The Logic of Comparative Cardinality
Yifeng Ding (voidprove.com) Joint work with Matthew Harrison-Trainer and Wesley Holliday Aug. 7, 2018 @ BLAST 2018
UC Berkeley Group of Logic and the Methodology of Science Introduction • The equational theory of Boolean algebras is also the equational theory of fields of sets, if we only care about Boolean operations. • But more information can be extracted from a field of sets. • We compare their sizes.
A field of sets
Definition A field of sets (X , F) is a pair where
1. X is a set and F ⊆ ℘(X ); 2. F is closed under intersection and complementation.
1 • But more information can be extracted from a field of sets. • We compare their sizes.
A field of sets
Definition A field of sets (X , F) is a pair where
1. X is a set and F ⊆ ℘(X ); 2. F is closed under intersection and complementation.
• The equational theory of Boolean algebras is also the equational theory of fields of sets, if we only care about Boolean operations.
1 • We compare their sizes.
A field of sets
Definition A field of sets (X , F) is a pair where
1. X is a set and F ⊆ ℘(X ); 2. F is closed under intersection and complementation.
• The equational theory of Boolean algebras is also the equational theory of fields of sets, if we only care about Boolean operations. • But more information can be extracted from a field of sets.
1 A field of sets
Definition A field of sets (X , F) is a pair where
1. X is a set and F ⊆ ℘(X ); 2. F is closed under intersection and complementation.
• The equational theory of Boolean algebras is also the equational theory of fields of sets, if we only care about Boolean operations. • But more information can be extracted from a field of sets. • We compare their sizes.
1 • A field of sets model is hX , F, V i where V :Φ → F. • Terms are evaluated by Vb on F in the obvious way. •| s| ≥ |t|: set s is at least as large as set t: there is an injection from Vb(t) to Vb(s).
Comparing the sizes
Definition Given a countably infinite set Φ of set labels, the language L is generated by the following grammar:
t ::= a | tc | (t ∩ t) ϕ ::= |t| ≥ |t| | ¬ϕ | (ϕ ∧ ϕ),
2 • Terms are evaluated by Vb on F in the obvious way. •| s| ≥ |t|: set s is at least as large as set t: there is an injection from Vb(t) to Vb(s).
Comparing the sizes
Definition Given a countably infinite set Φ of set labels, the language L is generated by the following grammar:
t ::= a | tc | (t ∩ t) ϕ ::= |t| ≥ |t| | ¬ϕ | (ϕ ∧ ϕ),
• A field of sets model is hX , F, V i where V :Φ → F.
2 •| s| ≥ |t|: set s is at least as large as set t: there is an injection from Vb(t) to Vb(s).
Comparing the sizes
Definition Given a countably infinite set Φ of set labels, the language L is generated by the following grammar:
t ::= a | tc | (t ∩ t) ϕ ::= |t| ≥ |t| | ¬ϕ | (ϕ ∧ ϕ),
• A field of sets model is hX , F, V i where V :Φ → F. • Terms are evaluated by Vb on F in the obvious way.
2 Comparing the sizes
Definition Given a countably infinite set Φ of set labels, the language L is generated by the following grammar:
t ::= a | tc | (t ∩ t) ϕ ::= |t| ≥ |t| | ¬ϕ | (ϕ ∧ ϕ),
• A field of sets model is hX , F, V i where V :Φ → F. • Terms are evaluated by Vb on F in the obvious way. •| s| ≥ |t|: set s is at least as large as set t: there is an injection from Vb(t) to Vb(s).
2 s t
• • For finite sets s, t, |s| ≥ |t| ↔ |s ∩ tc | ≥ |t ∩ sc |. • For infinite sets s, t, u •| s| ≥ |t| → |s ∩ tc | ≥ |t ∩ sc | is not valid; • (|s| ≥ |t| ∧ |s| ≥ |u|) → |s| ≥ |t ∪ u| is valid.
Finite sets and infinite sets
• Finite sets and infinite sets obey very different laws.
3 Finite sets and infinite sets
• Finite sets and infinite sets obey very different laws.
s t
• • For finite sets s, t, |s| ≥ |t| ↔ |s ∩ tc | ≥ |t ∩ sc |. • For infinite sets s, t, u •| s| ≥ |t| → |s ∩ tc | ≥ |t ∩ sc | is not valid; • (|s| ≥ |t| ∧ |s| ≥ |u|) → |s| ≥ |t ∪ u| is valid.
