Scribd
Upload
32 views

Lecture 2a: First-Order Logic Over Words

This document discusses first-order logic over words. It contains the following key points: 1. FO1(<) refers to first-order logic using only one variable. Languages definable in FO1(<) are boolean combinations of languages of the form A* for A ⊆ Σ. 2. A language L is definable in FO1(<) if and only if L is recognized by a commutative and idempotent monoid. 3. Checking if a regular language is definable in FO1(<) can be done by computing its syntactic monoid and checking that it has the desired properties.

Uploaded by

rineeth m
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
32 views

Lecture 2a: First-Order Logic Over Words

This document discusses first-order logic over words. It contains the following key points: 1. FO1(<) refers to first-order logic using only one variable. Languages definable in FO1(<) are boolean combinations of languages of the form A* for A ⊆ Σ. 2. A language L is definable in FO1(<) if and only if L is recognized by a commutative and idempotent monoid. 3. Checking if a regular language is definable in FO1(<) can be done by computing its syntactic monoid and checking that it has the desired properties.

Uploaded by

rineeth m
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Lecture 2a: First-order Logic over words

Here is a formula describing the language of words over {a, b, c} that contain ab as a
subword:

∃x. ∃y. (a(x) ∧ b(y) ∧ (x < y) ∧ ∀z. (x < z) ⇒ (y = z) ∨ (y < z))

However, if the alphabet is {a, b} then this can expressed for the following simpler formula:

∃x. ∃y. a(x) ∧ b(y) ∧ (x < y)

The first formula uses 3 variables (x, y and z) while the second one uses only 2 and one of
things that we hope to prove in this course is that if the alphabet has at least 3 letters then
the above language cannot be expressed with FO(<) formulas that use at most 2 variables.

Exercise: Write down a FO(<) formula for the language (ab)∗ .

Formal Semantics for FO(<)


All the examples we used in the previous class were sentences i.e., every variable that occured
in the formula occured within the scope of a quantifier. (A variable that is tied to quantifier
is called a bound variable. Every variable in a sentence is a bound variable.) Given a sentence
φ, any word w either satisfies φ or does not.
However, in order to reason about sentences, one has to reason about subformulas of
sentences and these need not be sentences. As a matter of fact, subformulas of sentences are
usually NOT sentences.
For example, consider formulas First(x) and y = x + 1 that were used in Lecture 2. The
former has x as a free variable while the latter has x and y as free variables. A free variable
is one that is not “captured” by a quantifier. It does not make sense to ask if w satisfies
First(x). Instead, one has to give a word w and a position i in the word w and then one
may ask if First(i) is true. Similarly to evaluate y = x + 1, one needs values (i.e. positions)
for the variables x and y before we can verify its truth.
To meaningfully discuss the truth or falsity of a formula with k free variables, we need
a word along with assignment of positions to the k variables. For example, the formula
φ = (x < y) ∧ a(x) ∧ b(y) is true of bacabc with x assigned position 2 and y assigned position
5. On the other hand the formula φ is not true of bacabc if x and y are assigned position 5.
Thus, to evaluate the formula ϕ whose set of free variables is contained in a set V , we
need a word w = a1 a2 . . . an along with a valuation σ which assigns to each variable in V
a position from the set {1, 2, . . . , n}. We write w, σ |= ϕ, to denote that the formula ϕ
evaluates to true w.r.t. to the pair consisting of the word w and the valuation σ and this
relation |= is defined inductively as follows:

1
a1 a2 . . . an , σ |= a(x) if aσ(x) = a
a1 a2 . . . an , σ |= x<y if σ(x) < σ(y)
a1 a2 . . . an σ |= φ ∧ φ0 if (a1 a2 . . . an , σ |= φ) and (a1 a2 . . . an , σ |= φ0 )
a1 a2 . . . an , σ |= ¬φ if (a1 a2 . . . an , σ 6|= φ)
a1 a2 . . . an , σ |= ∃x.φ if there is an i ∈ {1, 2, 3, . . . n} such a1 a2 . . . an , σ[x : i] |= φ

where σ[v : y](u) = σ(u) if u 6= v and σ[v : y](v) = y. Thus with this notation,

bacabc, [x 7→ 2, y 7→ 5] |= (x < y) ∧ a(x) ∧ b(y)

and
bacabc, [x 7→ 5, y 7→ 5] 6|= (x < y) ∧ a(x) ∧ b(y)
From the definition above, it is clear that if w is a word, ϕ is a formula and σ and σ 0 are
valuations such that for all variables x that occur as free variables in ϕ we have σ(x) = σ 0 (x)
then w, σ |= ϕ iff w, σ 0 |= ϕ. In particular, if the formula is a sentence, then the valuation
is irrelevant and we may instead write w |= ϕ. The language defined by a sentence ϕ is the
set {w | w |= ϕ}.

