Quelle  AC_in_L.thy

(*  Title:      ZF/Constructible/AC_in_L.thy
  Author: Lawrence C Paulson, Cambridge University Computer Laboratory
*)

section ‹The Axiom of Choice Holds in L!›

theory AC_in_L imports Formula Separation begin

subsection‹Extending a Wellordering over a List -- Lexicographic Power›

text‹This could be moved into a library.›

consts
  rlist   :: "[i,i]==>i"

inductive
  domains "rlist(A,r)" ⊆ "list(A) * list(A)"
  intros
    shorterI:
      "[length(l') < length(l); l' ∈ list(A); l ∈ list(A)]
       ==> ∈ rlist(A,r)"

    sameI:
      "[ ∈ rlist(A,r); a ∈ A]
       ==> ∈ rlist(A,r)"

    diffI:
      "[length(l') = length(l); ∈ r;
          l' ∈ list(A); l ∈ list(A); a' ∈ A; a ∈ A]
       ==> ∈ rlist(A,r)"
  type_intros list.intros


subsubsection‹Type checking›

lemmas rlist_type = rlist.dom_subset

lemmas field_rlist = rlist_type [THEN field_rel_subset]

subsubsection‹Linearity›

lemma rlist_Nil_Cons [intro]:
    "[a ∈ A; l ∈ list(A)] ==> <[], Cons(a,l)> ∈ rlist(A, r)"
by (simp add: shorterI)

lemma linear_rlist:
  assumes r: "linear(A,r)" shows "linear(list(A),rlist(A,r))"
proof -
  have "xs ∈ list(A) ==> ys ∈ list(A) ==> ⟨xs,ys⟩ ∈ rlist(A,r) ∨ xs = ys ∨ ⟨ys,xs⟩ ∈ rlist(A, r)"
    for xs ys
  proof (induct xs arbitrary: ys rule: list.induct)
    case Nil 
    thus ?case by (induct ys rule: list.induct) (auto simp add: shorterI)
  next
    case (Cons x xs)
    then have yConsCase: "⟨Cons(x,xs),Cons(y,ys)⟩ ∈ rlist(A,r) ∨ x=y ∧ xs = ys ∨ ⟨Cons(y,ys), Cons(x,xs)⟩ ∈ rlist(A,r)" 
      if "y ∈ A" and "ys ∈ list(A)" for y ys
      using that
      apply (rule_tac i = "length(xs)" and j = "length(ys)" in Ord_linear_lt)
      apply (simp_all add: shorterI)
      apply (rule linearE [OF r, of x y]) 
      apply (auto simp add: diffI intro: sameI) 
      done
    from ‹ys ∈ list(A)› show ?case
      by (cases rule: list.cases) (simp_all add: Cons rlist_Nil_Cons yConsCase)
  qed
  thus ?thesis by (simp add: linear_def) 
qed


subsubsection‹Well-foundedness›

text‹Nothing preceeds Nil in this ordering.›
inductive_cases rlist_NilE: " ∈ rlist(A,r)"

inductive_cases rlist_ConsE: " ∈ rlist(A,r)"

lemma not_rlist_Nil [simp]: " ∉ rlist(A,r)"
by (blast intro: elim: rlist_NilE)

lemma rlist_imp_length_le: " ∈ rlist(A,r) ==> length(l') ≤ length(l)"
apply (erule rlist.induct)
apply (simp_all add: leI)
done

lemma wf_on_rlist_n:
  "[n ∈ nat; wf[A](r)] ==> wf[{l ∈ list(A). length(l) = n}](rlist(A,r))"
apply (induct_tac n)
 apply (rule wf_onI2, simp)
apply (rule wf_onI2, clarify)
apply (erule_tac a=y in list.cases, clarify)
 apply (simp (no_asm_use))
apply clarify
apply (simp (no_asm_use))
apply (subgoal_tac "∀l2 ∈ list(A). length(l2) = x ⟶ Cons(a,l2) ∈ B", blast)
apply (erule_tac a=a in wf_on_induct, assumption)
apply (rule ballI)
apply (rule impI)
apply (erule_tac a=l2 in wf_on_induct, blast, clarify)
apply (rename_tac a' l2 l')
apply (drule_tac x="Cons(a',l')" in bspec, typecheck)
apply simp
apply (erule mp, clarify)
apply (erule rlist_ConsE, auto)
done