3 • The sentences in L valid on infinite sets have been axiomatized, with size interpreted as likelihood or possibilities. • We want to combine them: with no extra constraint on (X , F), what is the logic?
More background
• The sentences in L valid on finite sets have been axiomatized, with size interpreted as probability, credence, etc..
4 • We want to combine them: with no extra constraint on (X , F), what is the logic?
More background
• The sentences in L valid on finite sets have been axiomatized, with size interpreted as probability, credence, etc.. • The sentences in L valid on infinite sets have been axiomatized, with size interpreted as likelihood or possibilities.
4 More background
• The sentences in L valid on finite sets have been axiomatized, with size interpreted as probability, credence, etc.. • The sentences in L valid on infinite sets have been axiomatized, with size interpreted as likelihood or possibilities. • We want to combine them: with no extra constraint on (X , F), what is the logic?
4 Outline
Introduction
Laws common to finite and infinite sets
Logic with predicates for finite and infinite sets
Eliminating extra predicates
Further questions
5 Laws common to finite and infinite sets • if t = 0 is provable in the equational theory of Boolean algebras, then |∅| ≥ |t| is a theorem.
•| s| ≥ |t| ∨ |t| ≥ |s|;(|s| ≥ |t| ∧ |t| ≥ |u|) → |s| ≥ |u|; c •| ∅| ≥ |s ∩ t | → |t| ≥ |s|;
c •¬| ∅| ≥ |∅ |; • (|∅| ≥ |s| ∧ |∅| ≥ |t|) → |∅| ≥ |s ∪ t|;
BasicCompLogic
Definition (BasicCompLogic) Boolean reasoning on the sentence level. Boolean Reasoning on the set level.
≥ is a total preorder extending ⊇.
≥ works well with ∅.
6 • if t = 0 is provable in the equational theory of Boolean algebras, then |∅| ≥ |t| is a theorem.
•| s| ≥ |t| ∨ |t| ≥ |s|;(|s| ≥ |t| ∧ |t| ≥ |u|) → |s| ≥ |u|; c •| ∅| ≥ |s ∩ t | → |t| ≥ |s|;
c •¬| ∅| ≥ |∅ |; • (|∅| ≥ |s| ∧ |∅| ≥ |t|) → |∅| ≥ |s ∪ t|;
BasicCompLogic
Definition (BasicCompLogic) Boolean reasoning on the sentence level. Boolean Reasoning on the set level.
≥ is a total preorder extending ⊇.
≥ works well with ∅.
6 •| s| ≥ |t| ∨ |t| ≥ |s|;(|s| ≥ |t| ∧ |t| ≥ |u|) → |s| ≥ |u|; c •| ∅| ≥ |s ∩ t | → |t| ≥ |s|;
c •¬| ∅| ≥ |∅ |; • (|∅| ≥ |s| ∧ |∅| ≥ |t|) → |∅| ≥ |s ∪ t|;
BasicCompLogic
Definition (BasicCompLogic) Boolean reasoning on the sentence level. Boolean Reasoning on the set level. • if t = 0 is provable in the equational theory of Boolean algebras, then |∅| ≥ |t| is a theorem. ≥ is a total preorder extending ⊇.
≥ works well with ∅.
6 c •¬| ∅| ≥ |∅ |; • (|∅| ≥ |s| ∧ |∅| ≥ |t|) → |∅| ≥ |s ∪ t|;
BasicCompLogic
Definition (BasicCompLogic) Boolean reasoning on the sentence level. Boolean Reasoning on the set level. • if t = 0 is provable in the equational theory of Boolean algebras, then |∅| ≥ |t| is a theorem. ≥ is a total preorder extending ⊇. •| s| ≥ |t| ∨ |t| ≥ |s|;(|s| ≥ |t| ∧ |t| ≥ |u|) → |s| ≥ |u|; c •| ∅| ≥ |s ∩ t | → |t| ≥ |s|; ≥ works well with ∅.
