 5 2
6
|p| + 1 copies of the from the 1-cycles changed 1
original perm null cycle to 2-cycles
42
Average number of 1-cycles in a random permutation

! | | !
CGF. ( )= ( ) =
| |! ≥
!
∈P

| |+
Apply construction. = (| | ( )+ )
(| | + )!
P

| | | |
Differentiate. ( )= | | ( ) + = ( )+
| |! | |!
P P

Solve. ( )=
( )

Integrate. ( )=

cumulated cost

Expand. [ ] ( )= =
! average # 1-cycles in a
random permutation
43
Application: Students and rooms revisited

A group of N students who live in single rooms go to a party that leads to a state of inebriation.
When returning, they each end up in a random room.
Q. What is the average number of students who wind up in their own room?

A. One (!)

44
In-class exercises

Q. How many 2-cycles in a random permutation of size N ?

A. 1/2

Q. How many r-cycles in a random permutation of size N ?

A. 1/r

45
Inversions

Def. An inversion in a permutation is the number of pairs ii jj with i > j.

Equivalent: Sum number of entries larger and to the left of each entry.
0 1 2 3 4 1 2 4 3 1
Q. How many inversions in a random permutation of size N ? 1 2 1 3 4 2 1 4 3 2
2 3 1 2 4 3 1 4 2 3
2 4 1 2 3 4 1 3 2 4
1 2 3 0 1 2
1 3 2 4 1 3 4 2
2 1 3 1 2 2 3 1 4 2 3 4 1 4
1 2 0 3 1 2 2 3 3 2 1 4 3 2 4 1 4
1 0
2 1 1 1 3 2 1 3 4 2 1 3 4 2 3 1 4
P1 = 1 2 3 1
P2 = 2 2 2 1 4 2 3 1 4 3 2 3
B1 = 0
B2 = 0 + 1 = 1 3 2 1 3 2 4 1 3 2 4 3 1
B1/P1 = 0 3 4
B2/P2 = 1/2 = 0.5 P3 = 6 4 3 4 1 2 3 4 2 1 5
B3 = 2･1 + 2･2 + 1･3 = 9 5 4 3 1 2 4 3 2 1 6
B3/P3 = 9/6 = 1.5 P4 = 24
B4 = 3･1 + 7･2 + 5･3 + 6･4 + 2･5 + 1･6 = 72
B4/P4 = 72/24 = 3 46
Application: Insertion sort
i = 10
sorted before i untouched after i

public static void sort(Comparable[] a) A E G I N O R S T X M P L E D Q Z

{
int N = a.length; A E G I Nexchange
O R M S with
T entry
M X toPits LleftEif larger
D Q Z
for (int i = 1; i < N; i++)
{ A E G I N O R S M T X P L E D Q Z
for (int j = i; j > 0; j--)
if (less(a[j], a[j-1])) A E G I N O R M S T X P L E D Q Z
exch(a, j, j-1);
else break; A E G I N O M R S T X P L E D Q Z
}
} A E G I N M O R S T X P L E D Q Z

A E G I M N O R S T X P L E D Q Z

Q. How many exchanges during the sort? exchanges put M in place among elements to its left

A. The number of inversions in the permutation.

Q. How many inversions in a random permutation?

Section 2.1
47
Construction for inversions

Create |p|+1 perms from a perm p by "largest" construction.

3 6 2 5 1 4 7

3 6 2 5 1 7 4 adds 1 inversion

3 6 2 5 7 1 4 adds 2 inversions
3 6 2 5 1 4 ✶ 1 = 3 6 2 7 5 1 4 adds 3 inversions

3 6 7 2 5 1 4 adds 4 inversions

3 7 6 2 5 1 4 adds 5 inversions

7 3 6 2 5 1 4 adds 6 inversions

Original perm has inv(p) inversions.

Q. How many inversions in the set of constructed perms?
A. (|p| + 1) inv(p) + (|p| + 1) |p| / 2

|p| + 1 copies of the all the inversions

original perm caused by |p| + 1
48
Average number of inversions in a random permutation

! | | !
CGF. ( )= ( ) =
| |! ≥
!
∈P

| |+
Apply construction. = (| | + ) ( ) + (| | + )| |/
(| | + )!
P

| |+ | |+ !
Simplify. = ( ) + | | = ( )+
(| |)! (| |)!
P P ≥

Substitute. = ( )+
( − )

Solve. ( )=
( − ) ·
/ != =
cumulated cost
·
/ != = .
( − ) ·
✓
Expand. [ ] ( )= = / != = .
!
average # inversions in a
random permutation ·
/ != =
49
Parameters of permutations
all can be handled in a similar manner

50
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations
• Basics
• Sets of cycles
• Left-right-minima
OF
• Other parameters
• BGFs and distributions
http://aofa.cs.princeton.edu

7e.Perms.BGFs
Bivariate generating functions
are the method of choice in analyzing combinatorial parameters.

Definition. A combinatorial class is a set of combinatorial objects and an

associated size function that may have an associated parameter.

Definition. The bivariate generating function (BGF ) associated with a class

is the formal power series
| |
| | (( ))
( , )= (unlabelled)
(labelled)
| |!

where |a| is the size and cost(a) is the value of the parameter.

