Chapter 04 — Power Series
Proofs omitted. Hover over highlighted terms to see their Lean types.
Reading sections
4.1 — Formal Power Series
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Definition 4.1.1: For a ∈ ℝ, a formal power series centered at a is given by a sequence of real coefficients (c_n)_{n≥0}, corresponding to the formal expression ∑_{n=0}^∞ c_n (x - a)^n in the indeterminate x; its nth coefficient is coeff n.
structure FormalPowerSeriesCenteredAt (a : ℝ) where
coeff : ℕ → ℝ
Definition 4.1.2 (Radius of convergence): for a formal power series ∑_{n=0}^∞ c_n (x-a)^n, let L = limsup_{n→∞} |c_n|^(1/n) (valued in [0, +∞]); the radius of convergence is R = 1 / L, with the conventions 1/0 = +∞ and 1/(+∞) = 0.
noncomputable def FormalPowerSeriesCenteredAt.radiusOfConvergence {a : ℝ}
(f : FormalPowerSeriesCenteredAt a) : ENNReal :=
(FormalMultilinearSeries.ofScalars ℝ f.coeff).radius
Definition 4.1.3: Define the function f : ℝ \ {1} → ℝ by f(x) = 1 / (1 - x).
noncomputable def reciprocalOneSub (x : {x : ℝ // x ≠ 1}) : ℝ :=
1 / (1 - (x : ℝ))
Points of (-1, 1) are distinct from 1.
lemma ne_one_of_mem_Ioo_neg_one_one {x : ℝ} (hx : x ∈ Set.Ioo (-1 : ℝ) 1) : x ≠ 1
The total real extension of x ↦ 1 / (1 - x).
noncomputable def reciprocalOneSub_ext : ℝ → ℝ :=
fun y => 1 / (1 - y)
The total extension agrees with reciprocalOneSub away from 1.
lemma reciprocalOneSub_ext_eq (x : ℝ) (hx : x ≠ 1) :
reciprocalOneSub_ext x = reciprocalOneSub ⟨x, hx⟩
Lemma 4.1.1: For every x ∈ (-1, 1), one has ∑' n, x^n = 1/(1-x) = f(x) (with f = reciprocalOneSub), and in particular f is real analytic at 0.
lemma geometric_series_eq_reciprocalOneSub_and_analyticAt_zero :
(∀ x : ℝ, ∀ hx : x ∈ Set.Ioo (-1 : ℝ) 1,
(∑' n : ℕ, x ^ n) = 1 / (1 - x)
∧ 1 / (1 - x) = reciprocalOneSub_ext x)
∧ AnalyticAt ℝ reciprocalOneSub_ext 0
Points of (1, 3) are distinct from 1.
lemma ne_one_of_mem_Ioo_one_three {x : ℝ} (hx : x ∈ Set.Ioo (1 : ℝ) 3) : x ≠ 1
Lemma 4.1.2: For every x ∈ (1, 3), one has ∑' n, (-1)^(n+1) (x-2)^n = 1/(1-x) = f(x) (with f = reciprocalOneSub); in particular, f is real analytic at 2.
lemma shifted_geometric_series_eq_reciprocalOneSub_and_analyticAt_two :
(∀ x : ℝ, ∀ hx : x ∈ Set.Ioo (1 : ℝ) 3,
(∑' n : ℕ, (-1 : ℝ) ^ (n + 1) * (x - 2) ^ n) = 1 / (1 - x)
∧ 1 / (1 - x) = reciprocalOneSub ⟨x, ne_one_of_mem_Ioo_one_three hx⟩)
∧ AnalyticAt ℝ reciprocalOneSub_ext 2
Theorem 4.1.1: (Divergence outside the radius of convergence) for a power series with center a and coefficients c_n, if |x - a| > R where R is the radius of convergence, then ∑ n, c_n * (x - a)^n diverges at x.
theorem FormalPowerSeriesCenteredAt.diverges_outside_radius {a x : ℝ}
(f : FormalPowerSeriesCenteredAt a)
(hx : f.radiusOfConvergence < ENNReal.ofReal |x - a|) :
¬Summable (fun n : ℕ => f.coeff n * (x - a) ^ n)
Theorem 4.1.2: If |x - a| < R for a power series ∑ n, c_n (x-a)^n with radius of convergence R, then ∑ n, c_n (x-a)^n converges absolutely at x.
theorem FormalPowerSeriesCenteredAt.converges_absolutely_inside_radius {a x : ℝ}
(f : FormalPowerSeriesCenteredAt a)
(hx : ENNReal.ofReal |x - a| < f.radiusOfConvergence) :
Summable (fun n : ℕ => |f.coeff n * (x - a) ^ n|)
Theorem 4.1.3: (Uniform convergence on compact subintervals) for a power series centered at a, if 0 < r < R (with R the radius of convergence), then the partial sums converge uniformly on [a-r, a+r] to the series sum; in particular, the sum function is continuous at every point where |x-a| < R.
theorem FormalPowerSeriesCenteredAt.uniformly_converges_on_compact_subintervals {a r : ℝ}
(f : FormalPowerSeriesCenteredAt a)
(hr0 : 0 < r)
(hrR : ENNReal.ofReal r < f.radiusOfConvergence) :
TendstoUniformlyOn
(fun N x => (Finset.range N).sum (fun n => f.coeff n * (x - a) ^ n))
(fun x => ∑' n, f.coeff n * (x - a) ^ n)
Filter.atTop
(Set.Icc (a - r) (a + r))
∧
∀ x : ℝ,
ENNReal.ofReal |x - a| < f.radiusOfConvergence →
ContinuousAt (fun y => ∑' n, f.coeff n * (y - a) ^ n) x
Theorem 4.1.4: (Term-by-term differentiation of a power series): if a power series centered at a has positive radius of convergence R, then its sum function is differentiable at every point x with |x - a| < R; moreover, for every r with 0 < r < R, the derived series converges uniformly to the derivative on [a - r, a + r].
theorem FormalPowerSeriesCenteredAt.term_by_term_differentiation_of_power_series {a : ℝ}
(f : FormalPowerSeriesCenteredAt a)
(hRpos : 0 < f.radiusOfConvergence) :
DifferentiableOn ℝ
(fun x => ∑' n, f.coeff n * (x - a) ^ n)
{x : ℝ | ENNReal.ofReal |x - a| < f.radiusOfConvergence}
∧
∀ r : ℝ,
0 < r →
ENNReal.ofReal r < f.radiusOfConvergence →
TendstoUniformlyOn
(fun N x => (Finset.range N).sum (fun n => ((n + 1 : ℝ) * f.coeff (n + 1) * (x - a) ^ n))
)
(fun x => deriv (fun y => ∑' n, f.coeff n * (y - a) ^ n) x)
Filter.atTop
(Set.Icc (a - r) (a + r))
Theorem 4.1.5: (Term-by-term integration of a power series) let f(x) = ∑' n, c_n (x-a)^n be a power series with positive radius of convergence; for any closed interval [y, z] contained in the convergence region, ∫ x in y..z, f x equals the term-by-term antiderivative series evaluated at z and y.
theorem FormalPowerSeriesCenteredAt.term_by_term_integration_of_power_series {a y z : ℝ}
(f : FormalPowerSeriesCenteredAt a)
(hRpos : 0 < f.radiusOfConvergence)
(hsub : Set.uIcc y z ⊆ {x : ℝ | ENNReal.ofReal |x - a| < f.radiusOfConvergence}) :
(∫ x in y..z, (∑' n : ℕ, f.coeff n * (x - a) ^ n)) =
∑' n : ℕ,
f.coeff n * (((z - a) ^ (n + 1) - (y - a) ^ (n + 1)) / (n + 1 : ℝ))
Helper for Theorem 4.1.1: radiusOfConvergence is the radius of the associated scalar formal multilinear series.
lemma FormalPowerSeriesCenteredAt.radiusOfConvergence_eq_ofScalarsRadius {a : ℝ}
(f : FormalPowerSeriesCenteredAt a) :
f.radiusOfConvergence = (FormalMultilinearSeries.ofScalars ℝ f.coeff).radius
Helper for Lemma 4.1.2: for x ∈ (1,3), the shifted variable satisfies |x-2| < 1.
lemma helperForLemma_4_1_2_abs_shift_lt_one {x : ℝ} (hx : x ∈ Set.Ioo (1 : ℝ) 3) :
Helper for Lemma 4.1.2: rewrite each shifted alternating term as a negated geometric term.
lemma helperForLemma_4_1_2_alternating_term_as_neg_geometric (x : ℝ) (n : ℕ) :
(-1 : ℝ) ^ (n + 1) * (x - 2) ^ n = -((((-1 : ℝ) * (x - 2)) ^ n))
Helper for Lemma 4.1.2: the shifted alternating geometric series sums to 1/(1-x) on (1,3).
lemma helperForLemma_4_1_2_tsum_identity {x : ℝ} (hx : x ∈ Set.Ioo (1 : ℝ) 3) :
(∑' n : ℕ, (-1 : ℝ) ^ (n + 1) * (x - 2) ^ n) = 1 / (1 - x)
Helper for Lemma 4.1.2: the extension reciprocalOneSub_ext is real-analytic at 2.
lemma helperForLemma_4_1_2_analyticAt_two : AnalyticAt ℝ reciprocalOneSub_ext 2
Helper for Theorem 4.1.3: translate a scalar formal multilinear series to a ball expansion centered at a by composition with subtraction.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_3_shiftedHasFPowerSeriesOnBall
{a : ℝ} (p : FormalMultilinearSeries ℝ ℝ ℝ) (hpos : 0 < p.radius) :
HasFPowerSeriesOnBall (fun x : ℝ => p.sum (x - a)) p a p.radius
Helper for Theorem 4.1.3: pick an intermediate NNReal radius strictly between r and the convergence radius.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_3_existsStrictIntermediateRadius
{a r : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hrR : ENNReal.ofReal r < f.radiusOfConvergence) :
∃ ρ : NNReal, ENNReal.ofReal r < (ρ : ENNReal) ∧ (ρ : ENNReal) < f.radiusOfConvergence
Helper for Theorem 4.1.3: if r < ρ, then [a-r, a+r] lies in the open ball Metric.ball a ρ.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_3_Icc_subset_ball
{a r ρ : ℝ} (hrρ : r < ρ) :
Set.Icc (a - r) (a + r) ⊆ Metric.ball a ρ
Helper for Theorem 4.1.3: rewrite both partial sums and total sums in textbook coefficient form.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_3_rewriteSeriesForms
{a : ℝ} (f : FormalPowerSeriesCenteredAt a) :
((fun N x => (FormalMultilinearSeries.ofScalars ℝ f.coeff).partialSum N (x - a)) =
(fun N x => (Finset.range N).sum (fun n => f.coeff n * (x - a) ^ n)))
∧
((fun x => (FormalMultilinearSeries.ofScalars ℝ f.coeff).sum (x - a)) =
(fun x => ∑' n, f.coeff n * (x - a) ^ n))
Helper for Theorem 4.1.3: inside the convergence radius, the scalar-series sum function is continuous.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_3_continuousAt_inside_radius
{a x : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hx : ENNReal.ofReal |x - a| < f.radiusOfConvergence) :
ContinuousAt (fun y => ∑' n, f.coeff n * (y - a) ^ n) x
Helper for Theorem 4.1.1: identify the textbook radius with the formal multilinear-series radius.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_1_radius_eq_ofScalarsRadius {a : ℝ}
(f : FormalPowerSeriesCenteredAt a) :
f.radiusOfConvergence = (FormalMultilinearSeries.ofScalars ℝ f.coeff).radius
Helper for Theorem 4.1.1: summability of scalar terms forces weighted norms to tend to 0.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_1_tendsto_weightedNorm_zero_of_summable
{a x : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hs : Summable (fun n : ℕ => f.coeff n * (x - a) ^ n)) :
Filter.Tendsto (fun n : ℕ => ‖f.coeff n‖ * ‖x - a‖ ^ n) Filter.atTop (nhds 0)
Helper for Theorem 4.1.1: if the scalar series at x is summable, then |x-a| lies within the radius.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_1_ofRealAbs_le_radius_of_summable
{a x : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hs : Summable (fun n : ℕ => f.coeff n * (x - a) ^ n)) :
ENNReal.ofReal |x - a| ≤ f.radiusOfConvergence
Helper for Theorem 4.1.4: identify the convergence-radius set with the corresponding emetric ball.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_4_ball_eq_insideRadiusSet
{a : ℝ} (f : FormalPowerSeriesCenteredAt a) :
Metric.eball a (FormalMultilinearSeries.ofScalars ℝ f.coeff).radius =
{x : ℝ | ENNReal.ofReal |x - a| < f.radiusOfConvergence}
Helper for Theorem 4.1.4: derivative power-series expansion on the convergence ball.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_4_derivHasFPowerSeriesOnBall
{a : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hRpos : 0 < f.radiusOfConvergence) :
HasFPowerSeriesOnBall
(fun x => deriv (fun y => (FormalMultilinearSeries.ofScalars ℝ f.coeff).sum (y - a)) x)
((((ContinuousLinearMap.apply ℝ ℝ) (1 : ℝ)).compFormalMultilinearSeries
((FormalMultilinearSeries.ofScalars ℝ f.coeff).derivSeries)))
a
(FormalMultilinearSeries.ofScalars ℝ f.coeff).radius
Helper for Theorem 4.1.4: rewrite derivative-series partial sums in textbook coefficient form.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_4_rewriteDerivedPartialSums
{a : ℝ} (f : FormalPowerSeriesCenteredAt a) :
(fun N x =>
((((ContinuousLinearMap.apply ℝ ℝ) (1 : ℝ)).compFormalMultilinearSeries
((FormalMultilinearSeries.ofScalars ℝ f.coeff).derivSeries)).partialSum N (x - a)))
=
(fun N x =>
(Finset.range N).sum (fun n => ((n + 1 : ℝ) * f.coeff (n + 1) * (x - a) ^ n)) )
Helper for Theorem 4.1.4: differentiability of the power-series sum on its open convergence interval.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_4_differentiableOn_insideRadius
{a : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hRpos : 0 < f.radiusOfConvergence) :
DifferentiableOn ℝ
(fun x => ∑' n, f.coeff n * (x - a) ^ n)
{x : ℝ | ENNReal.ofReal |x - a| < f.radiusOfConvergence}
Helper for Theorem 4.1.5: endpoints of Set.uIcc y z satisfy the same radius bound inherited from the interval hypothesis.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_endpoint_radius_bounds
{a y z : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hsub : Set.uIcc y z ⊆ {x : ℝ | ENNReal.ofReal |x - a| < f.radiusOfConvergence}) :
ENNReal.ofReal |y - a| < f.radiusOfConvergence
∧ ENNReal.ofReal |z - a| < f.radiusOfConvergence
Helper for Theorem 4.1.5: every point of Set.uIcc y z has distance to a bounded by the maximum endpoint distance.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_abs_sub_le_maxEndpoint
{a y z x : ℝ} (hx : x ∈ Set.uIcc y z) :
Helper for Theorem 4.1.5: each monomial term x ↦ c_n (x-a)^n is continuous.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_term_continuous
{a : ℝ} (f : FormalPowerSeriesCenteredAt a) (n : ℕ) :
Continuous (fun x : ℝ => f.coeff n * (x - a) ^ n)
Helper for Theorem 4.1.5: package the nth term as a continuous map on ℝ.
noncomputable def FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_termContinuousMap
{a : ℝ} (f : FormalPowerSeriesCenteredAt a) (n : ℕ) : C(ℝ, ℝ) :=
⟨fun x => f.coeff n * (x - a) ^ n,
FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_term_continuous (a := a) f n⟩
Helper for Theorem 4.1.5: the supremum norms of restricted terms on Set.uIcc y z form a summable series.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_summable_restrictedTermNorms
{a y z : ℝ} (f : FormalPowerSeriesCenteredAt a)
(hsub : Set.uIcc y z ⊆ {x : ℝ | ENNReal.ofReal |x - a| < f.radiusOfConvergence}) :
Summable (fun n : ℕ =>
‖(FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_termContinuousMap (a := a) f n).restrict
(⟨Set.uIcc y z, isCompact_uIcc⟩ : TopologicalSpace.Compacts ℝ)‖)
Helper for Theorem 4.1.5: the interval integral of each term has the textbook antiderivative form.
lemma FormalPowerSeriesCenteredAt.helperForTheorem_4_1_5_intervalIntegral_term_formula
{a y z : ℝ} (f : FormalPowerSeriesCenteredAt a) (n : ℕ) :
(∫ x in y..z, f.coeff n * (x - a) ^ n)
= f.coeff n * (((z - a) ^ (n + 1) - (y - a) ^ (n + 1)) / (n + 1 : ℝ))4.2 — Real Analytic Functions — Part 1
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Definition 4.2.1 (Real analytic at a point): let E ⊆ ℝ, f : E → ℝ, and a ∈ E be an interior point of E. The function f is real analytic at a if there exist r > 0 and coefficients c : ℕ → ℝ such that (a-r, a+r) ⊆ E, the power series ∑ c n (x-a)^n converges for every x ∈ (a-r, a+r) (encoding radius of convergence at least r), and f x equals that series on (a-r, a+r).
open Classical
def IsRealAnalyticAt (E : Set ℝ) (f : E → ℝ) (a : E) : Prop :=
(a : ℝ) ∈ interior E ∧
∃ r : ℝ, 0 < r ∧
∃ hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E,
∃ c : ℕ → ℝ,
(∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n)) ∧
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n
A function on E is real analytic on E when E is open and the function is real analytic at each point of E.
open Classical
def IsRealAnalyticOn (E : Set ℝ) (f : E → ℝ) : Prop :=
IsOpen E ∧ ∀ a : E, IsRealAnalyticAt E f a
A set of real numbers has no isolated points if every x ∈ E is a limit point of E.
open Classical
def HasNoIsolatedPoints (E : Set ℝ) : Prop :=
∀ x ∈ E, x ∈ closure (E \ {x})
Canonical extension of f : E → ℝ to ℝ, taking value 0 off E.
open Classical
noncomputable def SubtypeExtension (E : Set ℝ) (f : E → ℝ) : ℝ → ℝ :=
fun x => if hx : x ∈ E then f ⟨x, hx⟩ else 0
The first derivative f' : E → ℝ, defined by the within-set derivative on E.
open Classical
noncomputable def FirstDerivativeSub (E : Set ℝ) (f : E → ℝ) : E → ℝ :=
fun x => derivWithin (SubtypeExtension E f) E x
Iterated derivatives on E: f^(0)=f and f^(n+1) is the first derivative of f^(n).
open Classical
noncomputable def IteratedDerivativeSub (E : Set ℝ) (f : E → ℝ) : ℕ → E → ℝ
A function f : E → ℝ is once differentiable on E when it is differentiable within E at every point of E.
open Classical
def IsOnceDifferentiableSub (E : Set ℝ) (f : E → ℝ) : Prop :=
∀ x : E, DifferentiableWithinAt ℝ (SubtypeExtension E f) E (x : ℝ)
Definition 4.2.2 (k-times differentiability on a set). (i) once differentiable means differentiable at every point of E; (ii) for k ≥ 2, k-times differentiable is defined recursively via the derivative; (iii) iterated derivatives are f^(0)=f, f^(1)=f', f^(k)=((f^(k-1))'); (iv) infinitely differentiable (smooth) means k-times differentiable for every k ≥ 0. The standing hypothesis that every point of E is a limit point of E is recorded by HasNoIsolatedPoints E as part of the definition.
open Classical
def IsKTimesDifferentiableSub (E : Set ℝ) (f : E → ℝ) : ℕ → Prop :=
fun k =>
HasNoIsolatedPoints E ∧
Nat.rec (motive := fun _ => Prop) True
(fun n ih => IsOnceDifferentiableSub E (IteratedDerivativeSub E f n) ∧ ih) k
A function on E is smooth when it is k-times differentiable for every k : ℕ.
open Classical
def IsSmoothSub (E : Set ℝ) (f : E → ℝ) : Prop :=
∀ k : ℕ, IsKTimesDifferentiableSub E f k
The function x ↦ |x|^3 on ℝ.
open Classical
def absCube : ℝ → ℝ :=
fun x => |x| ^ (3 : ℕ)
Proposition 4.2.1: for f(x) = |x|^3, one has f'(x) = 3x|x| and f''(x) = 6|x| for all x ∈ ℝ; hence f is twice differentiable on ℝ, while f'' is not differentiable at 0 (equivalently, f is not three times differentiable at 0).
