Files
agda-spa/lean/Spa/Analysis/Forward.lean

154 lines
6.3 KiB
Lean4
Raw Normal View History

import Spa.Analysis.Forward.Lattices
import Spa.Analysis.Forward.Evaluation
import Spa.Analysis.Forward.Adapters
import Spa.Fixedpoint
namespace Spa
namespace Forward
variable {L : Type} [FiniteHeightLattice L] {prog : Program} [E : StmtEvaluator L prog]
def updateVariablesForState (s : prog.State) (sv : StateVariables L prog) :
VariableValues L prog := E.eval s (variablesAt s sv)
lemma updateVariablesForState_mono (s : prog.State) :
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
Monotone (updateVariablesForState (L := L) s) := fun _ _ hle =>
E.eval_mono s (variablesAt_le hle s)
def updateAll (sv : StateVariables L prog) : StateVariables L prog :=
FiniteMap.generalizedUpdate id updateVariablesForState
prog.states sv
lemma updateAll_mono : Monotone (updateAll (L := L) (prog := prog)) :=
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
FiniteMap.generalizedUpdate_monotone monotone_id updateVariablesForState_mono
lemma updateAll_mem_eq {s : prog.State} {vs : VariableValues L prog}
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
{sv : StateVariables L prog} (hmem : (s, vs) updateAll sv) :
vs = updateVariablesForState s sv :=
FiniteMap.generalizedUpdate_mem_eq (prog.states_complete s) hmem
lemma variablesAt_updateAll (s : prog.State) (sv : StateVariables L prog) :
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
variablesAt s (updateAll sv) = updateVariablesForState s sv :=
updateAll_mem_eq (variablesAt_mem s (updateAll sv))
def analyze (sv : StateVariables L prog) : StateVariables L prog :=
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
updateAll (joinAll sv)
lemma analyze_mono : Monotone (analyze (L := L) (prog := prog)) := fun _ _ hle =>
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
updateAll_mono (joinAll_mono hle)
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
variable [DecidableEq L]
variable (L prog) in
def result : StateVariables L prog :=
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
Fixedpoint.aFix analyze analyze_mono
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
variable (L prog) in
lemma result_eq : result L prog = analyze (result L prog) :=
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
Fixedpoint.aFix_eq analyze analyze_mono
lemma joinForKey_initialState :
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
joinForKey prog.initialState (result L prog) = botV L prog := by
rw [joinForKey, prog.incoming_initialState_eq_nil]
rfl
class ValidStateEvaluator (L : Type) [FiniteHeightLattice L] (prog : Program)
[E : StmtEvaluator L prog] [S : StateInterpretation L prog] where
valid : (s₁ s₂ : prog.State) {ρ₁ ρ₂ ρ₃: Env}
{vs : VariableValues L prog},
(tr : Traceₗ prog.cfg s₁ s₂ ρ₁ ρ₂)
(hbs : EvalBasicStmtOpt ρ₂ (prog.cfg.nodes s₂) ρ₃) vs (S.Pre tr)
E.eval s₂ vs (S.Post (tr ++ hbs))
botV_init : botV L prog (S.Pre (Traceₗ.single prog.cfg prog.initialState []))
instance [LatticeInterpretation L] [ValidStmtEvaluator L prog] :
ValidStateEvaluator L prog where
valid := by intro _ _ _ _ _ _ tr hbs hvs; exact ValidStmtEvaluator.valid hbs hvs
botV_init := by intro k l _ v hmem; cases hmem
section
variable [S : StateInterpretation L prog] [V : ValidStateEvaluator L prog]
omit [DecidableEq L] in
lemma updateAll_matches {s₁ s₂ : prog.State} {sv : StateVariables L prog}
{ρ₁ ρ₂ ρ₃ : Env}
(tr : Traceₗ prog.cfg s₁ s₂ ρ₁ ρ₂)
(hnode : EvalBasicStmtOpt ρ₂ (prog.code s₂) ρ₃)
(hvs : variablesAt s₂ sv (S.Pre tr)) :
variablesAt s₂ (updateAll sv) (S.Post (tr ++ hnode)) := by
rw [variablesAt_updateAll]
exact V.valid s₁ s₂ tr hnode hvs
lemma stepTrace {s₁ s₂ : prog.State} {ρ₁ ρ₂ : Env}
(tr : Traceₗ prog.cfg s₁ s₂ ρ₁ ρ₂)
(hjoin : joinForKey s₂ (result L prog) (S.Pre tr))
(hnode : EvalBasicStmtOpt ρ₂ (prog.code s₂) ρ₃) :
variablesAt s₂ (result L prog) (S.Post (tr ++ hnode)) := by
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
rw [result_eq L prog]
refine updateAll_matches tr hnode ?_
rw [variablesAt_joinAll]
exact hjoin
2026-08-09 21:23:21 -05:00
/-- Soundness propagates along an execution prefix: if the analysis is sound at
