from lawvere to brandenburger keisler interactive forms
play

From Lawvere to Brandenburger-Keisler: interactive forms of - PowerPoint PPT Presentation

From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference Samson Abramsky and Jonathan Zvesper Department of Computer Science, University of Oxford Samson Abramsky and Jonathan Zvesper (Department of Computer


  1. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  2. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  3. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  4. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Symmetrically, elements of U b represent possible epistemic states of Bob. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  5. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Symmetrically, elements of U b represent possible epistemic states of Bob. The relations R a ⊆ U a × U b , R b ⊆ U b × U a specify these beliefs. Thus R a ( x , y ) expresses that in state x , Alice believes that state y is possible for Bob. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  6. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Symmetrically, elements of U b represent possible epistemic states of Bob. The relations R a ⊆ U a × U b , R b ⊆ U b × U a specify these beliefs. Thus R a ( x , y ) expresses that in state x , Alice believes that state y is possible for Bob. We say that a state x ∈ U a believes P ⊆ U b if R a ( x ) ⊆ P . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  7. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Symmetrically, elements of U b represent possible epistemic states of Bob. The relations R a ⊆ U a × U b , R b ⊆ U b × U a specify these beliefs. Thus R a ( x , y ) expresses that in state x , Alice believes that state y is possible for Bob. We say that a state x ∈ U a believes P ⊆ U b if R a ( x ) ⊆ P . Modally, ‘ x believes P ’ is just x | = ✷ a P . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  8. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Symmetrically, elements of U b represent possible epistemic states of Bob. The relations R a ⊆ U a × U b , R b ⊆ U b × U a specify these beliefs. Thus R a ( x , y ) expresses that in state x , Alice believes that state y is possible for Bob. We say that a state x ∈ U a believes P ⊆ U b if R a ( x ) ⊆ P . Modally, ‘ x believes P ’ is just x | = ✷ a P . We say that x assumes P if R a ( x ) = P . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  9. Setting for the BK argument The ‘real’ game theory applications involve probabilities; players’ beliefs are represented as various forms of probability measures (conditional, lexicographic etc.). To expose the essential structure of their core argument, Brandenburger and Keisler present it in a simplified, relational setting. Type spaces U a and U b for Alice and Bob: Elements of U a represent possible epistemic states of Alice in which she holds beliefs about Bob, Bob’s beliefs, etc. Symmetrically, elements of U b represent possible epistemic states of Bob. The relations R a ⊆ U a × U b , R b ⊆ U b × U a specify these beliefs. Thus R a ( x , y ) expresses that in state x , Alice believes that state y is possible for Bob. We say that a state x ∈ U a believes P ⊆ U b if R a ( x ) ⊆ P . Modally, ‘ x believes P ’ is just x | = ✷ a P . We say that x assumes P if R a ( x ) = P . This is x | = ⊞ a P , where ⊞ a is the modality defined by x | = ⊞ a φ ≡ ∀ y . R a ( x , y ) ⇔ y | = φ. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 4 / 30

  10. Assumption Completeness Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 5 / 30

  11. Assumption Completeness A structure ( U a , U b , R a , R b ) is assumption-complete with respect to a collection of predicates on U a and U b if for every predicate P on U b in the collection, there is a state x on U a such that x assumes P ; and similarly for the predicates on U a . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 5 / 30

  12. Assumption Completeness A structure ( U a , U b , R a , R b ) is assumption-complete with respect to a collection of predicates on U a and U b if for every predicate P on U b in the collection, there is a state x on U a such that x assumes P ; and similarly for the predicates on U a . The hypothesis of assumption completeness is needed to show the soundness of various solution concepts in games. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 5 / 30

  13. Assumption Completeness A structure ( U a , U b , R a , R b ) is assumption-complete with respect to a collection of predicates on U a and U b if for every predicate P on U b in the collection, there is a state x on U a such that x assumes P ; and similarly for the predicates on U a . The hypothesis of assumption completeness is needed to show the soundness of various solution concepts in games. Brandenburger and Keisler show that this hypothesis, in the case where the predicates include those definable in the first-order language of this structure, leads to a contradiction. (They also show the existence of assumption complete models for some other cases.) Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 5 / 30