open Classical
theorem absCube_twiceDifferentiable_and_not_threeTimesDifferentiableAt_zero :
Differentiable ℝ absCube ∧
Differentiable ℝ (deriv absCube) ∧
(∀ x : ℝ, deriv absCube x = 3 * x * |x|) ∧
(∀ x : ℝ, deriv (deriv absCube) x = 6 * |x|) ∧
¬DifferentiableAt ℝ (deriv (deriv absCube)) 0
Proposition 4.2.2 (Real analytic functions are k-times differentiable): if f : (a-r, a+r) → ℝ has a convergent power-series representation f(x) = ∑' n, c n * (x-a)^n on (a-r, a+r), then for every k : ℕ, f is k-times differentiable on (a-r, a+r), and the k-th iterated derivative has the termwise differentiated series ∑' n, c (n+k) * ((n+k)! / n!) * (x-a)^n (equivalently using (n+1)(n+2)...(n+k), with empty product 1 when k = 0).
open Classical
theorem realAnalyticOnInterval_iteratedDerivative_eq_powerSeries
{a r : ℝ} (hr : 0 < r) (f : Set.Ioo (a - r) (a + r) → ℝ)
(hseries :
∃ c : ℕ → ℝ,
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n) :
∃ c : ℕ → ℝ,
(∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n) ∧
∀ k : ℕ,
IsKTimesDifferentiableSub (Set.Ioo (a - r) (a + r)) f k ∧
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable
(fun n : ℕ =>
c (n + k) *
((Nat.factorial (n + k) : ℝ) / (Nat.factorial n : ℝ)) *
(x - a) ^ n) ∧
IteratedDerivativeSub (Set.Ioo (a - r) (a + r)) f k ⟨x, hx⟩ =
∑' n : ℕ,
c (n + k) *
((Nat.factorial (n + k) : ℝ) / (Nat.factorial n : ℝ)) *
(x - a) ^ n
Helper for Proposition 4.2.1: rewrite absCube using real powers.
open Classical
lemma helperForProposition_4_2_1_absCube_eq_absRpowThree :
absCube = fun x : ℝ => |x| ^ (3 : ℝ)
Helper for Proposition 4.2.1: derivative formula for x ↦ |x|^3.
open Classical
lemma helperForProposition_4_2_1_hasDerivAt_absCube (x : ℝ) :
HasDerivAt absCube (3 * x * |x|) x
Helper for Proposition 4.2.1: derivative formula for x ↦ x|x|.
open Classical
lemma helperForProposition_4_2_1_hasDerivAt_xMulAbs (x : ℝ) :
HasDerivAt (fun y : ℝ => y * |y|) (2 * |x|) x
Helper for Proposition 4.2.1: derivative formula for deriv absCube.
open Classical
lemma helperForProposition_4_2_1_hasDerivAt_derivAbsCube (x : ℝ) :
HasDerivAt (deriv absCube) (6 * |x|) x
Helper for Proposition 4.2.1: the second derivative x ↦ 6|x| is not differentiable at 0.
open Classical
lemma helperForProposition_4_2_1_notDifferentiableAt_zero_secondDeriv :
¬DifferentiableAt ℝ (fun x : ℝ => 6 * |x|) 0
Helper for Proposition 4.2.2: open intervals in ℝ have no isolated points.
open Classical
lemma helperForProposition_4_2_2_interval_hasNoIsolatedPoints (a r : ℝ) :
HasNoIsolatedPoints (Set.Ioo (a - r) (a + r))
Helper for Proposition 4.2.2: the k = 0 termwise coefficient factor is 1.
open Classical
lemma helperForProposition_4_2_2_kZero_seriesTerm_eq
(c : ℕ → ℝ) (a x : ℝ) :
(fun n : ℕ =>
c (n + 0) *
((Nat.factorial (n + 0) : ℝ) / (Nat.factorial n : ℝ)) *
(x - a) ^ n) =
(fun n : ℕ => c n * (x - a) ^ n)
Helper for Proposition 4.2.2: the target series identity for k = 0.
open Classical
lemma helperForProposition_4_2_2_seriesFormula_kZero
{a r : ℝ} (f : Set.Ioo (a - r) (a + r) → ℝ) (c : ℕ → ℝ)
(hc :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n) :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable
(fun n : ℕ =>
c (n + 0) *
((Nat.factorial (n + 0) : ℝ) / (Nat.factorial n : ℝ)) *
(x - a) ^ n) ∧
IteratedDerivativeSub (Set.Ioo (a - r) (a + r)) f 0 ⟨x, hx⟩ =
∑' n : ℕ,
c (n + 0) *
((Nat.factorial (n + 0) : ℝ) / (Nat.factorial n : ℝ)) *
(x - a) ^ n
Helper for Proposition 4.2.2: IteratedDerivativeSub agrees pointwise with iteratedDerivWithin on the underlying set.
open Classical
lemma helperForProposition_4_2_2_iteratedDerivativeSub_eq_iteratedDerivWithin
(E : Set ℝ) (f : E → ℝ) :
∀ k x (hx : x ∈ E),
IteratedDerivativeSub E f k ⟨x, hx⟩ =
iteratedDerivWithin k (SubtypeExtension E f) E x
Helper for Proposition 4.2.2: smoothness of the ambient extension implies that every iterated subtype derivative is once differentiable on the set.
open Classical
lemma helperForProposition_4_2_2_isOnceDifferentiableSub_of_contDiffOn
{E : Set ℝ} (f : E → ℝ)
(hcont : ContDiffOn ℝ ⊤ (SubtypeExtension E f) E)
(hunique : UniqueDiffOn ℝ E) :
∀ n : ℕ, IsOnceDifferentiableSub E (IteratedDerivativeSub E f n)
Helper for Proposition 4.2.2: if all iterated derivatives are once differentiable, then the recursive notion of k-times differentiability holds for every k.
open Classical
lemma helperForProposition_4_2_2_isKTimesDifferentiableSub_of_allOnceDifferentiable
{E : Set ℝ} (f : E → ℝ) (hNoIso : HasNoIsolatedPoints E)
(hOnce : ∀ n : ℕ, IsOnceDifferentiableSub E (IteratedDerivativeSub E f n)) :
∀ k : ℕ, IsKTimesDifferentiableSub E f k
Helper for Proposition 4.2.2: the interval series data gives HasFPowerSeriesWithinOnBall for the canonical extension.
open Classical
lemma helperForProposition_4_2_2_hasFPowerSeriesWithinOnBall_of_seriesData
{a r : ℝ} (hr : 0 < r) (f : Set.Ioo (a - r) (a + r) → ℝ) (c : ℕ → ℝ)
(hc :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n) :
HasFPowerSeriesWithinOnBall
(SubtypeExtension (Set.Ioo (a - r) (a + r)) f)
(FormalMultilinearSeries.ofScalars ℝ c)
(Set.Ioo (a - r) (a + r)) a (ENNReal.ofReal r)
Helper for Proposition 4.2.2: one-step coefficient recursion for iteratedFDerivSeries evaluated at all-ones.
open Classical
lemma helperForProposition_4_2_2_iteratedFDerivSeriesCoeffOnes_step
(c : ℕ → ℝ) (k n : ℕ) :
(((FormalMultilinearSeries.ofScalars ℝ c).iteratedFDerivSeries (k + 1)).coeff n)
(fun _ : Fin (k + 1) => (1 : ℝ))
=
(n + 1 : ℝ) *
(((FormalMultilinearSeries.ofScalars ℝ c).iteratedFDerivSeries k).coeff (n + 1))
(fun _ : Fin k => (1 : ℝ))
Helper for Proposition 4.2.2: closed-form coefficients for iteratedFDerivSeries of ofScalars, evaluated at all-ones.
open Classical
lemma helperForProposition_4_2_2_iteratedFDerivSeriesCoeffOnes_closedForm
(c : ℕ → ℝ) (k n : ℕ) :
(((FormalMultilinearSeries.ofScalars ℝ c).iteratedFDerivSeries k).coeff n)
(fun _ : Fin k => (1 : ℝ))
=
c (n + k) * ((Nat.factorial (n + k) : ℝ) / (Nat.factorial n : ℝ))
Helper for Proposition 4.2.2: explicit term formula for iteratedFDerivSeries of ofScalars.
open Classical
lemma helperForProposition_4_2_2_iteratedFDerivSeries_term_formula
(c : ℕ → ℝ) (k n : ℕ) (y : ℝ) :
(((FormalMultilinearSeries.ofScalars ℝ c).iteratedFDerivSeries k n)
(fun _ : Fin n => y))
(fun _ : Fin k => (1 : ℝ))
= c (n + k) * ((Nat.factorial (n + k) : ℝ) / (Nat.factorial n : ℝ)) * y ^ n
Helper for Proposition 4.2.2: series formula for iteratedDerivWithin on the interval.
open Classical
lemma helperForProposition_4_2_2_iteratedDerivWithin_series_formula
{a r : ℝ} (hr : 0 < r) (f : Set.Ioo (a - r) (a + r) → ℝ) (c : ℕ → ℝ)
(hc :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n) :
∀ k x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable
(fun n : ℕ =>
c (n + k) * ((Nat.factorial (n + k) : ℝ) / (Nat.factorial n : ℝ)) * (x - a) ^ n) ∧
iteratedDerivWithin k (SubtypeExtension (Set.Ioo (a - r) (a + r)) f)
(Set.Ioo (a - r) (a + r)) x =
∑' n : ℕ,
c (n + k) * ((Nat.factorial (n + k) : ℝ) / (Nat.factorial n : ℝ)) * (x - a) ^ n
Helper for Proposition 4.2.2: the interval power-series representation implies that all iterated subtype derivatives are once differentiable.
open Classical
lemma helperForProposition_4_2_2_onceDifferentiable_all_orders_of_analytic
{a r : ℝ} (hr : 0 < r) (f : Set.Ioo (a - r) (a + r) → ℝ) (c : ℕ → ℝ)
(hc :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n) :
∀ n : ℕ,
IsOnceDifferentiableSub (Set.Ioo (a - r) (a + r))
(IteratedDerivativeSub (Set.Ioo (a - r) (a + r)) f n)4.2 — Real Analytic Functions — Part 2
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Proposition 4.2.3: for f(x) = |x|, the function is not differentiable at 0, hence not real analytic at 0; moreover, for every a ≠ 0, f is real analytic at a.
open Classical
theorem abs_notDifferentiableAt_zero_notAnalyticAt_zero_analyticAt_nonzero :
¬DifferentiableAt ℝ (fun x : ℝ => |x|) 0 ∧
¬AnalyticAt ℝ (fun x : ℝ => |x|) 0 ∧
∀ a : ℝ, a ≠ 0 → AnalyticAt ℝ (fun x : ℝ => |x|) a
Theorem 4.2.1: if f : E → ℝ is real analytic on E, then f is smooth on E. Moreover, for every integer k ≥ 0, the iterated derivative f^(k) exists at every point of E and is real analytic on E.
open Classical
theorem realAnalyticOn_isSmooth_and_iteratedDerivative_realAnalytic
{E : Set ℝ} (f : E → ℝ) (hf : IsRealAnalyticOn E f) :
IsSmoothSub E f ∧
∀ k : ℕ,
IsOnceDifferentiableSub E (IteratedDerivativeSub E f k) ∧
IsRealAnalyticOn E (IteratedDerivativeSub E f k)
Theorem 4.2.2 (Taylor's formula): let E ⊆ ℝ, let a : E be an interior point, and let f : E → ℝ be real analytic at a. If for some r > 0 and coefficients c : ℕ → ℝ one has f(x) = ∑' n, c n (x-a)^n for all x ∈ (a-r, a+r), then for every k ≥ 0, f^(k)(a) = k! * c_k; in particular, f(x) = ∑' n, (f^(n)(a) / n!) (x-a)^n on (a-r, a+r).
open Classical
theorem taylorFormula_iteratedDerivativeSub_eq_factorial_mul_coeff
{E : Set ℝ} (f : E → ℝ) (a : E) (hf : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(c : ℕ → ℝ)
(hsummable :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hseries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n) :
(∀ k : ℕ, IteratedDerivativeSub E f k a = (Nat.factorial k : ℝ) * c k) ∧
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ =
∑' n : ℕ,
(IteratedDerivativeSub E f n a / (Nat.factorial n : ℝ)) *
(x - (a : ℝ)) ^ n
Theorem 4.2.3 (Uniqueness of power series): let E ⊆ ℝ, let a be an interior point of E, and let f : E → ℝ. If there are radii R_c, R_d > 0 and coefficient sequences c, d : ℕ → ℝ such that for every x ∈ E with |x-a| < R_c one has f(x) = ∑' n, c n * (x-a)^n with convergence, and for every x ∈ E with |x-a| < R_d one has f(x) = ∑' n, d n * (x-a)^n with convergence, then c n = d n for all n.
open Classical
theorem powerSeries_coefficients_unique
{E : Set ℝ} (f : E → ℝ) (a : E)
(ha : (a : ℝ) ∈ interior E)
(c d : ℕ → ℝ) {R_c R_d : ℝ} (hR_c : 0 < R_c) (hR_d : 0 < R_d)
(hc :
∀ x : ℝ, ∀ hxE : x ∈ E, |x - (a : ℝ)| < R_c →
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n) ∧
f ⟨x, hxE⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
(hd :
∀ x : ℝ, ∀ hxE : x ∈ E, |x - (a : ℝ)| < R_d →
Summable (fun n : ℕ => d n * (x - (a : ℝ)) ^ n) ∧
f ⟨x, hxE⟩ = ∑' n : ℕ, d n * (x - (a : ℝ)) ^ n) :
∀ n : ℕ, c n = d n
Proposition 4.2.4: for every real number x with |x| < 1, the geometric series ∑' n : ℕ, x^n converges and equals 1 / (1 - x).
open Classical
theorem geometricSeries_converges_and_tsum_eq_inv_one_sub
(x : ℝ) (hx : |x| < 1) :
Summable (fun n : ℕ => x ^ n) ∧
(∑' n : ℕ, x ^ n) = 1 / (1 - x)
Proposition 4.2.5: if x : ℝ satisfies |x - 1/2| < 1/2, then the geometric series ∑' n : ℕ, 2^(n+1) * (x - 1/2)^n converges and equals 1 / (1 - x).
open Classical
theorem centeredGeometricSeries_converges_and_tsum_eq_inv_one_sub
(x : ℝ) (hx : |x - (1 / 2 : ℝ)| < (1 / 2 : ℝ)) :
Summable (fun n : ℕ => (2 : ℝ) ^ (n + 1) * (x - (1 / 2 : ℝ)) ^ n) ∧
(∑' n : ℕ, (2 : ℝ) ^ (n + 1) * (x - (1 / 2 : ℝ)) ^ n) = 1 / (1 - x)
Helper for Proposition 4.2.3: |x| is not real analytic at 0.
open Classical
lemma helperForProposition_4_2_3_notAnalyticAt_zero_abs :
¬AnalyticAt ℝ (fun x : ℝ => |x|) 0
Helper for Proposition 4.2.3: at positive points, |x| agrees locally with x.
open Classical
lemma helperForProposition_4_2_3_eventuallyEq_abs_id_of_pos
{a : ℝ} (ha : 0 < a) :
(fun x : ℝ => |x|) =ᶠ[nhds a] (fun x : ℝ => x)
Helper for Proposition 4.2.3: at negative points, |x| agrees locally with -x.
open Classical
lemma helperForProposition_4_2_3_eventuallyEq_abs_negId_of_neg
{a : ℝ} (ha : a < 0) :
(fun x : ℝ => |x|) =ᶠ[nhds a] (fun x : ℝ => -x)
Helper for Proposition 4.2.3: |x| is real analytic at every nonzero real point.
open Classical
lemma helperForProposition_4_2_3_analyticAt_abs_of_ne_zero
{a : ℝ} (ha : a ≠ 0) :
AnalyticAt ℝ (fun x : ℝ => |x|) a
Helper for Theorem 4.2.1: every open subset of ℝ has no isolated points.
open Classical
lemma helperForTheorem_4_2_1_hasNoIsolatedPoints_of_isOpen
{E : Set ℝ} (hE : IsOpen E) : HasNoIsolatedPoints E
Helper for Theorem 4.2.1: local interval and ambient set are eventually equal at the center.
open Classical
lemma helperForTheorem_4_2_1_intervalEventuallyEq_set_at_center
{E : Set ℝ} {a r : ℝ} (hr : 0 < r)
(hI : Set.Ioo (a - r) (a + r) ⊆ E) :
Set.Ioo (a - r) (a + r) =ᶠ[nhds a] E
Helper for Theorem 4.2.1: a pointwise real-analytic witness yields analytic-within-at for the canonical extension.
open Classical
lemma helperForTheorem_4_2_1_analyticWithinAtSubtypeExtension_of_IsRealAnalyticAt
{E : Set ℝ} (f : E → ℝ) (a : E)
(ha : IsRealAnalyticAt E f a) :
AnalyticWithinAt ℝ (SubtypeExtension E f) E (a : ℝ)
Helper for Theorem 4.2.1: on open sets, analyticity of the canonical extension implies the custom pointwise real-analytic property.
open Classical
lemma helperForTheorem_4_2_1_isRealAnalyticAt_of_analyticWithinAtSubtypeExtension_open
{E : Set ℝ} (g : E → ℝ) (hE : IsOpen E) (a : E)
(ha : AnalyticWithinAt ℝ (SubtypeExtension E g) E (a : ℝ)) :
IsRealAnalyticAt E g a
Helper for Theorem 4.2.1: convert the custom IsRealAnalyticOn notion to mathlib AnalyticOn for the canonical extension.
open Classical
lemma helperForTheorem_4_2_1_analyticOnSubtypeExtension_of_IsRealAnalyticOn
{E : Set ℝ} (f : E → ℝ) (hf : IsRealAnalyticOn E f) :
AnalyticOn ℝ (SubtypeExtension E f) E
Helper for Theorem 4.2.1: convert analyticity of the canonical extension on an open set back to the custom IsRealAnalyticOn predicate.
open Classical
lemma helperForTheorem_4_2_1_isRealAnalyticOn_of_analyticOnSubtypeExtension
{E : Set ℝ} (g : E → ℝ) (hE : IsOpen E)
(hAnalytic : AnalyticOn ℝ (SubtypeExtension E g) E) :
IsRealAnalyticOn E g
Helper for Theorem 4.2.1: analyticity of an extension propagates to all iterated subtype derivatives.
open Classical
lemma helperForTheorem_4_2_1_analyticOnSubtypeExtension_iteratedDerivativeSub
{E : Set ℝ} (f : E → ℝ) (hE : IsOpen E)
(hAnalytic : AnalyticOn ℝ (SubtypeExtension E f) E) :
∀ k : ℕ, AnalyticOn ℝ (SubtypeExtension E (IteratedDerivativeSub E f k)) E
Helper for Theorem 4.2.1: real-analytic-on implies once differentiable on the set.
open Classical
lemma helperForTheorem_4_2_1_isOnceDifferentiableSub_of_IsRealAnalyticOn
{E : Set ℝ} (g : E → ℝ) (hg : IsRealAnalyticOn E g) :
IsOnceDifferentiableSub E g
Helper for Theorem 4.2.2: iterated derivatives within eventually equal sets agree at the center.
open Classical
lemma helperForTheorem_4_2_2_iteratedDerivWithin_setChange_at_center
{g : ℝ → ℝ} {a : ℝ} {s t : Set ℝ}
(hst : s =ᶠ[nhds a] t) :
∀ k : ℕ, iteratedDerivWithin k g s a = iteratedDerivWithin k g t a
Helper for Theorem 4.2.2: identify ambient and interval iterated subtype derivatives at the center.
open Classical
lemma helperForTheorem_4_2_2_iteratedDerivativeSub_eq_intervalRestriction_at_center
{E : Set ℝ} (f : E → ℝ) (a : E) {r : ℝ} (hr : 0 < r)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(haI : (a : ℝ) ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)) :
∀ k : ℕ,
IteratedDerivativeSub E f k a =
IteratedDerivativeSub
(Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(fun x : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) => f ⟨x, hI x.property⟩)
k
⟨(a : ℝ), haI⟩
Helper for Theorem 4.2.2: the interval center value of the iterated derivative extracts factorial times the coefficient.
open Classical
lemma helperForTheorem_4_2_2_interval_center_factorial_coeff
{a r : ℝ} (hr : 0 < r)
(f : Set.Ioo (a - r) (a + r) → ℝ) (c : ℕ → ℝ)
(hseries :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n)
(haI : a ∈ Set.Ioo (a - r) (a + r)) :
∀ k : ℕ,
IteratedDerivativeSub (Set.Ioo (a - r) (a + r)) f k ⟨a, haI⟩ =
(Nat.factorial k : ℝ) * c k
Helper for Theorem 4.2.2: divide the factorial identity to recover coefficients.