Advantages of BGFs:
• Carry full information.
• Easy to compute counting sequence and CGF (see next slide).
• Full distribution often available via analytic combinatorics.
52
Basic BGF calculations

Definition. The bivariate generating function (BGF ) associated with a labelled class
| | z marks size.
is the formal power series ( , )= ( )
u marks the parameter.
| |!

Define ANk to be the number of elements of size N with parameter value k.

| |
( )
Fundamental (elementary) identity ( , )= =
| |! !

Q. How many objects of size N with value k ?

A. ![ ][ ] ( , ) =
( , )=
!

Q. Average value of a parameter of a permutation ?

( , ) ( , ) =
=
!
A. [ ] ( , ) ( , ) =
[ ] ( , )= ( , ) =
=
!
53
Review: Average number of cycles in a random permutation with CGFs

! | | !
CGF. ( )= ( ) =
| |! !
∈P ≥

| |+
Decompose. = (| | + ) ( )+
(| | + )!
P

| |+ | |+
Simplify. = ( ) +
(| |)! (| | + )!
P P

! +
Substitute. = ( )+ = ( ) + ln
( + ) −
≥

Solve. ( )= ln OGF for the Harmonic numbers

cumulated cost

Expand. [ ] ( )= =
! average # cycles in a
random permutation
54
Average number of cycles in a random permutation with BGFs

! | |
( )
BGF. ( , )=
| |!
∈P

| |+
( )+ ( )
Apply construction . = +| |
(| | + )!
P

| | | |
( )+ ( )
Differentiate wrt z. ( , )= + | |
(| |)! (| |)!
P P

Substitute. = ( , )+ ( , )

Solve for Bz(z, u). ( , )= ( , )

−

Solve ODE. ( , )=
( − )

Average number of cycles. ( , )= ln

− −
[ ] ( , )= ✓
55
Average number of cycles in a random permutation with BGFs and the symbolic method

Combinatorial class. P, the class of all permutations

Construction. = ( ( ))

! " immediate from

BGF equation ( , ) = exp ln =
− ( − ) transfer theorem.

Average number of cycles. ( , )= ln

− −

[ ] ( , )= ✓

Bottom line: BGFs are the method of choice in analyzing parameters

56
Average number of cycles of a given size in a random permutation

Combinatorial class. P, the class of all permutations

Construction. = ( != + ( ))

ln +
BGF equation ( , )= immediate from
transfer theorem.

Average number of cycles. ( , )=

[ ] ( , )= ≥ ✓ Many, many
examples to follow.
Stay tuned for Part 2
BGFs are the method of choice in analyzing parameters.
57
Number of permutations of size N with k cycles
are known as Stirling numbers of the first kind.
all 2

Notation: 4 1
3

2 3
1

2 3
1 1
1
2 3
1 1
=
3
1
= =
1
3
1

3 1
= = =

= = = =

58
Stirling numbers of the first kind (cycle numbers)

| |
( )
Fundamental identity ( , )= = =
| |! ! ( − )

Distribution ( , )= ( + )...( + ) (Taylor's theorem)

!

k
N 1 2 3 4 5 6 7

1 1

2 1 1

3 2 3 1

[ ] ( + )( + )( + ) 4 6 11 6 1

5 24 50 35 10 1

6 120 274 225 85 15 1

59
Stirling numbers of the first kind (cycle numbers) distribution

.5

.333

.167

0
0 60
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations
• Basics
• Sets of cycles
• Left-right-minima
OF
• Other parameters
• BGFs and distributions
http://aofa.cs.princeton.edu • Exercises
7e.Perms.Exs
Exercise 7.29

Arrangements.

.
AN INTRODUCTION

ANALYSIS
TO THE

ALGORITHMS
OF

S E C O N D E D I T I O N

ROBERT SEDGEWICK
PHILIPPE FLAJOLET

62
Exercise 7.45

Inversions in involutions.

.
AN INTRODUCTION

ANALYSIS
TO THE

ALGORITHMS
OF

S E C O N D E D I T I O N

ROBERT SEDGEWICK
PHILIPPE FLAJOLET

63
Exercise 7.61

Cycle length distribution.

.
AN INTRODUCTION

ANALYSIS
TO THE

ALGORITHMS
OF

S E C O N D E D I T I O N

ROBERT SEDGEWICK
PHILIPPE FLAJOLET

64
Assignments for next lecture

1. Read pages 345-413 in text.

AN INTRODUCTION

ANALYSIS
TO THE

ALGORITHMS
OF

S E C O N D E D I T I O N

2. Run experiments to validate mathematical results.

ROBERT SEDGEWICK
PHILIPPE FLAJOLET

Experiment 1. Generate 1000 random permutations for

N = 100, 1000, and 10,000 and compare the average
number of cycles and 1-cycles with the values predicted
by analysis.

Experiment 2. Extra credit. Validate the results of Exercise 7.61

for N = 1000 and k = 10 by generating 10,000 random
permutations and plotting the histogram of occurences of cycles
of length 10.

3. Write up solutions to Exercises 7.29, 7.45, and 7.61.

65
 A N A LY T I C C O M B I N AT O R I C S

PART ONE

7. Permutations

http://aofa.cs.princeton.edu