(* Title: A Definitional Encoding of TLA in Isabelle/HOL Authors:GudmundGrov<ggrovatinf.ed.ac.uk> StephanMerz<Stephan.Merzatloria.fr> Year:2011 Maintainer:GudmundGrov<ggrovatinf.ed.ac.uk>
*)
section"A Proof System for TLA* "
theory Rules imports PreFormulas begin
text‹
We prove soundness of the proof system of \tlastar{}, from which the system
verification rules from Lamport's original TLA paper will be derived.
This theory is still state-independent, thus state-dependent enableness proofs,
required for proofs based on fairness assumptions, and flexible quantification,
are not discussed here.
The \tlastar{} paper cite‹"Merz99"› suggest both a \emph{hetereogeneous} and a \emph{homogenous} proof system for \tlastar{}.
The homogeneous version eliminates the auxiliary definitions from the ‹Preformula› theory, creating a single provability relation.
This axiomatisation is based on the fact that a pre-formula can only be used
via the ‹sq› rule. In a nutshell, ‹sq› is applied to ‹pax1› to ‹pax5›, and ‹nex›, ‹pre› and ‹pmp›
are changed to accommodate this. It is argued that while the hetereogenous version
is easier to understand, the homogenous system avoids the introduction of an
auxiliary provability relation. However, the price to pay is that reasoning about
pre-formulas (in particular, actions) has to be performed in the scope of
temporal operators such as ‹◻[P]_v›, which is notationally quite heavy,
We prefer here the heterogeneous approach, which exposes the pre-formulas and
lets us use standard HOL rules more directly. ›
subsection"The Basic Axioms"
theorem fmp: assumes"⊨ F"and"⊨ F ⟶ G"shows"⊨ G" using assms[unlifted] by auto
theorem pmp: assumes"|~ F"and"|~ F ⟶ G"shows"|~ G" using assms[unlifted] by auto
theorem sq: assumes"|~ P"shows"⊨◻[P]_v" using assms[unlifted] by (auto simp: action_def)
theorempre: assumes"⊨ F"shows"|~ F" using assms by auto
theorem nex: assumes h1: "⊨ F"shows"|~ ◯F" using assms by (auto simp: nexts_def)
theorem ax0: "⊨ # True" by auto
theorem ax1: "⊨◻F ⟶ F" proof (clarsimp simp: always_def) fix w assume"∀n. (w |s n) ⊨ F" hence"(w |s 0) ⊨ F" .. thus"w ⊨ F"by simp qed
theorem ax2: "⊨◻F ⟶◻[◻F]_v" by (auto simp: always_def action_def suffix_plus)
theorem ax3: assumes H: "|~ F ∧ Unchanged v ⟶◯F" shows"⊨◻[F ⟶◯F]_v ⟶ (F ⟶◻F)" proof (clarsimp simp: always_def) fix w n assume a1: "w ⊨◻[F ⟶◯F]_v"and a2: "w ⊨ F" show"(w |s n) ⊨ F" proof (induct n) from a2 show"(w |s 0) ⊨ F"by simp next fix m assume a3: "(w |s m) ⊨ F" with a1 H[unlifted] show"(w |s (Suc m)) ⊨ F" by (auto simp: nexts_def action_def tail_suffix_suc) qed qed
text‹
Theorem to show that universal quantification distributes over the always
operator. Since the \tlastar{} paper only addresses the propositional fragment,
this theorem does not appear there. ›
theorem allT: "⊨ (∀x. ◻(F x)) = (◻(∀x. F x))" by (auto simp: always_def)
theorem allActT: "⊨ (∀x. ◻[F x]_v) = (◻[(∀x. F x)]_v)" by (force simp: action_def)
subsection"Derived Theorems"
text‹
This section includes some derived theorems based on the axioms, taken
from the \tlastar{} paper~cite‹"Merz99"›. We mimic the proofs given there
and avoid semantic reasoning whenever possible.
The ‹alw› theorem of~cite‹"Merz99"› states that if F holds
in all worlds then it always holds, i.e. $F \vDash\Box F$. However,
the derivation of this theorem (using the proof rules above)
relies on access of the set of free variables (FV), which is not
available in a shallow encoding.
However, we can prove a similar rule ‹alw2› using an additional
hypothesis @{term "|~ F ∧ Unchanged v ⟶◯F"}. ›
theorem alw2: assumes h1: "⊨ F"and h2: "|~ F ∧ Unchanged v ⟶◯F" shows"⊨◻F" proof - from h1 have g2: "|~ ◯F"by (rule nex) hence g3: "|~ F ⟶◯F"by auto hence g4:"⊨◻[(F ⟶◯F)]_v"by (rule sq) from h2 have"⊨◻[(F ⟶◯F)]_v ⟶ F ⟶◻F"by (rule ax3) with g4[unlifted] have g5: "⊨ F ⟶◻F"by auto with h1[unlifted] show ?thesis by auto qed
text‹
Similar theorem, assuming that @{term "F"} is stuttering invariant. ›
