diff --git a/coq/QLearn/jaakkola_vector.v b/coq/QLearn/jaakkola_vector.v index 01a2d1c6..6f1aeea4 100644 --- a/coq/QLearn/jaakkola_vector.v +++ b/coq/QLearn/jaakkola_vector.v @@ -7016,15 +7016,23 @@ Proof. {isl2 : forall k i pf, IsLp prts 2 (vecrvnth i pf (XF k))} {rvXF : forall k, RandomVariable (F (S k)) (Rvector_borel_sa (S N)) (XF k)} : + (**r α and β are almost always non-negative *) almost prts (fun ω => forall k i pf, 0 <= vector_nth i pf (α k ω)) -> almost prts (fun ω => forall k i pf, 0 <= vector_nth i pf (β k ω)) -> + (**r α is almost always less than or equal to β, component-wise *) almost prts (fun ω => forall k i pf, vector_nth i pf (β k ω) <= vector_nth i pf (α k ω)) -> + (**r The sum of αs almost always increase without limit *) almost prts (fun ω => forall i pf, is_lim_seq (sum_n (fun k => vector_nth i pf (α k ω))) p_infty) -> + (**r The sum of βs almost always increases without limit *) almost prts (fun ω => forall i pf, is_lim_seq (sum_n (fun k => vector_nth i pf (β k ω))) p_infty) -> + + (**r The sum of α²s almost always converges to a finite limit *) (forall i pf, almost prts (fun ω => ex_series (fun n => Rsqr (vector_nth i pf (α n ω))))) -> + (**r The sum of β²s almost always converges to a finite limit *) (forall i pf, almost prts (fun ω => ex_series (fun n => Rsqr (vector_nth i pf (β n ω))))) -> + (exists (γ : R), 0 < γ < 1 /\ almost prts @@ -7032,21 +7040,26 @@ Proof. (forall k, (pos_scaled_Rvector_max_abs ((vector_FiniteConditionalExpectation prts (filt_sub k) (XF k) (rv := vec_rvXF_I filt_sub XF k) (isfe := vec_isfe XF (vec_rvXF_I := vec_rvXF_I filt_sub XF) k)) ω) W) < (γ * (pos_scaled_Rvector_max_abs (X k ω) W))))) -> + (forall i pf, is_lim_seq'_uniform_almost (fun n ω => sum_n (fun k => rvsqr (vecrvnth i pf (α k)) ω) n) (fun ω => Lim_seq (sum_n (fun k => rvsqr (vecrvnth i pf (α k)) ω)))) -> + (forall i pf, is_lim_seq'_uniform_almost (fun n ω => sum_n (fun k => rvsqr (vecrvnth i pf (β k)) ω) n) (fun ω => Lim_seq (sum_n (fun k => rvsqr (vecrvnth i pf (β k)) ω)))) -> - (exists (C : R), + + (exists (C : R), (0 < C) /\ almost prts (fun ω => (forall k i pf, ((FiniteConditionalVariance_new prts (filt_sub k) (vecrvnth i pf (XF k)) (rv := vecrvnth_rv i pf (XF k) (rv := vec_rvXF_I filt_sub XF k))) ω) <= (C * (1 + pos_scaled_Rvector_max_abs (X k ω) W)^2)))) -> + (forall k, rv_eq (X (S k)) (vecrvplus (vecrvminus (X k) (vecrvmult (α k) (X k))) (vecrvmult (β k) (XF k) ))) -> + almost prts (fun ω => forall i pf, is_lim_seq (fun m => vector_nth i pf (X m ω)) 0).