Université Paris Cité – LIPADE

Algorithmic Complexity
Introduction

Jean-Guy Mailly ([email protected])

2023
 Infos sur l’UE
1

I Jean-Guy Mailly : [email protected]

Bureau 814 I
I 15h de cours : lundi, 11h00–12h30, Curie A
I 15h de TD : lundi, 15h45–17h15, Vieussens B (IAD/VMI)
jeudi, 13h30–15h00, Cordier D (DCI/RSA)
I Modalités de contrôle de connaissances :
I Un contrôle continu vers la mi-semestre
I Un examen en fin de semestre
I Note finale : max(EX , EX +CC )
2
I Moodle: Cours IFFAX020 Complexité Algorithmique
https://moodle.u-paris.fr/course/view.php?id=5486

J.-G. Mailly | Intro to Complexity

 Aims
2

Algorithmic Complexity
Measure the hardness of a problem w.r.t. the efficiency of algorithms
to solve it
I Time
I Space

Goal of the Course

I Bases of complexity theory
I Main complexity classes (time and space)
I Complexity of usual problems
I Being able to determine the complexity of a problem

J.-G. Mailly | Intro to Complexity

 Resources
3

When solving a problem, two kinds of resources are used

I time
I number of seconds
I number of steps for the computation
I space
I number of bytes used to execute the program
I number of variables used to represent and solve the problem

Complexity Theory
I Classification of problems w.r.t. the resources required to solve
them
I The more we need time and/or space, the harder it is
I Comparison of problems (depending on the class they belong)
I Solving problems by translating them into other problems (with
the same complexity)

J.-G. Mailly | Intro to Complexity

 Outline
4

Mappings and Asymptotic Bounds

Problems and Languages

Graph Theory
Non-Directed Graphs
Directed Graphs

Logic

J.-G. Mailly | Intro to Complexity

 Mappings and Asymptotic Bounds
5

Integer mappings
We use mappings f : N → N to represent the time (or space) used to
solve a problem
I Intuition: If the problem entry has size n, f (n) steps are required
to compute the result
I If needed, we use the closest integer value (e.g. log(n) means
dlog(n)e)

J.-G. Mailly | Intro to Complexity

 Mappings and Asymptotic Bounds
5

Integer mappings
We use mappings f : N → N to represent the time (or space) used to
solve a problem
I Intuition: If the problem entry has size n, f (n) steps are required
to compute the result
I If needed, we use the closest integer value (e.g. log(n) means
dlog(n)e)

Asymptotic bounds – O notation

Given a mapping f , O(f (n)) is the set of mappings g s.t. ∃no , c,

∀n ≥ n0 , g(n) ≤ c × f (n)

I Intuition: When n is large enough, g is smaller than f modulo

some constant c
J.-G. Mailly | Intro to Complexity
 Example
6

g(n)

I We suppose that we
have an algorithm
which solves a graphs
problem in
g(n) = 4×n2 +3×n+2
steps, when n is the
size of the graph (i.e.
number of vertices)

n
J.-G. Mailly | Intro to Complexity
 Example
6

g(n) 5 × f (n)

I We suppose that we
have an algorithm
which solves a graphs
problem in
g(n) = 4×n2 +3×n+2
steps, when n is the
size of the graph (i.e.
number of vertices)
I When n becomes large
enough, g(n) ≤
5 × f (n) = 5 × n2 , so
g(n) ∈ O(n2 )
n
J.-G. Mailly | Intro to Complexity
 Example
6

g(n) 5 × f (n)

I We suppose that we
have an algorithm
which solves a graphs
problem in
g(n) = 4×n2 +3×n+2
steps, when n is the
size of the graph (i.e.
number of vertices)
I When n becomes large
enough, g(n) ≤
5 × f (n) = 5 × n2 , so
g(n) ∈ O(n2 )

n0 n
J.-G. Mailly | Intro to Complexity
 On Usual O Families
7

From Slow to Fast Increase

C is an arbitrary constant, and log is any logarithmic function
Family of Functions Name
O(1) Constant
O(log(n)) Logarithmic
O((log(n))c ) Polylogarithmic
O(n) Linear
O(n log(n)) Linearithmic
O(n2 ) Quadratic
O(nC ) Polynomial
O(C n ) Exponential
O(n!) Factorial

On the Sum of Functions

If f (n) = g(n) + h(n), O(f (n)) = max(O(g(n), O(h(n))
J.-G. Mailly | Intro to Complexity
 Explanation of the O “Hierarchy”
8

I O(1) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × 1}

Fast

O(1)

J.-G. Mailly | Intro to Complexity

 Explanation of the O “Hierarchy”
8

I O(1) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × 1}

I O(log(n)) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × log(n)}

Fast

Slower

O(1)

O(log(n))

J.-G. Mailly | Intro to Complexity

 Explanation of the O “Hierarchy”
8

I O(1) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × 1}

I O(log(n)) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × log(n)}
I ...