theorem alw3: assumes h1: "⊨ F"and h2: "stutinv F" shows"⊨◻F" proof - from h2 have"|~ F ∧ Unchanged id ⟶◯F"by (rule pre_id_unch) with h1 show ?thesis by (rule alw2) qed
text‹
In a deep embedding, we could prove that all (proper) \tlastar{}
formulas are stuttering invariant and then get rid of the second
hypothesis of rule ‹alw3›. In fact, the rule is even true
for pre-formulas, as shown by the following rule, whose proof relies
on semantical reasoning. › theorem alw: assumes H1: "⊨ F"shows"⊨◻F" using H1 by (auto simp: always_def)
theorem alw_valid_iff_valid: "(⊨◻F) = (⊨ F)" proof assume"⊨◻F" from this ax1 show"⊨ F"by (rule fmp) qed (rule alw)
text‹ cite‹"Merz99"› proves the following theorem using the deduction theorem of \tlastar{}: ‹(⊨ F ==>⊨ G) ==>⊨ []F ⟶ G›, which can only be
proved by induction on the formula structure, in a deep embedding. ›
theorem T1[simp_unl]: "⊨◻◻F = []F" proof (auto simp: always_def suffix_plus) fix w n assume"∀m k. (w |s (k+m)) ⊨ F" hence"(w |s (n+0)) ⊨ F"by blast thus"(w |s n) ⊨ F"by simp qed
theorem T2[simp_unl]: "⊨◻◻[P]_v = ◻[P]_v" proof - have1: "|~ ◻[P]_v ⟶◯◻[P]_v"using pax4 by force hence"⊨◻[◻[P]_v ⟶◯◻[P]_v]_v"by (rule sq) moreover have"⊨◻[ ◻[P]_v ⟶◯◻[P]_v ]_v ⟶◻[P]_v ⟶◻◻[P]_v" by (rule ax3) (auto elim: 1[unlift_rule]) moreover have"⊨◻◻[P]_v ⟶◻[P]_v"by (rule ax1) ultimatelyshow ?thesis by force qed
theorem T3[simp_unl]: "⊨◻[[P]_v]_v = ◻[P]_v" proof - have"|~ P ⟶ [P]_v"by (auto simp: actrans_def) hence"⊨◻[(P ⟶ [P]_v)]_v"by (rule sq) with ax4 have"⊨◻[P]_v ⟶◻[[P]_v]_v"by force moreover have"|~ [P]_v ⟶ v`≠ $v ⟶ P"by (auto simp: unch_def actrans_def) hence"⊨◻[[P]_v ⟶ v`≠ $v ⟶ P]_v"by (rule sq) with ax5 have"⊨◻[[P]_v]_v ⟶◻[P]_v"by (force intro: ax4[unlift_rule]) ultimatelyshow ?thesis by force qed
theorem M2: assumes h: "|~ F ⟶ G" shows"⊨◻[F]_v ⟶◻[G]_v" using sq[OF h] ax4 by force
theorem N1: assumes h: "⊨ F ⟶ G" shows"|~ ◯F ⟶◯G" by (rule pmp[OF nex[OF h] pax2])
theorem T4: "⊨◻[P]_v ⟶◻[[P]_v]_w" proof - have"⊨◻◻[P]_v ⟶◻[◻◻[P]_v]_w"by (rule ax2) moreover from pax4 have"|~ ◻◻[P]_v ⟶ [P]_v"unfolding T2[int_rewrite] by force hence"⊨◻[◻◻[P]_v]_w ⟶◻[[P]_v]_w"by (rule M2) ultimatelyshow ?thesis unfolding T2[int_rewrite] by (rule lift_imp_trans) qed
theorem T5: "⊨◻[[P]_w]_v ⟶◻[[P]_v]_w" proof - have"|~ [[P]_w]_v ⟶ [[P]_v]_w"by (auto simp: actrans_def) hence"⊨◻[[[P]_w]_v]_w ⟶◻[[[P]_v]_w]_w"by (rule M2) with T4 show ?thesis unfolding T3[int_rewrite] by (rule lift_imp_trans) qed
theorem T6: "⊨◻F ⟶◻[◯F]_v" proof - from ax1 have"|~ ◯(◻F ⟶ F)"by (rule nex) with pax2 have"|~ ◯◻F ⟶◯F"by force with pax3 have"|~ ◻F ⟶◯F"by (rule pref_imp_trans) hence"⊨◻[◻F]_v ⟶◻[◯F]_v"by (rule M2) with ax2 show ?thesis by (rule lift_imp_trans) qed
theorem T7: assumes h: "|~ F ∧ Unchanged v ⟶◯F" shows"|~ (F ∧◯◻F) = ◻F" proof - have"⊨◻[◯F ⟶ F ⟶◯F]_v"by (rule sq) auto with ax4 have"⊨◻[◯F]_v ⟶◻[(F ⟶◯F)]_v"by force with ax3[OF h, unlifted] have"⊨◻[◯F]_v ⟶ (F ⟶◻F)"by force with pax5 have"|~ F ∧◯◻F ⟶◻F"by force with ax1[of "TEMP F",unlifted] pax3[of "TEMP F",unlifted] show ?thesis by force qed
theorem T8: "|~ ◯(F ∧ G) = (◯F ∧◯G)" proof - have"|~ ◯(F ∧ G) ⟶◯F"by (rule N1) auto moreover have"|~ ◯(F ∧ G) ⟶◯G"by (rule N1) auto moreover have"⊨ F ⟶ G ⟶ F ∧ G"by auto from nex[OF this] have"|~ ◯F ⟶◯G ⟶◯(F ∧ G)" by (force intro: pax2[unlift_rule]) ultimatelyshow ?thesis by force qed
lemma T9: "|~ ◻[A]_v ⟶ [A]_v" using pax4 by force
theorem H1: assumes h1: "⊨◻[P]_v"and h2: "⊨◻[P ⟶ Q]_v" shows"⊨◻[Q]_v" using assms ax4[unlifted] by force
theorem H2: assumes h1: "⊨ F"shows"⊨◻[F]_v" using h1 by (blast dest: pre sq)
theorem H4: "⊨◻[[P]_v ⟶ P]_v" proof - have"|~ v` ≠ $v ⟶ ([P]_v ⟶ P)"by (auto simp: unch_def actrans_def) hence"⊨◻[v` ≠ $v ⟶ ([P]_v ⟶ P)]_v"by (rule sq) with ax5 show ?thesis by (rule H1) qed
theorem H5: "⊨◻[◻F ⟶◯◻F]_v" by (rule pax3[THEN sq])
subsection"Some other useful derived theorems"
theorem P1: "|~ ◻F ⟶◯F" proof - have"|~ ◯◻F ⟶◯F"by (rule N1[OF ax1]) with pax3 show ?thesis by (rule pref_imp_trans) qed
theorem P2: "|~ ◻F ⟶ F ∧◯F" using ax1[of F] P1[of F] by force