open Classical
lemma helperForTheorem_4_2_2_coeff_rewrite_from_factorial_identity
{E : Set ℝ} (f : E → ℝ) (a : E) (c : ℕ → ℝ)
(hfac : ∀ n : ℕ, IteratedDerivativeSub E f n a = (Nat.factorial n : ℝ) * c n) :
∀ n : ℕ, IteratedDerivativeSub E f n a / (Nat.factorial n : ℝ) = c n
Helper for Theorem 4.2.3: from interior membership and two positive radii, construct a common positive interval radius contained in E and bounded by both given radii.
open Classical
lemma helperForTheorem_4_2_3_exists_commonIntervalRadius
{E : Set ℝ} {a : ℝ} (ha : a ∈ interior E)
{R_c R_d : ℝ} (hR_c : 0 < R_c) (hR_d : 0 < R_d) :
∃ r : ℝ, 0 < r ∧ r ≤ R_c ∧ r ≤ R_d ∧ Set.Ioo (a - r) (a + r) ⊆ E
Helper for Theorem 4.2.3: transport power-series data from a radius condition |x-a| < R on E to data on a smaller symmetric interval around a.
open Classical
lemma helperForTheorem_4_2_3_seriesData_on_commonInterval
{E : Set ℝ} {f : E → ℝ} {a r R : ℝ} {u : ℕ → ℝ}
(hI : Set.Ioo (a - r) (a + r) ⊆ E) (hrR : r ≤ R)
(hu :
∀ x : ℝ, ∀ hxE : x ∈ E, |x - a| < R →
Summable (fun n : ℕ => u n * (x - a) ^ n) ∧
f ⟨x, hxE⟩ = ∑' n : ℕ, u n * (x - a) ^ n) :
∀ x : ℝ, ∀ hx : x ∈ Set.Ioo (a - r) (a + r),
Summable (fun n : ℕ => u n * (x - a) ^ n) ∧
f ⟨x, hI hx⟩ = ∑' n : ℕ, u n * (x - a) ^ n
Helper for Theorem 4.2.3: interval power-series data gives real analyticity at the center.
open Classical
lemma helperForTheorem_4_2_3_isRealAnalyticAt_of_intervalSeriesData
{E : Set ℝ} (f : E → ℝ) (a : E) (ha : (a : ℝ) ∈ interior E)
{r : ℝ} (hr : 0 < r)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(u : ℕ → ℝ)
(hseriesData :
∀ x : ℝ, ∀ hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => u n * (x - (a : ℝ)) ^ n) ∧
f ⟨x, hI hx⟩ = ∑' n : ℕ, u n * (x - (a : ℝ)) ^ n) :
IsRealAnalyticAt E f a
Helper for Theorem 4.2.3: cancel a nonzero factorial factor in two identities for the same iterated derivative sequence.
open Classical
lemma helperForTheorem_4_2_3_cancel_factorial
{E : Set ℝ} {f : E → ℝ} {a : E} {c d : ℕ → ℝ}
(hc : ∀ n : ℕ, IteratedDerivativeSub E f n a = (Nat.factorial n : ℝ) * c n)
(hd : ∀ n : ℕ, IteratedDerivativeSub E f n a = (Nat.factorial n : ℝ) * d n) :
∀ n : ℕ, c n = d n
Helper for Proposition 4.2.4: convert |x| < 1 into the norm inequality ‖x‖ < 1 used by geometric-series lemmas.
open Classical
lemma helperForProposition_4_2_4_norm_lt_one_of_abs_lt_one
{x : ℝ} (hx : |x| < 1) : ‖x‖ < 1
Helper for Proposition 4.2.4: under ‖x‖ < 1, the geometric series has sum (1 - x)⁻¹.
open Classical
lemma helperForProposition_4_2_4_hasSum_geometric
{x : ℝ} (hx : ‖x‖ < 1) :
HasSum (fun n : ℕ => x ^ n) ((1 - x)⁻¹)
Helper for Proposition 4.2.4: rewrite the geometric-series sum in the form 1 / (1 - x).
open Classical
lemma helperForProposition_4_2_4_tsum_eq_one_div_form
{x : ℝ} (hx : ‖x‖ < 1) :
(∑' n : ℕ, x ^ n) = 1 / (1 - x)
Helper for Proposition 4.2.5: convert |x - 1/2| < 1/2 into the norm bound ‖2 * (x - 1/2)‖ < 1 used by geometric-series lemmas.
open Classical
lemma helperForProposition_4_2_5_norm_lt_one_scaledShift
{x : ℝ} (hx : |x - (1 / 2 : ℝ)| < (1 / 2 : ℝ)) :
‖(2 : ℝ) * (x - (1 / 2 : ℝ))‖ < 1
Helper for Proposition 4.2.5: rewrite the centered-series term as a scaled geometric term in the ratio 2 * (x - 1/2).
open Classical
lemma helperForProposition_4_2_5_term_rewrite_to_scaledGeometric
(x : ℝ) (n : ℕ) :
(2 : ℝ) ^ (n + 1) * (x - (1 / 2 : ℝ)) ^ n =
2 * ((2 : ℝ) * (x - (1 / 2 : ℝ))) ^ n
Helper for Proposition 4.2.5: the centered geometric series has the closed-form sum 2 * (1 - 2*(x - 1/2))⁻¹.
open Classical
lemma helperForProposition_4_2_5_hasSum_centeredSeries
{x : ℝ} (hx : |x - (1 / 2 : ℝ)| < (1 / 2 : ℝ)) :
HasSum
(fun n : ℕ => (2 : ℝ) ^ (n + 1) * (x - (1 / 2 : ℝ)) ^ n)
(2 * (1 - (2 : ℝ) * (x - (1 / 2 : ℝ)))⁻¹)
Helper for Proposition 4.2.5: simplify the centered closed form to 1 / (1 - x).
open Classical
lemma helperForProposition_4_2_5_sumValue_rewrite
(x : ℝ) :
2 * (1 - (2 : ℝ) * (x - (1 / 2 : ℝ)))⁻¹ = 1 / (1 - x)4.2 — Real Analytic Functions — Part 3
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Proposition 4.2.6: let n : ℕ and a c : ℝ, and define f(x) = c * (x - a)^n. Then f is infinitely differentiable on ℝ. Moreover, for every k with k ≤ n, iteratedDeriv k f x = c * (n! / (n-k)!) * (x-a)^(n-k) for all x : ℝ, and for every k > n, iteratedDeriv k f x = 0 for all x : ℝ.
open Classical
theorem iteratedDeriv_monomial_shifted_const_mul
(n : ℕ) (a c : ℝ) :
ContDiff ℝ ⊤ (fun x : ℝ => c * (x - a) ^ n) ∧
(∀ k : ℕ, k ≤ n →
∀ x : ℝ,
iteratedDeriv k (fun y : ℝ => c * (y - a) ^ n) x =
c * ((Nat.factorial n : ℝ) / (Nat.factorial (n - k) : ℝ)) * (x - a) ^ (n - k)) ∧
(∀ k : ℕ, n < k →
∀ x : ℝ,
iteratedDeriv k (fun y : ℝ => c * (y - a) ^ n) x = 0)
Proposition 4.2.7: let a, b ∈ ℝ and n ≥ 0 be an integer. Then for every x ∈ ℝ, (x - a)^n = ∑_{m=0}^n (n! / (m!(n-m)!)) (b - a)^(n-m) (x - b)^m.
open Classical
theorem shiftedPower_eq_sum_factorial_ratio_mul_shiftedPowers
(a b : ℝ) (n : ℕ) (x : ℝ) :
(x - a) ^ n =
Finset.sum (Finset.range (n + 1))
(fun m =>
((Nat.factorial n : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial (n - m) : ℝ))) *
(b - a) ^ (n - m) * (x - b) ^ m)
Proposition 4.2.8: if r > 0 and |x| < r, then r / (r - x) = ∑' n = 0..∞, x^n r^{-n}.
open Classical
theorem scaledGeometricSeries_tsum_eq_r_div_r_sub
(r x : ℝ) (hr : 0 < r) (hx : |x| < r) :
r / (r - x) = ∑' n : ℕ, x ^ n * (r⁻¹) ^ n
Proposition 4.2.9: if r > 0, m : ℕ, and |x| < r, then r / (r - x)^(m+1) equals the binomial-coefficient power series, written in the reindexed form ∑' n, ((n+m)!/(m!n!)) x^n r^{-(n+m)} (equivalent to summing from n = m); moreover this series converges absolutely for every such x.
open Classical
theorem scaledGeometricSeries_higherDerivative_tsum_and_absConvergence
(r x : ℝ) (m : ℕ) (hr : 0 < r) (hx : |x| < r) :
r / (r - x) ^ (m + 1) =
∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
x ^ n * (r⁻¹) ^ (n + m) ∧
Summable
(fun n : ℕ =>
|((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
x ^ n * (r⁻¹) ^ (n + m)|)
Proposition 4.2.10: let E ⊆ ℝ, let a be an interior point of E, and let f : E → ℝ be real analytic at a with a power-series expansion f(x) = ∑' n, c n * (x-a)^n converging for all x ∈ (a-r, a+r) for some r > 0. If (b-s, b+s) with s > 0 is any open interval contained in (a-r, a+r), then |a-b| ≤ r-s; in particular |a-b| < r.
open Classical
theorem center_distance_le_radius_subradius_of_interval_subset
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)) :
Proposition 4.2.11: under the same hypotheses and notation as the preceding power-series setup, for every ε with 0 < ε < r there exists C > 0 such that |c n| ≤ C * (r - ε)^{-n} for all integers n ≥ 0 (encoded with n : ℕ).
open Classical
theorem powerSeries_coeff_norm_le_of_subradius
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
(ε : ℝ) (hε0 : 0 < ε) (hεr : ε < r) :
∃ C : ℝ, 0 < C ∧ ∀ n : ℕ, |c n| ≤ C * ((r - ε)⁻¹) ^ n
Proposition 4.2.12: under the hypotheses and notation of the local power-series setup, if (b-s, b+s) ⊆ (a-r, a+r) with s > 0, then for every m : ℕ the shifted coefficient series ∑' n, ((n+m)!/(m!n!)) * (b-a)^n * c (n+m) (equivalently the original sum from n = m) is absolutely convergent; hence the corresponding d_m is well-defined.
open Classical
theorem shifted_coeffSeries_absSummable_of_interval_subset
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)) :
∀ m : ℕ,
Summable
(fun n : ℕ =>
|((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)|)
Helper for Proposition 4.2.6: smoothness of the shifted monomial with constant factor.
open Classical
lemma helperForProposition_4_2_6_contDiff_shiftedMonomial_constMul
(n : ℕ) (a c : ℝ) :
ContDiff ℝ ⊤ (fun x : ℝ => c * (x - a) ^ n)
Helper for Proposition 4.2.6: explicit iterated derivative formula for (x - a)^n with falling-product coefficient.
open Classical
lemma helperForProposition_4_2_6_iteratedDeriv_shiftedMonomial_fallingProd
(n k : ℕ) (a x : ℝ) :
iteratedDeriv k (fun y : ℝ => (y - a) ^ n) x =
Finset.prod (Finset.range k) (fun i => ((n : ℝ) - i)) * (x - a) ^ (n - k)
Helper for Proposition 4.2.6: pull the constant factor c out of iterated derivatives of the shifted monomial.
open Classical
lemma helperForProposition_4_2_6_iteratedDeriv_constMul_shiftedMonomial_fallingProd
(n k : ℕ) (a c x : ℝ) :
iteratedDeriv k (fun y : ℝ => c * (y - a) ^ n) x =
c * (Finset.prod (Finset.range k) (fun i => ((n : ℝ) - i)) * (x - a) ^ (n - k))
Helper for Proposition 4.2.6: convert the falling-product coefficient to the factorial-ratio coefficient when k ≤ n.
open Classical
lemma helperForProposition_4_2_6_fallingProd_eq_factorialRatio
(n : ℕ) {k : ℕ} (hk : k ≤ n) :
Finset.prod (Finset.range k) (fun i => ((n : ℝ) - i)) =
(Nat.factorial n : ℝ) / (Nat.factorial (n - k) : ℝ)
Helper for Proposition 4.2.6: the falling-product coefficient vanishes when k > n.
open Classical
lemma helperForProposition_4_2_6_fallingProd_eq_zero_of_lt
(n : ℕ) {k : ℕ} (hnk : n < k) :
Finset.prod (Finset.range k) (fun i => ((n : ℝ) - i)) = 0
Helper for Proposition 4.2.8: convert |x| < r with r > 0 into the geometric-series ratio bound |x / r| < 1.
open Classical
lemma helperForProposition_4_2_8_absRatio_lt_one
(r x : ℝ) (hr : 0 < r) (hx : |x| < r) :
Helper for Proposition 4.2.8: rewrite powers of x / r into x^n * (r⁻¹)^n.
open Classical
lemma helperForProposition_4_2_8_geometricRatioPow_rewrite
(r x : ℝ) :
(fun n : ℕ => (x / r) ^ n) = (fun n : ℕ => x ^ n * (r⁻¹) ^ n)
Helper for Proposition 4.2.8: identify the scaled rational expression with the closed form of the geometric series.
open Classical
lemma helperForProposition_4_2_8_scaledFraction_eq_geometricClosedForm
(r x : ℝ) (hr : 0 < r) (hx : |x| < r) :
r / (r - x) = (1 - x / r)⁻¹
Helper for Proposition 4.2.9: rewrite the factorial-ratio summand into a constant multiple of the standard choose-geometric term.
open Classical
lemma helperForProposition_4_2_9_factorialRatioTerm_eq_scaledChooseGeometricTerm
(r x : ℝ) (m n : ℕ) :
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
x ^ n * (r⁻¹) ^ (n + m) =
(r⁻¹) ^ m * ((Nat.choose (n + m) m : ℝ) * (x / r) ^ n)
Helper for Proposition 4.2.9: rewrite the left-hand side as the scaled closed form matching the choose-geometric generating function.
open Classical
lemma helperForProposition_4_2_9_lhs_eq_scaledClosedForm
(r x : ℝ) (m : ℕ) (hr : 0 < r) (hx : |x| < r) :
r / (r - x) ^ (m + 1) = (r⁻¹) ^ m * (1 / (1 - x / r) ^ (m + 1))
Helper for Proposition 4.2.9: evaluate the reindexed factorial-ratio series as the same scaled closed form.
open Classical
lemma helperForProposition_4_2_9_tsum_targetTerm_eq_scaledClosedForm
(r x : ℝ) (m : ℕ) (hr : 0 < r) (hx : |x| < r) :
(∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
x ^ n * (r⁻¹) ^ (n + m)) =
(r⁻¹) ^ m * (1 / (1 - x / r) ^ (m + 1))
Helper for Proposition 4.2.9: absolute convergence of the reindexed factorial-ratio series for |x| < r.
open Classical
lemma helperForProposition_4_2_9_summable_abs_targetTerm
(r x : ℝ) (m : ℕ) (hr : 0 < r) (hx : |x| < r) :
Summable
(fun n : ℕ =>
|((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
x ^ n * (r⁻¹) ^ (n + m)|)
Helper for Proposition 4.2.10: if s > 0 then (b - s, b + s) is nonempty.
open Classical
lemma helperForProposition_4_2_10_interval_left_right_strict
(b s : ℝ) (hs : 0 < s) :
b - s < b + s
Helper for Proposition 4.2.10: inclusion of open intervals yields endpoint bounds.
open Classical
lemma helperForProposition_4_2_10_endpoint_bounds_of_Ioo_subset
(a0 b r s : ℝ) (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo (a0 - r) (a0 + r)) :
(a0 - r ≤ b - s) ∧ (b + s ≤ a0 + r)
Helper for Proposition 4.2.10: endpoint bounds control center distance by r - s.
open Classical
lemma helperForProposition_4_2_10_abs_le_radius_subradius
(a0 b r s : ℝ)
(hleft : a0 - r ≤ b - s) (hright : b + s ≤ a0 + r) :
Helper for Proposition 4.2.10: a bound by r - s improves to a strict bound by r.
open Classical
lemma helperForProposition_4_2_10_abs_lt_radius_of_subradius_bound
(a0 b r s : ℝ) (habs : |a0 - b| ≤ r - s) (hs : 0 < s) :
Helper for Proposition 4.2.11: the point a + (r - ε) lies in (a-r, a+r) whenever 0 < ε < r.
open Classical
lemma helperForProposition_4_2_11_subradius_point_mem_interval
(a r ε : ℝ) (hr : 0 < r) (hε0 : 0 < ε) (hεr : ε < r) :
a + (r - ε) ∈ Set.Ioo (a - r) (a + r)
Helper for Proposition 4.2.11: evaluating the interval summability hypothesis at x = a + (r - ε) yields summability of n ↦ c n * (r - ε)^n.
open Classical
lemma helperForProposition_4_2_11_summable_scaled_terms_at_subradius
{E : Set ℝ} (a : E) {r ε : ℝ} (c : ℕ → ℝ)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hr : 0 < r) (hε0 : 0 < ε) (hεr : ε < r) :
Summable (fun n : ℕ => c n * (r - ε) ^ n)
Helper for Proposition 4.2.11: every summable real sequence has a uniform absolute-value bound.
open Classical
lemma helperForProposition_4_2_11_exists_abs_bound_of_summable
(u : ℕ → ℝ) (hu : Summable u) :
∃ B : ℝ, ∀ n : ℕ, |u n| ≤ B
Helper for Proposition 4.2.11: from a uniform bound on |c n * ρ^n|, recover a coefficient bound with inverse powers.
open Classical
lemma helperForProposition_4_2_11_coeff_bound_from_scaled_bound
(c : ℕ → ℝ) {ρ K : ℝ} (hρ : 0 < ρ)
(hK : ∀ n : ℕ, |c n * ρ ^ n| ≤ K) :
∀ n : ℕ, |c n| ≤ K * (ρ⁻¹) ^ n4.2 — Real Analytic Functions — Part 4
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Proposition 4.2.13: assume the hypotheses and notation of the shifted power-series setup centered at a, and let d m = ∑' n, ((n+m)!/(m!n!)) * (b-a)^n * c (n+m), with this defining series convergent for each m. Then for every ε with 0 < ε < s, there exists C > 0 such that |d m| ≤ C * (s - ε)^{-m} for all integers m ≥ 0 (encoded as m : ℕ).
open Classical
theorem shifted_coefficients_norm_le_of_subradius
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(d : ℕ → ℝ)
(hdSummable :
∀ m : ℕ,
Summable
(fun n : ℕ =>
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)))
(hd :
∀ m : ℕ,
d m =
∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m))
(ε : ℝ) (hε0 : 0 < ε) (hεs : ε < s) :
∃ C : ℝ, 0 < C ∧ ∀ m : ℕ, |d m| ≤ C * ((s - ε)⁻¹) ^ m
Proposition 4.2.14: assume the hypotheses and notation of the shifted power-series setup (with coefficients d). Then for every x ∈ (b-s, b+s), the power series ∑' m, d m * (x-b)^m is absolutely convergent, and one has f(x) = ∑' m, d m * (x-b)^m on that interval.
open Classical
theorem shifted_powerSeries_absConvergent_and_eq_on_interval
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(d : ℕ → ℝ)
(hdSummable :
∀ m : ℕ,
Summable
(fun n : ℕ =>
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)))
(hd :
∀ m : ℕ,
d m =
∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)) :
∀ x (hx : x ∈ Set.Ioo (b - s) (b + s)),
Summable (fun m : ℕ => |d m * (x - b) ^ m|) ∧
f ⟨x, hI (hsub hx)⟩ =
∑' m : ℕ, d m * (x - b) ^ m
Proposition 4.2.15: assume the hypotheses and notation of the local power-series setup centered at a. Then f is real analytic at every b ∈ (a-r, a+r).
open Classical
theorem realAnalyticAt_every_point_of_localPowerSeriesInterval
{E : Set ℝ} (a : E) (f : E → ℝ)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n) :
∀ b (hb : b ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
IsRealAnalyticAt E f ⟨b, hI hb⟩
Helper for Proposition 4.2.14: for a point in (b-s, b+s), one can choose ε with 0 < ε < s and still keep a strict gap |x-b| < s-ε.