6 BasicCompLogic
Definition (BasicCompLogic) Boolean reasoning on the sentence level. Boolean Reasoning on the set level. • if t = 0 is provable in the equational theory of Boolean algebras, then |∅| ≥ |t| is a theorem. ≥ is a total preorder extending ⊇. •| s| ≥ |t| ∨ |t| ≥ |s|;(|s| ≥ |t| ∧ |t| ≥ |u|) → |s| ≥ |u|; c •| ∅| ≥ |s ∩ t | → |t| ≥ |s|; ≥ works well with ∅. c •¬| ∅| ≥ |∅ |; • (|∅| ≥ |s| ∧ |∅| ≥ |t|) → |∅| ≥ |s ∪ t|;
6 Any formula ϕ consistent with BasicCompLogic is satisfiable in a finite comparison algebra, with interpreting | · | ≥ | · |.
ϕ ⇒ Σ ⇒ hB, , V i ⇒ h V (T (var(ϕ))) , , V i | {z } relevant terms, a finite set
6⇒ hX , F, V i
From logic to algebra
Definition A comparison algebra is a pair hB, i where B is a Boolean algebra and is a total preorder on B such that
• for all a, b ∈ B, a ≥B b implies a b,
•⊥ B 6 b for all b ∈ B \ {⊥B }.
7 6⇒ hX , F, V i
From logic to algebra
Definition A comparison algebra is a pair hB, i where B is a Boolean algebra and is a total preorder on B such that
• for all a, b ∈ B, a ≥B b implies a b,
•⊥ B 6 b for all b ∈ B \ {⊥B }.
Any formula ϕ consistent with BasicCompLogic is satisfiable in a finite comparison algebra, with interpreting | · | ≥ | · |.
ϕ ⇒ Σ ⇒ hB, , V i ⇒ h V (T (var(ϕ))) , , V i | {z } relevant terms, a finite set
7 From logic to algebra
Definition A comparison algebra is a pair hB, i where B is a Boolean algebra and is a total preorder on B such that
• for all a, b ∈ B, a ≥B b implies a b,
•⊥ B 6 b for all b ∈ B \ {⊥B }.
Any formula ϕ consistent with BasicCompLogic is satisfiable in a finite comparison algebra, with interpreting | · | ≥ | · |.
ϕ ⇒ Σ ⇒ hB, , V i ⇒ h V (T (var(ϕ))) , , V i | {z } relevant terms, a finite set
6⇒ hX , F, V i
7 • (010) and (100) should be finite. • Then all must be finite. • But |(011)| = |(101)| while |(010)| < |(100)|.
Not enough constraints
(111) : 4
(011) : 4 (101) : 4 (110) : 4
(001) : 1 (010) : 2 (100) : 3
(000) : 0
8 Not enough constraints
(111) : 4
• (010) and (100) should be (011) : 4 (101) : 4 (110) : 4 finite. • Then all must be finite. • But |(011)| = |(101)| while (001) : 1 (010) : 2 (100) : 3 |(010)| < |(100)|.
(000) : 0
8 • But the ordering in this B might not be based on any cardinality comparison. • We need to know when the ordering arise from cardinality comparison, and add the constraints to the logic.
Message
• Any formula ϕ consistent with BasicCompLogic is satisfiable in a finite comparison algebra B, with interpreting | · | ≥ | · |.
9 • We need to know when the ordering arise from cardinality comparison, and add the constraints to the logic.
Message
• Any formula ϕ consistent with BasicCompLogic is satisfiable in a finite comparison algebra B, with interpreting | · | ≥ | · |. • But the ordering in this B might not be based on any cardinality comparison.
9 Message
• Any formula ϕ consistent with BasicCompLogic is satisfiable in a finite comparison algebra B, with interpreting | · | ≥ | · |. • But the ordering in this B might not be based on any cardinality comparison. • We need to know when the ordering arise from cardinality comparison, and add the constraints to the logic.
9 Plan
Introduction
Laws common to finite and infinite sets
A representation theorem
Logic with predicates for finite and infinite sets
Eliminating extra predicates
Further questions
10 Definition A comparison algebra hB, i is represented by a measure algebra hB, µi if for all a, b ∈ B, we have a b iff µ(a) ≥ µ(b).