theorem P4: "⊨◻F ⟶◻[F]_v" proof - have"⊨◻[◻F]_v ⟶◻[F]_v"by (rule M2[OF pre[OF ax1]]) with ax2 show ?thesis by (rule lift_imp_trans) qed
theorem P5: "⊨◻[P]_v ⟶◻[◻[P]_v]_w" proof - have"⊨◻◻[P]_v ⟶◻[◻[P]_v]_w"by (rule P4) thus ?thesis by (unfold T2[int_rewrite]) qed
theorem M0: "⊨◻F ⟶◻[F ⟶◯F]_v" proof - from P1 have"|~ ◻F ⟶ F ⟶◯F"by force hence"⊨◻[◻F]_v ⟶◻[F ⟶◯F]_v"by (rule M2) with ax2 show ?thesis by (rule lift_imp_trans) qed
theorem M1: "⊨◻F ⟶◻[F ∧◯F]_v" proof - have"|~ ◻F ⟶ F ∧◯F"by (rule P2) hence"⊨◻[◻F]_v ⟶◻[F ∧◯F]_v"by (rule M2) with ax2 show ?thesis by (rule lift_imp_trans) qed
theorem M3: assumes h: "⊨ F"shows"⊨◻[◯F]_v" using alw[OF h] T6 by (rule fmp)
theorem M6: "⊨◻[F ∧ G]_v ⟶◻[F]_v ∧◻[G]_v" proof - have"⊨◻[F ∧ G]_v ⟶◻[F]_v"by (rule M2) auto moreover have"⊨◻[F ∧ G]_v ⟶◻[G]_v"by (rule M2) auto ultimatelyshow ?thesis by force qed
theorem M7: "⊨◻[F]_v ∧◻[G]_v ⟶◻[F ∧ G]_v" proof - have"|~ F ⟶ G ⟶ F ∧ G"by auto hence"⊨◻[F]_v ⟶◻[G ⟶ F ∧ G]_v"by (rule M2) with ax4 show ?thesis by force qed
theorem M9: "|~ ◻F ⟶ F ∧◯◻F" usingpre[OF ax1[of "F"]] pax3[of "F"] by force
theorem M10: assumes h: "|~ F ∧ Unchanged v ⟶◯F" shows"|~ F ∧◯◻F ⟶◻F" using T7[OF h] by auto
theorem M11: assumes h: "|~ [A]_f ⟶ [B]_g" shows"⊨◻[A]_f ⟶◻[B]_g" proof - from h have"⊨◻[[A]_f]_g ⟶◻[[B]_g]_g"by (rule M2) with T4 show ?thesis by force qed
theorem M12: "⊨ (◻[A]_f ∧◻[B]_g) = ◻[[A]_f ∧ [B]_g]_(f,g)" proof - have"⊨◻[A]_f ∧◻[B]_g ⟶◻[[A]_f ∧ [B]_g]_(f,g)" by (auto simp: M8[int_rewrite] elim: T4[unlift_rule]) moreover have"|~ [[A]_f ∧ [B]_g]_(f,g) ⟶ [A]_f" by (auto simp: actrans_def unch_def all_before_eq all_after_eq) hence"⊨◻[[A]_f ∧ [B]_g]_(f,g) ⟶◻[A]_f"by (rule M11) moreover have"|~ [[A]_f ∧ [B]_g]_(f,g) ⟶ [B]_g" by (auto simp: actrans_def unch_def all_before_eq all_after_eq) hence"⊨◻[[A]_f ∧ [B]_g]_(f,g) ⟶◻[B]_g" by (rule M11) ultimatelyshow ?thesis by force qed
text‹
We now derive Lamport's 6 simple temporal logic rules (STL1)-(STL6) cite‹"Lamport94"›.
Firstly, STL1 is the same as @{thm alw} derived above. ›
lemmas STL1 = alw
text‹
STL2 and STL3 have also already been derived. ›
lemmas STL2 = ax1
lemmas STL3 = T1
text‹
As with the derivation of @{thm alw}, a purely syntactic derivation of
(STL4) relies on an additional argument -- either using ‹Unchanged›
or ‹STUTINV›. ›
theorem STL4_2: assumes h1: "⊨ F ⟶ G"and h2: "|~ G ∧ Unchanged v ⟶◯G" shows"⊨◻F ⟶◻G" proof - from ax1[of F] h1 have"⊨◻F ⟶ G"by (rule lift_imp_trans) moreover from h1 have"|~ ◯F ⟶◯G"by (rule N1) hence"|~ ◯F ⟶ G ⟶◯G"by auto hence"⊨◻[◯F]_v ⟶◻[G ⟶◯G]_v"by (rule M2) with T6 have"⊨◻F ⟶◻[G ⟶◯G]_v"by (rule lift_imp_trans) moreover from h2 have"⊨◻[G ⟶◯G]_v ⟶ G ⟶◻G"by (rule ax3) ultimately show ?thesis by force qed
lemma STL4_3: assumes h1: "⊨ F ⟶ G"and h2: "STUTINV G" shows"⊨◻F ⟶◻G" using h1 h2[THEN pre_id_unch] by (rule STL4_2)
text‹Of course, the original rule can be derived semantically›
lemma STL4: assumes h: "⊨ F ⟶ G"shows"⊨◻F ⟶◻G" using h by (force simp: always_def)
text‹Dual rule for ‹♢››
lemma STL4_eve: assumes h: "⊨ F ⟶ G"shows"⊨♢F ⟶♢G" using h by (force simp: eventually_defs)
text‹
Similarly, a purely syntactic derivation of (STL5) requires extra hypotheses. ›
theorem STL5_2: assumes h1: "|~ F ∧ Unchanged f ⟶◯F" and h2: "|~ G ∧ Unchanged g ⟶◯G" shows"⊨◻(F ∧ G) = (◻F ∧◻G)" proof (rule int_iffI) have"⊨ F ∧ G ⟶ F"by auto from this h1 have"⊨◻(F ∧ G) ⟶◻F"by (rule STL4_2) moreover have"⊨ F ∧ G ⟶ G"by auto from this h2 have"⊨◻(F ∧ G) ⟶◻G"by (rule STL4_2) ultimatelyshow"⊨◻(F ∧ G) ⟶◻F ∧◻G"by force next have"|~ Unchanged (f,g) ⟶ Unchanged f ∧ Unchanged g"by (auto simp: unch_defs) with h1[unlifted] h2[unlifted] T8[of F G, unlifted] have h3: "|~ (F ∧ G) ∧ Unchanged (f,g) ⟶◯(F ∧ G)"by force from ax1[of F] ax1[of G] have"⊨◻F ∧◻G ⟶ F ∧ G"by force moreover from ax2[of F] ax2[of G] have"⊨◻F ∧◻G ⟶◻[◻F]_(f,g) ∧◻[◻G]_(f,g)"by force with M8 have"⊨◻F ∧◻G ⟶◻[◻F ∧◻G]_(f,g)"by force moreover from P1[of F] P1[of G] have"|~ ◻F ∧◻G ⟶ F ∧ G ⟶◯(F ∧ G)" unfolding T8[int_rewrite] by force hence"⊨◻[ ◻F ∧◻G ]_(f,g) ⟶◻[F ∧ G ⟶◯(F ∧ G)]_(f,g)"by (rule M2) from this ax3[OF h3] have"⊨◻[ ◻F ∧◻G ]_(f,g) ⟶ F ∧ G ⟶◻(F ∧ G)" by (rule lift_imp_trans) ultimatelyshow"⊨◻F ∧◻G ⟶◻(F ∧ G)"by force qed