open Classical
lemma helperForProposition_4_2_14_exists_subradius_gap_at_point
{b s x : ℝ} (hs : 0 < s)
(hx : x ∈ Set.Ioo (b - s) (b + s)) :
∃ ε : ℝ, 0 < ε ∧ ε < s ∧ |x - b| < s - ε
Helper for Proposition 4.2.14: the shifted series is absolutely summable at every point of (b-s, b+s).
open Classical
lemma helperForProposition_4_2_14_absSummable_shifted_series_at_point
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(d : ℕ → ℝ)
(hdSummable :
∀ m : ℕ,
Summable
(fun n : ℕ =>
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)))
(hd :
∀ m : ℕ,
d m =
∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)) :
∀ x (hx : x ∈ Set.Ioo (b - s) (b + s)),
Summable (fun m : ℕ => |d m * (x - b) ^ m|)
Helper for Proposition 4.2.14: expand the original power series around a at a point x ∈ (b-s, b+s) into a triangular finite sum in powers of (x - b).
open Classical
lemma helperForProposition_4_2_14_expand_original_series_termwise
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(x : ℝ) (hx : x ∈ Set.Ioo (b - s) (b + s)) :
f ⟨x, hI (hsub hx)⟩ =
∑' n : ℕ,
Finset.sum (Finset.range (n + 1)) (fun m =>
((Nat.factorial n : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial (n - m) : ℝ))) *
(b - (a : ℝ)) ^ (n - m) * c n * (x - b) ^ m)
Helper for Proposition 4.2.14: each row sum in the triangular rearrangement is exactly d m * (x - b)^m.
open Classical
lemma helperForProposition_4_2_14_row_tsum_eq_d_times_shiftedPower
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(d : ℕ → ℝ)
(hdSummable :
∀ m : ℕ,
Summable
(fun n : ℕ =>
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)))
(hd :
∀ m : ℕ,
d m =
∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m))
(x : ℝ) (m : ℕ) :
(∑' n : ℕ,
(((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)) *
(x - b) ^ m) =
d m * (x - b) ^ m
Helper for Proposition 4.2.14: if a kernel on ℕ × ℕ is summable, then the triangular range-indexed tsum can be rewritten as the corresponding product-indexed tsum via antidiagonal reindexing.
open Classical
lemma helperForProposition_4_2_14_triangular_reindex_tsum_via_prod
(K : ℕ → ℕ → ℝ)
(hK : Summable (fun p : ℕ × ℕ => K p.1 p.2)) :
(∑' n : ℕ, Finset.sum (Finset.range (n + 1)) (fun m => K m (n - m))) =
∑' p : ℕ × ℕ, K p.1 p.2
Helper for Proposition 4.2.14: once a product-indexed kernel is summable, its tsum can be identified with the shifted-coefficient series by evaluating row tsums.
open Classical
lemma helperForProposition_4_2_14_prod_rows_to_shifted_coeff_tsum
(K : ℕ → ℕ → ℝ) (d : ℕ → ℝ) (x b : ℝ)
(hK : Summable (fun p : ℕ × ℕ => K p.1 p.2))
(hRows : ∀ m : ℕ, (∑' n : ℕ, K m n) = d m * (x - b) ^ m) :
(∑' p : ℕ × ℕ, K p.1 p.2) = ∑' m : ℕ, d m * (x - b) ^ m
Helper for Proposition 4.2.14: for fixed x ∈ (b-s, b+s), the product kernel (m,k) ↦ ((m+k)!/(m!k!)) * (b-a)^k * c (m+k) * (x-b)^m is summable on ℕ × ℕ.
open Classical
lemma helperForProposition_4_2_14_summable_prod_shifted_kernel_at_point
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(x : ℝ) (hx : x ∈ Set.Ioo (b - s) (b + s)) :
Summable
(fun p : ℕ × ℕ =>
((Nat.factorial (p.1 + p.2) : ℝ) /
((Nat.factorial p.1 : ℝ) * (Nat.factorial p.2 : ℝ))) *
(b - (a : ℝ)) ^ p.2 * c (p.1 + p.2) * (x - b) ^ p.1)
Helper for Proposition 4.2.14: the triangular expansion obtained from the original series can be reordered into the shifted coefficient series.
open Classical
lemma helperForProposition_4_2_14_triangular_tsum_to_shifted_coeff_tsum
{E : Set ℝ} (a : E) (f : E → ℝ)
(hanalytic : IsRealAnalyticAt E f a)
{r : ℝ} (hr : 0 < r) (c : ℕ → ℝ)
(hI : Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r) ⊆ E)
(hsummable :
∀ x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r),
Summable (fun n : ℕ => c n * (x - (a : ℝ)) ^ n))
(hpowerSeries :
∀ x (hx : x ∈ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r)),
f ⟨x, hI hx⟩ = ∑' n : ℕ, c n * (x - (a : ℝ)) ^ n)
{b s : ℝ} (hs : 0 < s)
(hsub : Set.Ioo (b - s) (b + s) ⊆ Set.Ioo ((a : ℝ) - r) ((a : ℝ) + r))
(d : ℕ → ℝ)
(hdSummable :
∀ m : ℕ,
Summable
(fun n : ℕ =>
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m)))
(hd :
∀ m : ℕ,
d m =
∑' n : ℕ,
((Nat.factorial (n + m) : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial n : ℝ))) *
(b - (a : ℝ)) ^ n * c (n + m))
(x : ℝ) (hx : x ∈ Set.Ioo (b - s) (b + s)) :
(∑' n : ℕ,
Finset.sum (Finset.range (n + 1)) (fun m =>
((Nat.factorial n : ℝ) /
((Nat.factorial m : ℝ) * (Nat.factorial (n - m) : ℝ))) *
(b - (a : ℝ)) ^ (n - m) * c n * (x - b) ^ m)) =
∑' m : ℕ, d m * (x - b) ^ m4.3 — Abel's Theorem
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Lemma 4.3.1 (Summation by parts formula): Let a, b : ℕ → ℝ with Tendsto a atTop (𝓝 A) and Tendsto b atTop (𝓝 B). If the series ∑' n, (a (n + 1) - a n) * b n converges, then the ordered partial sums of ∑ a (n + 1) * (b (n + 1) - b n) converge to A * B - a 0 * b 0 - ∑' n, (a (n + 1) - a n) * b n.
open scoped BigOperators
lemma summation_by_parts_formula
(a b : ℕ → ℝ) (A B : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n)) :
Filter.Tendsto
(fun N : ℕ => ∑ n ∈ Finset.range N, a (n + 1) * (b (n + 1) - b n))
Filter.atTop
(nhds (A * B - a 0 * b 0 - ∑' n : ℕ, (a (n + 1) - a n) * b n))
Theorem 4.3.1 (Abel's theorem): let f(x) = ∑' n, c_n (x-a)^n be a real power series centered at a with radius of convergence R and 0 < R. (1) If ∑' n, c_n R^n converges, then x ↦ ∑' n, c_n (x-a)^n tends to ∑' n, c_n R^n as x → a + R within (a - R, a + R). (2) If ∑' n, c_n (-R)^n converges, then x ↦ ∑' n, c_n (x-a)^n tends to ∑' n, c_n (-R)^n as x → a - R within (a - R, a + R).
open scoped BigOperators
theorem FormalPowerSeriesCenteredAt.abel_theorem {a R : ℝ}
(f : FormalPowerSeriesCenteredAt a)
(hRpos : 0 < R)
(hRadius : f.radiusOfConvergence = ENNReal.ofReal R) :
(Summable (fun n : ℕ => f.coeff n * R ^ n) →
Filter.Tendsto (fun x : ℝ => ∑' n : ℕ, f.coeff n * (x - a) ^ n)
(nhdsWithin (a + R) (Set.Ioo (a - R) (a + R)))
(nhds (∑' n : ℕ, f.coeff n * R ^ n))) ∧
(Summable (fun n : ℕ => f.coeff n * (-R) ^ n) →
Filter.Tendsto (fun x : ℝ => ∑' n : ℕ, f.coeff n * (x - a) ^ n)
(nhdsWithin (a - R) (Set.Ioo (a - R) (a + R)))
(nhds (∑' n : ℕ, f.coeff n * (-R) ^ n)))
Lemma 4.3.2: For all z₁, z₂ ∈ ℂ, we have z₁ * z₂ = z₂ * z₁ (equation (4.230)).
open scoped BigOperators
lemma complex_mul_comm (z₁ z₂ : ℂ) : z₁ * z₂ = z₂ * z₁
Lemma 4.3.3: For all z₁, z₂, z₃ ∈ ℂ, we have (z₁ * z₂) * z₃ = z₁ * (z₂ * z₃) (equation (4.u230)).
open scoped BigOperators
lemma complex_mul_assoc (z₁ z₂ z₃ : ℂ) : (z₁ * z₂) * z₃ = z₁ * (z₂ * z₃)
Lemma 4.3.4: For all z ∈ ℂ, we have z * 1_ℂ = z and 1_ℂ * z = z (equation (4.u230)).
open scoped BigOperators
lemma complex_mul_one_and_one_mul (z : ℂ) : z * (1 : ℂ) = z ∧ (1 : ℂ) * z = z
Lemma 4.3.5: For all z₁, z₂, z₃ ∈ ℂ, we have z₁ * (z₂ + z₃) = z₁ * z₂ + z₁ * z₃ (equation (4.u230)).
open scoped BigOperators
lemma complex_mul_add_left_distrib (z₁ z₂ z₃ : ℂ) :
z₁ * (z₂ + z₃) = z₁ * z₂ + z₁ * z₃
Lemma 4.3.6: For all z₁, z₂, z₃ ∈ ℂ, we have (z₂ + z₃) * z₁ = z₂ * z₁ + z₃ * z₁ (equation (4.u230)).
open scoped BigOperators
lemma complex_add_mul_right_distrib (z₁ z₂ z₃ : ℂ) :
(z₂ + z₃) * z₁ = z₂ * z₁ + z₃ * z₁
Helper for Lemma 4.3.1: one-step algebraic summation-by-parts identity.
open scoped BigOperators
lemma helperForLemma_4_3_1_stepIdentity (a b : ℕ → ℝ) (n : ℕ) :
(a (n + 1) - a n) * b n + a (n + 1) * (b (n + 1) - b n)
= a (n + 1) * b (n + 1) - a n * b n
Helper for Lemma 4.3.1: telescoping finite sum of the two-by-parts terms.
open scoped BigOperators
lemma helperForLemma_4_3_1_telescoping (a b : ℕ → ℝ) :
∀ N : ℕ,
(∑ n ∈ Finset.range N,
((a (n + 1) - a n) * b n + a (n + 1) * (b (n + 1) - b n)))
= a N * b N - a 0 * b 0
Helper for Lemma 4.3.1: finite summation-by-parts identity over Finset.range N.
open scoped BigOperators
lemma helperForLemma_4_3_1_finiteFormula (a b : ℕ → ℝ) (N : ℕ) :
(∑ n ∈ Finset.range N, (a (n + 1) - a n) * b n)
= a N * b N - a 0 * b 0 - ∑ n ∈ Finset.range N, a (n + 1) * (b (n + 1) - b n)
Helper for Lemma 4.3.1: convergence of ordered partial sums of a (n+1) * (b (n+1) - b n) to the summation-by-parts limit value.
open scoped BigOperators
lemma helperForLemma_4_3_1_tendsto_partialSums
(a b : ℕ → ℝ) (A B : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n)) :
Filter.Tendsto
(fun N : ℕ => ∑ n ∈ Finset.range N, a (n + 1) * (b (n + 1) - b n))
Filter.atTop
(nhds (A * B - a 0 * b 0 - ∑' n : ℕ, (a (n + 1) - a n) * b n))
Helper for Lemma 4.3.1: if the forward-difference series has a prescribed sum S, the summation-by-parts identity holds with that same S.
open scoped BigOperators
lemma helperForLemma_4_3_1_identity_of_hasSumForwardDifference
(a b : ℕ → ℝ) (A B S : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n))
(hhasSumForward : HasSum (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) S) :
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - S
Helper for Lemma 4.3.1: a HasSum hypothesis for the forward-difference series immediately yields summability and the corresponding summation-by-parts identity.
open scoped BigOperators
lemma helperForLemma_4_3_1_formula_with_hasSumForwardDifference
(a b : ℕ → ℝ) (A B S : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n))
(hhasSumForward : HasSum (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) S) :
Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - S
Helper for Lemma 4.3.1: an explicit forward-difference HasSum hypothesis already yields the textbook identity written with tsum.
open scoped BigOperators
lemma helperForLemma_4_3_1_tsumIdentity_of_hasSumForwardDifference
(a b : ℕ → ℝ) (A B S : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n))
(hhasSumForward : HasSum (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) S) :
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n)
Helper for Lemma 4.3.1: a corrected summation-by-parts statement with an explicit forward HasSum hypothesis gives both forward summability and the textbook tsum identity.
open scoped BigOperators
lemma helperForLemma_4_3_1_formula_with_explicitForwardHasSum
(a b : ℕ → ℝ) (A B S : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n))
(hhasSumForward : HasSum (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) S) :
Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n)
Helper for Lemma 4.3.1: unconditional forward-difference summability supplies the canonical HasSum witness at its tsum.
open scoped BigOperators
lemma helperForLemma_4_3_1_hasSumForwardDifference_of_summable
(a b : ℕ → ℝ)
(hsummableForward : Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))) :
HasSum (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))
(∑' n : ℕ, a (n + 1) * (b (n + 1) - b n))
Helper for Lemma 4.3.1: assuming unconditional summability of the forward-difference series, the target summation-by-parts identity follows from uniqueness of the partial-sum limit.
open scoped BigOperators
lemma helperForLemma_4_3_1_identity_of_summableForwardDifference
(a b : ℕ → ℝ) (A B : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n))
(hsummableForward : Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))) :
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n)
Helper for Lemma 4.3.1: if forward-difference summability is given explicitly, the full summation-by-parts conclusion follows.
open scoped BigOperators
lemma helperForLemma_4_3_1_formula_with_forwardSummable
(a b : ℕ → ℝ) (A B : ℝ)
(ha : Filter.Tendsto a Filter.atTop (nhds A))
(hb : Filter.Tendsto b Filter.atTop (nhds B))
(hconv : Summable (fun n : ℕ => (a (n + 1) - a n) * b n))
(hsummableForward : Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))) :
Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n)
Helper for Lemma 4.3.1: the alternating harmonic sequence tends to 0.
open scoped BigOperators
lemma helperForLemma_4_3_1_tendsto_alternatingHarmonic :
Filter.Tendsto (fun n : ℕ => (-1 : ℝ) ^ n / (n + 1)) Filter.atTop (nhds 0)
Helper for Lemma 4.3.1: the forward difference of the alternating harmonic sequence is not unconditionally summable in ℝ.
open scoped BigOperators
lemma helperForLemma_4_3_1_notSummable_alternatingHarmonicForwardDifference :
¬ Summable (fun n : ℕ => ((-1 : ℝ) ^ (n + 1) / (n + 2) - (-1 : ℝ) ^ n / (n + 1)))
Helper for Lemma 4.3.1: for b n = (-1)^n/(n+1), the forward-difference series ∑ (b (n+1) - b n) is not summable.
open scoped BigOperators
lemma helperForLemma_4_3_1_notSummable_forwardDifference_of_explicitB :
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
¬ Summable (fun n : ℕ => b (n + 1) - b n)
Helper for Lemma 4.3.1: for the explicit constant-one and alternating-harmonic data, the weighted forward-difference series is not summable.
open scoped BigOperators
lemma helperForLemma_4_3_1_notSummable_weightedForwardDifference_of_explicitData :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
¬ Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))
Helper for Lemma 4.3.1: the explicit constant-one and alternating-harmonic data with concrete limits A = 1, B = 0 satisfy the hypotheses while the weighted forward-difference series is not summable.
open scoped BigOperators
lemma helperForLemma_4_3_1_explicitDataWithConcreteLimits :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
Filter.Tendsto a Filter.atTop (nhds (1 : ℝ)) ∧
Filter.Tendsto b Filter.atTop (nhds (0 : ℝ)) ∧
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) ∧
¬ Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))
Helper for Lemma 4.3.1: for the explicit constant-one and alternating-harmonic data, the ordered partial sums of the weighted forward-difference series converge to -1.
open scoped BigOperators
lemma helperForLemma_4_3_1_explicitWeightedForwardDifference_partialSums_tendsto :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
Filter.Tendsto
(fun N : ℕ => ∑ n ∈ Finset.range N, a (n + 1) * (b (n + 1) - b n))
Filter.atTop
(nhds (-1 : ℝ))
Helper for Lemma 4.3.1: there exists an explicit sequence whose ordered partial sums converge, while the sequence is not Summable in Lean's finite-subset summability sense.
open scoped BigOperators
lemma helperForLemma_4_3_1_existsOrderedPartialSumsLimitWithoutSummable :
∃ (f : ℕ → ℝ) (L : ℝ),
Filter.Tendsto (fun N : ℕ => ∑ n ∈ Finset.range N, f n) Filter.atTop (nhds L) ∧
¬ Summable f
Helper for Lemma 4.3.1: explicit data satisfy the hypotheses while the forward-difference series fails to be unconditionally summable.
open scoped BigOperators
lemma helperForLemma_4_3_1_existsForwardCounterexample :
∃ (a b : ℕ → ℝ) (A B : ℝ),
Filter.Tendsto a Filter.atTop (nhds A) ∧
Filter.Tendsto b Filter.atTop (nhds B) ∧
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) ∧
¬ Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n))
Helper for Lemma 4.3.1: the stated hypotheses do not force unconditional summability of the forward-difference series in Lean's Summable sense.
open scoped BigOperators
lemma helperForLemma_4_3_1_universalForwardSummableFalse :
¬
(∀ (a b : ℕ → ℝ) (A B : ℝ),
Filter.Tendsto a Filter.atTop (nhds A) →
Filter.Tendsto b Filter.atTop (nhds B) →
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) →
Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)))
Helper for Lemma 4.3.1: for the explicit constant-one and alternating-harmonic data at concrete limits, even the forward-summability implication alone is false.
open scoped BigOperators
lemma helperForLemma_4_3_1_explicitForwardImplicationFalseAtConcreteLimits :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
¬
(Filter.Tendsto a Filter.atTop (nhds (1 : ℝ)) →
Filter.Tendsto b Filter.atTop (nhds (0 : ℝ)) →
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) →
Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)))
Helper for Lemma 4.3.1: the full universal statement of the textbook conclusion would imply a universal forward-summability implication, which is false.
open scoped BigOperators
lemma helperForLemma_4_3_1_universalConclusionFalse :
¬
(∀ (a b : ℕ → ℝ) (A B : ℝ),
Filter.Tendsto a Filter.atTop (nhds A) →
Filter.Tendsto b Filter.atTop (nhds B) →
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) →
(Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n)))
Helper for Lemma 4.3.1: there exists explicit data satisfying the hypotheses for which the full textbook conclusion fails in Lean's unconditional Summable sense.
open scoped BigOperators
lemma helperForLemma_4_3_1_existsCounterexampleToConclusion :
∃ (a b : ℕ → ℝ) (A B : ℝ),
Filter.Tendsto a Filter.atTop (nhds A) ∧
Filter.Tendsto b Filter.atTop (nhds B) ∧
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) ∧
¬
(Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n))
Helper for Lemma 4.3.1: the explicit constant-one and alternating-harmonic data at concrete limits A = 1, B = 0 refute the full textbook conclusion.
open scoped BigOperators
lemma helperForLemma_4_3_1_explicitDataRefutesConclusionAtConcreteLimits :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
¬
(Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
(1 : ℝ) * 0 - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n))
Helper for Lemma 4.3.1: for the explicit constant-one and alternating-harmonic data, convergence of ∑ (a_{n+1} - a_n)b_n does not imply unconditional summability of ∑ a_{n+1}(b_{n+1} - b_n) in Lean.
open scoped BigOperators
lemma helperForLemma_4_3_1_explicitForwardSummableImplicationFalse :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
¬
(Summable (fun n : ℕ => (a (n + 1) - a n) * b n) →
Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)))
Helper for Lemma 4.3.1: any universal proof of the textbook conclusion contradicts the explicit counterexample already established above.
open scoped BigOperators
lemma helperForLemma_4_3_1_universalProofLeadsToFalse
(hUniversal :
∀ (a b : ℕ → ℝ) (A B : ℝ),
Filter.Tendsto a Filter.atTop (nhds A) →
Filter.Tendsto b Filter.atTop (nhds B) →
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) →
(Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
A * B - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n))) :
False
Helper for Lemma 4.3.1: for the explicit constant-one and alternating-harmonic data at concrete limits, the implication from the stated hypotheses to the full textbook conclusion is false.