Fast

Slower

O(1) •••

O(log(n))

J.-G. Mailly | Intro to Complexity

 Explanation of the O “Hierarchy”
8

I O(1) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × 1}

I O(log(n)) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × log(n)}
I ...
I O(n!) = {g | ∃n0 , c, ∀n ≥ n0 , g(n) ≤ c × n!}
Fast

Slower

O(1) ••• Slower again

O(log(n))

O(n!)

J.-G. Mailly | Intro to Complexity

 Outline
9

Mappings and Asymptotic Bounds

Problems and Languages

Graph Theory
Non-Directed Graphs
Directed Graphs

Logic

J.-G. Mailly | Intro to Complexity

 Different Kinds of Problems (1/2)
10

Decision Problem, Function Problem

A decision problem over a set of input data E is a mapping from any
element in E to a value in {false, true} (' {0, 1} ' {NO,YES}).
A function problem over a set of input data E is a mapping from any
element in E to a single outcome.

Solving an Equation
Decision Problem Does the equation f (x) = y have a solution?
Function Problem Give a solution of the equation f (x) = y

J.-G. Mailly | Intro to Complexity

 Different Kinds of Problems (2/2)
11

Enumeration Problem, Optimization Problem

Given a function problem P,
The enumeration problem ENUM-P over E is a mapping from any
element ei ∈ E to the set of all outcome of P over ei .
The optimization problem OPT-P over E is a mapping from any
element in E to a single outcome which minimizes a given criterion.

Solving an Equation
Enumeration Problem Give all the solutions of the equation f (x) = y
Optimization Problem Give a minimal solution of the equation
f (x) = y

J.-G. Mailly | Intro to Complexity

 Languages
12

Definition
I Set of symbols Σ called vocabulary or alphabet.
I A word w is a sequence of symbols w1 w2 . . . wk , with wi ∈ Σ for
all i.
I Σ∗ = {w1 w2 . . . wk | wi ∈ Σ, k ∈ N}
I a language L is any subset of Σ∗ , the complement of L is
L̄ = Σ∗ \ L

Language ' Decision Problem

I For any language L, P(L) is the decision problem:
Given x ∈ Σ∗ , does x belong to L?
I For any decision problem P, L(P) is the language:
{x ∈ instances of P | x is a positive instance of P}

J.-G. Mailly | Intro to Complexity

 Outline
13

Mappings and Asymptotic Bounds

Problems and Languages

Graph Theory
Non-Directed Graphs
Directed Graphs

Logic

J.-G. Mailly | Intro to Complexity

 Non-Directed Graphs
14

Definition
A non-directed graph is a pair G = hN, Ei where N is the set of nodes
and E ⊆ pairs(N) is the set of edges, with
pairs(N) = {{xi , xj } | xi , xj ∈ N}

J.-G. Mailly | Intro to Complexity

 Non-Directed Graphs
14

Definition
A non-directed graph is a pair G = hN, Ei where N is the set of nodes
and E ⊆ pairs(N) is the set of edges, with
pairs(N) = {{xi , xj } | xi , xj ∈ N}

G = hN, Ei, with N = {x1 , x2 , x3 , x4 , x5 , x6 , x7 } and

E = {{x1 , x2 }, {x2 , x3 }, {x3 , x4 }, {x4 , x5 }, {x5 , x6 }, {x6 , x7 }, {x7 , x5 }}

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
15

Definition
A subgraph in a non-directed graph G = hN, Ei is a pair
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E. If
E 0 = pairs(N 0 ) ∩ E, we say that G0 is the subgraph of G induced by N 0

J.-G. Mailly | Intro to Complexity

 Subgraphs
15

Definition
A subgraph in a non-directed graph G = hN, Ei is a pair
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E. If
E 0 = pairs(N 0 ) ∩ E, we say that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {{x1 , x2 }}i is a subgraph of G

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
15

Definition
A subgraph in a non-directed graph G = hN, Ei is a pair
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E. If
E 0 = pairs(N 0 ) ∩ E, we say that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {{x1 , x2 }}i is a subgraph of G

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
15

Definition
A subgraph in a non-directed graph G = hN, Ei is a pair
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E. If
E 0 = pairs(N 0 ) ∩ E, we say that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {{x1 , x2 }}i is a subgraph of G

G0 = h{x1 , x2 , x3 }, {{x1 , x2 }, {x2 , x3 }}i is the subgraph of G induced
by {x1 , x2 , x3 }

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
15

Definition
A subgraph in a non-directed graph G = hN, Ei is a pair
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E. If
E 0 = pairs(N 0 ) ∩ E, we say that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {{x1 , x2 }}i is a subgraph of G

G0 = h{x1 , x2 , x3 }, {{x1 , x2 }, {x2 , x3 }}i is the subgraph of G induced
by {x1 , x2 , x3 }

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Chain
16

Definition
A chain in a non-directed graph G = hN, Ei is a vector of nodes
(n1 , . . . , nk ) such that ∀0 ≤ i < k , {ni , ni+1 } ∈ E.