lemma STL6: "⊨♢◻(F ∧ G) = (♢◻F ∧♢◻G)" proof auto fix w assume a1: "w ⊨♢◻F"and a2: "w ⊨♢◻G" from a1 obtain nf where nf: "(w |s nf) ⊨◻F"by (auto simp: eventually_defs) from a2 obtain ng where ng: "(w |s ng) ⊨◻G"by (auto simp: eventually_defs) let ?n = "max nf ng" have"nf ≤ ?n"by simp from this nf have"(w |s ?n) ⊨◻F"by (rule linalw) moreover have"ng ≤ ?n"by simp from this ng have"(w |s ?n) ⊨◻G"by (rule linalw) ultimately have"(w |s ?n) ⊨◻(F ∧ G)"by (rule box_conjE) thus"w ⊨♢◻(F ∧ G)"by (auto simp: eventually_defs) next fix w assume h: "w ⊨♢◻(F ∧ G)" have"⊨ F ∧ G ⟶ F"by auto hence"⊨♢◻(F ∧ G) ⟶♢◻F"by (rule STL4_eve[OF STL4]) with h show"w ⊨♢◻F"by auto next fix w assume h: "w ⊨♢◻(F ∧ G)" have"⊨ F ∧ G ⟶ G"by auto hence"⊨♢◻(F ∧ G) ⟶♢◻G"by (rule STL4_eve[OF STL4]) with h show"w ⊨♢◻G"by auto qed
lemma MM0: "⊨◻(F ⟶ G) ⟶◻F ⟶◻G" proof - have"⊨◻(F ∧ (F ⟶ G)) ⟶◻G"by (rule STL4) auto thus ?thesis by (auto simp: STL5[int_rewrite]) qed
lemma MM1: assumes h: "⊨ F = G"shows"⊨◻F = ◻G" by (auto simp: h[int_rewrite])
theorem MM2: "⊨◻A ∧◻(B ⟶ C) ⟶◻(A ∧ B ⟶ C)" proof - have"⊨◻(A ∧ (B ⟶ C)) ⟶◻(A ∧ B ⟶ C)"by (rule STL4) auto thus ?thesis by (auto simp: STL5[int_rewrite]) qed
theorem MM3: "⊨◻¬A ⟶◻(A ∧ B ⟶ C)" by (rule STL4) auto
theorem MM4[simp_unl]: "⊨◻#F = #F" proof (cases "F") assume"F" hence1: "⊨ #F"by auto hence"⊨◻#F"by (rule alw) with1show ?thesis by force next assume"¬F" hence1: "⊨¬ #F"by auto from ax1 have"⊨¬ #F ⟶¬◻#F"by (rule lift_imp_neg) with1show ?thesis by force qed
theorem MM4b[simp_unl]: "⊨◻¬#F = ¬#F" proof - have"⊨¬#F = #(¬F)"by auto hence"⊨◻¬#F = ◻#(¬F)"by (rule MM1) thus ?thesis by auto qed
theorem MM5: "⊨◻F ∨◻G ⟶◻(F ∨ G)" proof - have"⊨◻F ⟶◻(F ∨ G)"by (rule STL4) auto moreover have"⊨◻G ⟶◻(F ∨ G)"by (rule STL4) auto ultimatelyshow ?thesis by force qed
theorem MM6: "⊨◻F ∨◻G ⟶◻(◻F ∨◻G)" proof - have"⊨◻◻F ∨◻◻G ⟶◻(◻F ∨◻G)"by (rule MM5) thus ?thesis by simp qed
lemma MM10: assumes h: "|~ F = G"shows"⊨◻[F]_v = ◻[G]_v" by (auto simp: h[int_rewrite])
lemma MM9: assumes h: "⊨ F = G"shows"⊨◻[F]_v = ◻[G]_v" by (rule MM10[OF pre[OF h]])
theorem MM11: "⊨◻[¬(P ∧ Q)]_v ⟶◻[P]_v ⟶◻[P ∧¬Q]_v" proof - have"⊨◻[¬(P ∧ Q)]_v ⟶◻[P ⟶ P ∧¬Q]_v"by (rule M2) auto from this ax4 show ?thesis by (rule lift_imp_trans) qed
theorem E1: "⊨♢(F ∨ G) = (♢F ∨♢G)" proof - have"⊨◻¬(F ∨ G) = ◻(¬F ∧¬G)"by (rule MM1) auto thus ?thesis unfolding eventually_def STL5[int_rewrite] by force qed
theorem E3: "⊨ F ⟶♢F" unfolding eventually_def by (force dest: ax1[unlift_rule])
theorem E4: "⊨◻F ⟶♢F" by (rule lift_imp_trans[OF ax1 E3])
theorem E5: "⊨◻F ⟶◻♢F" proof - have"⊨◻◻F ⟶◻♢F"by (rule STL4[OF E4]) thus ?thesis by simp qed
theorem E6: "⊨◻F ⟶♢◻F" using E4[of "TEMP ◻F"] by simp
theorem E7: assumes h: "|~ ¬F ∧ Unchanged v ⟶◯¬F" shows"|~ ♢F ⟶ F ∨◯♢F" proof - from h have"|~ ¬F ∧◯◻¬F ⟶◻¬F"by (rule M10) thus ?thesis by (auto simp: eventually_def) qed
theorem E8: "⊨♢(F ⟶ G) ⟶◻F ⟶♢G" proof - have"⊨◻(F ∧¬G) ⟶◻¬(F ⟶ G)"by (rule STL4) auto thus ?thesis unfolding eventually_def STL5[int_rewrite] by auto qed
theorem E9: "⊨◻(F ⟶ G) ⟶♢F ⟶♢G" proof - have"⊨◻(F ⟶ G) ⟶◻(¬G ⟶¬F)"by (rule STL4) auto with MM0[of "TEMP ¬G""TEMP ¬F"] show ?thesis unfolding eventually_def by force qed
theorem E10[simp_unl]: "⊨♢♢F = ♢F" by (simp add: eventually_def)
theorem E22: assumes h: "⊨ F = G" shows"⊨♢F = ♢G" by (auto simp: h[int_rewrite])
theorem E15[simp_unl]: "⊨♢#F = #F" by (simp add: eventually_def)
theorem E15b[simp_unl]: "⊨♢¬#F = ¬#F" by (simp add: eventually_def)
theorem E16: "⊨♢◻F ⟶♢F" by (rule STL4_eve[OF ax1])
lemma E23: "|~ ◯F ⟶♢F" using P1 by (force simp: eventually_def)
lemma E24: "⊨♢◻Q ⟶◻[♢Q]_v" by (rule lift_imp_trans[OF E20 P4])
lemma E25: "⊨♢⟨A⟩_v ⟶♢A" using P4 by (force simp: eventually_def angle_action_def)
lemma E26: "⊨◻♢⟨A⟩_v ⟶◻♢A" by (rule STL4[OF E25])
lemma allBox: "(s ⊨◻(∀x. F x)) = (∀x. s ⊨◻(F x))" unfolding allT[unlifted] ..