Definitions
Definition
A measure algebra is a pair hB, µi, where B is a Boolean algebra and µ is a function assigning a cardinal to each element of B such that
• if a ∧ b = ⊥, then µ(a ∨ b) = µ(a) + µ(b), and • µ(b) = 0 iff b = ⊥.
11 Definitions
Definition
A measure algebra is a pair hB, µi, where B is a Boolean algebra and µ is a function assigning a cardinal to each element of B such that
• if a ∧ b = ⊥, then µ(a ∨ b) = µ(a) + µ(b), and • µ(b) = 0 iff b = ⊥.
Definition A comparison algebra hB, i is represented by a measure algebra hB, µi if for all a, b ∈ B, we have a b iff µ(a) ≥ µ(b).
11 We call this condition “finite cancellation”
A representation theorem for finite sets
Theorem (Kraft, Pratt, Seidenberg) For any finite comparison algebra hB, i, it is represented by a measure algebra hB, µi where Range(µ) = ω if and only if:
• for any two sequences of elements a1, a2,..., an and b1, b2,..., bn from B, if every atom of B is below (in the order of the Boolean algebra) exactly as many a’s as b’s, and
if ai bi for all i ∈ {1,..., n − 1}, then bn an.
12 A representation theorem for finite sets
Theorem (Kraft, Pratt, Seidenberg) For any finite comparison algebra hB, i, it is represented by a measure algebra hB, µi where Range(µ) = ω if and only if:
• for any two sequences of elements a1, a2,..., an and b1, b2,..., bn from B, if every atom of B is below (in the order of the Boolean algebra) exactly as many a’s as b’s, and
if ai bi for all i ∈ {1,..., n − 1}, then bn an.
We call this condition “finite cancellation”
12 • Then |a0| + |a1| + |a2| = |b0| + |b1| + |b2|.
• Then if |a0| ≥ |b0| and |a1| ≥ |b1|, we can’t have |a2| > |b2|, which means |b2| ≥ |a2|.
Finite cancellation illustrated
a0 + a1 + a2 = b0 + b1 + b2 1 1 0 2 0 1 1 0 1 0 1 1 0 0 + + = = + + 1 1 1 3 1 1 1 0 1 1 2 1 1 0
13 • Then if |a0| ≥ |b0| and |a1| ≥ |b1|, we can’t have |a2| > |b2|, which means |b2| ≥ |a2|.
Finite cancellation illustrated
a0 + a1 + a2 = b0 + b1 + b2 1 1 0 2 0 1 1 0 1 0 1 1 0 0 + + = = + + 1 1 1 3 1 1 1 0 1 1 2 1 1 0
• Then |a0| + |a1| + |a2| = |b0| + |b1| + |b2|.
13 Finite cancellation illustrated
a0 + a1 + a2 = b0 + b1 + b2 1 1 0 2 0 1 1 0 1 0 1 1 0 0 + + = = + + 1 1 1 3 1 1 1 0 1 1 2 1 1 0
• Then |a0| + |a1| + |a2| = |b0| + |b1| + |b2|.
• Then if |a0| ≥ |b0| and |a1| ≥ |b1|, we can’t have |a2| > |b2|, which means |b2| ≥ |a2|.
13 Suppose there is an F ⊆ B such that the following conditions are satisfied:
1. F is an ideal; 2. elements in F satisfy the finite cancellation condition; 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c; 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
A representation theorem
Let B = hB, i be a finite comparison algebra.
14 1. F is an ideal; 2. elements in F satisfy the finite cancellation condition; 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c; 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
A representation theorem
Let B = hB, i be a finite comparison algebra. Suppose there is an F ⊆ B such that the following conditions are satisfied:
14 2. elements in F satisfy the finite cancellation condition; 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c; 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
A representation theorem
Let B = hB, i be a finite comparison algebra. Suppose there is an F ⊆ B such that the following conditions are satisfied:
1. F is an ideal;
14 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c; 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
A representation theorem
Let B = hB, i be a finite comparison algebra. Suppose there is an F ⊆ B such that the following conditions are satisfied:
1. F is an ideal; 2. elements in F satisfy the finite cancellation condition;
14 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
A representation theorem
Let B = hB, i be a finite comparison algebra. Suppose there is an F ⊆ B such that the following conditions are satisfied:
1. F is an ideal; 2. elements in F satisfy the finite cancellation condition; 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c;
14 Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
A representation theorem
Let B = hB, i be a finite comparison algebra. Suppose there is an F ⊆ B such that the following conditions are satisfied:
1. F is an ideal; 2. elements in F satisfy the finite cancellation condition; 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c; 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
14 A representation theorem
Let B = hB, i be a finite comparison algebra. Suppose there is an F ⊆ B such that the following conditions are satisfied:
1. F is an ideal; 2. elements in F satisfy the finite cancellation condition; 3. for any a, b, c ∈ B such that a 6∈ F , if a b and a c then
a b ∨B c; 4. for any a, b ∈ B, if a ∈ F and b 6∈ F , then b a and not a b.