J.-G. Mailly | Intro to Complexity

 Chain
16

Definition
A chain in a non-directed graph G = hN, Ei is a vector of nodes
(n1 , . . . , nk ) such that ∀0 ≤ i < k , {ni , ni+1 } ∈ E.

(x3 , x4 , x5 , x6 ) and (x3 , x4 , x5 , x7 ) are chains of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Chain
16

Definition
A chain in a non-directed graph G = hN, Ei is a vector of nodes
(n1 , . . . , nk ) such that ∀0 ≤ i < k , {ni , ni+1 } ∈ E.

(x3 , x4 , x5 , x6 ) and (x3 , x4 , x5 , x7 ) are chains of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Chain
16

Definition
A chain in a non-directed graph G = hN, Ei is a vector of nodes
(n1 , . . . , nk ) such that ∀0 ≤ i < k , {ni , ni+1 } ∈ E.

(x3 , x4 , x5 , x6 ) and (x3 , x4 , x5 , x7 ) are chains of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Chain
16

Definition
A chain in a non-directed graph G = hN, Ei is a vector of nodes
(n1 , . . . , nk ) such that ∀0 ≤ i < k , {ni , ni+1 } ∈ E.

(x3 , x4 , x5 , x6 ) and (x3 , x4 , x5 , x7 ) are chains of G

(x1 , x6 ) is not a chain of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Chain
16

Definition
A chain in a non-directed graph G = hN, Ei is a vector of nodes
(n1 , . . . , nk ) such that ∀0 ≤ i < k , {ni , ni+1 } ∈ E.

(x3 , x4 , x5 , x6 ) and (x3 , x4 , x5 , x7 ) are chains of G

(x1 , x6 ) is not a chain of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Clique
17

Definition
A clique in a non-directed graph G = hN, Ei is subgraph
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E, such that
∀xi , xj ∈ N 0 , xi 6= xj , {xi , xj } ∈ E 0

J.-G. Mailly | Intro to Complexity

 Clique
17

Definition
A clique in a non-directed graph G = hN, Ei is subgraph
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E, such that
∀xi , xj ∈ N 0 , xi 6= xj , {xi , xj } ∈ E 0

h{x5 , x6 , x7 }, {{x5 , x6 }, {x6 , x7 }, {x7 , x5 }, }i is a clique in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Clique
17

Definition
A clique in a non-directed graph G = hN, Ei is subgraph
G0 = hN 0 , E 0 i, with N 0 ⊆ N and E 0 ⊆ pairs(N 0 ) ∩ E, such that
∀xi , xj ∈ N 0 , xi 6= xj , {xi , xj } ∈ E 0

h{x5 , x6 , x7 }, {{x5 , x6 }, {x6 , x7 }, {x7 , x5 }, }i is a clique in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
18

Definition
A cycle in a non-directed graph G = hN, Ei is a chain (n1 , . . . , nk )
such that n1 = nk . The length of the cycle is k − 1

J.-G. Mailly | Intro to Complexity

 Cycle
18

Definition
A cycle in a non-directed graph G = hN, Ei is a chain (n1 , . . . , nk )
such that n1 = nk . The length of the cycle is k − 1

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
18

Definition
A cycle in a non-directed graph G = hN, Ei is a chain (n1 , . . . , nk )
such that n1 = nk . The length of the cycle is k − 1

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Directed Graphs
19

Definition
A directed graph is a pair G = hN, Ei where N is the set of nodes and
E ⊆ N × N is the set of edges

J.-G. Mailly | Intro to Complexity

 Directed Graphs
19

Definition
A directed graph is a pair G = hN, Ei where N is the set of nodes and
E ⊆ N × N is the set of edges

G = hN, Ei, with N = {x1 , x2 , x3 , x4 , x5 , x6 , x7 } and E =

{(x1 , x2 ), (x2 , x3 ), (x3 , x4 ), (x4 , x3 ), (x4 , x5 ), (x5 , x6 ), (x6 , x7 ), (x7 , x5 )}

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
20

Definition
A subgraph in a directed graph G = hN, Ei is a pair G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E. If E 0 = (N 0 × N 0 ) ∩ E, we say
that G0 is the subgraph of G induced by N 0

J.-G. Mailly | Intro to Complexity

 Subgraphs
20

Definition
A subgraph in a directed graph G = hN, Ei is a pair G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E. If E 0 = (N 0 × N 0 ) ∩ E, we say
that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {(x1 , x2 )}i is a subgraph of G

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
20

Definition
A subgraph in a directed graph G = hN, Ei is a pair G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E. If E 0 = (N 0 × N 0 ) ∩ E, we say
that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {(x1 , x2 )}i is a subgraph of G

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
20

Definition
A subgraph in a directed graph G = hN, Ei is a pair G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E. If E 0 = (N 0 × N 0 ) ∩ E, we say
that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {(x1 , x2 )}i is a subgraph of G

G0 = h{x1 , x2 , x3 }, {(x1 , x2 ), (x2 , x3 )}i is the subgraph of G induced by
{x1 , x2 , x3 }

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Subgraphs
20

Definition
A subgraph in a directed graph G = hN, Ei is a pair G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E. If E 0 = (N 0 × N 0 ) ∩ E, we say
that G0 is the subgraph of G induced by N 0

G0 = h{x1 , x2 , x3 }, {(x1 , x2 )}i is a subgraph of G

G0 = h{x1 , x2 , x3 }, {(x1 , x2 ), (x2 , x3 )}i is the subgraph of G induced by
{x1 , x2 , x3 }

x6
⇒ x1 x2 x3
x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Path
21

Definition
A path in a directed graph G = hN, Ei is a vector of nodes (n1 , . . . , nk )
such that ∀0 ≤ i < k , (ni , ni+1 ) ∈ E.