lemma E29: "|~ ◯♢F ⟶♢F" unfolding eventually_def using pax3 by force
lemma E30: assumes h1: "⊨ F ⟶◻F"and h2: "⊨♢F" shows"⊨♢◻F" using h2 h1[THEN STL4_eve] by (rule fmp)
lemma E31: "⊨◻(F ⟶◻F) ∧♢F ⟶♢◻F" proof - have"⊨◻(F ⟶◻F) ∧♢F ⟶♢(◻(F ⟶◻F) ∧ F)"by (rule SE1) moreover have"⊨◻(F ⟶◻F) ∧ F ⟶◻F"using ax1[of "TEMP F ⟶◻F"] by auto hence"⊨♢(◻(F ⟶◻F) ∧ F) ⟶♢◻F"by (rule STL4_eve) ultimatelyshow ?thesis by (rule lift_imp_trans) qed
lemma allActBox: "(s ⊨◻[(∀x. F x)]_v) = (∀x. s ⊨◻[(F x)]_v)" unfolding allActT[unlifted] ..
theorem exEE: "⊨ (∃x. ♢(F x)) = ♢(∃x. F x)" proof - have"⊨¬(∃ x. ♢(F x)) = ¬♢(∃ x. F x)" by (auto simp: eventually_def Not_Rex[int_rewrite] allBox) thus ?thesis by force qed
theorem exActE: "⊨ (∃x. ♢⟨F x⟩_v) = ♢⟨(∃x. F x)⟩_v" proof - have"⊨¬(∃x. ♢⟨F x⟩_v) = ¬♢⟨(∃x. F x)⟩_v" by (auto simp: angle_action_def Not_Rex[int_rewrite] allActBox) thus ?thesis by force qed
subsection"Theorems about the leadsto operator"
theorem LT1: "⊨ F ↝ F" unfolding leadsto_def by (rule alw[OF E3])
theorem LT2: assumes h: "⊨ F ⟶ G"shows"⊨ F ⟶♢G" by (rule lift_imp_trans[OF h E3])
theorem LT3: assumes h: "⊨ F ⟶ G"shows"⊨ F ↝ G" unfolding leadsto_def by (rule alw[OF LT2[OF h]])
theorem LT4: "⊨ F ⟶ (F ↝ G) ⟶♢G" unfolding leadsto_def using ax1[of "TEMP F ⟶♢G"] by auto
theorem LT5: "⊨◻(F ⟶♢G) ⟶♢F ⟶♢G" using E9[of "F""TEMP ♢G"] by simp
theorem LT6: "⊨♢F ⟶ (F ↝ G) ⟶♢G" unfolding leadsto_def using LT5[of "F""G"] by auto
theorem LT10: assumes h: "⊨ (F ∧¬G) ↝ G" shows"⊨ F ↝ G" proof - from h have"⊨ ((F ∧¬G) ∨ G) ↝ G" by (auto simp: LT17[int_rewrite] LT1[int_rewrite]) moreover have"⊨ F ↝ ((F ∧¬G) ∨ G)"by (rule LT3, auto) ultimately show ?thesis by (force elim: LT13[unlift_rule]) qed
theorem LT18: "⊨ (A ↝ (B ∨ C)) ⟶ (B ↝ D) ⟶ (C ↝ D) ⟶ (A ↝ D)" proof - have"⊨ (B ↝ D) ⟶ (C ↝ D) ⟶ ((B ∨ C) ↝ D)"by (rule LT16) thus ?thesis by (force elim: LT13[unlift_rule]) qed
theorem LT19: "⊨ (A ↝ (D ∨ B)) ⟶ (B ↝ D) ⟶ (A ↝ D)" using LT18[of "A""D""B""D"] LT1[of "D"] by force
theorem LT20: "⊨ (A ↝ (B ∨ D)) ⟶ (B ↝ D) ⟶ (A ↝ D)" using LT18[of "A""B""D""D"] LT1[of "D"] by force
theorem LT21: "⊨ ((∃x. F x) ↝ G) = (∀x. (F x ↝ G))" proof - have"⊨◻((∃x. F x) ⟶♢G) = ◻(∀x. (F x ⟶♢G))"by (rule MM1) auto thus ?thesis by (unfold leadsto_def allT[int_rewrite]) qed
theorem LT22: "⊨ (F ↝ (G ∨ H)) ⟶◻¬G ⟶ (F ↝ H)" proof - have"⊨◻¬G ⟶ (G ↝ H)"unfolding leadsto_def by (rule STL4) auto thus ?thesis by (force elim: LT20[unlift_rule]) qed
theorem LT24: "⊨◻I ⟶ ((P ∧ I) ↝ Q) ⟶ P ↝ Q" proof - have"⊨◻I ⟶◻((P ∧ I ⟶♢Q) ⟶ (P ⟶♢Q))"by (rule STL4) auto from lift_imp_trans[OF this MM0] show ?thesis by (unfold leadsto_def) qed
lemma LT28: assumes h: "|~ P ⟶◯P ∨◯Q" shows"|~ (P ⟶◯P) ∨♢Q" using h E23[of Q] by force
lemma LT29: assumes h1: "|~ P ⟶◯P ∨◯Q"and h2: "|~ P ∧ Unchanged v ⟶◯P" shows"⊨ P ⟶◻P ∨♢Q" proof - from h1[THEN LT28] have"|~ ◻¬Q ⟶ (P ⟶◯P)"unfolding eventually_def by auto hence"⊨◻[◻¬Q]_v ⟶◻[P ⟶◯P]_v"by (rule M2) moreover have"⊨¬♢Q ⟶◻[◻¬Q]_v"unfolding dualization_rew by (rule ax2) moreover note ax3[OF h2] ultimately show ?thesis by force qed
