• Keine Ergebnisse gefunden

Mathematics for linguists

N/A
N/A
Protected

Academic year: 2022

Aktie "Mathematics for linguists"

Copied!
17
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Mathematics for linguists

Gerhard J¨ager

gerhard.jaeger@uni-tuebingen.de

Uni T¨ubingen, WS 2009/2010

November 3, 2009

(2)

Composition of relations and functions

let R⊆A×B andS ⊆B×C be relations

new relation S◦R⊆A×C is formed as

S◦R:={hx, yi|there is a z such that R(x, z) and S(z, y)}

IfR andS are functions, S◦R is also a function.

(3)

The identity function

functionF :A→A is called identity functioniff F ={hx, xi ∈A×A|x∈A}

notation: F =idA

composition of a relationR with an identity function with the appropriate domain yieldsR again:

IfRA×B, then

idBR=R

RidA=R

IfF :A→B is an injective function, then:

FF−1idB

F−1F =idA

Generally, it holds that:

idπ0[R]R−1R

idπ1[R]RR−1

(4)

Properties of relations

Many relations that are used in linguistic applications have certain structural properties. The most important ones are briefly discussed here; the apply to relations on a given set, i.e.R⊆A×A.

Reflexivity

R⊆A×A is reflexiveiff for all x∈A it holds that R(x, x).

(5)

Reflexivity

Example

A={1,2,3}

R1={h1,1i,h2,2i,h3,3i,h3,1i}

R2={h1,1i,h2,2i}

R1 is reflexive but R2 is not (because h3,3i 6∈R2).

R⊆A×A is reflexive iffidA⊆R.

More examples

“has the same final digit as”

“has birthday on the same day as”

“greater or equal”

(6)

Irreflexivity

Relations that are not reflexive are called non-reflexive

Relations that never connect an object to itself are called irreflexive.

Irreflexivity

R is irreflexive iff there is no objectx such thatR(x, x).

In other words:R is irreflexive iffR∩idA=∅.

Examples

R3={h1,2i,h3,2i}

R4={ha, bi ∈N×N|a < b}

(7)

Symmetry

Symmetry

A relationR is symmetriciff for all x, y with R(x, y) it holds that R(y, x).

In other words:R is symmetric iffR=R−1. Examples

“married to”

“relatively prime”

“happened in the same year as”

“cousin of”

{h1,2i,h2,1i,h3,2i,h2,3i}

{h1,3i,h3,1i}

{h2,2i}

(8)

Asymmetry

Asymmetry

A relationR is asymmetric iff it never holds that bothR(x, y) andR(y, x).

Every asymmetric relation must be irreflexive.

examples:

{h2,3i,h1,2i}

{h1,3i,h2,3i,h1,2i}

{h1,2i}

(9)

Anti-symmetry

Anti-symmetry

A relationR is anti-symmetric iff whenever bothR(x, y) and R(y, x), then x=y.

R need not be reflexive to be anti-symmetric!

Every asymmetric relation is also anti-symmetric.

IfR is anti-symmetric, then R−idAis asymmetric.

(10)

Anti-symmetry

Examples for anti-symmetric relations

“greater or equal”

“divisible by”

“is a subset of”

{h2,3i,h1,1i}

{h1,1i,h2,2i}

{h1,2i,h2,3i}

(11)

Transitivity

Transitivity

A relationR is transitiveiff whenever R(x, y) andR(y, z), then R(x, z).

examples:

“older than”

“richer than”

“greater than”

“ancestor of”

equality

{h2,2i}

{h1,2i,h2,3i,h1,3i}

{h1,2i,h2,1i,h1,1i,h2,2i}

{h1,2i,h2,3i,h1,3i,h3,2i,h2,1i,h3,1i,h1,1i,h2,2i,h3,3i}

R is transitive iff R◦R⊆R.

(12)

Connectedness

Connectedness

A relationR⊆A×Ais connectediff for all x, y∈A withx6=y:

R(x, y)or R(y, x) (or both).

Examples

“greater than” (as applied to the natural numbers)

{h1,2i,h3,1i,h3,2i}

{h1,1i,h2,3i,h1,2i,h3,1i,h2,2i}

(13)

Properties of the inverse and the complement

R (6=∅) R−1 R

reflexive reflexive irreflexive irreflexive irreflexive reflexive

symmetric symmetric symmetric

asymmetric asymmetric not symmetric anti-symmetric anti-symmetric depends on R transitive transitive depends on R connected connected depends on R

(14)

Equivalence relations and partitions

Equivalence relations

A relationR is anequivalence relationiff R is reflexive, transitive and symmetric.

examples:

equality

“has the same hair color as”

“is of the same age as”

“has the same remainder if divided by”

notation:

[[a]]R:={x|R(a, x)}

[[a]]Ris the set of all objects which are reachable fromavia R.

(15)

Equivalence relations and partitions

Partition

LetA be a set. The setP ⊆℘(A) is a partition ofA iff

S

P =A, and

for all X, Y ∈P with X6=Y:X∩Y =∅.

Examples: Let A={a, b, c, d, e}

The following sets are partitions ofA:

P1={{a, c, d},{b, e}}

P2={{a},{b},{c},{d},{e}}

P3={{a, b, c, d, e}}

The following sets are not partitions ofA:

C={{a, b, c},{b, d},{e}}

D={{a},{b, e},{c}}

(16)

Equivalence relations and partitions

There is a tight connection between equivalence relations and partitions.

Let R⊆A×Abe an equivalence relation. Then the following set is a partition of π0[R]:

PR={x∈℘(A)|there is a y∈A such thatx= [[y]]R}

Let P be a partition of A. Then the following is an equivalence relation:

RP ={hx, yi ∈A×A|there is anX ∈P with x∈X andy∈X}

(Instead of “RP(a, b)” we sometimes write “a≡P b”.)

(17)

Equivalence relations and partitions

example:

A={1,2,3,4,5}

R = {h1,1i,h1,3i,h3,1i,h3,3i,h2,2i,h2,4i, h4,2i,h4,5i,h4,4i,h5,2i,h5,4i,h5,5i,h2,5i}

The corresponding partition is:

PR={{1,3},{2,4,5}}

Exercise: Let

R = {h1,1i,h1,2i,h2,1i,h2,2i,h3,3i,h3,5i h5,3i,h5,5i,h4,4i}

What is the corresponding partition PR?

Referenzen

ÄHNLICHE DOKUMENTE

An ordering is linear (or total) iff it is connected.. If an ordering R is not linear, there are two elements a and b such that neither R(a, b) nor

• in each derivation step, only the substring that corresponds to the right hand side of the rule applied is written into the next line, and.. • all new new symbols are connected by

• Per temporal unit, the automaton reads a symbol α on the input tape and moves along an arrow with the label α towards a new state. • If the automaton is in a final state after

• Regular languages are closed under the deletion of single elements from Σ: If I delete all occurrences of a given symbol.. — let’s say a — in all words of a regular language L,

• mildly context-sensitive grammars: family of grammar formalisms that are only slightly more powerful than type-2 grammars, but are able to express crossing dependencies. •

John comes to the party just in case Paul

direct extension of natural deduction for statement logic four new rules: one introduction rule and one. elimination rule for

two-valued semantics (every statement is either true or false) there is a sound and complete syntactic description of logical inference; there are several systems of syntactic