open scoped BigOperators
lemma helperForLemma_4_3_1_explicitImplicationFalseAtConcreteLimits :
let a : ℕ → ℝ := fun _ => 1
let b : ℕ → ℝ := fun n => (-1 : ℝ) ^ n / (n + 1)
¬
(Filter.Tendsto a Filter.atTop (nhds (1 : ℝ)) →
Filter.Tendsto b Filter.atTop (nhds (0 : ℝ)) →
Summable (fun n : ℕ => (a (n + 1) - a n) * b n) →
(Summable (fun n : ℕ => a (n + 1) * (b (n + 1) - b n)) ∧
(∑' n : ℕ, (a (n + 1) - a n) * b n) =
(1 : ℝ) * 0 - a 0 * b 0 - ∑' n : ℕ, a (n + 1) * (b (n + 1) - b n)))
Helper for Theorem 4.3.1: the right-endpoint affine normalization x ↦ (x-a)/R tends to 1 from below when x → a+R inside (a-R,a+R).
open scoped BigOperators
lemma helperForTheorem_4_3_1_tendsto_affineToUnit_right {a R : ℝ}
(hRpos : 0 < R) :
Filter.Tendsto (fun x : ℝ => (x - a) / R)
(nhdsWithin (a + R) (Set.Ioo (a - R) (a + R)))
(nhdsWithin 1 (Set.Iio 1))
Helper for Theorem 4.3.1: the left-endpoint affine normalization x ↦ (a-x)/R tends to 1 from below when x → a-R inside (a-R,a+R).
open scoped BigOperators
lemma helperForTheorem_4_3_1_tendsto_affineToUnit_left {a R : ℝ}
(hRpos : 0 < R) :
Filter.Tendsto (fun x : ℝ => (a - x) / R)
(nhdsWithin (a - R) (Set.Ioo (a - R) (a + R)))
(nhdsWithin 1 (Set.Iio 1))
Helper for Theorem 4.3.1: right-endpoint term normalization identity.
open scoped BigOperators
lemma helperForTheorem_4_3_1_termRewrite_right {a R : ℝ}
(hRpos : 0 < R) (f : FormalPowerSeriesCenteredAt a) (x : ℝ) (n : ℕ) :
f.coeff n * R ^ n * (((x - a) / R) ^ n) = f.coeff n * (x - a) ^ n
Helper for Theorem 4.3.1: left-endpoint term normalization identity.
open scoped BigOperators
lemma helperForTheorem_4_3_1_termRewrite_left {a R : ℝ}
(hRpos : 0 < R) (f : FormalPowerSeriesCenteredAt a) (x : ℝ) (n : ℕ) :
f.coeff n * (-R) ^ n * (((a - x) / R) ^ n) = f.coeff n * (x - a) ^ n4.4 — Multiplication of Power Series
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
The Cauchy-product coefficient sequence attached to c, d : ℕ → ℝ.
def cauchyProductCoeff (c d : ℕ → ℝ) (n : ℕ) : ℝ :=
∑ m ∈ Finset.range (n + 1), c m * d (n - m)
Theorem 4.4.1: if f, g : (a-r, a+r) → ℝ are analytic and have power-series expansions around a with coefficients c_n and d_n on (a-r, a+r), then fg is analytic on (a-r, a+r), and for every x in that interval, f(x)g(x) = ∑' n, e_n (x-a)^n where e_n = ∑_{m=0}^n c_m d_{n-m}.
theorem realAnalyticOnInterval_mul_eq_cauchyProduct
{a r : ℝ} (hr : 0 < r)
(f g : Set.Ioo (a - r) (a + r) → ℝ)
(hf_analytic : IsRealAnalyticOn (Set.Ioo (a - r) (a + r)) f)
(hg_analytic : IsRealAnalyticOn (Set.Ioo (a - r) (a + r)) g)
(c d : ℕ → ℝ)
(hf_series :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (x - a) ^ n) ∧
f ⟨x, hx⟩ = ∑' n : ℕ, c n * (x - a) ^ n)
(hg_series :
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => d n * (x - a) ^ n) ∧
g ⟨x, hx⟩ = ∑' n : ℕ, d n * (x - a) ^ n) :
IsRealAnalyticOn (Set.Ioo (a - r) (a + r)) (fun x => f x * g x) ∧
∀ x (hx : x ∈ Set.Ioo (a - r) (a + r)),
f ⟨x, hx⟩ * g ⟨x, hx⟩ =
∑' n : ℕ, cauchyProductCoeff c d n * (x - a) ^ n
Helper for Theorem 4.4.1: at any point of (a-r, a+r), the corresponding power-series terms are absolutely summable.
lemma helperForTheorem_4_4_1_absSummable_seriesTerm_at_point
{a r x : ℝ} (c : ℕ → ℝ)
(hc :
∀ y (hy : y ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (y - a) ^ n))
(hx : x ∈ Set.Ioo (a - r) (a + r)) :
Summable (fun n : ℕ => |c n * (x - a) ^ n|)
Helper for Theorem 4.4.1: rewrite the inner Cauchy-product finite sum into cauchyProductCoeff c d n * t^n.
lemma helperForTheorem_4_4_1_cauchyInnerSum_rewrite
(c d : ℕ → ℝ) (t : ℝ) (n : ℕ) :
(∑ k ∈ Finset.range (n + 1),
(c k * t ^ k) * (d (n - k) * t ^ (n - k))) =
cauchyProductCoeff c d n * t ^ n
Helper for Theorem 4.4.1: at each point in (a-r,a+r), the pointwise product equals the Cauchy-product power-series sum.
lemma helperForTheorem_4_4_1_pointwise_product_eq_tsum
{a r x : ℝ} (hr : 0 < r)
(f g : Set.Ioo (a - r) (a + r) → ℝ)
(c d : ℕ → ℝ)
(hf_series :
∀ y (hy : y ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => c n * (y - a) ^ n) ∧
f ⟨y, hy⟩ = ∑' n : ℕ, c n * (y - a) ^ n)
(hg_series :
∀ y (hy : y ∈ Set.Ioo (a - r) (a + r)),
Summable (fun n : ℕ => d n * (y - a) ^ n) ∧
g ⟨y, hy⟩ = ∑' n : ℕ, d n * (y - a) ^ n)
(hx : x ∈ Set.Ioo (a - r) (a + r)) :
f ⟨x, hx⟩ * g ⟨x, hx⟩ =
∑' n : ℕ, cauchyProductCoeff c d n * (x - a) ^ n4.5 — The Exponential and Logarithm Functions — Part 1
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Definition 4.5.1 (Exponential function): for every real number x, the exponential function is given by the power series exp(x) := ∑' n : ℕ, x^n / n!.
open scoped Topology
noncomputable def realExpSeries (x : ℝ) : ℝ :=
∑' n : ℕ, x ^ n / (n.factorial : ℝ)
Definition 4.5.2 (Euler's number): Euler's number e is the real number defined by e := exp(1), with series expansion e = ∑' n : ℕ, 1 / n!.
open scoped Topology
noncomputable def realEulerNumber : ℝ :=
Real.exp 1
Definition 4.5.3 (Natural logarithm): the natural logarithm is the function log : (0, ∞) → ℝ, also denoted ln, defined as the inverse of exp : ℝ → (0, ∞).
open scoped Topology
noncomputable def naturalLog : Set.Ioi (0 : ℝ) → ℝ :=
fun x => Real.log x
The restricted natural logarithm and exponential satisfy exp (naturalLog x) = x for x > 0 and naturalLog (exp y) = y.
open scoped Topology
theorem naturalLog_inverse_exp :
(∀ x : Set.Ioi (0 : ℝ), Real.exp (naturalLog x) = x.1) ∧
(∀ y : ℝ, naturalLog ⟨Real.exp y, Real.exp_pos y⟩ = y)
The defining series ∑ n, 1 / n! for Euler's number is summable in ℝ.
open scoped Topology
lemma summable_one_div_factorial :
Summable (fun n : ℕ => (1 : ℝ) / (n.factorial : ℝ))
Series characterization of Euler's number via the exponential series at x = 1.
open scoped Topology
lemma realEulerNumber_eq_tsum :
realEulerNumber = ∑' n : ℕ, (1 : ℝ) / (n.factorial : ℝ)
Proposition 4.5.1: for every real number x, the exponential function satisfies Real.exp x = realEulerNumber ^ x (that is, exp(x) = e^x).
open scoped Topology
theorem real_exp_eq_eulerNumber_rpow :
∀ x : ℝ, Real.exp x = realEulerNumber ^ x
Theorem 4.5.1 (Basic properties of exp, part (a)): for every real number x, the series ∑' n : ℕ, x^n / n! is absolutely convergent. In particular, Real.exp x = ∑' n : ℕ, x^n / n! for all x, the associated power series has infinite radius of convergence, and Real.exp is real analytic on ℝ.
open scoped Topology
theorem exp_basic_properties_part_a :
(∀ x : ℝ, Summable (fun n : ℕ => |x ^ n / (n.factorial : ℝ)|)) ∧
(∀ x : ℝ, Real.exp x = ∑' n : ℕ, x ^ n / (n.factorial : ℝ)) ∧
(NormedSpace.expSeries ℝ ℝ).radius = ⊤ ∧
AnalyticOn ℝ Real.exp Set.univ
Theorem 4.5.2 (Basic properties of exp, part (b)): the function exp : ℝ → ℝ is differentiable on ℝ, and for every x : ℝ, we have exp'(x) = exp(x).
open scoped Topology
theorem exp_basic_properties_part_b :
Differentiable ℝ Real.exp ∧ ∀ x : ℝ, deriv Real.exp x = Real.exp x
Theorem 4.5.3 (Basic properties of exp, part (c)): the function exp is continuous on ℝ. Moreover, for every a, b : ℝ, ∫ x in a..b, exp x = exp b - exp a.
open scoped Topology
theorem exp_basic_properties_part_c :
Continuous Real.exp ∧
∀ a b : ℝ, (∫ x in a..b, Real.exp x) = Real.exp b - Real.exp a
Theorem 4.5.4 (Basic properties of exp, part (d)): for all x, y : ℝ, exp (x + y) = exp x * exp y.
open scoped Topology
theorem exp_basic_properties_part_d :
∀ x y : ℝ, Real.exp (x + y) = Real.exp x * Real.exp y
Theorem 4.5.5 (Basic properties of exp, part (e)): exp 0 = 1. Moreover, for every x : ℝ, we have exp x > 0 and exp (-x) = 1 / exp x.
open scoped Topology
theorem exp_basic_properties_part_e :
Real.exp 0 = 1 ∧ ∀ x : ℝ, Real.exp x > 0 ∧ Real.exp (-x) = 1 / Real.exp x
Theorem 4.5.6 (Basic properties of exp, part (f)): the function exp is strictly increasing on ℝ. Equivalently, for all x, y : ℝ, y > x ↔ Real.exp y > Real.exp x.
open scoped Topology
theorem exp_basic_properties_part_f :
StrictMono Real.exp ∧ ∀ x y : ℝ, y > x ↔ Real.exp y > Real.exp x
Theorem 4.5.7 (Logarithm properties, part (a)): for every x > 0, the derivative of ln is ln'(x) = 1 / x. In particular, for any 0 < a < b, ∫ x in a..b, (1 / x) = ln(b) - ln(a).
open scoped Topology
theorem logarithm_properties_part_a :
(∀ x : ℝ, 0 < x → deriv Real.log x = 1 / x) ∧
(∀ a b : ℝ, 0 < a → a < b → (∫ x in a..b, (1 / x)) = Real.log b - Real.log a)
Theorem 4.5.8 (Logarithm properties, part (b)): for all x, y > 0, ln (xy) = ln x + ln y.
open scoped Topology
theorem logarithm_properties_part_b :
∀ x y : ℝ, 0 < x → 0 < y → Real.log (x * y) = Real.log x + Real.log y
Theorem 4.5.9 (Logarithm properties, part (c)): ln(1) = 0. Moreover, for every x > 0, ln(1 / x) = -ln(x).
open scoped Topology
theorem logarithm_properties_part_c :
Real.log 1 = 0 ∧ ∀ x : ℝ, 0 < x → Real.log (1 / x) = -Real.log x
Theorem 4.5.10 (Logarithm properties, part (d)): for any x > 0 and y : ℝ, one has ln (x^y) = y ln x, where x^y is defined by x^y = exp (y ln x).
open scoped Topology
theorem logarithm_properties_part_d :
∀ x y : ℝ, 0 < x → Real.log (x ^ y) = y * Real.log x
open scoped Topology
theorem logarithm_properties_part_e :
(∀ x : ℝ, x ∈ Set.Ioo (-1 : ℝ) 1 →
Real.log (1 - x) =
-∑' n : ℕ, x ^ (n + 1) / ((n + 1 : ℕ) : ℝ)) ∧
AnalyticAt ℝ Real.log 1 ∧
(∀ x : ℝ, x ∈ Set.Ioo (0 : ℝ) 2 →
Real.log x =
∑' n : ℕ,
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1)) ∧
(∀ x : ℝ, |x - 1| < 1 →
Summable
(fun n : ℕ =>
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1))) ∧
(∀ x : ℝ, 1 < |x - 1| →
¬Summable
(fun n : ℕ =>
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1)))
Proposition 4.5.2: the alternating harmonic series converges to ln(2), encoded in Lean as convergence of ordered partial sums: ∑_{n=1}^∞ (-1)^{n+1}/n = ln(2) becomes Tendsto (fun N => ∑ n < N, (-1)^n/(n+1)) atTop (𝓝 (log 2)).
open scoped Topology
theorem alternating_harmonic_summable_and_tsum_eq_log_two :
Filter.Tendsto
(fun N : ℕ => ∑ n ∈ Finset.range N, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ))
Filter.atTop
(nhds (Real.log 2))
Helper for Proposition 4.5.1: rewrite Real.exp x as (Real.exp 1) ^ x.
open scoped Topology
lemma helperForProposition_4_5_1_exp_eq_expOne_rpow (x : ℝ) :
Real.exp x = (Real.exp 1) ^ x
Helper for Proposition 4.5.1: rewrite (Real.exp 1) ^ x using realEulerNumber.
open scoped Topology
lemma helperForProposition_4_5_1_realEulerNumber_rpow_rewrite (x : ℝ) :
(Real.exp 1) ^ x = realEulerNumber ^ x
Helper for Theorem 4.5.7: derivative of Real.log at positive inputs.
open scoped Topology
lemma helperForTheorem_4_5_7_deriv_log_of_pos (x : ℝ) (_hx : 0 < x) :
deriv Real.log x = 1 / x
Helper for Theorem 4.5.7: positivity of the right endpoint from 0 < a < b.
open scoped Topology
lemma helperForTheorem_4_5_7_rightEndpoint_pos
(a b : ℝ) (ha : 0 < a) (hab : a < b) :
0 < b
Helper for Theorem 4.5.7: integral of 1 / x on a positive interval.
open scoped Topology
lemma helperForTheorem_4_5_7_integral_one_div_eq_log_div
(a b : ℝ) (ha : 0 < a) (hab : a < b) :
(∫ x in a..b, (1 / x)) = Real.log (b / a)
Helper for Theorem 4.5.7: convert log (b / a) into log b - log a.
open scoped Topology
lemma helperForTheorem_4_5_7_log_div_eq_sub
(a b : ℝ) (ha : 0 < a) (hab : a < b) :
Real.log (b / a) = Real.log b - Real.log a
Helper for Theorem 4.5.9: the logarithm of 1 is 0.
open scoped Topology
lemma helperForTheorem_4_5_9_log_one : Real.log 1 = 0
Helper for Theorem 4.5.9: reciprocal inside log gives negation.
open scoped Topology
lemma helperForTheorem_4_5_9_log_one_div_eq_neg_log (x : ℝ) :
Real.log (1 / x) = -Real.log x
Helper for Theorem 4.5.9: the reciprocal-log identity in the x > 0 form.
open scoped Topology
lemma helperForTheorem_4_5_9_log_one_div_eq_neg_log_of_pos (x : ℝ) (_hx : 0 < x) :
Real.log (1 / x) = -Real.log x
Helper for Theorem 4.5.10: rewrite x ^ y with the exp (y * log x) definition when x > 0.
open scoped Topology
lemma helperForTheorem_4_5_10_rpow_def_of_pos {x y : ℝ} (hx : 0 < x) :
x ^ y = Real.exp (y * Real.log x)
Helper for Theorem 4.5.10: the logarithm of a real power for positive base.
open scoped Topology
lemma helperForTheorem_4_5_10_log_rpow_of_pos {x y : ℝ} (hx : 0 < x) :
Real.log (x ^ y) = y * Real.log x
Helper for Theorem 4.5.11: the logarithm series identity on (-1, 1).
open scoped Topology
lemma helperForTheorem_4_5_11_log_one_sub_eq_neg_tsum_of_mem_Ioo_neg_one_one
{x : ℝ} (hx : x ∈ Set.Ioo (-1 : ℝ) 1) :
Real.log (1 - x) = -∑' n : ℕ, x ^ (n + 1) / ((n + 1 : ℕ) : ℝ)
Helper for Theorem 4.5.11: rewrite the centered logarithm-series term.
open scoped Topology
lemma helperForTheorem_4_5_11_term_rewrite_center_one (x : ℝ) (n : ℕ) :
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1) =
-((1 - x) ^ (n + 1) / ((n + 1 : ℕ) : ℝ))
Helper for Theorem 4.5.11: on (0, 2), log equals its centered power series at 1, and that series is summable.
open scoped Topology
lemma helperForTheorem_4_5_11_log_eq_and_summable_inside_radius
{x : ℝ} (hx : x ∈ Set.Ioo (0 : ℝ) 2) :
(Real.log x = ∑' n : ℕ,
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1)) ∧
Summable (fun n : ℕ =>
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1))
Helper for Theorem 4.5.11: outside radius 1 about 1, the centered logarithm series is not summable.
open scoped Topology
lemma helperForTheorem_4_5_11_not_summable_outside_radius
{x : ℝ} (hx : 1 < |x - 1|) :
¬Summable (fun n : ℕ =>
((-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) * (x - 1) ^ (n + 1))
Helper for Proposition 4.5.2: each alternating-harmonic term has absolute value 1 / (n + 1).
open scoped Topology
lemma helperForProposition_4_5_2_abs_alternating_harmonic_term
(n : ℕ) :
Helper for Proposition 4.5.2: unconditional summability of the alternating harmonic series implies summability of the harmonic series.
open scoped Topology
lemma helperForProposition_4_5_2_summable_alternating_implies_summable_harmonic :
Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) →
Summable (fun n : ℕ => (1 : ℝ) / (n : ℝ))
Helper for Proposition 4.5.2: under Lean's default (unconditional) Summable, the alternating harmonic series is not summable.
open scoped Topology
lemma helperForProposition_4_5_2_not_summable_unconditional_alternating_harmonic :
¬ Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ))
Helper for Proposition 4.5.2: any putative witness of the target conjunction would force a Summable claim that is already refuted under Lean's unconditional Summable semantics on ℕ.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_implies_not_summable :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) →
¬ Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ))
Helper for Proposition 4.5.2: Real.log 2 is nonzero.
open scoped Topology
lemma helperForProposition_4_5_2_log_two_ne_zero :
Real.log 2 ≠ (0 : ℝ)
Helper for Proposition 4.5.2: under Lean's default unconditional Summable semantics on ℕ, the alternating-harmonic tsum is forced to be 0.
open scoped Topology
lemma helperForProposition_4_5_2_tsum_eq_zero_unconditional :
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = 0
Helper for Proposition 4.5.2: under Lean's default unconditional Summable semantics on ℕ, the asserted tsum = log 2 identity cannot hold.
open scoped Topology
lemma helperForProposition_4_5_2_not_tsum_eq_log_two_unconditional :
¬ ((∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2)
Helper for Proposition 4.5.2: any proof of the target conjunction forces Real.log 2 = 0 under unconditional tsum semantics.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_implies_log_two_eq_zero :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) →
Real.log 2 = 0
Helper for Proposition 4.5.2: any proof of the target conjunction directly contradicts Real.log 2 ≠ 0.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_implies_false_via_log_two :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) → False
Helper for Proposition 4.5.2: the target conjunction is inconsistent under Lean's default (unconditional) Summable on ℕ.