lemma LT30: assumes h: "|~ P ∧ N ⟶◯P ∨◯Q" shows"|~ N ⟶ (P ⟶◯P) ∨♢Q" using h E23 by force
lemma LT31: assumes h1: "|~ P ∧ N ⟶◯P ∨◯Q"and h2: "|~ P ∧ Unchanged v ⟶◯P" shows"⊨◻N ⟶ P ⟶◻P ∨♢Q" proof - from h1[THEN LT30] have"|~ N ⟶◻¬Q ⟶ P ⟶◯P"unfolding eventually_def by auto hence"⊨◻[N ⟶◻¬Q ⟶ P ⟶◯P]_v"by (rule sq) hence"⊨◻[N]_v ⟶◻[◻¬Q]_v ⟶◻[P ⟶◯P]_v" by (force intro: ax4[unlift_rule]) with P4 have"⊨◻N ⟶◻[◻¬Q]_v ⟶◻[P ⟶◯P]_v"by (rule lift_imp_trans) moreover have"⊨¬♢Q ⟶◻[◻¬Q]_v"unfolding dualization_rew by (rule ax2) moreover note ax3[OF h2] ultimately show ?thesis by force qed
lemma LT33: "⊨ ((#P ∧ F) ↝ G) = (#P ⟶ (F ↝ G))" by (cases "P", auto simp: leadsto_def)
lemma AA1: "⊨◻[#False]_v ⟶¬♢⟨Q⟩_v" unfolding dualization_rew by (rule M2) auto
lemma AA3: "⊨◻P ∧◻[P ⟶ Q]_v ∧♢⟨A⟩_v ⟶♢Q" proof - have"⊨◻P ∧◻[P ⟶ Q]_v ⟶◻[P ∧ (P ⟶ Q)]_v" by (auto dest: P4[unlift_rule] simp: M8[int_rewrite]) moreover have"⊨◻[P ∧ (P ⟶ Q)]_v ⟶◻[Q]_v"by (rule M2) auto ultimatelyhave"⊨◻P ∧◻[P ⟶ Q]_v ⟶◻[Q]_v"by (rule lift_imp_trans) moreover have"⊨♢(Q ∧ A) ⟶♢Q"by (rule STL4_eve) auto hence"⊨♢⟨Q ∧ A⟩_v ⟶♢Q"by (force dest: E25[unlift_rule]) with AA2 have"⊨◻[Q]_v ∧♢⟨A⟩_v ⟶♢Q"by (rule lift_imp_trans) ultimatelyshow ?thesis by force qed
lemma AA4: "⊨♢⟨⟨A⟩_v⟩_w ⟶♢⟨⟨A⟩_w⟩_v" unfolding angle_action_def angle_actrans_def using T5 by force
lemma AA7: assumes h: "|~ F ⟶ G"shows"⊨♢⟨F⟩_v ⟶♢⟨G⟩_v" proof - from h have"⊨◻[¬G]_v ⟶◻[¬F]_v"by (intro M2) auto thus ?thesis unfolding angle_action_def by force qed
lemma AA6: "⊨◻[P ⟶ Q]_v ∧♢⟨P⟩_v ⟶♢⟨Q⟩_v" proof - have"⊨♢⟨(P ⟶ Q) ∧ P⟩_v ⟶♢⟨Q⟩_v"by (rule AA7) auto with AA2 show ?thesis by (rule lift_imp_trans) qed
lemma AA8: "⊨◻[P]_v ∧♢⟨A⟩_v ⟶♢⟨◻[P]_v ∧ A⟩_v" proof - have"⊨◻[◻[P]_v]_v ∧♢⟨A⟩_v ⟶♢⟨◻[P]_v ∧ A⟩_v"by (rule AA2) with P5 show ?thesis by force qed
lemma AA10: "⊨¬(◻[P]_v ∧♢⟨¬P⟩_v)" unfolding angle_action_def by auto
lemma AA11: "⊨¬♢⟨v$ = $v⟩_v" unfolding dualization_rew by (rule ax5)
lemma AA15: "⊨♢⟨P ∧ Q⟩_v ⟶♢⟨P⟩_v" by (rule AA7) auto
lemma AA16: "⊨♢⟨P ∧ Q⟩_v ⟶♢⟨Q⟩_v" by (rule AA7) auto
lemma AA13: "⊨♢⟨P⟩_v ⟶♢⟨v$ ≠ $v⟩_v" proof - have"⊨◻[v$ ≠ $v]_v ∧♢⟨P⟩_v ⟶♢⟨v$ ≠ $v ∧ P⟩_v"by (rule AA2) hence"⊨♢⟨P⟩_v ⟶♢⟨v$ ≠ $v ∧ P⟩_v"by (simp add: ax5[int_rewrite]) from this AA15 show ?thesis by (rule lift_imp_trans) qed
lemma AA14: "⊨♢⟨P ∨ Q⟩_v = (♢⟨P⟩_v ∨♢⟨Q⟩_v)" proof - have"⊨◻[¬(P ∨ Q)]_v = ◻[¬P ∧¬Q]_v"by (rule MM10) auto hence"⊨◻[¬(P ∨ Q)]_v = (◻[¬P]_v ∧◻[¬Q]_v)"by (unfold M8[int_rewrite]) thus ?thesis unfolding angle_action_def by auto qed
lemma AA17: "⊨♢⟨[P]_v ∧ A⟩_v ⟶♢⟨P ∧ A⟩_v" proof - have"⊨◻[v$ ≠ $v ∧¬(P ∧ A)]_v ⟶◻[¬([P]_v ∧ A)]_v" by (rule M2) (auto simp: actrans_def unch_def) with ax5[of "v"] show ?thesis unfolding angle_action_def M8[int_rewrite] by force qed
lemma AA19: "⊨◻P ∧♢⟨A⟩_v ⟶♢⟨P ∧ A⟩_v" using P4 by (force intro: AA2[unlift_rule])