J.-G. Mailly | Intro to Complexity

 Path
21

Definition
A path in a directed graph G = hN, Ei is a vector of nodes (n1 , . . . , nk )
such that ∀0 ≤ i < k , (ni , ni+1 ) ∈ E.

(x3 , x4 , x5 , x6 ) is a path of G, but (x3 , x4 , x5 , x7 ) is not

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Path
21

Definition
A path in a directed graph G = hN, Ei is a vector of nodes (n1 , . . . , nk )
such that ∀0 ≤ i < k , (ni , ni+1 ) ∈ E.

(x3 , x4 , x5 , x6 ) is a path of G, but (x3 , x4 , x5 , x7 ) is not

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Path
21

Definition
A path in a directed graph G = hN, Ei is a vector of nodes (n1 , . . . , nk )
such that ∀0 ≤ i < k , (ni , ni+1 ) ∈ E.

(x3 , x4 , x5 , x6 ) is a path of G, but (x3 , x4 , x5 , x7 ) is not

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Path
21

Definition
A path in a directed graph G = hN, Ei is a vector of nodes (n1 , . . . , nk )
such that ∀0 ≤ i < k , (ni , ni+1 ) ∈ E.

(x3 , x4 , x5 , x6 ) is a path of G, but (x3 , x4 , x5 , x7 ) is not

(x1 , x6 ) is not a path of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Path
21

Definition
A path in a directed graph G = hN, Ei is a vector of nodes (n1 , . . . , nk )
such that ∀0 ≤ i < k , (ni , ni+1 ) ∈ E.

(x3 , x4 , x5 , x6 ) is a path of G, but (x3 , x4 , x5 , x7 ) is not

(x1 , x6 ) is not a path of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Clique
22

Definition
A clique in a directed graph G = hN, Ei is subgraph G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E, such that ∀xi , xj ∈ N 0 , xi 6= xj ,
(xi , xj ) ∈ E 0

J.-G. Mailly | Intro to Complexity

 Clique
22

Definition
A clique in a directed graph G = hN, Ei is subgraph G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E, such that ∀xi , xj ∈ N 0 , xi 6= xj ,
(xi , xj ) ∈ E 0

h{x5 , x6 , x7 }, {(x5 , x6 ), (x6 , x7 ), (x7 , x5 )}i is not a clique in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Clique
22

Definition
A clique in a directed graph G = hN, Ei is subgraph G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E, such that ∀xi , xj ∈ N 0 , xi 6= xj ,
(xi , xj ) ∈ E 0

h{x5 , x6 , x7 }, {(x5 , x6 ), (x6 , x7 ), (x7 , x5 )}i is not a clique in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Clique
22

Definition
A clique in a directed graph G = hN, Ei is subgraph G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E, such that ∀xi , xj ∈ N 0 , xi 6= xj ,
(xi , xj ) ∈ E 0

h{x5 , x6 , x7 }, {(x5 , x6 ), (x6 , x7 ), (x7 , x5 )}i is not a clique in G

h{x3 , x4 }, {(x3 , x4 ), (x4 , x3 )}i is a clique in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Clique
22

Definition
A clique in a directed graph G = hN, Ei is subgraph G0 = hN 0 , E 0 i,
with N 0 ⊆ N and E 0 ⊆ (N 0 × N 0 ) ∩ E, such that ∀xi , xj ∈ N 0 , xi 6= xj ,
(xi , xj ) ∈ E 0

h{x5 , x6 , x7 }, {(x5 , x6 ), (x6 , x7 ), (x7 , x5 )}i is not a clique in G

h{x3 , x4 }, {(x3 , x4 ), (x4 , x3 )}i is a clique in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
23

Definition
A cycle in a directed graph G = hN, Ei is a path (n1 , . . . , nk ) such that
n1 = nk . The length of the cycle is k − 1

J.-G. Mailly | Intro to Complexity

 Cycle
23

Definition
A cycle in a directed graph G = hN, Ei is a path (n1 , . . . , nk ) such that
n1 = nk . The length of the cycle is k − 1

(x3 , x4 , x3 ) is a cycle of length 2 in G

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
23

Definition
A cycle in a directed graph G = hN, Ei is a path (n1 , . . . , nk ) such that
n1 = nk . The length of the cycle is k − 1