open scoped Topology
lemma helperForProposition_4_5_2_not_target_conjunction_unconditional :
¬ (Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2)
Helper for Proposition 4.5.2: any proof of the target conjunction would force summability of the (unshifted) harmonic series.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_implies_summable_harmonic :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) →
Summable (fun n : ℕ => (1 : ℝ) / (n : ℝ))
Helper for Proposition 4.5.2: the target conjunction yields a contradiction with Real.not_summable_one_div_natCast.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_implies_false_via_harmonic :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) → False
Helper for Proposition 4.5.2: contradiction with harmonic divergence yields direct negation of the target conjunction.
open scoped Topology
lemma helperForProposition_4_5_2_not_target_conjunction_via_harmonic_divergence :
¬ (Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2)
Helper for Proposition 4.5.2: under Lean's default unconditional Summable semantics on ℕ, the target conjunction is equivalent to False.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_iff_false :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) ↔ False
Helper for Proposition 4.5.2: under Lean's default unconditional Summable semantics on ℕ, both asserted claims in the target conjunction are blocked.
open scoped Topology
lemma helperForProposition_4_5_2_unconditional_obstructions :
(¬ Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ))) ∧
¬ ((∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2)
Helper for Proposition 4.5.2: the alternating harmonic partial sums converge conditionally (as a limit of finite sums), even though unconditional Summable on ℕ fails for this series in Lean.
open scoped Topology
lemma helperForProposition_4_5_2_partial_sums_tendsto_some_limit :
∃ l : ℝ, Filter.Tendsto
(fun n : ℕ => ∑ i ∈ Finset.range n, (-1 : ℝ) ^ i / ((i + 1 : ℕ) : ℝ))
Filter.atTop (nhds l)
Helper for Proposition 4.5.2: if the target conjunction held, then the ordered partial sums would converge to Real.log 2.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_implies_partial_sums_tendsto_log_two :
(Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2) →
Filter.Tendsto
(fun n : ℕ => ∑ i ∈ Finset.range n, (-1 : ℝ) ^ i / ((i + 1 : ℕ) : ℝ))
Filter.atTop (nhds (Real.log 2))
Helper for Proposition 4.5.2: the limit of alternating-harmonic partial sums, when it exists, is unique.
open scoped Topology
lemma helperForProposition_4_5_2_partial_sums_limit_unique
{l₁ l₂ : ℝ}
(h₁ : Filter.Tendsto
(fun n : ℕ => ∑ i ∈ Finset.range n, (-1 : ℝ) ^ i / ((i + 1 : ℕ) : ℝ))
Filter.atTop (nhds l₁))
(h₂ : Filter.Tendsto
(fun n : ℕ => ∑ i ∈ Finset.range n, (-1 : ℝ) ^ i / ((i + 1 : ℕ) : ℝ))
Filter.atTop (nhds l₂)) :
l₁ = l₂
Helper for Proposition 4.5.2: under the target conjunction, any candidate limit of alternating-harmonic partial sums must equal Real.log 2.
open scoped Topology
lemma helperForProposition_4_5_2_target_conjunction_forces_unique_limit_log_two
(h : Summable (fun n : ℕ => (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) ∧
(∑' n : ℕ, (-1 : ℝ) ^ n / ((n + 1 : ℕ) : ℝ)) = Real.log 2)
{l : ℝ}
(hl : Filter.Tendsto
(fun n : ℕ => ∑ i ∈ Finset.range n, (-1 : ℝ) ^ i / ((i + 1 : ℕ) : ℝ))
Filter.atTop (nhds l)) :
l = Real.log 24.5 — The Exponential and Logarithm Functions — Part 2
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Proposition 4.5.3: for every integer n ≥ 3, the factorial tail sum satisfies 0 < ∑' k : ℕ, 1 / (n + (k + 1))! < 1 / n!.
open scoped Topology
theorem factorial_tail_sum_pos_and_lt_inv_factorial
(n : ℕ) (hn : 3 ≤ n) :
0 < ∑' k : ℕ, (1 : ℝ) / ((n + (k + 1)).factorial : ℝ) ∧
(∑' k : ℕ, (1 : ℝ) / ((n + (k + 1)).factorial : ℝ)) < (1 : ℝ) / (n.factorial : ℝ)
Proposition 4.5.4: for every integer n ≥ 3, the real number n! * e is not an integer.
open scoped Topology
theorem factorial_mul_e_not_integer
(n : ℕ) (hn : 3 ≤ n) :
¬ ∃ m : ℤ, (n.factorial : ℝ) * realEulerNumber = (m : ℝ)
Proposition 4.5.5: define f : ℝ → ℝ by f x = exp (-1 / x) for x > 0 and f x = 0 for x ≤ 0. Then f is smooth on ℝ, all derivatives vanish at 0 (equivalently, for every integer k ≥ 0, f^(k)(0) = 0), and f is not real analytic at 0.
open scoped Topology
theorem flat_exp_cutoff_smooth_all_derivs_zero_not_analytic_at_zero :
let f : ℝ → ℝ := fun x => if 0 < x then Real.exp (-1 / x) else 0
ContDiff ℝ (↑(⊤ : ℕ∞)) f ∧ (∀ k : ℕ, iteratedDeriv k f 0 = 0) ∧ ¬ AnalyticAt ℝ f 0
Proposition 4.5.6: if f : ℝ → (0, ∞) is real analytic and satisfies f'(x) = f(x) for all x ∈ ℝ, then there exists a constant C > 0 such that f(x) = C e^x for all x ∈ ℝ.
open scoped Topology
theorem analytic_deriv_eq_self_exists_pos_const_mul_exp
{f : ℝ → ℝ}
(hf_pos : ∀ x : ℝ, 0 < f x)
(hf_analytic : AnalyticOn ℝ f Set.univ)
(hf_deriv : ∀ x : ℝ, deriv f x = f x) :
∃ C : ℝ, 0 < C ∧ ∀ x : ℝ, f x = C * Real.exp x
Proposition 4.5.7: for every natural number m ≥ 1, lim_{x→+∞} (exp x / x^m) = +∞.
open scoped Topology
theorem exp_div_pow_tendsto_atTop
(m : ℕ) (hm : 1 ≤ m) :
Filter.Tendsto (fun x : ℝ => Real.exp x / x ^ m) Filter.atTop Filter.atTop
Proposition 4.5.8: if P is a real polynomial and c > 0, then there exists N : ℝ such that for all real x > N, one has exp (c x) > |P(x)|.
open scoped Topology
theorem exp_mul_gt_abs_polynomial_eventually
(P : Polynomial ℝ) {c : ℝ} (hc : 0 < c) :
∃ N : ℝ, ∀ x : ℝ, x > N → Real.exp (c * x) > |P.eval x|
Proposition 4.5.9: define f : (0,+∞) × ℝ → ℝ by f(x,y) = x^y, where x^y = exp (y ln x) for x > 0 and y ∈ ℝ; then f is continuous on (0,+∞) × ℝ.
open scoped Topology
theorem positive_rpow_map_continuous_on_Ioi_prod :
Continuous (fun p : Set.Ioi (0 : ℝ) × ℝ => ((p.1 : ℝ) ^ p.2))
Theorem 4.5.12: The number e is irrational.
open scoped Topology
theorem realEulerNumber_irrational :
Irrational (Real.exp 1)
Theorem 4.5.13: The natural logarithm function ln : (0,+∞) → ℝ is real analytic on (0,+∞).
open scoped Topology
theorem logarithm_real_analytic_on_Ioi :
AnalyticOn ℝ Real.log (Set.Ioi (0 : ℝ))
Helper for Proposition 4.5.3: factorial tail denominator dominates a geometric factor.
open scoped Topology
lemma helperForProposition_4_5_3_factorial_lower_bound_nat
(n k : ℕ) :
(n + 1).factorial * 2 ^ k ≤ (n + (k + 1)).factorial
Helper for Proposition 4.5.3: each factorial-tail term is bounded by a geometric term.
open scoped Topology
lemma helperForProposition_4_5_3_term_le_geometric
(n k : ℕ) :
(1 : ℝ) / ((n + (k + 1)).factorial : ℝ)
≤ ((1 : ℝ) / ((n + 1).factorial : ℝ)) * ((1 / 2 : ℝ) ^ k)
Helper for Proposition 4.5.3: the shifted factorial reciprocal tail is summable.
open scoped Topology
lemma helperForProposition_4_5_3_summable_tail
(n : ℕ) :
Summable (fun k : ℕ => (1 : ℝ) / ((n + (k + 1)).factorial : ℝ))
Helper for Proposition 4.5.3: the shifted factorial reciprocal tail is bounded by a geometric sum equal to 2 / (n+1)!.
open scoped Topology
lemma helperForProposition_4_5_3_tsum_le_two_div_factorial_succ
(n : ℕ) :
(∑' k : ℕ, (1 : ℝ) / ((n + (k + 1)).factorial : ℝ))
≤ (2 : ℝ) / ((n + 1).factorial : ℝ)
Helper for Proposition 4.5.3: the geometric upper bound is strictly below 1 / n! when n ≥ 3.
open scoped Topology
lemma helperForProposition_4_5_3_two_div_factorial_succ_lt_inv_factorial
(n : ℕ) (hn : 3 ≤ n) :
(2 : ℝ) / ((n + 1).factorial : ℝ) < (1 : ℝ) / (n.factorial : ℝ)
Helper for Proposition 4.5.4: split the reciprocal-factorial series at index n+1.
open scoped Topology
lemma helperForProposition_4_5_4_tsum_split_at_factorial_index
(n : ℕ) :
(∑' j : ℕ, (1 : ℝ) / (j.factorial : ℝ))
= (∑ j ∈ Finset.range (n + 1), (1 : ℝ) / (j.factorial : ℝ))
+ ∑' k : ℕ, (1 : ℝ) / ((n + (k + 1)).factorial : ℝ)
Helper for Proposition 4.5.4: after scaling the finite prefix by n!, one gets a natural-number cast.
open scoped Topology
lemma helperForProposition_4_5_4_factorialScaledFinitePart_isNatCast
(n : ℕ) :
∃ A : ℕ,
(n.factorial : ℝ) * (∑ i ∈ Finset.range (n + 1), (1 : ℝ) / (i.factorial : ℝ)) = (A : ℝ)
Helper for Proposition 4.5.4: scaling the factorial tail by n! yields a remainder in (0,1).
open scoped Topology
lemma helperForProposition_4_5_4_factorialScaledTail_between_zero_one
(n : ℕ) (hn : 3 ≤ n) :
0 < (n.factorial : ℝ) * (∑' k : ℕ, (1 : ℝ) / ((n + (k + 1)).factorial : ℝ)) ∧
(n.factorial : ℝ) * (∑' k : ℕ, (1 : ℝ) / ((n + (k + 1)).factorial : ℝ)) < 1
Helper for Proposition 4.5.4: no integer cast lies strictly between two consecutive integer casts.
open scoped Topology
lemma helperForProposition_4_5_4_noInt_between_consecutiveIntCasts
(z m : ℤ) :
¬ ((z : ℝ) < (m : ℝ) ∧ (m : ℝ) < ((z + 1 : ℤ) : ℝ))
Helper for Proposition 4.5.5: the cutoff definition coincides with expNegInvGlue.
open scoped Topology
lemma helperForProposition_4_5_5_cutoff_eq_expNegInvGlue :
(fun x : ℝ => if 0 < x then Real.exp (-1 / x) else 0) = expNegInvGlue
Helper for Proposition 4.5.5: ContDiff ℝ ⊤ implies analyticity at 0.
open scoped Topology
lemma helperForProposition_4_5_5_contDiffTop_implies_analyticAt_zero
{f : ℝ → ℝ} (hcont : ContDiff ℝ ⊤ f) :
AnalyticAt ℝ f 0
Helper for Proposition 4.5.5: as currently encoded, the target conjunction is inconsistent.
open scoped Topology
lemma helperForProposition_4_5_5_target_conjunction_inconsistent :
¬ (let f : ℝ → ℝ := fun x => if 0 < x then Real.exp (-1 / x) else 0
ContDiff ℝ ⊤ f ∧ (∀ k : ℕ, iteratedDeriv k f 0 = 0) ∧ ¬ AnalyticAt ℝ f 0)
Helper for Proposition 4.5.5: all iterated derivatives of the cutoff vanish on the negative half-line.
open scoped Topology
lemma helperForProposition_4_5_5_iteratedDeriv_zero_on_Iio :
∀ k : ℕ,
Set.EqOn
(iteratedDeriv k (fun x : ℝ => if 0 < x then Real.exp (-1 / x) else 0))
(fun _ => (0 : ℝ))
(Set.Iio (0 : ℝ))
Helper for Proposition 4.5.5: smoothness forces every iterated derivative of the cutoff to vanish at 0.
open scoped Topology
lemma helperForProposition_4_5_5_iteratedDeriv_zero_at_zero
(hcont : ContDiff ℝ (↑(⊤ : ℕ∞))
(fun x : ℝ => if 0 < x then Real.exp (-1 / x) else 0)) :
∀ k : ℕ,
iteratedDeriv k (fun x : ℝ => if 0 < x then Real.exp (-1 / x) else 0) 0 = 0
Helper for Proposition 4.5.5: the cutoff is not analytic at 0.
open scoped Topology
lemma helperForProposition_4_5_5_not_analytic_at_zero :
¬ AnalyticAt ℝ (fun x : ℝ => if 0 < x then Real.exp (-1 / x) else 0) 0
Helper for Proposition 4.5.6: analyticity on all of ℝ implies global differentiability.
open scoped Topology
lemma helperForProposition_4_5_6_differentiable_of_analyticOn_univ
{f : ℝ → ℝ}
(hf_analytic : AnalyticOn ℝ f Set.univ) :
Differentiable ℝ f
Helper for Proposition 4.5.6: the derivative of log ∘ f is constantly 1.
open scoped Topology
lemma helperForProposition_4_5_6_deriv_log_eq_one
{f : ℝ → ℝ}
(hf_pos : ∀ x : ℝ, 0 < f x)
(hf_diff : Differentiable ℝ f)
(hf_deriv : ∀ x : ℝ, deriv f x = f x) :
∀ x : ℝ, deriv (fun y : ℝ => Real.log (f y)) x = 1
Helper for Proposition 4.5.6: x ↦ log (f x) - x is constant on ℝ.
open scoped Topology
lemma helperForProposition_4_5_6_log_sub_id_constant
{f : ℝ → ℝ}
(hf_pos : ∀ x : ℝ, 0 < f x)
(hf_diff : Differentiable ℝ f)
(hf_deriv : ∀ x : ℝ, deriv f x = f x) :
∀ x : ℝ, Real.log (f x) - x = Real.log (f 0)
Helper for Proposition 4.5.6: the solution has the form f x = f 0 * exp x.
open scoped Topology
lemma helperForProposition_4_5_6_exp_form
{f : ℝ → ℝ}
(hf_pos : ∀ x : ℝ, 0 < f x)
(hlog_const : ∀ x : ℝ, Real.log (f x) - x = Real.log (f 0)) :
∀ x : ℝ, f x = f 0 * Real.exp x
Helper for Proposition 4.5.8: rewriting the composed polynomial at a scaled input.
open scoped Topology
lemma helperForProposition_4_5_8_scaledEval_rewrite
(P : Polynomial ℝ) {c : ℝ} (hc : 0 < c) (x : ℝ) :
let Q : Polynomial ℝ := P.comp (Polynomial.C (1 / c) * Polynomial.X)
Q.eval (c * x) = P.eval x
Helper for Proposition 4.5.8: the ratio P(x) / exp(c x) tends to 0 at +∞.
open scoped Topology
lemma helperForProposition_4_5_8_tendsto_div_exp_mul_atTop
(P : Polynomial ℝ) {c : ℝ} (hc : 0 < c) :
Filter.Tendsto (fun x : ℝ => P.eval x / Real.exp (c * x)) Filter.atTop (nhds 0)
Helper for Proposition 4.5.8: eventually, |P(x)| is strictly below exp(c x).
open scoped Topology
lemma helperForProposition_4_5_8_eventually_abs_lt_exp_mul
(P : Polynomial ℝ) {c : ℝ} (hc : 0 < c) :
∀ᶠ x : ℝ in Filter.atTop, |P.eval x| < Real.exp (c * x)
Helper for Proposition 4.5.8: convert an eventual atTop property into a threshold statement.
open scoped Topology
lemma helperForProposition_4_5_8_eventually_atTop_to_threshold
{R : ℝ → Prop} (hR : ∀ᶠ x : ℝ in Filter.atTop, R x) :
∃ N : ℝ, ∀ x : ℝ, x > N → R x
Helper for Proposition 4.5.9: continuity of the positive-base coordinate map.
open scoped Topology
lemma helperForProposition_4_5_9_continuous_baseCoordinate :
Continuous (fun p : Set.Ioi (0 : ℝ) × ℝ => ((p.1 : Set.Ioi (0 : ℝ)) : ℝ))
Helper for Proposition 4.5.9: the base in (0,+∞) is always nonzero.
open scoped Topology
lemma helperForProposition_4_5_9_rpow_sideCondition :
∀ p : Set.Ioi (0 : ℝ) × ℝ, ((p.1 : ℝ) ≠ 0 ∨ 0 < p.2)
Helper for Theorem 4.5.12: the denominator of a rational divides (max 3 q.den)!.
open scoped Topology
lemma helperForTheorem_4_5_12_factorialDenominator_dvd_factorialMax
(q : ℚ) :
q.den ∣ Nat.factorial (max 3 q.den)
Helper for Theorem 4.5.12: multiplying a rational by a suitable factorial gives an integer cast in ℝ.
open scoped Topology
lemma helperForTheorem_4_5_12_factorialScaledRat_isIntCast
(q : ℚ) :
∃ m : ℤ, (((max 3 q.den).factorial : ℝ) * (q : ℝ)) = (m : ℝ)
Helper for Theorem 4.5.12: exp 1 is not equal to any rational cast.
open scoped Topology
lemma helperForTheorem_4_5_12_exp_one_ne_ratCast
(q : ℚ) :
Real.exp 1 ≠ (q : ℝ)4.6 — A Digression on Complex Numbers
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Definition 4.6.1 (Complex numbers): the set of complex numbers is {a + b i | a, b ∈ ℝ}, where i is a symbol satisfying i^2 = -1. In Lean, this is identified with the existing type ℂ.
abbrev complexNumbers : Type := ℂ
Every complex number can be written in the form a + b i with a b : ℝ.
theorem complexNumbers_exists_real_imag (z : ℂ) :
∃ a b : ℝ, z = (a : ℂ) + (b : ℂ) * Complex.I
Lemma 4.6.2: every z : ℂ has a unique decomposition z = a + b i with a, b : ℝ; moreover i^2 = -1 and -z = (-1)z for all z : ℂ.
theorem complex_unique_decomposition_i_sq_and_neg_mul :
(∀ z : ℂ, ∃! p : ℝ × ℝ, z = (p.1 : ℂ) + (p.2 : ℂ) * Complex.I) ∧
Complex.I ^ 2 = (-1 : ℂ) ∧
(∀ z : ℂ, -z = ((-1 : ℂ) * z))
Definition 4.6.2 (Complex numbers): (i) a complex number is an ordered pair (a, b) with a, b ∈ ℝ; (ii) equality is componentwise, i.e. (a, b) = (c, d) iff a = c and b = d; (iii) the set of all such complex numbers is ℝ × ℝ.
abbrev complexNumbersAsPairs : Type := ℝ × ℝ
Componentwise addition on ordered pairs of real numbers.
abbrev complexPairAdd (z w : complexNumbersAsPairs) : complexNumbersAsPairs :=
(z.1 + w.1, z.2 + w.2)
Componentwise negation on ordered pairs of real numbers.
abbrev complexPairNeg (z : complexNumbersAsPairs) : complexNumbersAsPairs :=
(-z.1, -z.2)
The zero ordered pair (0, 0) in ℝ × ℝ.
abbrev complexPairZero : complexNumbersAsPairs := (0, 0)
Definition 4.6.3 (Complex addition, negation, and zero): let ℂ be the set of ordered pairs of real numbers. For z = (a, b) and w = (c, d), define z + w := (a + c, b + d), -z := (-a, -b), and 0_C := (0, 0).
def complexPairAddNegZeroData :
(complexNumbersAsPairs → complexNumbersAsPairs → complexNumbersAsPairs) ×
(complexNumbersAsPairs → complexNumbersAsPairs) × complexNumbersAsPairs :=
(complexPairAdd, complexPairNeg, complexPairZero)
Lemma 4.6.1 (Addition on ℂ satisfies the abelian group axioms): for all z₁ z₂ z₃ : ℂ, (i) z₁ + z₂ = z₂ + z₁, (ii) (z₁ + z₂) + z₃ = z₁ + (z₂ + z₃), (iii) z₁ + 0 = 0 + z₁ = z₁, and (iv) z₁ + (-z₁) = (-z₁) + z₁ = 0. In particular, (ℂ, +) is an abelian group with identity element 0.