lemma AA20: assumes h1: "|~ P ⟶◯P ∨◯Q" and h2: "|~ P ∧ A ⟶◯Q" and h3: "|~ P ∧ Unchanged w ⟶◯P" shows"⊨◻(◻P ⟶♢⟨A⟩_v) ⟶ (P ↝ Q)" proof - from h2 E23 have"|~ P ∧ A ⟶♢Q"by force hence"⊨♢⟨P ∧ A⟩_v ⟶♢⟨♢Q⟩_v"by (rule AA7) with E25[of "TEMP ♢Q""v"] have"⊨♢⟨P ∧ A⟩_v ⟶♢Q"by force with AA19 have"⊨◻P ∧♢⟨A⟩_v ⟶♢Q"by (rule lift_imp_trans) with LT29[OF h1 h3] have"⊨ (◻P ⟶♢⟨A⟩_v) ⟶ (P ⟶♢Q)"by force thus ?thesis unfolding leadsto_def by (rule STL4) qed
lemma AA21: "|~ ♢⟨◯F⟩_v ⟶◯♢F" using pax5[of "TEMP ¬F""v"] unfolding angle_action_def eventually_def by auto
theorem AA24[simp_unl]: "⊨♢⟨⟨P⟩_f⟩_f = ♢⟨P⟩_f" unfolding angle_action_def angle_actrans_def by simp
lemma AA22: assumes h1: "|~ P ∧ N ⟶◯P ∨◯Q" and h2: "|~ P ∧ N ∧⟨A⟩_v ⟶◯Q" and h3: "|~ P ∧ Unchanged w ⟶◯P" shows"⊨◻N ∧◻(◻P ⟶♢⟨A⟩_v) ⟶ (P ↝ Q)" proof - from h2 have"|~ ⟨(N ∧ P) ∧ A⟩_v ⟶◯Q"by (auto simp: angle_actrans_sem[int_rewrite]) from pref_imp_trans[OF this E23] have"⊨♢⟨⟨(N ∧ P) ∧ A⟩_v⟩_v ⟶♢⟨♢Q⟩_v"by (rule AA7) hence"⊨♢⟨(N ∧ P) ∧ A⟩_v ⟶♢Q"by (force dest: E25[unlift_rule]) with AA19 have"⊨◻(N ∧ P) ∧♢⟨A⟩_v ⟶♢Q"by (rule lift_imp_trans) hence"⊨◻N ∧◻P ∧♢⟨A⟩_v ⟶♢Q"by (auto simp: STL5[int_rewrite]) with LT31[OF h1 h3] have"⊨◻N ∧ (◻P ⟶♢⟨A⟩_v) ⟶ (P ⟶♢Q)"by force hence"⊨◻(◻N ∧ (◻P ⟶♢⟨A⟩_v)) ⟶◻(P ⟶♢Q)"by (rule STL4) thus ?thesis by (simp add: leadsto_def STL5[int_rewrite]) qed
lemma AA23: assumes"|~ P ∧ N ⟶◯P ∨◯Q" and"|~ P ∧ N ∧⟨A⟩_v ⟶◯Q" and"|~ P ∧ Unchanged w ⟶◯P" shows"⊨◻N ∧◻♢⟨A⟩_v ⟶ (P ↝ Q)" proof - have"⊨◻♢⟨A⟩_v ⟶◻(◻P ⟶♢⟨A⟩_v)"by (rule STL4) auto with AA22[OF assms] show ?thesis by force qed
lemma AA25: assumes h: "|~ ⟨P⟩_v ⟶⟨Q⟩_w" shows"⊨♢⟨P⟩_v ⟶♢⟨Q⟩_w" proof - from h have"⊨♢⟨⟨P⟩_v⟩_v ⟶♢⟨⟨P⟩_w⟩_v" by (intro AA7) (auto simp: angle_actrans_def actrans_def) with AA4 have"⊨♢⟨P⟩_v ⟶♢⟨⟨P⟩_v⟩_w"by force from this AA7[OF h] have"⊨♢⟨P⟩_v ⟶♢⟨⟨Q⟩_w⟩_w"by (rule lift_imp_trans) thus ?thesis by simp qed
lemma AA26: assumes h: "|~ ⟨A⟩_v = ⟨B⟩_w" shows"⊨♢⟨A⟩_v = ♢⟨B⟩_w" proof - from h have"|~ ⟨A⟩_v ⟶⟨B⟩_w"by auto hence"⊨♢⟨A⟩_v ⟶♢⟨B⟩_w"by (rule AA25) moreover from h have"|~ ⟨B⟩_w ⟶⟨A⟩_v"by auto hence"⊨♢⟨B⟩_w ⟶♢⟨A⟩_v"by (rule AA25) ultimately show ?thesis by force qed
theorem AA28[simp_unl]: "⊨♢♢⟨A⟩_v = ♢⟨A⟩_v" unfolding eventually_def angle_action_def by simp
lemma next_const[simp_unl]: "|~ ◯#P = #P" proof (cases "P") assume"P" hence1: "⊨ #P"by auto hence"|~ ◯#P"by (rule nex) with1show ?thesis by force next assume"¬P" hence1: "⊨¬#P"by auto hence"|~ ◯¬#P"by (rule nex) with1show ?thesis by force qed
text‹
The following are proved semantically because they are essentially
first-order theorems. › lemma next_fun1: "|~ ◯f<x> = f<◯x>" by (auto simp: nexts_def)
lemma next_fun2: "|~ ◯f<x,y> = f<◯x,◯y>" by (auto simp: nexts_def)
lemma next_fun3: "|~ ◯f<x,y,z> = f<◯x,◯y,◯z>" by (auto simp: nexts_def)
lemma next_fun4: "|~ ◯f<x,y,z,zz> = f<◯x,◯y,◯z,◯zz>" by (auto simp: nexts_def)
lemma next_forall: "|~ ◯(∀ x. P x) = (∀ x. ◯ P x)" by (auto simp: nexts_def)
lemma next_exists: "|~ ◯(∃ x. P x) = (∃ x. ◯ P x)" by (auto simp: nexts_def)
lemma next_exists1: "|~ ◯(∃! x. P x) = (∃! x. ◯ P x)" by (auto simp: nexts_def)
text‹
Rewrite rules to push the ``next'' operator inward over connectives.