(x3 , x4 , x3 ) is a cycle of length 2 in G

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
23

Definition
A cycle in a directed graph G = hN, Ei is a path (n1 , . . . , nk ) such that
n1 = nk . The length of the cycle is k − 1

(x3 , x4 , x3 ) is a cycle of length 2 in G

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
23

Definition
A cycle in a directed graph G = hN, Ei is a path (n1 , . . . , nk ) such that
n1 = nk . The length of the cycle is k − 1

(x3 , x4 , x3 ) is a cycle of length 2 in G

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G
(x2 , x3 , x4 , x3 ) is not a cycle of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Cycle
23

Definition
A cycle in a directed graph G = hN, Ei is a path (n1 , . . . , nk ) such that
n1 = nk . The length of the cycle is k − 1

(x3 , x4 , x3 ) is a cycle of length 2 in G

(x5 , x6 , x7 , x5 ) is a cycle of length 3 in G
(x2 , x3 , x4 , x3 ) is not a cycle of G

x6

x1 x2 x3 x4 x5

x7

J.-G. Mailly | Intro to Complexity

 Outline
24

Mappings and Asymptotic Bounds

Problems and Languages

Graph Theory
Non-Directed Graphs
Directed Graphs

Logic

J.-G. Mailly | Intro to Complexity

 Propositional Logic
25

Definition
Propositional logic is a mathematical formalism which studies
propositions, i.e. logical objects which can be true or false. A
propositional formula is the combinations of propositions with
connectives

J.-G. Mailly | Intro to Complexity

 Propositional Logic
25

Definition
Propositional logic is a mathematical formalism which studies
propositions, i.e. logical objects which can be true or false. A
propositional formula is the combinations of propositions with
connectives
Examples of propositions:
I it’s raining
I it’s sunny
I I need an umbrella
I I need sunscreen

J.-G. Mailly | Intro to Complexity

 Propositional Logic
25

Definition
Propositional logic is a mathematical formalism which studies
propositions, i.e. logical objects which can be true or false. A
propositional formula is the combinations of propositions with
connectives
Examples of propositions:
I it’s raining
I it’s sunny
I I need an umbrella
I I need sunscreen
Propositions are usually represented by Boolean variables, i.e.
mathematical objects which can be assigned a value from a binary
set B = {0, 1} (sometimes written {False, True} or {⊥, >})

J.-G. Mailly | Intro to Complexity

 Syntax of Propositional Logic
26

Let V = {x1 , . . . , xn } be a set of Boolean variables

atom ∀xi ∈ V , xi is a (well-formed) formula
negation if ϕ is a formula, then ¬ϕ is a formula
conjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∧ ϕ2 is a formula

J.-G. Mailly | Intro to Complexity

 Syntax of Propositional Logic
26

Let V = {x1 , . . . , xn } be a set of Boolean variables

atom ∀xi ∈ V , xi is a (well-formed) formula
negation if ϕ is a formula, then ¬ϕ is a formula
conjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∧ ϕ2 is a formula
disjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∨ ϕ2 is a formula

J.-G. Mailly | Intro to Complexity

 Syntax of Propositional Logic
26

Let V = {x1 , . . . , xn } be a set of Boolean variables

atom ∀xi ∈ V , xi is a (well-formed) formula
negation if ϕ is a formula, then ¬ϕ is a formula
conjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∧ ϕ2 is a formula
disjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∨ ϕ2 is a formula
implication if ϕ1 and ϕ2 are formulas, then ϕ1 ⇒ ϕ2 is a formula

J.-G. Mailly | Intro to Complexity

 Syntax of Propositional Logic
26

Let V = {x1 , . . . , xn } be a set of Boolean variables

atom ∀xi ∈ V , xi is a (well-formed) formula
negation if ϕ is a formula, then ¬ϕ is a formula
conjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∧ ϕ2 is a formula
disjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∨ ϕ2 is a formula
implication if ϕ1 and ϕ2 are formulas, then ϕ1 ⇒ ϕ2 is a formula
equivalence if ϕ1 and ϕ2 are formulas, then ϕ1 ⇔ ϕ2 is a formula

J.-G. Mailly | Intro to Complexity

 Syntax of Propositional Logic
26

Let V = {x1 , . . . , xn } be a set of Boolean variables

atom ∀xi ∈ V , xi is a (well-formed) formula
negation if ϕ is a formula, then ¬ϕ is a formula
conjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∧ ϕ2 is a formula
disjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∨ ϕ2 is a formula
implication if ϕ1 and ϕ2 are formulas, then ϕ1 ⇒ ϕ2 is a formula
equivalence if ϕ1 and ϕ2 are formulas, then ϕ1 ⇔ ϕ2 is a formula
We use parentheses to simplify the reading and clarify priorities