`s₂` for the run so far (`trₗ`), then it is sound wherever the further prefix
`mid` ends up. -/
lemma walkPrefix : {s₂ s : prog.State} {ρ₂ ρin : Env}
(mid : Traceₗ prog.cfg s₂ s ρ₂ ρin) {s₁ : prog.State} {ρ₁ : Env}
(trₗ : Traceₗ prog.cfg s₁ s₂ ρ₁ ρ₂),
joinForKey s₂ (result L prog) (S.Pre trₗ)
joinForKey s (result L prog) (S.Pre (trₗ ++ mid)) := by
intro s₂ s ρ₂ ρin mid
induction mid with
| nil => intro s₁ ρ₁ trₗ hjoin; simpa [HAppend.hAppend, Traceₗ.append] using hjoin
| cons hnode hedge rest ih =>
intro s₁ ρ₁ trₗ hjoin
have hstep := stepTrace trₗ hjoin hnode
have hmem := FiniteMap.mem_valuesAt prog.states_nodup
(prog.mem_incoming_of_edge hedge) (variablesAt_mem _ (result L prog))
simpa [HAppend.hAppend, Traceₗ.append] using
ih ((trₗ ++ hnode).addEdge hedge)
(interp_foldr (S.post_pre (trₗ ++ hnode) hedge hstep) hmem)
omit [DecidableEq L] in
/-- The final node of a trace is always reached, with the environment/state the trace
2026-08-09 21:23:21 -05:00
ends in. Used to recover the final-state soundness theorem from `walkPrefix`. -/
def reaches_final {s : prog.State} {ρ : Env}
(tr : Trace prog.cfg prog.initialState s [] ρ) : Σ ρin, Reaches s ρin ρ :=
_, tr.split.2.1, tr.split.2.2
omit [DecidableEq L] in
2026-08-09 21:23:21 -05:00
@[simp] lemma reaches_final_post {s : prog.State} {ρ : Env}
(tr : Trace prog.cfg prog.initialState s [] ρ) :
(reaches_final tr).2.post = tr := Trace.split_append tr
variable (L prog) in
2026-08-09 21:23:21 -05:00
/-- Soundness at every program point an execution actually visits: the analysis
over-approximates both the environment entering that point and the one leaving
it. -/
theorem analyze_correct_at {s : prog.State} {ρin ρout : Env} (hr : Reaches s ρin ρout) :
joinForKey s (result L prog) (S.Pre hr.pre)
2026-08-09 21:23:21 -05:00
variablesAt s (result L prog) (S.Post hr.post) :=
have hpre := walkPrefix hr.pre Traceₗ.nil
(by rw [joinForKey_initialState]; exact ValidStateEvaluator.botV_init)
hpre, stepTrace hr.pre hpre hr.step
variable (L prog) in
theorem analyze_correct'
{ρ : Env} (hrun : EvalStmt [] prog.rootStmt ρ) :
variablesAt prog.finalState (result L prog) (S.Post (prog.trace hrun)) := by
2026-08-09 21:23:21 -05:00
have h := (analyze_correct_at L prog (reaches_final (prog.trace hrun)).2).2
rwa [reaches_final_post] at h
end
Lean migration: typeclass-based parameter passing, as in the Agda original The port had flattened Agda's instance arguments ({{flA}}, {{evaluator}}, {{latticeInterpretation}}, {{validEvaluator}}) into explicitly threaded values (fhL, E, I, hE). Restore them as typeclasses: - Spa.FiniteHeightLattice: now actually used — Fixedpoint takes the instance instead of a FixedHeight value; FiniteMap gets the missing instance (height = ks.length * height B), so varsFixedHeight / statesFixedHeight / signFixedHeight / constFixedHeight plumbing disappears (instance bottoms are defeq to the old ones) - Spa.Analysis.Forward.Evaluation: StmtEvaluator/ExprEvaluator become classes; the Valid* Props become Prop-classes, as in Agda - Spa.Analysis.Forward.Adapters: the expr→stmt adapter and its validity are instances (Agda: the ExprToStmtAdapter instances) - LatticeInterpretation is a class; sign/const interpretations, evaluators and validity proofs are instances; use sites read like the Agda module applications: result SignLattice prog Proof simplifications (same theorems, proofs factored): - Spa.Lattice.AboveBelow.monotone₂_of_strict: any ⊥-strict/⊤-dominated operation on a flat lattice is monotone — replaces the four near- identical case bashes per analysis (postulates in Agda) - Spa.Lattice.AboveBelow.interp_sup_of/interp_inf_of: the shared flat- lattice interpretation case analysis, making interpSign_sup/inf and interpConst_sup/inf one-liners lake build green with zero warnings; lake exe spa output verified byte-identical (diff) to the previous, Agda-verified output. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-09 23:32:38 -07:00
variable (L prog) in
theorem analyze_correct [LatticeInterpretation L] [ValidStmtEvaluator L prog]
{ρ : Env} (hrun : EvalStmt [] prog.rootStmt ρ) :
variablesAt prog.finalState (result L prog) ρ :=
analyze_correct' L prog hrun
end Forward
end Spa