Then B is represented by a finite measure algebra m(B) = hB, µi such that a ∈ F iff µ(a) is finite.
14 Plan
Introduction
Laws common to finite and infinite sets
A representation theorem
Logic with predicates for finite and infinite sets
Eliminating extra predicates
Further questions
15 The logic with extra predicates (With FC)
Add the following to BasicCompLogic, CardCompLogicFin,Inf is sound and complete w.r.t. measure algebras and fields of sets.
1. Fin(s) ⊕ Inf(s); 2.( Fin(s) ∧ Fin(t)) → Fin(s ∪ t); (Fin(t) ∧ s ⊆ t) → Fin(s); 3.( Fin(s) ∧ Inf(t)) → |t| > |s|; 4. Inf(s) → ((|s| ≥ |t| ∧ |s| ≥ |u|) → |s| ≥ |t ∪ u|); Vn 5. i=1(Fin(si ) ∧ Fin(ti )) → FC(s1, ··· , sn, t1, ··· , tn),
16 ϕ(0), ϕ(1), ϕ(2), ϕ(3), ϕ(4), ϕ(5), ϕ(6), ...
The logic with extra predicates (With Polarizability rule)
Add the following to BasicCompLogic, CardCompLogicFin,Inf is sound and complete w.r.t. measure algebras and fields of sets.
1. Fin(s) ⊕ Inf(s); 2.( Fin(s) ∧ Fin(t)) → Fin(s ∪ t); (Fin(t) ∧ s ⊆ t) → Fin(s); 3.( Fin(s) ∧ Inf(t)) → |t| > |s|; 4. Inf(s) → ((|s| ≥ |t| ∧ |s| ≥ |u|) → |s| ≥ |t ∪ u|); 5.( Fin(s) ∧ Fin(t)) → (|s| ≥ |t| ↔ |s ∩ tc | ≥ |t ∩ sc |); 6. where a|t abbreviates |t ∩ a| = |t ∩ ac | for a ∈ Φ, if a|t → φ is derivable, then φ is derivable, assuming that a does not occur in t or in φ.
17 The logic with extra predicates (With Polarizability rule)
Add the following to BasicCompLogic, CardCompLogicFin,Inf is sound and complete w.r.t. measure algebras and fields of sets.
1. Fin(s) ⊕ Inf(s); 2.( Fin(s) ∧ Fin(t)) → Fin(s ∪ t); (Fin(t) ∧ s ⊆ t) → Fin(s); 3.( Fin(s) ∧ Inf(t)) → |t| > |s|; 4. Inf(s) → ((|s| ≥ |t| ∧ |s| ≥ |u|) → |s| ≥ |t ∪ u|); 5.( Fin(s) ∧ Fin(t)) → (|s| ≥ |t| ↔ |s ∩ tc | ≥ |t ∩ sc |); 6. where a|t abbreviates |t ∩ a| = |t ∩ ac | for a ∈ Φ, if a|t → φ is derivable, then φ is derivable, assuming that a does not occur in t or in φ. ϕ(0), ϕ(1), ϕ(2), ϕ(3), ϕ(4), ϕ(5), ϕ(6), ...
17 ϕ(1), ϕ(3), ϕ(5),
The logic with extra predicates (With Polarizability rule)
Add the following to BasicCompLogic, CardCompLogicFin,Inf is sound and complete w.r.t. measure algebras and fields of sets.