J.-G. Mailly | Intro to Complexity

 Syntax of Propositional Logic
26

Let V = {x1 , . . . , xn } be a set of Boolean variables

atom ∀xi ∈ V , xi is a (well-formed) formula
negation if ϕ is a formula, then ¬ϕ is a formula
conjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∧ ϕ2 is a formula
disjunction if ϕ1 and ϕ2 are formulas, then ϕ1 ∨ ϕ2 is a formula
implication if ϕ1 and ϕ2 are formulas, then ϕ1 ⇒ ϕ2 is a formula
equivalence if ϕ1 and ϕ2 are formulas, then ϕ1 ⇔ ϕ2 is a formula
We use parentheses to simplify the reading and clarify priorities

Formulas Not Formulas

I x1 , ¬x1 , ¬¬x1 I x1 ¬, ¬x1 ¬

I x1 ∧ x2 , x1 ∨ x2 I x1 ∧, ∨x1
I x1 ∧ (x3 ∨ (x4 ⇒ x1 )) I x1 ⇒ (∨x2 )
I ϕ∧ψ I ∧ψ
J.-G. Mailly | Intro to Complexity
 Semantics of Propositional Logic
27

Let V = {x1 , . . . , xn } be a set of Boolean variables. An interpretation

ω is a mapping from V to B. Interpretations can be extended to
formulas:
atom if ϕ = xi , then ω(ϕ) = ω(xi )
negation if ϕ = ¬ϕ1 , then ω(ϕ) = 1 − ω(ϕ1 )
conjunction if ϕ = ϕ1 ∧ ϕ2 , then ω(ϕ) = min(ω(ϕ1 ), ω(ϕ2 ))

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
27

Let V = {x1 , . . . , xn } be a set of Boolean variables. An interpretation

ω is a mapping from V to B. Interpretations can be extended to
formulas:
atom if ϕ = xi , then ω(ϕ) = ω(xi )
negation if ϕ = ¬ϕ1 , then ω(ϕ) = 1 − ω(ϕ1 )
conjunction if ϕ = ϕ1 ∧ ϕ2 , then ω(ϕ) = min(ω(ϕ1 ), ω(ϕ2 ))
disjunction if ϕ = ϕ1 ∨ ϕ2 , then ω(ϕ) = max(ω(ϕ1 ), ω(ϕ2 ))

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
27

Let V = {x1 , . . . , xn } be a set of Boolean variables. An interpretation

ω is a mapping from V to B. Interpretations can be extended to
formulas:
atom if ϕ = xi , then ω(ϕ) = ω(xi )
negation if ϕ = ¬ϕ1 , then ω(ϕ) = 1 − ω(ϕ1 )
conjunction if ϕ = ϕ1 ∧ ϕ2 , then ω(ϕ) = min(ω(ϕ1 ), ω(ϕ2 ))
disjunction if ϕ = ϕ1 ∨ ϕ2 , then ω(ϕ) = max(ω(ϕ1 ), ω(ϕ2 ))
implication if ϕ = ϕ1 ⇒ ϕ2 , then ω(ϕ) = ω(¬ϕ1 ∨ ϕ2 )

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
27

Let V = {x1 , . . . , xn } be a set of Boolean variables. An interpretation

ω is a mapping from V to B. Interpretations can be extended to
formulas:
atom if ϕ = xi , then ω(ϕ) = ω(xi )
negation if ϕ = ¬ϕ1 , then ω(ϕ) = 1 − ω(ϕ1 )
conjunction if ϕ = ϕ1 ∧ ϕ2 , then ω(ϕ) = min(ω(ϕ1 ), ω(ϕ2 ))
disjunction if ϕ = ϕ1 ∨ ϕ2 , then ω(ϕ) = max(ω(ϕ1 ), ω(ϕ2 ))
implication if ϕ = ϕ1 ⇒ ϕ2 , then ω(ϕ) = ω(¬ϕ1 ∨ ϕ2 )
equivalence if ϕ = ϕ1 ⇔ ϕ2 , then ω(ϕ) = ω((ϕ1 ⇒ ϕ2 ) ∧ (ϕ2 ⇒ ϕ1 ))

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0

Intuitively:
I the negation takes the opposite of the original value

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0
I ω(x1 ∧ x2 ) = min(ω(x1 ), ω(x2 )) = min(0, 1) = 0

Intuitively:
I the negation takes the opposite of the original value

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0
I ω(x1 ∧ x2 ) = min(ω(x1 ), ω(x2 )) = min(0, 1) = 0
I ω(x2 ∧ x3 ) = min(ω(x2 ), ω(x3 )) = min(1, 1) = 1

Intuitively:
I the negation takes the opposite of the original value

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0
I ω(x1 ∧ x2 ) = min(ω(x1 ), ω(x2 )) = min(0, 1) = 0
I ω(x2 ∧ x3 ) = min(ω(x2 ), ω(x3 )) = min(1, 1) = 1

Intuitively:
I the negation takes the opposite of the original value
I the conjunction is true iff both values are true

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0
I ω(x1 ∧ x2 ) = min(ω(x1 ), ω(x2 )) = min(0, 1) = 0
I ω(x2 ∧ x3 ) = min(ω(x2 ), ω(x3 )) = min(1, 1) = 1
I ω(x1 ∨ x2 ) = max(ω(x1 ), ω(x2 )) = max(0, 1) = 1