  14. Assumption Completeness A structure ( U a , U b , R a , R b ) is assumption-complete with respect to a collection of predicates on U a and U b if for every predicate P on U b in the collection, there is a state x on U a such that x assumes P ; and similarly for the predicates on U a . The hypothesis of assumption completeness is needed to show the soundness of various solution concepts in games. Brandenburger and Keisler show that this hypothesis, in the case where the predicates include those definable in the first-order language of this structure, leads to a contradiction. (They also show the existence of assumption complete models for some other cases.) Our aim is to understand the general structures underlying this argument. Our first step is to recast their result as a positive one — a fixpoint lemma. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 5 / 30

  15. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  16. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . We assume that for ‘all’ (in some ‘definable’ class of) predicates p on U a there is x 0 such that: Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  17. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . We assume that for ‘all’ (in some ‘definable’ class of) predicates p on U a there is x 0 such that: (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  18. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . We assume that for ‘all’ (in some ‘definable’ class of) predicates p on U a there is x 0 such that: (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} (2) ∃ y . R a ( x 0 , y ). Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  19. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . We assume that for ‘all’ (in some ‘definable’ class of) predicates p on U a there is x 0 such that: (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} (2) ∃ y . R a ( x 0 , y ). Modally: x 0 | = ✷ a ⊞ b p & ✸ a ⊤ . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  20. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . We assume that for ‘all’ (in some ‘definable’ class of) predicates p on U a there is x 0 such that: (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} (2) ∃ y . R a ( x 0 , y ). Modally: x 0 | = ✷ a ⊞ b p & ✸ a ⊤ . Remark We can read (1) as saying: ‘ x 0 believes that ( y assumes that p )’, in the terminology of Brandenburger and Keisler. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  21. The Basic Lemma A 2-universe is a structure ( U a , U b , R a , R b ) where R a ⊆ U a × U b , R b ⊆ U b × U a . We assume that for ‘all’ (in some ‘definable’ class of) predicates p on U a there is x 0 such that: (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} (2) ∃ y . R a ( x 0 , y ). Modally: x 0 | = ✷ a ⊞ b p & ✸ a ⊤ . Remark We can read (1) as saying: ‘ x 0 believes that ( y assumes that p )’, in the terminology of Brandenburger and Keisler. Lemma (Basic Lemma) From (1) and (2) we have: p ( x 0 ) ⇐ ⇒ ∃ y . [ R a ( x 0 , y ) ∧ R b ( y , x 0 )] . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 6 / 30

  22. The BK Fixpoint Lemma Lemma (BK Fixpoint Lemma) Under our assumptions, every unary propositional operator O has a fixpoint. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 7 / 30

  23. The BK Fixpoint Lemma Lemma (BK Fixpoint Lemma) Under our assumptions, every unary propositional operator O has a fixpoint. Proof Since p was arbitrary, we can define q ( x ) ≡ ∃ y . [ R a ( x , y ) ∧ R b ( y , x )] p ( x ) ≡ O ( q ( x )) . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 7 / 30

  24. The BK Fixpoint Lemma Lemma (BK Fixpoint Lemma) Under our assumptions, every unary propositional operator O has a fixpoint. Proof Since p was arbitrary, we can define q ( x ) ≡ ∃ y . [ R a ( x , y ) ∧ R b ( y , x )] p ( x ) ≡ O ( q ( x )) . N.B. It is important that p is defined without reference to x 0 to avoid circularity. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 7 / 30