1. Fin(s) ⊕ Inf(s); 2.( Fin(s) ∧ Fin(t)) → Fin(s ∪ t); (Fin(t) ∧ s ⊆ t) → Fin(s); 3.( Fin(s) ∧ Inf(t)) → |t| > |s|; 4. Inf(s) → ((|s| ≥ |t| ∧ |s| ≥ |u|) → |s| ≥ |t ∪ u|); 5.( Fin(s) ∧ Fin(t)) → (|s| ≥ |t| ↔ |s ∩ tc | ≥ |t ∩ sc |); 6. where a|t abbreviates |t ∩ a| = |t ∩ ac | for a ∈ Φ, if a|t → φ is derivable, then φ is derivable, assuming that a does not occur in t or in φ. ϕ(0), ϕ(2), ϕ(4), ϕ(6), ...
17 • Axioms (rules) for finite sets. • Axioms for infinite sets. • Some simple interaction between finite and infinite sets.
But that’s assuming that we can distinguish finite and infinite sets, which uses two extra predicates.
Summary
In sum, the a complete logic can be made from the following:
• Basic comparison rules.
18 • Axioms for infinite sets. • Some simple interaction between finite and infinite sets.
But that’s assuming that we can distinguish finite and infinite sets, which uses two extra predicates.
Summary
In sum, the a complete logic can be made from the following:
• Basic comparison rules. • Axioms (rules) for finite sets.
18 • Some simple interaction between finite and infinite sets.
But that’s assuming that we can distinguish finite and infinite sets, which uses two extra predicates.
Summary
In sum, the a complete logic can be made from the following:
• Basic comparison rules. • Axioms (rules) for finite sets. • Axioms for infinite sets.
18 But that’s assuming that we can distinguish finite and infinite sets, which uses two extra predicates.
Summary
In sum, the a complete logic can be made from the following:
• Basic comparison rules. • Axioms (rules) for finite sets. • Axioms for infinite sets. • Some simple interaction between finite and infinite sets.
18 Summary
In sum, the a complete logic can be made from the following:
• Basic comparison rules. • Axioms (rules) for finite sets. • Axioms for infinite sets. • Some simple interaction between finite and infinite sets.
But that’s assuming that we can distinguish finite and infinite sets, which uses two extra predicates.
18 Plan
Introduction
Laws common to finite and infinite sets
A representation theorem
Logic with predicates for finite and infinite sets
Eliminating extra predicates
Further questions
19 No. There are models that satisfy exactly the same formulas in L, but one has only infinite sets and the other has a finite set.
Defining finiteness and infiniteness
Can we define Fin and Inf in the language of pure cardinality comparison?
20 Defining finiteness and infiniteness
Can we define Fin and Inf in the language of pure cardinality comparison?
No. There are models that satisfy exactly the same formulas in L, but one has only infinite sets and the other has a finite set.
20 Undefinability
|(111)| = ℵ3
|(011)| = ℵ2 |(101)| = ℵ3 |(110)| = ℵ3
|(001)| = ℵ1 |(010)| = ℵ2 |(100)| = ℵ3
|(000)| = 0
21 Undefinability
|(111)| = ℵ3
|(011)| = ℵ2 |(101)| = ℵ3 |(110)| = ℵ3
|(001)| = 1 |(010)| = ℵ2 |(100)| = ℵ3
|(000)| = 0
22 For any flexible measure algebra hB, µi and any cardinal κ, there exists a flexible measure algebra hB, µ0i such that
• µ(the smallest atom) = κ; • For any a, b ∈ B, µ(a) ≥ µ(b) iff µ0(a) ≥ µ0(b); 0 •h B, µ, V i ≡L hB, µ , V i for any valuation V .
Flexible algebras
We call a finite measure algebra hB, µi flexible when:
• There is a strictly smallest atom. • Any element (except the bottom element) is equally large to the largest atom below it. • Equivalently, there is at most one finite but non-empty element (which must be the strictly smallest atom).
23 Flexible algebras
We call a finite measure algebra hB, µi flexible when:
• There is a strictly smallest atom. • Any element (except the bottom element) is equally large to the largest atom below it. • Equivalently, there is at most one finite but non-empty element (which must be the strictly smallest atom).
For any flexible measure algebra hB, µi and any cardinal κ, there exists a flexible measure algebra hB, µ0i such that
• µ(the smallest atom) = κ; • For any a, b ∈ B, µ(a) ≥ µ(b) iff µ0(a) ≥ µ0(b); 0 •h B, µ, V i ≡L hB, µ , V i for any valuation V .