Intuitively:
I the negation takes the opposite of the original value
I the conjunction is true iff both values are true

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0
I ω(x1 ∧ x2 ) = min(ω(x1 ), ω(x2 )) = min(0, 1) = 0
I ω(x2 ∧ x3 ) = min(ω(x2 ), ω(x3 )) = min(1, 1) = 1
I ω(x1 ∨ x2 ) = max(ω(x1 ), ω(x2 )) = max(0, 1) = 1
I ω(x1 ∨ x4 ) = max(ω(x1 ), ω(x4 )) = max(0, 0) = 0
Intuitively:
I the negation takes the opposite of the original value
I the conjunction is true iff both values are true

J.-G. Mailly | Intro to Complexity

 Semantics of Propositional Logic
Example of Interpretation
28

V = {x1 , x2 , x3 , x4 }, ω(x1 ) = ω(x4 ) = 0, ω(x2 ) = ω(x3 ) = 1

I ω(¬x1 ) = 1 − ω(x1 ) = 1, ω(¬x2 ) = 1 − ω(x2 ) = 0
I ω(x1 ∧ x2 ) = min(ω(x1 ), ω(x2 )) = min(0, 1) = 0
I ω(x2 ∧ x3 ) = min(ω(x2 ), ω(x3 )) = min(1, 1) = 1
I ω(x1 ∨ x2 ) = max(ω(x1 ), ω(x2 )) = max(0, 1) = 1
I ω(x1 ∨ x4 ) = max(ω(x1 ), ω(x4 )) = max(0, 0) = 0
Intuitively:
I the negation takes the opposite of the original value
I the conjunction is true iff both values are true
I the disjunction is true iff at least one of the values is true

J.-G. Mailly | Intro to Complexity

 Other Notations for Interpretations
29

Sets of Atoms
I An interpretation can be written as the set of atoms which are
assigned 1

J.-G. Mailly | Intro to Complexity

 Other Notations for Interpretations
29

Sets of Atoms
I An interpretation can be written as the set of atoms which are
assigned 1
I Example: V = {x1 , x2 , x3 }, ω(x1 ) = 0, ω(x2 ) = ω(x3 ) = 1 is
equivalent to ω = {x2 , x3 }

J.-G. Mailly | Intro to Complexity

 Other Notations for Interpretations
29

Sets of Atoms
I An interpretation can be written as the set of atoms which are
assigned 1
I Example: V = {x1 , x2 , x3 }, ω(x1 ) = 0, ω(x2 ) = ω(x3 ) = 1 is
equivalent to ω = {x2 , x3 }

Vectors of Bits
I An interpretation can be written as a vector of 0 and 1; the order
has to be fixed (usually lexicographical order)

J.-G. Mailly | Intro to Complexity

 Other Notations for Interpretations
29

Sets of Atoms
I An interpretation can be written as the set of atoms which are
assigned 1
I Example: V = {x1 , x2 , x3 }, ω(x1 ) = 0, ω(x2 ) = ω(x3 ) = 1 is
equivalent to ω = {x2 , x3 }

Vectors of Bits
I An interpretation can be written as a vector of 0 and 1; the order
has to be fixed (usually lexicographical order)
I Example: V = {x1 , x2 , x3 }, ω(x1 ) = 0, ω(x2 ) = ω(x3 ) = 1 is
equivalent to ω = 011

J.-G. Mailly | Intro to Complexity

 Satisfaction of Formulas
30

I If ω(ϕ) = 1, then ω is a model of ϕ. We also say that ϕ is

satisfied by ω
I If ω(ϕ) = 0, then ω is a counter-model of ϕ
I mod(ϕ) is the set of models of ϕ
I If mod(ϕ) = ∅, then ϕ is inconsistent (or insatisfiable)
I If mod(ϕ) 6= ∅, then ϕ is consistent (or satisfiable)
I If mod(ϕ) is the set of all possible interpretations, then ϕ is valid

J.-G. Mailly | Intro to Complexity

 Satisfaction, Consequence, Equivalence
31

We define some meta-language symbols for reasoning about

interpretations and formulas:
I ω |= ϕ means that (the interpretation) ω satisfies (the formula) ϕ,
i.e. ω ∈ mod(ϕ)
I ϕ ` ψ means that (the formula) ψ is a consequence of (the
formula) ϕ, formally defined as mod(ϕ) ⊆ mod(ψ)
I ϕ ≡ ψ means that ϕ and ψ are equivalent, formally defined as
ϕ ` ψ and ψ ` ϕ (which implies mod(ϕ) = mod(ψ))

J.-G. Mailly | Intro to Complexity

 Normal Forms
32

I A formula is a literal iff it is an atom or the negation of an atom

x1 and ¬x1 are literals; x1 ∨ x2 is not a literal

J.-G. Mailly | Intro to Complexity

 Normal Forms
32

I A formula is a literal iff it is an atom or the negation of an atom

x1 and ¬x1 are literals; x1 ∨ x2 is not a literal
I A formula is a clause iff it is a disjunction of literals
x1 ∨ x2 ∨ ¬x3 is a clause ; x1 ∨ (x2 ∧ x3 ) is not a clause