  25. The BK Fixpoint Lemma Lemma (BK Fixpoint Lemma) Under our assumptions, every unary propositional operator O has a fixpoint. Proof Since p was arbitrary, we can define q ( x ) ≡ ∃ y . [ R a ( x , y ) ∧ R b ( y , x )] p ( x ) ≡ O ( q ( x )) . N.B. It is important that p is defined without reference to x 0 to avoid circularity. By the Basic Lemma, this yields q ( x 0 ) ⇐ ⇒ O ( q ( x 0 )) . � Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 7 / 30

  26. The BK Fixpoint Lemma Lemma (BK Fixpoint Lemma) Under our assumptions, every unary propositional operator O has a fixpoint. Proof Since p was arbitrary, we can define q ( x ) ≡ ∃ y . [ R a ( x , y ) ∧ R b ( y , x )] p ( x ) ≡ O ( q ( x )) . N.B. It is important that p is defined without reference to x 0 to avoid circularity. By the Basic Lemma, this yields q ( x 0 ) ⇐ ⇒ O ( q ( x 0 )) . � Taking O ≡ ¬ yields the BK ‘paradox’. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 7 / 30

  27. The BK Fixpoint Lemma Lemma (BK Fixpoint Lemma) Under our assumptions, every unary propositional operator O has a fixpoint. Proof Since p was arbitrary, we can define q ( x ) ≡ ∃ y . [ R a ( x , y ) ∧ R b ( y , x )] p ( x ) ≡ O ( q ( x )) . N.B. It is important that p is defined without reference to x 0 to avoid circularity. By the Basic Lemma, this yields q ( x 0 ) ⇐ ⇒ O ( q ( x 0 )) . � Taking O ≡ ¬ yields the BK ‘paradox’. (In fact ¬ q ( x ) is equivalent to their ‘diagonal formula’ D ). Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 7 / 30

  28. Some questions Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 8 / 30

  29. Some questions How can this be related to standard fixpoint notions. In particular, we aim to relate it to Lawvere’s categorical formulation of diagonal arguments. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 8 / 30

  30. Some questions How can this be related to standard fixpoint notions. In particular, we aim to relate it to Lawvere’s categorical formulation of diagonal arguments. Where does this particular form “believes . . . assumes . . . ” come from? Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 8 / 30

  31. Some questions How can this be related to standard fixpoint notions. In particular, we aim to relate it to Lawvere’s categorical formulation of diagonal arguments. Where does this particular form “believes . . . assumes . . . ” come from? How do these ideas generalize? Is there some general idea of many-person versions of classical one-person notions? Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 8 / 30

  32. Some questions How can this be related to standard fixpoint notions. In particular, we aim to relate it to Lawvere’s categorical formulation of diagonal arguments. Where does this particular form “believes . . . assumes . . . ” come from? How do these ideas generalize? Is there some general idea of many-person versions of classical one-person notions? Under what circumstances can “sufficiently complete type spaces” be constructed? Coalgebra can be used here! Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 8 / 30