23 Perhaps flexible models are the only models where finiteness can’t be defined?
Defining Fin and Inf
We must do our best.
24 Defining Fin and Inf
We must do our best. Perhaps flexible models are the only models where finiteness can’t be defined?
24 Defining Fin and Inf
u is the
When ∆ ⊆ Φ is finite, define Fin∆(u) forunion any of setri ’s term u ∈ T (∆) as: |R| u = S r i=1 i _ V|R| |si ∪ ti | > |si | ≥ |ti | R⊆T0(∆) i=1 |R| |si ∪ ti | ≥ |ri | S,T ∈T0(∆)
Here ri ranges over elements in R, and si , ti range over the si , ti and also elementsThere exists in sequences S and T , respectively. ri , si , ti ’s ri ’s are finite
25 Defining Fin and Inf
u is the
When ∆ ⊆ Φ is finite, define Fin∆(u) forunion any of setri ’s term u ∈ T (∆) as: |R| u = S r i=1 i _ V|R| |si ∪ ti | > |si | ≥ |ti | R⊆T0(∆) i=1 |R| |si ∪ ti | ≥ |ri | S,T ∈T0(∆)
Here ri ranges over elements in R, and si , ti range over the si , ti and also elementsThere exists in sequences S and T , respectively. ri , si , ti ’s ri ’s are finite
25 Defining Fin and Inf
u is the
When ∆ ⊆ Φ is finite, define Fin∆(u) forunion any of setri ’s term u ∈ T (∆) as: |R| u = S r i=1 i _ V|R| |si ∪ ti | > |si | ≥ |ti | R⊆T0(∆) i=1 |R| |si ∪ ti | ≥ |ri | S,T ∈T0(∆)
Here ri ranges over elements in R, and si , ti range over the si , ti and also elementsThere exists in sequences S and T , respectively. ri , si , ti ’s ri ’s are finite
25 Defining Fin and Inf
u is the
When ∆ ⊆ Φ is finite, define Fin∆(u) forunion any of setri ’s term u ∈ T (∆) as: |R| u = S r i=1 i _ V|R| |si ∪ ti | > |si | ≥ |ti | R⊆T0(∆) i=1 |R| |si ∪ ti | ≥ |ri | S,T ∈T0(∆)
Here ri ranges over elements in R, and si , ti range over the si , ti and also elementsThere exists in sequences S and T , respectively. ri , si , ti ’s ri ’s are finite
25 Defining Fin and Inf
When ∆ ⊆ Φ is finite, define Inf∆(u) := for any set term u ∈ T (∆) as: _ (t 6⊆ s ∧ |u| ≥ |s| ≥ |s ∪ t|)
s,t∈T0(∆)
26 Definition works
For any measure algebra model hB, µ, V i such that every element is named by a term in T (∆), namely V (T (∆)) = B:
• If Fin∆(u) is true, then µ(Vb(u)) is finite.
• If Inf∆(u) is true, then µ(Vb(u)) is infinite.
• Fin∆(u) and Inf∆(u) can’t be both true. • If they are both false, then hB, µi is flexible, and Vb(u) is the smallest atom. • (s ⊆ t ∧ Fin(t)) → Fin(s) and (Fin(s) ∧ Fin(t)) → Fin(s ∪ t) are derivable in BasicCompLogic.
27 The axioms
Definition Where ∆ ⊆ Φ is finite, define Axiom(∆) as the set containing all
of the following formulas for all u, s, t ∈ T0(∆):
1. ¬(Fin∆(u) ∧ Inf∆(u));
2.( ¬Fin∆(u) ∧ ¬Inf∆(u)) → V (|u| ≥ |t| → (t = ∨ t = u)); t∈T0(∆) ∅ c c 3. (Fin∆(s) ∧ Fin∆(t)) → (|s| ≥ |t| ↔ |s ∩ t | ≥ |t ∩ s |);
4. Inf∆(u) → ((|u| ≥ |s| ∧ |u| ≥ |t|) → |u| ≥ |s ∪ t|);
5. (Inf∆(s) ∧ Fin∆(t)) → |s| > |t|.
