Tv Dist bind left le const
tvDist_bind_left_le_const
Plain-language statement
Total-variation distance is convex over a shared bind: if tvDist (f a) (g a) ≤ c for every a ∈ support mx, then tvDist (mx >>= f) (mx >>= g) ≤ c. The real-valued root of the const bound, with ℝ≥0∞ companion ofReal_tvDist_bind_left_le_const.
Exact Lean statement
theorem tvDist_bind_left_le_const
{m : Type u → Type v} [Monad m] [LawfulMonad m] [MonadLiftT m PMF] [LawfulMonadLiftT m PMF]
[MonadLiftT m SetM] [EvalDistCompatible m]
{α β : Type u} (mx : m α) (f g : α → m β) (c : ℝ)
(hfg : ∀ a, a ∈ support mx → tvDist (f a) (g a) ≤ c) :
tvDist (mx >>= f) (mx >>= g) ≤ cFormal artifact
Lean source
theorem tvDist_bind_left_le_const {m : Type u → Type v} [Monad m] [LawfulMonad m] [MonadLiftT m PMF] [LawfulMonadLiftT m PMF] [MonadLiftT m SetM] [EvalDistCompatible m] {α β : Type u} (mx : m α) (f g : α → m β) (c : ℝ) (hfg : ∀ a, a ∈ support mx → tvDist (f a) (g a) ≤ c) : tvDist (mx >>= f) (mx >>= g) ≤ c := by classical have hprob_ne_top : ∀ a : α, Pr[= a | mx] ≠ ⊤ := fun a => ne_top_of_le_ne_top one_ne_top (probOutput_le_one (mx := mx) (x := a)) have hp_sum_ne_top : (∑' a : α, Pr[= a | mx]) ≠ ⊤ := by rw [tsum_probOutput_of_liftM_PMF]; exact one_ne_top have hp_summable : Summable (fun a : α => Pr[= a | mx].toReal) := ENNReal.summable_toReal hp_sum_ne_top have hp_sum_toReal : (∑' a : α, Pr[= a | mx].toReal) = 1 := by rw [← ENNReal.tsum_toReal_eq hprob_ne_top, tsum_probOutput_of_liftM_PMF, ENNReal.toReal_one] have hlhs_nonneg : ∀ a : α, 0 ≤ Pr[= a | mx].toReal * tvDist (f a) (g a) := fun _ => mul_nonneg ENNReal.toReal_nonneg (tvDist_nonneg _ _) have hlhs_le_p : ∀ a : α, Pr[= a | mx].toReal * tvDist (f a) (g a) ≤ Pr[= a | mx].toReal := fun _ => mul_le_of_le_one_right ENNReal.toReal_nonneg (tvDist_le_one _ _) have hlhs_summable : Summable (fun a : α => Pr[= a | mx].toReal * tvDist (f a) (g a)) := Summable.of_nonneg_of_le hlhs_nonneg hlhs_le_p hp_summable have hrhs_summable : Summable (fun a : α => Pr[= a | mx].toReal * c) := Summable.mul_right _ hp_summable refine (tvDist_bind_left_le mx f g).trans ?_ calc (∑' a : α, Pr[= a | mx].toReal * tvDist (f a) (g a)) ≤ ∑' a : α, Pr[= a | mx].toReal * c := Summable.tsum_le_tsum (fun a => by by_cases ha : a ∈ support mx · exact mul_le_mul_of_nonneg_left (hfg a ha) ENNReal.toReal_nonneg · rw [probOutput_eq_zero_of_not_mem_support ha]; simp) hlhs_summable hrhs_summable _ = (∑' a : α, Pr[= a | mx].toReal) * c := Summable.tsum_mul_right _ hp_summable _ = c := by rw [hp_sum_toReal, one_mul]- Project
- VCVio
- License
- Apache-2.0
- Commit
- 2ceb2d825ee3
- Source
- VCVio/EvalDist/TVDist.lean:228-264
Reuse this declaration
Bring the exact result into your workflow
The import identifies the source module. Your project still needs the pinned package dependency shown on this page.
What this badge means
This completion status comes from the project or community source. It has not yet been represented here as an independent rebuild and axiom audit.
Continue in this project
Related declarations
Expected Cost Nat eq sum tail probs of pathwise Cost At Most
AddWriterT.expectedCostNat_eq_sum_tail_probs_of_pathwiseCostAtMost
Plain-language statement
Finite tail-sum formula for natural-valued writer cost under a pathwise upper bound. If every execution path of oa incurs cost at most n, then the tail probabilities vanish above n, so the infinite tail sum truncates to Finset.range n.
Source project: VCVio
Person-level attribution pending.
IND CPA advantage to Real le sum step signed Advantage Real abs
AsymmEncAlg.IND_CPA_advantage_toReal_le_sum_step_signedAdvantageReal_abs
Plain-language statement
Planned generic one-time-to-many-time lift: bounded multi-query IND-CPA advantage is at most the sum of the extracted one-time signed advantages over the first q fresh LR queries.
Source project: VCVio
Person-level attribution pending.
IND CPA LR hybrid Game q eval Dist eq left of Makes At Most Queries
AsymmEncAlg.IND_CPA_LR_hybridGame_q_evalDist_eq_left_of_MakesAtMostQueries
Plain-language statement
If an adversary makes at most q fresh LR queries, then the leftUntil = q LR-hybrid is the all-left endpoint game.
Source project: VCVio
Person-level attribution pending.