  33. Lawvere fixpoint lemma: concrete formulation Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  34. Lawvere fixpoint lemma: concrete formulation We start off concretely working in Set . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  35. Lawvere fixpoint lemma: concrete formulation We start off concretely working in Set . Basic situation: a function g : X → V X or equivalently, by cartesian closure: g : X × X → V ˆ Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  36. Lawvere fixpoint lemma: concrete formulation We start off concretely working in Set . Basic situation: a function g : X → V X or equivalently, by cartesian closure: g : X × X → V ˆ Think of V as a set of ‘truth values’: V X is the set of ‘ V -valued predicates’. Then g is showing how predicates on X can be represented by elements of X . In terms of ˆ g : a predicate p : X → V is representable by x ∈ X if for all y ∈ X : p ( y ) = ˆ g ( x , y ) Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  37. Lawvere fixpoint lemma: concrete formulation We start off concretely working in Set . Basic situation: a function g : X → V X or equivalently, by cartesian closure: g : X × X → V ˆ Think of V as a set of ‘truth values’: V X is the set of ‘ V -valued predicates’. Then g is showing how predicates on X can be represented by elements of X . In terms of ˆ g : a predicate p : X → V is representable by x ∈ X if for all y ∈ X : p ( y ) = ˆ g ( x , y ) If predicates ‘talk about’ X , then representable predicates allow X to ‘talk about itself’. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  38. Lawvere fixpoint lemma: concrete formulation We start off concretely working in Set . Basic situation: a function g : X → V X or equivalently, by cartesian closure: g : X × X → V ˆ Think of V as a set of ‘truth values’: V X is the set of ‘ V -valued predicates’. Then g is showing how predicates on X can be represented by elements of X . In terms of ˆ g : a predicate p : X → V is representable by x ∈ X if for all y ∈ X : p ( y ) = ˆ g ( x , y ) If predicates ‘talk about’ X , then representable predicates allow X to ‘talk about itself’. If g is surjective , then every predicate on X is representable in X . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  39. Lawvere fixpoint lemma: concrete formulation We start off concretely working in Set . Basic situation: a function g : X → V X or equivalently, by cartesian closure: g : X × X → V ˆ Think of V as a set of ‘truth values’: V X is the set of ‘ V -valued predicates’. Then g is showing how predicates on X can be represented by elements of X . In terms of ˆ g : a predicate p : X → V is representable by x ∈ X if for all y ∈ X : p ( y ) = ˆ g ( x , y ) If predicates ‘talk about’ X , then representable predicates allow X to ‘talk about itself’. If g is surjective , then every predicate on X is representable in X . When can this happen? Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 9 / 30

  40. The Fixpoint Lemma Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  41. The Fixpoint Lemma Proposition Suppose that g : X → V X is surjective. Then every function α : V → V has a fixpoint: v ∈ V such that α ( v ) = v. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  42. The Fixpoint Lemma Proposition Suppose that g : X → V X is surjective. Then every function α : V → V has a fixpoint: v ∈ V such that α ( v ) = v. Define a predicate p by Proof ˆ g ✲ V X × X ✻ ∆ α ❄ ✲ V X p There is x ∈ X which represents p : then p ( x ) = α (ˆ g (∆( x ))) = α (ˆ g ( x , x )) = α ( p ( x )) so p ( x ) is a fixpoint of α . � Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  43. The Fixpoint Lemma Proposition Suppose that g : X → V X is surjective. Then every function α : V → V has a fixpoint: v ∈ V such that α ( v ) = v. Define a predicate p by Proof ˆ g ✲ V X × X ✻ ∆ α ❄ ✲ V X p There is x ∈ X which represents p : then p ( x ) = α (ˆ g (∆( x ))) = α (ˆ g ( x , x )) = α ( p ( x )) so p ( x ) is a fixpoint of α . � Some comments on the proof. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  44. The Fixpoint Lemma Proposition Suppose that g : X → V X is surjective. Then every function α : V → V has a fixpoint: v ∈ V such that α ( v ) = v. Define a predicate p by Proof ˆ g ✲ V X × X ✻ ∆ α ❄ ✲ V X p There is x ∈ X which represents p : then p ( x ) = α (ˆ g (∆( x ))) = α (ˆ g ( x , x )) = α ( p ( x )) so p ( x ) is a fixpoint of α . � Some comments on the proof. (i) Constructive. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  45. The Fixpoint Lemma Proposition Suppose that g : X → V X is surjective. Then every function α : V → V has a fixpoint: v ∈ V such that α ( v ) = v. Define a predicate p by Proof ˆ g ✲ V X × X ✻ ∆ α ❄ ✲ V X p There is x ∈ X which represents p : then p ( x ) = α (ˆ g (∆( x ))) = α (ˆ g ( x , x )) = α ( p ( x )) so p ( x ) is a fixpoint of α . � Some comments on the proof. (i) Constructive. (ii) Uses two descriptions of p . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  46. The Fixpoint Lemma Proposition Suppose that g : X → V X is surjective. Then every function α : V → V has a fixpoint: v ∈ V such that α ( v ) = v. Define a predicate p by Proof ˆ g ✲ V X × X ✻ ∆ α ❄ ✲ V X p There is x ∈ X which represents p : then p ( x ) = α (ˆ g (∆( x ))) = α (ˆ g ( x , x )) = α ( p ( x )) so p ( x ) is a fixpoint of α . � Some comments on the proof. (i) Constructive. (ii) Uses two descriptions of p . (iii) Since x represents p , p ( x ) is (indirect) self-application . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 10 / 30