lemma list_eq_UN_length: "list(A) = (∪n∈nat. {l ∈ list(A). length(l) = n})"
by (blast intro: length_type)


lemma wf_on_rlist: "wf[A](r) ==> wf[list(A)](rlist(A,r))"
apply (subst list_eq_UN_length)
apply (rule wf_on_Union)
  apply (rule wf_imp_wf_on [OF wf_Memrel [of nat]])
 apply (simp add: wf_on_rlist_n)
apply (frule rlist_type [THEN subsetD])
apply (simp add: length_type)
apply (drule rlist_imp_length_le)
apply (erule leE)
apply (simp_all add: lt_def)
done


lemma wf_rlist: "wf(r) ==> wf(rlist(field(r),r))"
apply (simp add: wf_iff_wf_on_field)
apply (rule wf_on_subset_A [OF _ field_rlist])
apply (blast intro: wf_on_rlist)
done

lemma well_ord_rlist:
     "well_ord(A,r) ==> well_ord(list(A), rlist(A,r))"
apply (rule well_ordI)
apply (simp add: well_ord_def wf_on_rlist)
apply (simp add: well_ord_def tot_ord_def linear_rlist)
done


subsection‹An Injection from Formulas into the Natural Numbers›

text‹There is a well-known bijection between 🍋‹nat*nat› and 🍋‹nat› given by the expression f(m,n) = triangle(m+n) + m, where triangle(k)
 enumerates the triangular numbers and can be defined by triangle(0)=0,
 triangle(succ(k)) = succ(k + triangle(k)). Some small amount of effort is
 needed to show that f is a bijection. We already know that such a bijection exists by the theorem ‹well_ord_InfCard_square_eq›:
 @{thm[display] well_ord_InfCard_square_eq[no_vars]}
 
 However, this result merely states that there is a bijection between the two
 sets. It provides no means of naming a specific bijection. Therefore, we
 conduct the proofs under the assumption that a bijection exists. The simplest
 way to organize this is to use a locale.›

text‹Locale for any arbitrary injection between 🍋‹nat*nat›
  and 🍋‹nat›\›
locale Nat_Times_Nat =
  fixes fn
  assumes fn_inj: "fn ∈ inj(nat*nat, nat)"


consts   enum :: "[i,i]==>i"
primrec
  "enum(f, Member(x,y)) = f ` <0, f ` ⟨x,y⟩>"
  "enum(f, Equal(x,y)) = f ` <1, f ` ⟨x,y⟩>"
  "enum(f, Nand(p,q)) = f ` <2, f ` >"
  "enum(f, Forall(p)) = f ` "

lemma (in Nat_Times_Nat) fn_type [TC,simp]:
    "[x ∈ nat; y ∈ nat] ==> fn`⟨x,y⟩ ∈ nat"
by (blast intro: inj_is_fun [OF fn_inj] apply_funtype)

lemma (in Nat_Times_Nat) fn_iff:
    "[x ∈ nat; y ∈ nat; u ∈ nat; v ∈ nat]
     ==> (fn`⟨x,y⟩ = fn`⟨u,v⟩) ⟷ (x=u ∧ y=v)"
by (blast dest: inj_apply_equality [OF fn_inj])

lemma (in Nat_Times_Nat) enum_type [TC,simp]:
    "p ∈ formula ==> enum(fn,p) ∈ nat"
by (induct_tac p, simp_all)

lemma (in Nat_Times_Nat) enum_inject [rule_format]:
    "p ∈ formula ==> ∀q∈formula. enum(fn,p) = enum(fn,q) ⟶ p=q"
apply (induct_tac p, simp_all)
   apply (rule ballI)
   apply (erule formula.cases)
   apply (simp_all add: fn_iff)
  apply (rule ballI)
  apply (erule formula.cases)
  apply (simp_all add: fn_iff)
 apply (rule ballI)
 apply (erule_tac a=qa in formula.cases)
 apply (simp_all add: fn_iff)
 apply blast
apply (rule ballI)
apply (erule_tac a=q in formula.cases)
apply (simp_all add: fn_iff, blast)
done