J.-G. Mailly | Intro to Complexity

 Normal Forms
32

I A formula is a literal iff it is an atom or the negation of an atom

x1 and ¬x1 are literals; x1 ∨ x2 is not a literal
I A formula is a clause iff it is a disjunction of literals
x1 ∨ x2 ∨ ¬x3 is a clause ; x1 ∨ (x2 ∧ x3 ) is not a clause
I A formula is a cube iff it is a conjunction of literals
x1 ∧ x2 ∧ ¬x3 is a cube ; x1 ∧ (x2 ∨ x3 ) is not a cube

J.-G. Mailly | Intro to Complexity

 Normal Forms
32

I A formula is a literal iff it is an atom or the negation of an atom

x1 and ¬x1 are literals; x1 ∨ x2 is not a literal
I A formula is a clause iff it is a disjunction of literals
x1 ∨ x2 ∨ ¬x3 is a clause ; x1 ∨ (x2 ∧ x3 ) is not a clause
I A formula is a cube iff it is a conjunction of literals
x1 ∧ x2 ∧ ¬x3 is a cube ; x1 ∧ (x2 ∨ x3 ) is not a cube

Conjunctive Normal Form

A formula is in Conjunctive Normal Form (CNF) iff it is a conjunction
of clauses

J.-G. Mailly | Intro to Complexity

 Normal Forms
32

I A formula is a literal iff it is an atom or the negation of an atom

x1 and ¬x1 are literals; x1 ∨ x2 is not a literal
I A formula is a clause iff it is a disjunction of literals
x1 ∨ x2 ∨ ¬x3 is a clause ; x1 ∨ (x2 ∧ x3 ) is not a clause
I A formula is a cube iff it is a conjunction of literals
x1 ∧ x2 ∧ ¬x3 is a cube ; x1 ∧ (x2 ∨ x3 ) is not a cube

Conjunctive Normal Form

A formula is in Conjunctive Normal Form (CNF) iff it is a conjunction
of clauses

Disjunctive Normal Form

A formula is in Disjunctive Normal Form (DNF) iff it is a disjunction of
cubes

J.-G. Mailly | Intro to Complexity

 Normal Forms
Examples
33

Examples of CNF formulas

I (x1 ∨ x2 ) ∧ (x3 ∨ x4 )
I (x1 ∨ ¬x2 ∨ x5 ) ∧ (x3 ∨ x5 )

Examples of DNF formulas

I (x1 ∧ x2 ) ∨ (x3 ∧ x4 )
I x1 ∨ (x2 ∧ x4 ) ∨ (x3 ∧ x4 ∧ ¬x5 )

J.-G. Mailly | Intro to Complexity

 Normal Forms
Semantics of CNF Formulas 34

I A clause is satisfied if at least one of its literals is satisfied

I A CNF formula is satisfied if all its clauses are satisfied

Example
With ω(x1 ) = ω(x2 ) = 1 and ω(x3 ) = ω(x4 ) = 0
I (x1 ∨ x2 ) ∧ (x3 ∨ x4 ) is not satisfied

J.-G. Mailly | Intro to Complexity

 Normal Forms
Semantics of CNF Formulas 34

I A clause is satisfied if at least one of its literals is satisfied

I A CNF formula is satisfied if all its clauses are satisfied

Example
With ω(x1 ) = ω(x2 ) = 1 and ω(x3 ) = ω(x4 ) = 0
I (x1 ∨ x2 ) ∧ (x3 ∨ x4 ) is not satisfied
With ω(x1 ) = ω(x3 ) = 1 and ω(x2 ) = ω(x4 ) = 0
I (x1 ∨ x2 ) ∧ (x3 ∨ x4 ) is satisfied

J.-G. Mailly | Intro to Complexity

 Normal Forms
Semantics of DNF Formulas 35

I A cube is satisfied if all its literals are satisfied

I A DNF formula is satisfied if at least one of its cubes is satisfied

Example
With ω(x1 ) = ω(x3 ) = 1 and ω(x2 ) = ω(x4 ) = 0
I (x1 ∧ x2 ) ∨ (x3 ∧ x4 ) is not satisfied

J.-G. Mailly | Intro to Complexity

 Normal Forms
Semantics of DNF Formulas 35

I A cube is satisfied if all its literals are satisfied

I A DNF formula is satisfied if at least one of its cubes is satisfied

Example
With ω(x1 ) = ω(x3 ) = 1 and ω(x2 ) = ω(x4 ) = 0
I (x1 ∧ x2 ) ∨ (x3 ∧ x4 ) is not satisfied
With ω(x1 ) = ω(x2 ) = 1 and ω(x3 ) = ω(x4 ) = 0
I (x1 ∧ x2 ) ∨ (x3 ∧ x4 ) is satisfied

J.-G. Mailly | Intro to Complexity