(Note that axiom ‹pax1› and theorem ‹next_const› are anyway active
as rewrite rules.) › lemmas next_commutes[int_rewrite] =
next_and next_or next_imp next_eq
next_fun1 next_fun2 next_fun3 next_fun4
next_forall next_exists next_exists1
text‹
Theorem ‹next_eventually› should not be used "blindly". › lemma next_eventually: assumes h: "stutinv F" shows"|~ ♢F ⟶¬F ⟶◯♢F" proof - from h have1: "stutinv (TEMP ¬F)"by (rule stut_not) have"|~ ◻¬F = (¬F ∧◯◻¬F)"unfolding T7[OF pre_id_unch[OF 1], int_rewrite] by simp thus ?thesis by (auto simp: eventually_def) qed
lemma next_action: "|~ ◻[P]_v ⟶◯◻[P]_v" using pax4[of P v] by auto
subsection"Higher Level Derived Rules"
text‹
In most verification tasks the low-level rules discussed above are not used directly.
Here, we derive some higher-level rules more suitable for verification. In particular,
variants of Lamport's rules ‹TLA1›, ‹TLA2›, ‹INV1› and ‹INV2›
are derived, where ‹|~› is used where appropriate. ›
theorem TLA1: assumes H: "|~ P ∧ Unchanged f ⟶◯P" shows"⊨◻P = (P ∧◻[P ⟶◯P]_f)" proof (rule int_iffI) from ax1[of P] M0[of P f] show"⊨◻P ⟶ P ∧◻[P ⟶◯P]_f"by force next from ax3[OF H] show"⊨ P ∧◻[P ⟶◯P]_f ⟶◻P"by auto qed
theorem TLA2: assumes h1: "⊨ P ⟶ Q" and h2: "|~ P ∧◯P ∧ [A]_f ⟶ [B]_g" shows"⊨◻P ∧◻[A]_f ⟶◻Q ∧◻[B]_g" proof - from h2 have"⊨◻[P ∧◯P ∧ [A]_f]_g ⟶◻[[B]_g]_g"by (rule M2) hence"⊨◻[P ∧◯P]_g ∧◻[[A]_f]_g ⟶◻[B]_g"by (auto simp add: M8[int_rewrite]) with M1[of P g] T4[of A f g] have"⊨◻P ∧◻[A]_f ⟶◻[B]_g"by force with h1[THEN STL4] show ?thesis by force qed
theorem INV1: assumes H: "|~ I ∧ [N]_f ⟶◯I" shows"⊨ I ∧◻[N]_f ⟶◻I" proof - from H have"|~ [N]_f ⟶ I ⟶◯I"by auto hence"⊨◻[[N]_f]_f ⟶◻[I ⟶◯I]_f"by (rule M2) moreover from H have"|~ I ∧ Unchanged f ⟶◯I"by (auto simp: actrans_def) hence"⊨◻[I ⟶◯I]_f ⟶ I ⟶◻I"by (rule ax3) ultimatelyshow ?thesis by force qed
theorem INV2: "⊨◻I ⟶◻[N]_f = ◻[N ∧ I ∧◯I]_f" proof - from M1[of I f] have"⊨◻I ⟶ (◻[N]_f = ◻[N]_f ∧◻[I ∧◯I]_f)"by auto thus ?thesis by (auto simp: M8[int_rewrite]) qed
lemma R1: assumes H: "|~ Unchanged w ⟶ Unchanged v" shows"⊨◻[F]_w ⟶◻[F]_v" proof - from H have"|~ [F]_w ⟶ [F]_v"by (auto simp: actrans_def) thus ?thesis by (rule M11) qed
theorem invmono: assumes h1: "⊨ I ⟶ P" and h2: "|~ P ∧ [N]_f ⟶◯P" shows"⊨ I ∧◻[N]_f ⟶◻P" using h1 INV1[OF h2] by force
theorem preimpsplit: assumes"|~ I ∧ N ⟶ Q" and"|~ I ∧ Unchanged v ⟶ Q" shows"|~ I ∧ [N]_v ⟶ Q" using assms[unlift_rule] by (auto simp: actrans_def)
theorem refinement1: assumes h1: "⊨ P ⟶ Q" and h2: "|~ I ∧◯I ∧ [A]_f ⟶ [B]_g" shows"⊨ P ∧◻I ∧◻[A]_f ⟶ Q ∧◻[B]_g" proof - have"⊨ I ⟶ #True"by simp from this h2 have"⊨◻I ∧◻[A]_f ⟶◻#True ∧◻[B]_g"by (rule TLA2) with h1 show ?thesis by force qed
theorem inv_join: assumes"⊨ P ⟶◻Q"and"⊨ P ⟶◻R" shows"⊨ P ⟶◻(Q ∧ R)" using assms[unlift_rule] unfolding STL5[int_rewrite] by force
lemma inv_cases: "⊨◻(A ⟶ B) ∧◻(¬A ⟶ B) ⟶◻B" proof - have"⊨◻((A ⟶ B) ∧ (¬A ⟶ B)) ⟶◻B"by (rule STL4) auto thus ?thesis by (simp add: STL5[int_rewrite]) qed
end
Messung V0.5 in Prozent
¤ Dauer der Verarbeitung: 0.34 Sekunden
(vorverarbeitet am 2026-06-10)
¤
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.