lemma (in Nat_Times_Nat) inj_formula_nat:
    "(λp ∈ formula. enum(fn,p)) ∈ inj(formula, nat)"
apply (simp add: inj_def lam_type)
apply (blast intro: enum_inject)
done

lemma (in Nat_Times_Nat) well_ord_formula:
    "well_ord(formula, measure(formula, enum(fn)))"
apply (rule well_ord_measure, simp)
apply (blast intro: enum_inject)
done

lemmas nat_times_nat_lepoll_nat =
    InfCard_nat [THEN InfCard_square_eqpoll, THEN eqpoll_imp_lepoll]


text‹Not needed--but interesting?›
theorem formula_lepoll_nat: "formula < nat"
apply (insert nat_times_nat_lepoll_nat)
  unfolding lepoll_def
apply (blast intro: Nat_Times_Nat.inj_formula_nat Nat_Times_Nat.intro)
done


subsection‹Defining the Wellordering on 🍋‹DPow(A)›\›

text‹The objective is to build a wellordering on 🍋‹DPow(A)› from a
 given one on 🍋‹A›. We first introduce wellorderings for environments,
 which are lists built over 🍋‹A›. We combine it with the enumeration of
 formulas. The order type of the resulting wellordering gives us a map from
 (environment, formula) pairs into the ordinals. For each member of 🍋‹DPow(A)›, we take the minimum such ordinal.›

definition
  env_form_r :: "[i,i,i]==>i" where
    🍋 ‹wellordering on (environment, formula) pairs›
   "env_form_r(f,r,A) ≡
      rmult(list(A), rlist(A, r),
            formula, measure(formula, enum(f)))"

definition
  env_form_map :: "[i,i,i,i]==>i" where
    🍋 ‹map from (environment, formula) pairs to ordinals›
   "env_form_map(f,r,A,z)
      ≡ ordermap(list(A) * formula, env_form_r(f,r,A)) ` z"

definition
  DPow_ord :: "[i,i,i,i,i]==>o" where
    🍋 ‹predicate that holds if 🍋‹k› is a valid index for 🍋‹X›\›
   "DPow_ord(f,r,A,X,k) ≡
           ∃env ∈ list(A). ∃p ∈ formula.
             arity(p) ≤ succ(length(env)) ∧
             X = {x∈A. sats(A, p, Cons(x,env))} ∧
             env_form_map(f,r,A,⟨env,p⟩) = k"

definition
  DPow_least :: "[i,i,i,i]==>i" where
    🍋 ‹function yielding the smallest index for 🍋‹X›\›
   "DPow_least(f,r,A,X) ≡ μ k. DPow_ord(f,r,A,X,k)"

definition
  DPow_r :: "[i,i,i]==>i" where
    🍋 ‹a wellordering on 🍋‹DPow(A)›\›
   "DPow_r(f,r,A) ≡ measure(DPow(A), DPow_least(f,r,A))"


lemma (in Nat_Times_Nat) well_ord_env_form_r:
    "well_ord(A,r)
     ==> well_ord(list(A) * formula, env_form_r(fn,r,A))"
by (simp add: env_form_r_def well_ord_rmult well_ord_rlist well_ord_formula)

lemma (in Nat_Times_Nat) Ord_env_form_map:
    "[well_ord(A,r); z ∈ list(A) * formula]
     ==> Ord(env_form_map(fn,r,A,z))"
by (simp add: env_form_map_def Ord_ordermap well_ord_env_form_r)

lemma DPow_imp_ex_DPow_ord:
    "X ∈ DPow(A) ==> ∃k. DPow_ord(fn,r,A,X,k)"
apply (simp add: DPow_ord_def)
apply (blast dest!: DPowD)
done

lemma (in Nat_Times_Nat) DPow_ord_imp_Ord:
     "[DPow_ord(fn,r,A,X,k); well_ord(A,r)] ==> Ord(k)"
apply (simp add: DPow_ord_def, clarify)
apply (simp add: Ord_env_form_map)
done

lemma (in Nat_Times_Nat) DPow_imp_DPow_least:
    "[X ∈ DPow(A); well_ord(A,r)]
     ==> DPow_ord(fn, r, A, X, DPow_least(fn,r,A,X))"
apply (simp add: DPow_least_def)
apply (blast dest: DPow_imp_ex_DPow_ord intro: DPow_ord_imp_Ord LeastI)
done