theorem complex_addition_abelian_group_axioms :
(∀ z₁ z₂ : ℂ, z₁ + z₂ = z₂ + z₁) ∧
(∀ z₁ z₂ z₃ : ℂ, (z₁ + z₂) + z₃ = z₁ + (z₂ + z₃)) ∧
(∀ z₁ : ℂ, z₁ + 0 = z₁ ∧ 0 + z₁ = z₁) ∧
(∀ z₁ : ℂ, z₁ + (-z₁) = 0 ∧ (-z₁) + z₁ = 0)
Definition 4.6.4 (Complex multiplication): for complex numbers viewed as ordered pairs z = (a, b) and w = (c, d), their product is zw = (ac - bd, ad + bc), and the complex multiplicative identity is 1_C = (1, 0).
def complexPairMulOneData :
(complexNumbersAsPairs → complexNumbersAsPairs → complexNumbersAsPairs) ×
complexNumbersAsPairs :=
(fun z w => (z.1 * w.1 - z.2 * w.2, z.1 * w.2 + z.2 * w.1), (1, 0))
Definition 4.6.5 (Embedding of real numbers into complex numbers): define ι : ℝ → ℂ by identifying x : ℝ with (x, 0) in ℂ.
def realToComplexEmbedding : ℝ → ℂ :=
fun x => (x : ℂ)
The embedding of real numbers into complex numbers preserves equality, addition, negation, and multiplication.
theorem realToComplexEmbedding_respects_operations :
(∀ x y : ℝ, x = y ↔ realToComplexEmbedding x = realToComplexEmbedding y) ∧
(∀ x₁ x₂ : ℝ,
realToComplexEmbedding (x₁ + x₂) =
realToComplexEmbedding x₁ + realToComplexEmbedding x₂) ∧
(∀ x : ℝ, realToComplexEmbedding (-x) = -realToComplexEmbedding x) ∧
(∀ x₁ x₂ : ℝ,
realToComplexEmbedding (x₁ * x₂) =
realToComplexEmbedding x₁ * realToComplexEmbedding x₂)
Definition 4.6.6 (Imaginary unit): the imaginary unit is the complex number i ∈ ℂ defined by i := (0, 1).
def imaginaryUnit : ℂ :=
⟨0, 1⟩
A complex number is real when its imaginary part vanishes.
def IsRealComplex (z : ℂ) : Prop :=
z.im = 0
A complex number is purely imaginary when its real part vanishes.
def IsPurelyImaginaryComplex (z : ℂ) : Prop :=
z.re = 0
Definition 4.6.7 (Real and imaginary parts; complex conjugate): for z : ℂ, z = Re(z) + i Im(z), a number is real iff Im(z) = 0, a number is purely imaginary iff Re(z) = 0, and the complex conjugate is conj z = Re(z) - i Im(z).
theorem complex_real_imaginary_parts_and_conjugate (z : ℂ) :
z = (z.re : ℂ) + Complex.I * (z.im : ℂ) ∧
(IsRealComplex z ↔ z.im = 0) ∧
(IsPurelyImaginaryComplex z ↔ z.re = 0) ∧
star z = (z.re : ℂ) - Complex.I * (z.im : ℂ)
Proposition 4.6.2: for every complex number z ∈ ℂ, Re(z) = (z + \overline{z}) / 2 and Im(z) = (z - \overline{z}) / (2i).
theorem complex_re_and_im_eq_conjugate_formulas (z : ℂ) :
((z.re : ℂ) = (z + star z) / (2 : ℂ)) ∧
((z.im : ℂ) = (z - star z) / ((2 : ℂ) * Complex.I))
Lemma 4.6.3 (Complex conjugation is an involution): for z w : ℂ, complex conjugation satisfies ` \overline{z + w} = \bar z + \bar w, \overline{-z} = -\bar z, \overline{z w} = \bar z \, \bar w, \overline{\bar z} = z, ` and moreover \bar z = \bar w iff z = w, while \bar z = z iff z is real.
theorem complex_conjugation_involution (z w : ℂ) :
star (z + w) = star z + star w ∧
star (-z) = -star z ∧
star (z * w) = star z * star w ∧
star (star z) = z ∧
(star z = star w ↔ z = w) ∧
(star z = z ↔ IsRealComplex z)
Definition 4.6.8 (Complex absolute value): for z = a + bi with a b : ℝ, the complex absolute value is the nonnegative real number |z| = sqrt (a^2 + b^2) = (a^2 + b^2)^(1/2). In Lean this is represented by the complex norm ‖z‖.
noncomputable abbrev complexAbsoluteValue (z : ℂ) : ℝ :=
‖z‖
Lemma 4.6.4: for every z ∈ ℂ, the complex absolute value |z| is a non-negative real number.
theorem complexAbsoluteValue_nonnegative (z : ℂ) :
0 ≤ complexAbsoluteValue z
Lemma 4.6.5: for every z ∈ ℂ, we have |z| = 0 if and only if z = 0.
theorem complexAbsoluteValue_eq_zero_iff (z : ℂ) :
complexAbsoluteValue z = 0 ↔ z = 0
Lemma 4.6.6: for every z ∈ ℂ, we have z * \overline{z} = |z|^2. In particular, |z| = sqrt (z * \overline{z}), interpreted in Lean as |z| = sqrt ((z * \overline{z}).re).
theorem complex_mul_conj_eq_abs_sq_and_abs_eq_sqrt (z : ℂ) :
z * star z = ((complexAbsoluteValue z) ^ 2 : ℂ) ∧
complexAbsoluteValue z = Real.sqrt ((z * star z).re)
Lemma 4.6.7: for all z, w ∈ ℂ, the complex absolute value is multiplicative: |zw| = |z| |w|.
lemma complexAbsoluteValue_mul (z w : ℂ) :
complexAbsoluteValue (z * w) = complexAbsoluteValue z * complexAbsoluteValue w
Lemma 4.6.8: for every z ∈ ℂ, the complex absolute value is invariant under complex conjugation: |\overline{z}| = |z|.
lemma complexAbsoluteValue_conj (z : ℂ) :
complexAbsoluteValue (star z) = complexAbsoluteValue z
Proposition 4.6.3: if z, w ∈ ℂ and w ≠ 0, then |z / w| = |z| / |w|.
theorem complexAbsoluteValue_div_of_ne_zero (z w : ℂ) :
w ≠ 0 → complexAbsoluteValue (z / w) = complexAbsoluteValue z / complexAbsoluteValue w
Definition 4.6.9 (Complex reciprocal and quotient): for z : ℂ, the reciprocal is undefined at 0 and is |z|^{-2} \overline{z} otherwise; for z w : ℂ, the quotient z / w is defined when w ≠ 0 by z * w^{-1}.
noncomputable def complexReciprocalQuotientPartialData :
(ℂ → Option ℂ) × (ℂ → ℂ → Option ℂ) :=
let reciprocal : ℂ → Option ℂ :=
fun z =>
if z = 0 then
none
else
some (((((complexAbsoluteValue z) ^ (2 : ℕ))⁻¹ : ℝ) : ℂ) * star z)
(reciprocal, fun z w => (reciprocal w).map (fun wInv => z * wInv))
Partial complex reciprocal: undefined at 0, and otherwise given by |z|^{-2} \overline{z}.
noncomputable def complexReciprocalPartial (z : ℂ) : Option ℂ :=
complexReciprocalQuotientPartialData.1 z
Partial complex quotient: for z w : ℂ, this is defined by mapping w to its partial reciprocal and multiplying by z.
noncomputable def complexQuotientPartial (z w : ℂ) : Option ℂ :=
complexReciprocalQuotientPartialData.2 z w
Definition 4.6.10 (Complex exponential): for each z ∈ ℂ, the complex exponential is defined by the power series exp(z) = ∑' n : ℕ, z^n / n!.
noncomputable def complexExponential (z : ℂ) : ℂ :=
∑' n : ℕ, z ^ n / (Nat.factorial n : ℂ)
Theorem 4.6.1: for all z, w ∈ ℂ, one has exp(z + w) = exp(z) exp(w).
theorem complex_exp_add (z w : ℂ) :
Complex.exp (z + w) = Complex.exp z * Complex.exp w
Proposition 4.6.1: for every z ∈ ℂ, the complex exponential commutes with complex conjugation, i.e. exp(conj z) = conj (exp z).
theorem complex_exp_conj (z : ℂ) :
Complex.exp (star z) = star (Complex.exp z)
Lemma 4.6.9: for every complex number z, its real part is bounded by its complex absolute value: -|z| ≤ Re(z) ≤ |z|.
theorem realPart_le_complexAbsoluteValue (z : ℂ) :
-complexAbsoluteValue z ≤ z.re ∧ z.re ≤ complexAbsoluteValue z
Lemma 4.6.10: for every z ∈ ℂ, the imaginary part is bounded by the complex absolute value: -|z| ≤ Im(z) ≤ |z|.
lemma imagPart_le_complexAbsoluteValue (z : ℂ) :
-complexAbsoluteValue z ≤ z.im ∧ z.im ≤ complexAbsoluteValue z
Lemma 4.6.11: for every z ∈ ℂ, the complex absolute value is bounded by the sum of the absolute values of its real and imaginary parts: |z| ≤ |Re(z)| + |Im(z)|.
lemma complexAbsoluteValue_le_abs_re_add_abs_im (z : ℂ) :
complexAbsoluteValue z ≤ |z.re| + |z.im|
Lemma 4.6.12: for all z, w ∈ ℂ, the complex absolute value satisfies the triangle inequality |z + w| ≤ |z| + |w|.
lemma complexAbsoluteValue_add_le (z w : ℂ) :
complexAbsoluteValue (z + w) ≤ complexAbsoluteValue z + complexAbsoluteValue w
Proposition 4.6.4: for nonzero complex numbers z and w, |z + w| = |z| + |w| if and only if there exists a real number c > 0 such that z = c w.
theorem complexAbsoluteValue_add_eq_add_iff_exists_pos_real_mul (z w : ℂ)
(hz : z ≠ 0) (hw : w ≠ 0) :
complexAbsoluteValue (z + w) = complexAbsoluteValue z + complexAbsoluteValue w ↔
∃ c : ℝ, 0 < c ∧ z = (c : ℂ) * w
Proposition 4.6.5: there does not exist a partition of ℂ into positive, negative, and zero classes such that for every z, w : ℂ: (i) exactly one of Pos z, Neg z, or z = 0 holds, (ii) positivity/negativity are exchanged by negation, (iii) positive numbers are closed under addition, and (iv) positive numbers are closed under multiplication.
theorem no_complex_positive_negative_zero_partition :
¬ ∃ Pos Neg : ℂ → Prop,
(∀ z : ℂ,
(Pos z ∨ Neg z ∨ z = 0) ∧
¬ (Pos z ∧ Neg z) ∧ ¬ (Pos z ∧ z = 0) ∧ ¬ (Neg z ∧ z = 0)) ∧
(∀ z : ℂ, Pos z → Neg (-z)) ∧
(∀ z : ℂ, Neg z → Pos (-z)) ∧
(∀ z w : ℂ, Pos z → Pos w → Pos (z + w)) ∧
(∀ z w : ℂ, Pos z → Pos w → Pos (z * w))
Lemma 4.6.13: with d(z,w) = |z - w| on ℂ, this gives the usual metric on ℂ; moreover, a sequence z_n converges to z in this metric iff both real and imaginary parts converge, i.e. Re(z_n) → Re(z) and Im(z_n) → Im(z).
lemma complex_tendsto_iff_re_im_tendsto (zSeq : ℕ → ℂ) (z : ℂ) :
let d : ℂ → ℂ → ℝ := fun z₁ z₂ => ‖z₁ - z₂‖;
((∀ z₁ z₂ : ℂ, 0 ≤ d z₁ z₂) ∧
(∀ z₁ z₂ : ℂ, d z₁ z₂ = 0 ↔ z₁ = z₂) ∧
(∀ z₁ z₂ : ℂ, d z₁ z₂ = d z₂ z₁) ∧
(∀ z₁ z₂ z₃ : ℂ, d z₁ z₃ ≤ d z₁ z₂ + d z₂ z₃)) ∧
((∀ ε > 0, ∃ N : ℕ, ∀ n ≥ N, d (zSeq n) z < ε) ↔
Filter.Tendsto (fun n => (zSeq n).re) Filter.atTop (nhds z.re) ∧
Filter.Tendsto (fun n => (zSeq n).im) Filter.atTop (nhds z.im))
Lemma 4.6.14 (Complex limit laws): if zₙ → z and wₙ → w in ℂ, then zₙ + wₙ → z + w, zₙ - wₙ → z - w, c zₙ → c z, zₙ wₙ → z w, and conj zₙ → conj z; moreover, if wₙ ≠ 0 for all n and w ≠ 0, then zₙ / wₙ → z / w.
lemma complex_limit_laws (zSeq wSeq : ℕ → ℂ) (z w c : ℂ)
(hz : Filter.Tendsto zSeq Filter.atTop (nhds z))
(hw : Filter.Tendsto wSeq Filter.atTop (nhds w)) :
Filter.Tendsto (fun n => zSeq n + wSeq n) Filter.atTop (nhds (z + w)) ∧
Filter.Tendsto (fun n => zSeq n - wSeq n) Filter.atTop (nhds (z - w)) ∧
Filter.Tendsto (fun n => c * zSeq n) Filter.atTop (nhds (c * z)) ∧
Filter.Tendsto (fun n => zSeq n * wSeq n) Filter.atTop (nhds (z * w)) ∧
Filter.Tendsto (fun n => star (zSeq n)) Filter.atTop (nhds (star z)) ∧
((∀ n : ℕ, wSeq n ≠ 0) → w ≠ 0 →
Filter.Tendsto (fun n => zSeq n / wSeq n) Filter.atTop (nhds (z / w)))
Helper for Lemma 4.6.2: every complex number admits a real-imaginary pair representation z = a + bi.
lemma helperForLemma_4_6_2_canonicalDecompositionExists (z : ℂ) :
∃ p : ℝ × ℝ, z = (p.1 : ℂ) + (p.2 : ℂ) * Complex.I
Helper for Lemma 4.6.2: any representation z = a + bi must use a = Re(z) and b = Im(z).
lemma helperForLemma_4_6_2_coordinatesFromRepresentation {z : ℂ} {p : ℝ × ℝ}
(h : z = (p.1 : ℂ) + (p.2 : ℂ) * Complex.I) :
p.1 = z.re ∧ p.2 = z.im
Helper for Lemma 4.6.2: the pair (Re z, Im z) is the unique coefficient pair in a decomposition z = a + bi.
lemma helperForLemma_4_6_2_uniqueDecompositionPair (z : ℂ) (p : ℝ × ℝ)
(h : z = (p.1 : ℂ) + (p.2 : ℂ) * Complex.I) :
p = (z.re, z.im)
Helper for Lemma 4.6.2: the imaginary unit squares to -1.
lemma helperForLemma_4_6_2_iSquared :
Complex.I ^ 2 = (-1 : ℂ)
Helper for Lemma 4.6.2: negation agrees with multiplication by -1.
lemma helperForLemma_4_6_2_negEqualsNegOneMul (z : ℂ) :
-z = ((-1 : ℂ) * z)
Helper for Proposition 4.6.2: Re(z) equals (z + \bar z)/2 written with star.
lemma helperForProposition_4_6_2_re_eq_add_star (z : ℂ) :
(z.re : ℂ) = (z + star z) / (2 : ℂ)
Helper for Proposition 4.6.2: Im(z) equals (z - \bar z)/(2i) written with star.
lemma helperForProposition_4_6_2_im_eq_sub_star (z : ℂ) :
(z.im : ℂ) = (z - star z) / ((2 : ℂ) * Complex.I)
Helper for Lemma 4.6.3: complex conjugation preserves addition, negation, and multiplication.
lemma helperForLemma_4_6_3_conj_add_neg_mul (z w : ℂ) :
star (z + w) = star z + star w ∧
star (-z) = -star z ∧
star (z * w) = star z * star w
Helper for Lemma 4.6.3: complex conjugation is involutive.
lemma helperForLemma_4_6_3_conj_involutive (z : ℂ) :
star (star z) = z
Helper for Lemma 4.6.3: complex conjugation is injective.
lemma helperForLemma_4_6_3_conj_eq_iff_eq (z w : ℂ) :
star z = star w ↔ z = w
Helper for Lemma 4.6.3: fixed points of conjugation are exactly the real complex numbers.
lemma helperForLemma_4_6_3_conj_fixed_iff_isReal (z : ℂ) :
star z = z ↔ IsRealComplex z
Helper for Lemma 4.6.6: z * conjugate(z) is the real scalar Complex.normSq z, viewed in ℂ.
lemma helperForLemma_4_6_6_mulConj_eq_normSqCast (z : ℂ) :
z * star z = (Complex.normSq z : ℂ)
Helper for Lemma 4.6.6: the real part of z * conjugate(z) is Complex.normSq z.
lemma helperForLemma_4_6_6_mulConj_re_eq_normSq (z : ℂ) :
(z * star z).re = Complex.normSq z
Helper for Lemma 4.6.6: (z * conjugate(z)).re equals |z|^2.
lemma helperForLemma_4_6_6_mulConj_re_eq_absSq (z : ℂ) :
(z * star z).re = (complexAbsoluteValue z) ^ 2
Helper for Lemma 4.6.6: √(|z|^2) = |z|.
lemma helperForLemma_4_6_6_sqrt_absSq_eq_absValue (z : ℂ) :
Real.sqrt ((complexAbsoluteValue z) ^ 2) = complexAbsoluteValue z
Helper for Lemma 4.6.9: the absolute value of the real part is bounded by the complex absolute value.
lemma helperForLemma_4_6_9_absRe_le_complexAbs (z : ℂ) :
Helper for Lemma 4.6.9: the real part is bounded below by the negated complex absolute value.
lemma helperForLemma_4_6_9_negComplexAbs_le_re (z : ℂ) :
-complexAbsoluteValue z ≤ z.re
Helper for Lemma 4.6.9: the real part is bounded above by the complex absolute value.
lemma helperForLemma_4_6_9_re_le_complexAbs (z : ℂ) :
z.re ≤ complexAbsoluteValue z
Helper for Lemma 4.6.10: the absolute value of the imaginary part is bounded by the complex absolute value.
lemma helperForLemma_4_6_10_absIm_le_complexAbs (z : ℂ) :
Helper for Lemma 4.6.10: the imaginary part is bounded below by the negated complex absolute value.
lemma helperForLemma_4_6_10_negComplexAbs_le_im (z : ℂ) :
-complexAbsoluteValue z ≤ z.im
Helper for Lemma 4.6.10: the imaginary part is bounded above by the complex absolute value.
lemma helperForLemma_4_6_10_im_le_complexAbs (z : ℂ) :
z.im ≤ complexAbsoluteValue z
Helper for Proposition 4.6.4: equality in the triangle inequality for complex norm is equivalent to the two numbers lying on the same real ray.
lemma helperForProposition_4_6_4_normEq_iff_sameRay (z w : ℂ) :
complexAbsoluteValue (z + w) = complexAbsoluteValue z + complexAbsoluteValue w ↔
SameRay ℝ z w
Helper for Proposition 4.6.4: two nonzero complex numbers on the same real ray differ by multiplication by a positive real scalar.
lemma helperForProposition_4_6_4_sameRay_to_existsPosMul (z w : ℂ)
(hz : z ≠ 0) (hw : w ≠ 0) (hzw : SameRay ℝ z w) :
∃ c : ℝ, 0 < c ∧ z = (c : ℂ) * w
Helper for Proposition 4.6.4: if one complex number is a positive real multiple of another, then equality holds in the triangle inequality.
lemma helperForProposition_4_6_4_existsPosMul_to_normEq (z w : ℂ) :
(∃ c : ℝ, 0 < c ∧ z = (c : ℂ) * w) →
complexAbsoluteValue (z + w) = complexAbsoluteValue z + complexAbsoluteValue w
Helper for Proposition 4.6.5: if i is positive and positivity is closed under multiplication, then -1 is positive.
lemma helperForProposition_4_6_5_posNegOne_from_posI
(Pos : ℂ → Prop)
(hMulPos : ∀ z w : ℂ, Pos z → Pos w → Pos (z * w))
(hPosI : Pos Complex.I) :
Pos (-(1 : ℂ))
Helper for Proposition 4.6.5: if i is negative, negation sends negatives to positives, and positivity is multiplicatively closed, then -1 is positive.