28 The logic
Definition
Let CardCompLogic be the logic for L with the following axioms and rules:
1. all axioms and rules in BasicCompLogic; 2. for any finite ∆ ⊆ Φ, all formulas in Axioms(∆); 3. the polarizability rule (A7).
29 Proof sketch
Pick a ϕ consistent with CardCompLogic, take ∆ = var(ϕ):
1. Extend it to Σ maximally consistent in CardCompLogic. 2. Σ is also maximally consistent with BasicCompLogic. Get canonical comparison model C. 3. Restrict C to terms in T (∆), get B. 4. B |= Axiom(∆) and also Fin(~s) → FC(~s). Use the terms with Fin as finite elements. Apply representation theorem and get
measure algebra model M ≡L B. 5. M |= ϕ. So ϕ is satisfiable.
30 Conclusion
The logic of cardinal comparison on arbitrary fields of sets can be axiomatized by putting together
• a basic system for orderings extending the inclusion ordering; • a working definition for finiteness and infiniteness based on witnesses; • characteristic axioms and rules for finite and infinite sets.
The axiomatization is weak; the logic is non-compact. We use finite Boolean algebras in an essential way.
31 Plan
Introduction
Laws common to finite and infinite sets
A representation theorem
Logic with predicates for finite and infinite sets
Eliminating extra predicates
Further questions
32 • A field of sets hX , Fi (F possibly infinite) naturally give rise to a measure algebra hB, µi. The B part can be arbitrary due to Stone duality. But what about µ? • Same question for hB, i. What conditions must satisfy?
Our logic is not strongly complete, as it is finitary but not compact. What is the strongly complete logic?
Further questions
Representation theorems in the infinite:
33 • Same question for hB, i. What conditions must satisfy?
Our logic is not strongly complete, as it is finitary but not compact. What is the strongly complete logic?
Further questions
Representation theorems in the infinite:
• A field of sets hX , Fi (F possibly infinite) naturally give rise to a measure algebra hB, µi. The B part can be arbitrary due to Stone duality. But what about µ?
33 Our logic is not strongly complete, as it is finitary but not compact. What is the strongly complete logic?
Further questions
Representation theorems in the infinite:
• A field of sets hX , Fi (F possibly infinite) naturally give rise to a measure algebra hB, µi. The B part can be arbitrary due to Stone duality. But what about µ? • Same question for hB, i. What conditions must satisfy?
33 Further questions
Representation theorems in the infinite:
• A field of sets hX , Fi (F possibly infinite) naturally give rise to a measure algebra hB, µi. The B part can be arbitrary due to Stone duality. But what about µ? • Same question for hB, i. What conditions must satisfy?
Our logic is not strongly complete, as it is finitary but not compact. What is the strongly complete logic?
33 Thank You.
34 Non-compactness
The problem of finiteness:
{|sn| < |sn+1| | n ∈ ω} ∪ {|sn| ≤ |t| | n ∈ ω} ∪ {Fin(t)}.
The problem of well-foundedness:
{|sn+1| < |sn| | n ∈ ω}.
The problem of discreteness:
Disjoint{ti , si | i ∈ ω}∪
{|ti | = |tj |, |si | = |sj | | i, j ∈ ω}∪ m m √ {| ∪ t | < | ∪ s | < | ∪ t | | ( 1 , 2 ) →lim 2}. i 35 Representation in the infinite Can we have a reasonably nice list of conditions for an arbitrary comparison algebra (finite or infinite) to be represented by a cardinal-valued measure function? 36 Definition of FC Definition n For each sequence of n terms ~s = hs0, ··· , sn−1i and f ∈ 2, define \ \ c ~s[f ] = {si | f (i) = 1} ∩ {si | f (i) = 0}, [ −1 Nm(~s) = {~s[f ] | f : n → 2 and |f (1)| = m}. Given two sequences ~s and ~t of n terms, define ^ ~s E~t = (Ni (~s) = Ni (~t)), 0≤i≤n ^ FC(~s,~t) = ~s E~t → (( |si | ≥ |ti |) → |tn−1| ≥ |sn−1|). i 37 Polarization With polarization, we can almost do set addition: 38 Polarization With polarization, we can almost do set addition: 39 Polarization With polarization, we can almost do set addition: 40