lemma (in Nat_Times_Nat) env_form_map_inject:
    "[env_form_map(fn,r,A,u) = env_form_map(fn,r,A,v); well_ord(A,r);
       u ∈ list(A) * formula; v ∈ list(A) * formula]
     ==> u=v"
apply (simp add: env_form_map_def)
apply (rule inj_apply_equality [OF bij_is_inj, OF ordermap_bij,
                                OF well_ord_env_form_r], assumption+)
done

lemma (in Nat_Times_Nat) DPow_ord_unique:
    "[DPow_ord(fn,r,A,X,k); DPow_ord(fn,r,A,Y,k); well_ord(A,r)]
     ==> X=Y"
apply (simp add: DPow_ord_def, clarify)
apply (drule env_form_map_inject, auto)
done

lemma (in Nat_Times_Nat) well_ord_DPow_r:
    "well_ord(A,r) ==> well_ord(DPow(A), DPow_r(fn,r,A))"
apply (simp add: DPow_r_def)
apply (rule well_ord_measure)
 apply (simp add: DPow_least_def)
apply (drule DPow_imp_DPow_least, assumption)+
apply simp
apply (blast intro: DPow_ord_unique)
done

lemma (in Nat_Times_Nat) DPow_r_type:
    "DPow_r(fn,r,A) ⊆ DPow(A) * DPow(A)"
by (simp add: DPow_r_def measure_def, blast)


subsection‹Limit Construction for Well-Orderings›

text‹Now we work towards the transfinite definition of wellorderings for
 🍋‹Lset(i)›. We assume as an inductive hypothesis that there is a family
 of wellorderings for smaller ordinals.›

definition
  rlimit :: "[i,i==>i]==>i" where
  🍋 ‹Expresses the wellordering at limit ordinals. The conditional
  lets us remove the premise 🍋‹Limit(i)› from some theorems.›
    "rlimit(i,r) ≡
       if Limit(i) then
         {z: Lset(i) * Lset(i).
          ∃x' x. z = ∧
                 (lrank(x') < lrank(x) |
                  (lrank(x') = lrank(x) ∧ ∈ r(succ(lrank(x)))))}
       else 0"

definition
  Lset_new :: "i==>i" where
  🍋 ‹This constant denotes the set of elements introduced at level
  🍋‹succ(i)›\›
    "Lset_new(i) ≡ {x ∈ Lset(succ(i)). lrank(x) = i}"

lemma Limit_Lset_eq2:
    "Limit(i) ==> Lset(i) = (∪j∈i. Lset_new(j))"
apply (simp add: Limit_Lset_eq)
apply (rule equalityI)
 apply safe
 apply (subgoal_tac "Ord(y)")
  prefer 2 apply (blast intro: Ord_in_Ord Limit_is_Ord)
 apply (simp_all add: Limit_is_Ord Lset_iff_lrank_lt Lset_new_def
                      Ord_mem_iff_lt)
 apply (blast intro: lt_trans)
apply (rule_tac x = "succ(lrank(x))" in bexI)
 apply (simp)
apply (blast intro: Limit_has_succ ltD)
done

lemma wf_on_Lset:
    "wf[Lset(succ(j))](r(succ(j))) ==> wf[Lset_new(j)](rlimit(i,r))"
apply (simp add: wf_on_def Lset_new_def)
apply (erule wf_subset)
apply (simp add: rlimit_def, force)
done

lemma wf_on_rlimit:
    "(∀j==> wf[Lset(i)](rlimit(i,r))"
apply (case_tac "Limit(i)") 
 prefer 2
 apply (simp add: rlimit_def wf_on_any_0)
apply (simp add: Limit_Lset_eq2)
apply (rule wf_on_Union)
  apply (rule wf_imp_wf_on [OF wf_Memrel [of i]])
 apply (blast intro: wf_on_Lset Limit_has_succ Limit_is_Ord ltI)
apply (force simp add: rlimit_def Limit_is_Ord Lset_iff_lrank_lt Lset_new_def
                       Ord_mem_iff_lt)
done

