math / data /lean /OAI /Probability /SATComputability /EffectiveSums.lean
introvoyz045's picture
Migrated from GitHub
4a56cd6 verified
Raw History Blame Contribute Delete
2.08 kB
import OAI.Probability.SATComputability.EffectiveLog
namespace OAI
namespace PeriodicLattice.CertifiedReal
open Encodable Filter
open scoped Topology BigOperators
local instance effectiveSumsRatPrimcodable : Primcodable ℚ := RecursiveArithmetic.ratPrimcodable
variable {A : Type*} [Primcodable A]
theorem Effective.sum_range {f : A → ℕ → ℝ}
(hf : Effective (fun p : A × ℕ => f p.1 p.2))
{m : A → ℕ} (hm : Computable m) :
Effective (fun a => ∑ i ∈ Finset.range (m a), f a i) := by
obtain ⟨q, hq, herror⟩ := hf
apply of_enclosures
(q := fun an => ∑ i ∈ Finset.range (m an.1), q ((an.1,i),an.2))
(e := fun an => (m an.1 : ℚ)*qerror an.2)
· apply computable_sum (hm.comp Computable.fst)
unfold Computable₂
fun_prop
· fun_prop
· intro a n
simp only [Rat.cast_sum, Rat.cast_mul, Rat.cast_natCast, cast_qerror]
rw [← Finset.sum_sub_distrib]
calc
_ ≤ ∑ i ∈ Finset.range (m a), |f a i - (q ((a,i),n) : ℝ)| :=
Finset.abs_sum_le_sum_abs _ _
_ ≤ ∑ _i ∈ Finset.range (m a), error n :=
Finset.sum_le_sum (fun i _ => herror (a,i) n)
_ = _ := by simp
· intro a
simpa only [Rat.cast_mul, Rat.cast_natCast, cast_qerror, mul_zero] using
error_tendsto.const_mul (m a : ℝ)
theorem effective_limit {f : A → ℝ} {g : A → ℕ → ℝ}
(hg : Effective (fun p : A × ℕ => g p.1 p.2))
(tail : A × ℕ → ℚ) (htail : Computable tail)
(hbound : ∀ a n, |f a - g a n| ≤ (tail (a,n) : ℝ))
(hzero : ∀ a, Tendsto (fun n => (tail (a,n) : ℝ)) atTop (nhds 0)) :
Effective f := by
obtain ⟨q, hq, he⟩ := hg
apply of_enclosures (q := fun an => q ((an.1,an.2),an.2))
(e := fun an => tail an + qerror an.2)
· fun_prop
· fun_prop
· intro a n
simp only [Rat.cast_add, cast_qerror]
exact (abs_sub_le _ _ _).trans (add_le_add (hbound a n) (he (a,n) n))
· intro a
simpa only [Rat.cast_add, cast_qerror, add_zero] using (hzero a).add error_tendsto
end PeriodicLattice.CertifiedReal
end OAI