Quelle  Collecting1.thy

subsection "A small step semantics on annotated commands"

theory Collecting1
imports Collecting
begin

text‹The idea: the state is propagated through the annotated command as an
 annotation 🍋‹{s}›, all other annotations are 🍋‹{}›. It is easy
 to show that this semantics approximates the collecting semantics.›

lemma step_preserves_le:
  "[ step S cs = cs; S' ⊆ S; cs' ≤ cs ] ==>
   step S' cs' ≤ cs"
by (metis mono2_step)

lemma steps_empty_preserves_le: assumes "step S cs = cs"
shows "cs' ≤ cs ==> (step {} ^^ n) cs' ≤ cs"
proof(induction n arbitrary: cs')
  case 0 thus ?case by simp
next
  case (Suc n) thus ?case
    using Suc.IH[OF step_preserves_le[OF assms empty_subsetI Suc.prems]]
    by(simp add:funpow_swap1)
qed


definition steps :: "state ==> com ==> nat ==> state set acom" where
"steps s c n = ((step {})^^n) (step {s} (annotate (λp. {}) c))"

lemma steps_approx_fix_step: assumes "step S cs = cs" and "s ∈ S"
shows "steps s (strip cs) n ≤ cs"
proof-
  let ?bot = "annotate (λp. {}) (strip cs)"
  have "?bot ≤ cs" by(induction cs) auto
  from step_preserves_le[OF assms(1)_ this, of "{s}"] ‹s ∈ S›
  have 1: "step {s} ?bot ≤ cs" by simp
  from steps_empty_preserves_le[OF assms(1) 1]
  show ?thesis by(simp add: steps_def)
qed

theorem steps_approx_CS: "steps s c n ≤ CS c"
by (metis CS_unfold UNIV_I steps_approx_fix_step strip_CS)

end