  47. Does this make sense? Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  48. Does this make sense? Say that X has the fixpoint property (fpp) if every endofunction on X has a fixpoint. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  49. Does this make sense? Say that X has the fixpoint property (fpp) if every endofunction on X has a fixpoint. Of course, no set with more than one element has the fixpoint property! Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  50. Does this make sense? Say that X has the fixpoint property (fpp) if every endofunction on X has a fixpoint. Of course, no set with more than one element has the fixpoint property! Basic example: 2 = { 0 , 1 } . The negation ¬ 0 = 1 , ¬ 1 = 0 does not have a fixpoint. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  51. Does this make sense? Say that X has the fixpoint property (fpp) if every endofunction on X has a fixpoint. Of course, no set with more than one element has the fixpoint property! Basic example: 2 = { 0 , 1 } . The negation ¬ 0 = 1 , ¬ 1 = 0 does not have a fixpoint. So the meaning of the theorem in Set must be taken contrapositively : Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  52. Does this make sense? Say that X has the fixpoint property (fpp) if every endofunction on X has a fixpoint. Of course, no set with more than one element has the fixpoint property! Basic example: 2 = { 0 , 1 } . The negation ¬ 0 = 1 , ¬ 1 = 0 does not have a fixpoint. So the meaning of the theorem in Set must be taken contrapositively : For all sets X , V where V has more than one element, there is no surjective map X → V X Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  53. Does this make sense? Say that X has the fixpoint property (fpp) if every endofunction on X has a fixpoint. Of course, no set with more than one element has the fixpoint property! Basic example: 2 = { 0 , 1 } . The negation ¬ 0 = 1 , ¬ 1 = 0 does not have a fixpoint. So the meaning of the theorem in Set must be taken contrapositively : For all sets X , V where V has more than one element, there is no surjective map X → V X Suitably formulated, this is valid in any elementary topos. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 11 / 30

  54. Two Applications Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 12 / 30

  55. Two Applications Cantor’s Theorem . Take V = 2 . There is no surjective map X → 2 X and hence | P ( X ) | �≤ | X | . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 12 / 30

  56. Two Applications Cantor’s Theorem . Take V = 2 . There is no surjective map X → 2 X and hence | P ( X ) | �≤ | X | . We can apply the fixpoint lemma to any putative such map, with α = ¬ , to get the usual ‘diagonalization argument’. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 12 / 30

  57. Two Applications Cantor’s Theorem . Take V = 2 . There is no surjective map X → 2 X and hence | P ( X ) | �≤ | X | . We can apply the fixpoint lemma to any putative such map, with α = ¬ , to get the usual ‘diagonalization argument’. Russell’s Paradox . Let S be a ‘universe’ (set) of sets. Let g : S × S → 2 ˆ define the membership relation: ˆ g ( x , y ) ⇔ y ∈ x Then there is a predicate which can be defined on S , and which is not representable by any element of S . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 12 / 30

  58. Two Applications Cantor’s Theorem . Take V = 2 . There is no surjective map X → 2 X and hence | P ( X ) | �≤ | X | . We can apply the fixpoint lemma to any putative such map, with α = ¬ , to get the usual ‘diagonalization argument’. Russell’s Paradox . Let S be a ‘universe’ (set) of sets. Let g : S × S → 2 ˆ define the membership relation: ˆ g ( x , y ) ⇔ y ∈ x Then there is a predicate which can be defined on S , and which is not representable by any element of S . Such a predicate is given by the standard Russell set, which arises by applying the fixpoint lemma. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 12 / 30