First-order Logic with 1 variable:


As a warm up to understanding FO over words, let us consider the simple fragment consisting
1
all the formulas which use only one variable,
W say x. We write FO (<) to refer ∗to this fragment.
Let A ⊆ Σ. Then the formula ∀x. a∈A a(x) describes the language A . Further, since
1
FO (<) has the logical operators ¬ and ∨, the class of languages definable using formulas
in this logic is closed under boolean operations. Thus, every language that is a boolean
combination of languages of the form A∗ is definable in FO1 (<). What about the converse?
First of all we observe that in FO1 (<) the operators = and < are useless (i.e. they can
be eliminated). The only formulas that may use these operators are x < x which is always
false and x = x is always true. Thus we may assume that the atomic formulas are only of
the form a(x), a ∈ Σ. We will now show that if u ≡α v then for any sentence in ϕ ∈ FO1 (<),
u |= ϕ if and only if v |= ϕ. The obvious way to prove this is by induction on the size of
the formula, but alas that will lead us to consider formulas with free variables and hence we
have to strengthen the hypothesis of our inductive argument. The details are as follows:

Lemma 1 Let ϕ be any formula in FO1 (<) and let u and v be words with u ≡α v. Then, if
i and j are any two positions in u and v respectively, such that ui = vj , then u, [x 7→ i] |= ϕ
if and only if v, [x 7→ j] |= ϕ.

Proof: The proof is by induction on the structure of the formula.


Case 1: ϕ = a(x)
u, [x 7→ i] |= a(x) iff ui = a iff vj = a iff v, [x 7→ j] |= a(x).

2
Case 2: ϕ = ¬ψ
u, [x 7→ i] |= ¬ψ iff u, [x 7→ i] 6|= ψ iff v, [x 7→ j] 6|= ψ iff v, [x 7→ j] |= ¬ψ
Case 3: ϕ = ϕ1 ∨ ϕ2
u, [x 7→ i] |= ϕ1 ∨ ϕ2 iff u, [x 7→ i] |= ϕ1 or u, [x 7→ i] |= ϕ2 and by induction hypothesis,
u, [x 7→ i] |= ϕk iff v, [x 7→ j] |= ϕk for k ∈ 1, 2. The result follows.
Case 4: ϕ = ∃x. ψ
If u, [x 7→ i] |= ∃x. ψ then there is a position i0 in u such that u, [x 7→ i0 ] |= ψ. Further
since u ≡α v, there is a position j 0 in v such that ui0 = vj 0 . Then, by induction hypothesis,
v, [x 7→ j 0 ] |= ψ. Therefore, v, [x 7→ j] |= ∃x. ψ. The converse is proved symmetrically and
this completes the proof of this lemma.
As a consequence of the above lemma we have that if u ≡α v and ϕ is any formula in
FO1 (<) then u |= ϕ iff v |= ϕ.
But any equivalence class of ≡a lpha is a boolean combination of languages of the form
A∗ (see Lecture 02a). Thus we have the following theorem:

Theorem 2 Let Σ be any alphabet. For any language L over Σ the following are equivalent:

1. L is a boolean combination of languages of the form A∗ , A ⊆ Σ.

2. L is recognised by a commutative, idempotent monoid.

3. L is definable in FO1 (<).

The equivalence of the first two is proved above and the equivalence of the last two can
be found in Lecture 02a. As a consequence of these equivalences, checking if a given regular
language is definable in FO1 (<), reduces to computing its syntactic monoid and checking
that it is commutative and idempotent.

References
[1] V. Diekart, P. Gastin and M. Kufleitner: A Survey on Small Fragments of First-Order
Logic over Finite Words, Int. J. Found. Comput. Sci., Vol 19, 2008.

You might also like