lemma linear_rlimit:
    "[Limit(i); ∀j]
     ==> linear(Lset(i), rlimit(i,r))"
apply (frule Limit_is_Ord)
apply (simp add: Limit_Lset_eq2 Lset_new_def)
apply (simp add: linear_def rlimit_def Ball_def lt_Ord Lset_iff_lrank_lt)
apply (simp add: ltI, clarify)
apply (rename_tac u v)
apply (rule_tac i="lrank(u)" and j="lrank(v)" in Ord_linear_lt, simp_all) 
apply (drule_tac x="succ(lrank(u) ∪ lrank(v))" in ospec)
 apply (simp add: ltI)
apply (drule_tac x=u in spec, simp)
apply (drule_tac x=v in spec, simp)
done

lemma well_ord_rlimit:
    "[Limit(i); ∀j]
     ==> well_ord(Lset(i), rlimit(i,r))"
by (blast intro: well_ordI wf_on_rlimit well_ord_is_wf
                           linear_rlimit well_ord_is_linear)

lemma rlimit_cong:
     "(∧j. j==> r'(j) = r(j)) ==> rlimit(i,r) = rlimit(i,r')"
apply (simp add: rlimit_def, clarify) 
apply (rule refl iff_refl Collect_cong ex_cong conj_cong)+
apply (simp add: Limit_is_Ord Lset_lrank_lt)
done


subsection‹Transfinite Definition of the Wellordering on 🍋‹L›\›

definition
  L_r :: "[i, i] ==> i" where
  "L_r(f) ≡ λi.
      transrec3(i, 0, λx r. DPow_r(f, r, Lset(x)),
                λx r. rlimit(x, λy. r`y))"

subsubsection‹The Corresponding Recursion Equations›
lemma [simp]: "L_r(f,0) = 0"
by (simp add: L_r_def)

lemma [simp]: "L_r(f, succ(i)) = DPow_r(f, L_r(f,i), Lset(i))"
by (simp add: L_r_def)

text‹The limit case is non-trivial because of the distinction between
 object-level and meta-level abstraction.›
lemma [simp]: "Limit(i) ==> L_r(f,i) = rlimit(i, L_r(f))"
by (simp cong: rlimit_cong add: transrec3_Limit L_r_def ltD)

lemma (in Nat_Times_Nat) L_r_type:
    "Ord(i) ==> L_r(fn,i) ⊆ Lset(i) * Lset(i)"
apply (induct i rule: trans_induct3)
  apply (simp_all add: Lset_succ DPow_r_type well_ord_DPow_r rlimit_def
                       Transset_subset_DPow [OF Transset_Lset], blast)
done

lemma (in Nat_Times_Nat) well_ord_L_r:
    "Ord(i) ==> well_ord(Lset(i), L_r(fn,i))"
apply (induct i rule: trans_induct3)
apply (simp_all add: well_ord0 Lset_succ L_r_type well_ord_DPow_r
                     well_ord_rlimit ltD)
done

lemma well_ord_L_r:
    "Ord(i) ==> ∃r. well_ord(Lset(i), r)"
apply (insert nat_times_nat_lepoll_nat)
  unfolding lepoll_def
apply (blast intro: Nat_Times_Nat.well_ord_L_r Nat_Times_Nat.intro)
done


text‹Every constructible set is well-ordered! Therefore the Wellordering Theorem and
  the Axiom of Choice hold in 🍋‹L›!›
theorem L_implies_AC: assumes x: "L(x)" shows "∃r. well_ord(x,r)"
  using Transset_Lset x
apply (simp add: Transset_def L_def)
apply (blast dest!: well_ord_L_r intro: well_ord_subset)
done

interpretation L: M_basic L by (rule M_basic_L)

theorem "∀x[L]. ∃r. wellordered(L,x,r)"
proof 
  fix x
  assume "L(x)"
  then obtain r where "well_ord(x,r)" 
    by (blast dest: L_implies_AC) 
  thus "∃r. wellordered(L,x,r)" 
    by (blast intro: L.well_ord_imp_relativized)
qed

text‹In order to prove 🍋‹ ∃r[L]. wellordered(L,x,r)›, it's necessary to know
 that 🍋‹r› is actually constructible. It follows from the assumption ``🍋‹V› equals 🍋‹L''›,
 but this reasoning doesn't appear to work in Isabelle.›

end