  59. The general case Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 13 / 30

  60. The general case Lawvere’s argument was in the setting of cartesian (closed) categories. Amazingly, it only needs finite products! Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 13 / 30

  61. The general case Lawvere’s argument was in the setting of cartesian (closed) categories. Amazingly, it only needs finite products! (In fact, even less suffices: just monoidal structure and a ‘diagonal’ satisfying only point naturality and monoidality.) Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 13 / 30

  62. The general case Lawvere’s argument was in the setting of cartesian (closed) categories. Amazingly, it only needs finite products! (In fact, even less suffices: just monoidal structure and a ‘diagonal’ satisfying only point naturality and monoidality.) Let C be a category with finite products. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 13 / 30

  63. The general case Lawvere’s argument was in the setting of cartesian (closed) categories. Amazingly, it only needs finite products! (In fact, even less suffices: just monoidal structure and a ‘diagonal’ satisfying only point naturality and monoidality.) Let C be a category with finite products. (Lawvere) An arrow f : A × A → V is weakly point surjective (wps) if for every p : A → V there is an x : 1 → A such that, for all y : 1 → A : p ◦ y = f ◦ � x , y � : 1 → V In this case, we say that p is represented by x . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 13 / 30

  64. Abstract Fixpoint Lemma Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 14 / 30

  65. Abstract Fixpoint Lemma Proposition (Abstract Fixpoint Lemma) Let C be a category with finite products. If f : A × A → V is weakly point surjective, then every endomorphism α : V → V has a fixpoint v : 1 → V such that α ◦ v = v. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 14 / 30

  66. Abstract Fixpoint Lemma Proposition (Abstract Fixpoint Lemma) Let C be a category with finite products. If f : A × A → V is weakly point surjective, then every endomorphism α : V → V has a fixpoint v : 1 → V such that α ◦ v = v. Define p : A → V by Proof f ✲ V A × A ✻ ∆ A α ❄ ✲ V A p Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 14 / 30

  67. Abstract Fixpoint Lemma Proposition (Abstract Fixpoint Lemma) Let C be a category with finite products. If f : A × A → V is weakly point surjective, then every endomorphism α : V → V has a fixpoint v : 1 → V such that α ◦ v = v. Define p : A → V by Proof f ✲ V A × A ✻ ∆ A α ❄ ✲ V A p Suppose p is represented by x : 1 → A . Then p ◦ x = α ◦ f ◦ ∆ A ◦ x def of p = α ◦ f ◦ � x , x � diagonal = α ◦ p ◦ x x represents p . so p ◦ x is a fixpoint of α . � Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 14 / 30

  68. Can we reduce BK to Lawvere? Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 15 / 30

  69. Can we reduce BK to Lawvere? There are many applications of Lawvere’s result. The very nice article by Noson Yanofsky A universal approach to self-referential paradoxes, incompleteness and fixed points, (BSL 2003) covers: semantic paradozes (Liar, Berry, Richard), the Halting Problem, existence of an oracle B such that P B � = NP B , Parikh sentences, L¨ ob’s paradox, the Recursion theorem, Rice’s theorem, von Neumann’s self-reproducing automata, . . . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 15 / 30

  70. Can we reduce BK to Lawvere? There are many applications of Lawvere’s result. The very nice article by Noson Yanofsky A universal approach to self-referential paradoxes, incompleteness and fixed points, (BSL 2003) covers: semantic paradozes (Liar, Berry, Richard), the Halting Problem, existence of an oracle B such that P B � = NP B , Parikh sentences, L¨ ob’s paradox, the Recursion theorem, Rice’s theorem, von Neumann’s self-reproducing automata, . . . However, the question of using it to prove the BK result remained open. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 15 / 30

  71. Can we reduce BK to Lawvere? There are many applications of Lawvere’s result. The very nice article by Noson Yanofsky A universal approach to self-referential paradoxes, incompleteness and fixed points, (BSL 2003) covers: semantic paradozes (Liar, Berry, Richard), the Halting Problem, existence of an oracle B such that P B � = NP B , Parikh sentences, L¨ ob’s paradox, the Recursion theorem, Rice’s theorem, von Neumann’s self-reproducing automata, . . . However, the question of using it to prove the BK result remained open. We shall present a way of doing this. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 15 / 30

  72. Can we reduce BK to Lawvere? There are many applications of Lawvere’s result. The very nice article by Noson Yanofsky A universal approach to self-referential paradoxes, incompleteness and fixed points, (BSL 2003) covers: semantic paradozes (Liar, Berry, Richard), the Halting Problem, existence of an oracle B such that P B � = NP B , Parikh sentences, L¨ ob’s paradox, the Recursion theorem, Rice’s theorem, von Neumann’s self-reproducing automata, . . . However, the question of using it to prove the BK result remained open. We shall present a way of doing this. This needs the two results to be put on a common footing — yet they look very different! Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 15 / 30

  73. Can we reduce BK to Lawvere? There are many applications of Lawvere’s result. The very nice article by Noson Yanofsky A universal approach to self-referential paradoxes, incompleteness and fixed points, (BSL 2003) covers: semantic paradozes (Liar, Berry, Richard), the Halting Problem, existence of an oracle B such that P B � = NP B , Parikh sentences, L¨ ob’s paradox, the Recursion theorem, Rice’s theorem, von Neumann’s self-reproducing automata, . . . However, the question of using it to prove the BK result remained open. We shall present a way of doing this. This needs the two results to be put on a common footing — yet they look very different! The first step is to analyze exactly what logical resources are needed to carry through the BK argument. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 15 / 30

  74. Towards a categorical version of the BK argument Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 16 / 30

  75. Towards a categorical version of the BK argument First observation: this argument is valid in regular logic , comprising sequents φ ⊢ X ψ where φ and ψ are built from atomic formulas by conjunction and existential quantification. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 16 / 30

  76. Towards a categorical version of the BK argument First observation: this argument is valid in regular logic , comprising sequents φ ⊢ X ψ where φ and ψ are built from atomic formulas by conjunction and existential quantification. The intended meaning of such a sequent is ∀ x 1 · · · ∀ x n [ φ ⇒ ψ ] where X = { x 1 , . . . , x n } . Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 16 / 30

  77. Towards a categorical version of the BK argument First observation: this argument is valid in regular logic , comprising sequents φ ⊢ X ψ where φ and ψ are built from atomic formulas by conjunction and existential quantification. The intended meaning of such a sequent is ∀ x 1 · · · ∀ x n [ φ ⇒ ψ ] where X = { x 1 , . . . , x n } . This is a common fragment of intuitionistic and classical logic. It plays a core rˆ ole in categorical logic. Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 16 / 30

  78. Formal version of the BK argument Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 17 / 30

  79. Formal version of the BK argument The assumptions given in the informal argument: Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 17 / 30

  80. Formal version of the BK argument The assumptions given in the informal argument: For each p there is x 0 such that Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 17 / 30

  81. Formal version of the BK argument The assumptions given in the informal argument: For each p there is x 0 such that (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} (2) ∃ y . R a ( x 0 , y ). Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 17 / 30

  82. Formal version of the BK argument The assumptions given in the informal argument: For each p there is x 0 such that (1) R a ( x 0 ) ⊆ { y | R b ( y ) = { x | p ( x ) }} (2) ∃ y . R a ( x 0 , y ). can be expressed as regular sequents as follows. ( A 1) R a ( c , y ) & R b ( y , x ) ⊢ { x , y } p ( x ) ( A 2) R a ( c , y ) & p ( x ) ⊢ { x , y } R b ( y , x ) ( A 3) ⊢ ∃ y . R a ( c , y ) Samson Abramsky and Jonathan Zvesper (Department of Computer Science, University of Oxford) From Lawvere to Brandenburger-Keisler: interactive forms of diagonalization and self-reference 17 / 30

Recommend


More recommend