lemma helperForProposition_4_6_5_posNegOne_from_negI
(Pos Neg : ℂ → Prop)
(hNegToPos : ∀ z : ℂ, Neg z → Pos (-z))
(hMulPos : ∀ z w : ℂ, Pos z → Pos w → Pos (z * w))
(hNegI : Neg Complex.I) :
Pos (-(1 : ℂ))
Helper for Proposition 4.6.5: trichotomy at i, together with the negation rule Neg z → Pos (-z) and multiplicative closure of positivity, forces -1 to be positive.
lemma helperForProposition_4_6_5_posNegOne
(Pos Neg : ℂ → Prop)
(hTrichotomy : ∀ z : ℂ, Pos z ∨ Neg z ∨ z = 0)
(hNegToPos : ∀ z : ℂ, Neg z → Pos (-z))
(hMulPos : ∀ z w : ℂ, Pos z → Pos w → Pos (z * w)) :
Pos (-(1 : ℂ))
Helper for Proposition 4.6.5: if -1 is positive, positivity is closed under multiplication, and positives negate to negatives, then exclusivity of positive/negative classes yields a contradiction.
lemma helperForProposition_4_6_5_contradiction_from_posNegOne
(Pos Neg : ℂ → Prop)
(hPosNegExclusive : ∀ z : ℂ, ¬ (Pos z ∧ Neg z))
(hPosToNeg : ∀ z : ℂ, Pos z → Neg (-z))
(hMulPos : ∀ z w : ℂ, Pos z → Pos w → Pos (z * w))
(hPosNegOne : Pos (-(1 : ℂ))) :
False
Helper for Lemma 4.6.14: bundled arithmetic limit laws in ℂ for sum, difference, scalar multiplication, and product.
lemma helperForLemma_4_6_14_arith_tendsto_bundle (zSeq wSeq : ℕ → ℂ) (z w c : ℂ)
(hz : Filter.Tendsto zSeq Filter.atTop (nhds z))
(hw : Filter.Tendsto wSeq Filter.atTop (nhds w)) :
Filter.Tendsto (fun n => zSeq n + wSeq n) Filter.atTop (nhds (z + w)) ∧
Filter.Tendsto (fun n => zSeq n - wSeq n) Filter.atTop (nhds (z - w)) ∧
Filter.Tendsto (fun n => c * zSeq n) Filter.atTop (nhds (c * z)) ∧
Filter.Tendsto (fun n => zSeq n * wSeq n) Filter.atTop (nhds (z * w))
Helper for Lemma 4.6.14: conjugation preserves sequence limits in ℂ.
lemma helperForLemma_4_6_14_star_tendsto (zSeq : ℕ → ℂ) (z : ℂ)
(hz : Filter.Tendsto zSeq Filter.atTop (nhds z)) :
Filter.Tendsto (fun n => star (zSeq n)) Filter.atTop (nhds (star z))
Helper for Lemma 4.6.14: quotient limit law in ℂ under nonzero limit.
lemma helperForLemma_4_6_14_div_tendsto (zSeq wSeq : ℕ → ℂ) (z w : ℂ)
(hz : Filter.Tendsto zSeq Filter.atTop (nhds z))
(hw : Filter.Tendsto wSeq Filter.atTop (nhds w)) :
(∀ n : ℕ, wSeq n ≠ 0) → w ≠ 0 →
Filter.Tendsto (fun n => zSeq n / wSeq n) Filter.atTop (nhds (z / w))4.7 — Trigonometric Functions — Part 1
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Definition 4.7.1 (Trigonometric functions): for each z : ℂ, cos z = (exp (i z) + exp (-i z)) / 2 and sin z = (exp (i z) - exp (-i z)) / (2i).
theorem complex_trigonometric_function_definitions (z : ℂ) :
Complex.cos z = (Complex.exp (Complex.I * z) + Complex.exp (-(Complex.I * z))) / (2 : ℂ) ∧
Complex.sin z = (Complex.exp (Complex.I * z) - Complex.exp (-(Complex.I * z))) /
((2 : ℂ) * Complex.I)
Proposition 4.7.1 (Reality and absolute convergence on ℝ): for every x : ℝ, the series ∑' n, (-1)^n * x^(2n) / (2n)! and ∑' n, (-1)^n * x^(2n+1) / (2n+1)! converge absolutely. Consequently, defining cos(x) and sin(x) by these series gives real-valued quantities for each real input.
theorem real_trigonometric_series_absolutely_convergent (x : ℝ) :
let a : ℕ → ℝ :=
fun n => (((-1 : ℝ) ^ n) * x ^ (2 * n)) / (Nat.factorial (2 * n) : ℝ)
let b : ℕ → ℝ :=
fun n => (((-1 : ℝ) ^ n) * x ^ (2 * n + 1)) / (Nat.factorial (2 * n + 1) : ℝ)
Summable (fun n : ℕ => |a n|) ∧
Summable (fun n : ℕ => |b n|) ∧
Summable a ∧ Summable b
Proposition 4.7.2: If f : ℝ → ℝ is differentiable at x₀, with f x₀ = 0 and f' x₀ ≠ 0, then there exists c > 0 such that for every y : ℝ, 0 < |y - x₀| < c implies f y ≠ 0.
theorem isolated_zero_of_differentiableAt_deriv_ne_zero
(f : ℝ → ℝ) (x₀ : ℝ)
(hf_diff : DifferentiableAt ℝ f x₀)
(hf_zero : f x₀ = 0)
(hf_deriv_ne_zero : deriv f x₀ ≠ 0) :
∃ c > 0, ∀ y : ℝ, 0 < |y - x₀| ∧ |y - x₀| < c → f y ≠ 0
Proposition 4.7.3: There exists a real number c > 0 such that for every real x with 0 < x < c, one has sin x ≠ 0.
theorem exists_pos_radius_sine_ne_zero_right_of_zero :
∃ c > 0, ∀ x : ℝ, 0 < x ∧ x < c → Real.sin x ≠ 0
Theorem 4.7.1: For every real number x, one has sin(x)^2 + cos(x)^2 = 1; in particular sin(x), cos(x) ∈ [-1, 1].
theorem real_sine_cosine_pythagorean_identity_and_bounds (x : ℝ) :
Real.sin x ^ 2 + Real.cos x ^ 2 = 1 ∧
Real.sin x ∈ Set.Icc (-1) 1 ∧
Real.cos x ∈ Set.Icc (-1) 1
Theorem 4.7.2: For every real number x, (Real.sin)'(x) = Real.cos x and (Real.cos)'(x) = -Real.sin x.
theorem real_sine_cosine_derivative_formulas (x : ℝ) :
deriv Real.sin x = Real.cos x ∧
deriv Real.cos x = -Real.sin x
Theorem 4.7.3: For every real number x, sin(-x) = -sin(x) and cos(-x) = cos(x).
theorem real_sine_odd_and_cosine_even (x : ℝ) :
Real.sin (-x) = -Real.sin x ∧
Real.cos (-x) = Real.cos x
Theorem 4.7.4: For all real numbers x, y, cos(x + y) = cos(x)cos(y) - sin(x)sin(y) and sin(x + y) = sin(x)cos(y) + cos(x)sin(y).
theorem real_sine_cosine_addition_formulas (x y : ℝ) :
Real.cos (x + y) = Real.cos x * Real.cos y - Real.sin x * Real.sin y ∧
Real.sin (x + y) = Real.sin x * Real.cos y + Real.cos x * Real.sin y
Theorem 4.7.5: We have sin(0) = 0 and cos(0) = 1.
theorem real_sine_and_cosine_at_zero :
Real.sin 0 = 0 ∧ Real.cos 0 = 1
Lemma 4.7.1: There exists a real number x > 0 such that sin x = 0.
lemma exists_positive_real_with_sine_zero :
∃ x : ℝ, 0 < x ∧ Real.sin x = 0
Definition 4.7.2: Define π to be inf {x ∈ (0, ∞) : Real.sin x = 0}.
noncomputable def firstPositiveSineZero : ℝ :=
sInf {x : ℝ | x ∈ Set.Ioi (0 : ℝ) ∧ Real.sin x = 0}
Definition 4.7.3: The tangent function is the map tan : (-π/2, π/2) → ℝ defined by tan(x) = sin x / cos x.
noncomputable def tangentFunction : Set.Ioo (-(firstPositiveSineZero / 2)) (firstPositiveSineZero / 2) → ℝ :=
fun x => Real.sin x / Real.cos x
Definition 4.7.4: Define f : ℝ → ℝ by f(x) = ∑' n = 1..∞, 4^{-n} * cos(32^n * π * x). Equivalently (reindexing by n : ℕ), f(x) = ∑' n, (1 / 4^(n+1)) * cos(32^(n+1) * π * x).
noncomputable def scaledCosineSeriesFunction : ℝ → ℝ :=
fun x =>
∑' n : ℕ,
(1 / (4 : ℝ) ^ (n + 1)) * Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x)
Proposition 4.7.8: Define tan : (-π/2, π/2) → ℝ by tan(x) = sin x / cos x. Then tan is differentiable on (-π/2, π/2) and (d/dx) tan(x) = 1 + tan(x)^2 for x ∈ (-π/2, π/2). In particular, tan is strictly increasing on (-π/2, π/2), lim_{x→(π/2)^-} tan(x) = +∞, lim_{x→(-π/2)^+} tan(x) = -∞, and hence tan is a bijection from (-π/2, π/2) onto ℝ.
theorem tangent_differentiableOn_deriv_strictMono_limits_and_bijective :
DifferentiableOn ℝ (fun x : ℝ => Real.sin x / Real.cos x)
(Set.Ioo (-(firstPositiveSineZero / 2)) (firstPositiveSineZero / 2)) ∧
(∀ x ∈ Set.Ioo (-(firstPositiveSineZero / 2)) (firstPositiveSineZero / 2),
deriv (fun y : ℝ => Real.sin y / Real.cos y) x =
1 + (Real.sin x / Real.cos x) ^ 2) ∧
StrictMonoOn (fun x : ℝ => Real.sin x / Real.cos x)
(Set.Ioo (-(firstPositiveSineZero / 2)) (firstPositiveSineZero / 2)) ∧
Filter.Tendsto (fun x : ℝ => Real.sin x / Real.cos x)
(nhdsWithin (firstPositiveSineZero / 2) (Set.Iio (firstPositiveSineZero / 2))) Filter.atTop ∧
Filter.Tendsto (fun x : ℝ => Real.sin x / Real.cos x)
(nhdsWithin (-(firstPositiveSineZero / 2)) (Set.Ioi (-(firstPositiveSineZero / 2)))) Filter.atBot ∧
Function.Bijective
(fun x : Set.Ioo (-(firstPositiveSineZero / 2)) (firstPositiveSineZero / 2) =>
Real.sin x.1 / Real.cos x.1)
Proposition 4.7.9: Let tan⁻¹ : ℝ → (-π/2, π/2) be the inverse of tan : (-π/2, π/2) → ℝ. Then Real.arctan is differentiable on ℝ and for every x : ℝ, (d/dx) arctan(x) = 1 / (1 + x^2).
theorem arctan_differentiable_and_deriv :
Differentiable ℝ Real.arctan ∧
∀ x : ℝ, deriv Real.arctan x = 1 / (1 + x ^ 2)
Proposition 4.7.10: For every real number x with |x| < 1, the arctangent function has the power series expansion arctan x = ∑' n, (-1)^n * x^(2n+1) / (2n+1).
theorem arctan_eq_tsum_powerSeries_of_abs_lt_one
{x : ℝ} (hx : |x| < 1) :
Real.arctan x =
∑' n : ℕ, (((-1 : ℝ) ^ n) * x ^ (2 * n + 1)) / (2 * n + 1 : ℝ)
Proposition 4.7.4: If x, y : ℝ satisfy x^2 + y^2 = 1, then there exists a unique θ ∈ (-π, π] such that sin θ = x and cos θ = y.
theorem existsUnique_angle_in_Ioc_neg_pi_pi_of_sq_add_sq_eq_one
(x y : ℝ) (hxy : x ^ 2 + y ^ 2 = 1) :
∃! θ : ℝ,
θ ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero ∧
Real.sin θ = x ∧ Real.cos θ = y
Proposition 4.7.5: if r, s > 0 and θ, α : ℝ satisfy (r : ℂ) * exp(iθ) = (s : ℂ) * exp(iα), then r = s and θ = α + 2πk for some k : ℤ.
theorem complex_polar_exp_eq_iff
(r s θ α : ℝ)
(hr : 0 < r)
(hs : 0 < s)
(hEq : (r : ℂ) * Complex.exp (Complex.I * (θ : ℂ)) =
(s : ℂ) * Complex.exp (Complex.I * (α : ℂ))) :
r = s ∧ ∃ k : ℤ, θ = α + 2 * Real.pi * (k : ℝ)
Proposition 4.7.6: for every nonzero complex number z, there exists a unique pair (r, θ) with r > 0 and θ ∈ (-π, π] (where π is firstPositiveSineZero) such that z = r * exp(iθ).
theorem complex_existsUnique_polar_coordinates_Ioc
(z : ℂ) (hz : z ≠ 0) :
∃! p : ℝ × ℝ,
0 < p.1 ∧
p.2 ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero ∧
z = (p.1 : ℂ) * Complex.exp (Complex.I * (p.2 : ℂ))
Theorem 4.7.6: for every real number x, exp(ix) = cos(x) + i sin(x) and exp(-ix) = cos(x) - i sin(x). In particular, cos(x) = Re(exp(ix)) and sin(x) = Im(exp(ix)).
theorem real_euler_formula_and_real_imag_parts (x : ℝ) :
Complex.exp (Complex.I * (x : ℂ)) = (Real.cos x : ℂ) + Complex.I * (Real.sin x : ℂ) ∧
Complex.exp (-Complex.I * (x : ℂ)) = (Real.cos x : ℂ) - Complex.I * (Real.sin x : ℂ) ∧
Real.cos x = Complex.re (Complex.exp (Complex.I * (x : ℂ))) ∧
Real.sin x = Complex.im (Complex.exp (Complex.I * (x : ℂ)))
Proposition 4.7.7 (de Moivre identities): for every θ : ℝ and n : ℤ, cos(nθ) and sin(nθ) are the real and imaginary parts of (cos θ + i sin θ)^n; equivalently, (cos θ + i sin θ)^n = cos(nθ) + i sin(nθ).
theorem de_moivre_identities_integer_powers (θ : ℝ) (n : ℤ) :
Real.cos ((n : ℝ) * θ) =
Complex.re (((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) ^ n) ∧
Real.sin ((n : ℝ) * θ) =
Complex.im (((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) ^ n) ∧
(((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) ^ n) =
(Real.cos ((n : ℝ) * θ) : ℂ) + Complex.I * (Real.sin ((n : ℝ) * θ) : ℂ)
Theorem 4.7.7: For every real number x (with π from Definition 4.7.2), cos (x + π) = -cos x and sin (x + π) = -sin x. In particular, cos (x + 2π) = cos x and sin (x + 2π) = sin x, so sin and cos are periodic with period 2π.
theorem real_sine_cosine_pi_antiperiodicity_and_two_pi_periodicity :
(∀ x : ℝ,
Real.cos (x + firstPositiveSineZero) = -Real.cos x ∧
Real.sin (x + firstPositiveSineZero) = -Real.sin x) ∧
(∀ x : ℝ,
Real.cos (x + 2 * firstPositiveSineZero) = Real.cos x ∧
Real.sin (x + 2 * firstPositiveSineZero) = Real.sin x) ∧
Function.Periodic Real.cos (2 * firstPositiveSineZero) ∧
Function.Periodic Real.sin (2 * firstPositiveSineZero)
Theorem 4.7.8: For every real number x (with π given by Definition 4.7.2), Real.sin x = 0 if and only if x / π is an integer.
theorem real_sine_eq_zero_iff_div_firstPositiveSineZero_integer (x : ℝ) :
Real.sin x = 0 ↔ ∃ n : ℤ, x / firstPositiveSineZero = (n : ℝ)
Helper for Proposition 4.7.1: the absolute-value even cosine-series coefficients are summable over ℝ.
lemma helperForProposition_4_7_1_abs_even_series_summable (x : ℝ) :
Summable (fun n : ℕ => |(((-1 : ℝ) ^ n * x ^ (2 * n)) / (Nat.factorial (2 * n) : ℝ))|)
Helper for Proposition 4.7.1: the absolute-value odd sine-series coefficients are summable over ℝ.
lemma helperForProposition_4_7_1_abs_odd_series_summable (x : ℝ) :
Summable
(fun n : ℕ => |(((-1 : ℝ) ^ n * x ^ (2 * n + 1)) / (Nat.factorial (2 * n + 1) : ℝ))|)
Helper for Proposition 4.7.1: the signed cosine and sine coefficient series are summable over ℝ.
lemma helperForProposition_4_7_1_signed_series_summable (x : ℝ) :
Summable (fun n : ℕ => (((-1 : ℝ) ^ n) * x ^ (2 * n)) / (Nat.factorial (2 * n) : ℝ)) ∧
Summable (fun n : ℕ => (((-1 : ℝ) ^ n) * x ^ (2 * n + 1)) / (Nat.factorial (2 * n + 1) : ℝ))
Helper for Proposition 4.7.1: assemble the final conjunction for the let-bound series definitions.
lemma helperForProposition_4_7_1_assemble_let_bound_goal (x : ℝ) :
let a : ℕ → ℝ :=
fun n => (((-1 : ℝ) ^ n) * x ^ (2 * n)) / (Nat.factorial (2 * n) : ℝ)
let b : ℕ → ℝ :=
fun n => (((-1 : ℝ) ^ n) * x ^ (2 * n + 1)) / (Nat.factorial (2 * n + 1) : ℝ)
Summable (fun n : ℕ => |a n|) ∧
Summable (fun n : ℕ => |b n|) ∧
Summable a ∧ Summable b
Helper for Proposition 4.7.2: a nonzero derivative at x₀ forces nearby punctured points to have function value different from f x₀.
lemma helperForProposition_4_7_2_eventually_ne_value_at_x0
(f : ℝ → ℝ) (x₀ : ℝ)
(hf_diff : DifferentiableAt ℝ f x₀)
(hf_deriv_ne_zero : deriv f x₀ ≠ 0) :
∀ᶠ y in nhdsWithin x₀ (({x₀} : Set ℝ)ᶜ), f y ≠ f x₀
Helper for Proposition 4.7.2: rewriting local inequality against f x₀ as local inequality against 0 when f x₀ = 0.
lemma helperForProposition_4_7_2_eventually_ne_zero
(f : ℝ → ℝ) (x₀ : ℝ)
(hf_zero : f x₀ = 0)
(h_ne_value_at_x0 : ∀ᶠ y in nhdsWithin x₀ (({x₀} : Set ℝ)ᶜ), f y ≠ f x₀) :
∀ᶠ y in nhdsWithin x₀ (({x₀} : Set ℝ)ᶜ), f y ≠ 0
Helper for Proposition 4.7.2: an eventual punctured-neighborhood property yields a positive radius with the textbook 0 < |y - x₀| < c form.
lemma helperForProposition_4_7_2_radius_from_eventually_punctured
(x₀ : ℝ) {P : ℝ → Prop}
(hP : ∀ᶠ y in nhdsWithin x₀ (({x₀} : Set ℝ)ᶜ), P y) :
∃ c > 0, ∀ y : ℝ, 0 < |y - x₀| ∧ |y - x₀| < c → P y
Helper for Theorem 4.7.1: every real sine value lies in [-1, 1].
lemma helperForTheorem_4_7_1_sine_mem_Icc (x : ℝ) :
Real.sin x ∈ Set.Icc (-1) 1
Helper for Theorem 4.7.1: every real cosine value lies in [-1, 1].
lemma helperForTheorem_4_7_1_cosine_mem_Icc (x : ℝ) :
Real.cos x ∈ Set.Icc (-1) 1
Helper for Theorem 4.7.2: pointwise derivative formula for Real.sin.
lemma helperForTheorem_4_7_2_sin_deriv_pointwise (x : ℝ) :
deriv Real.sin x = Real.cos x
Helper for Theorem 4.7.2: pointwise derivative formula for Real.cos.
lemma helperForTheorem_4_7_2_cos_deriv_pointwise (x : ℝ) :
deriv Real.cos x = -Real.sin x
Helper for Proposition 4.7.8: every positive real zero of Real.sin is at least π.
lemma helperForProposition_4_7_8_pi_le_of_positive_sine_zero
(t : ℝ) (ht_pos : 0 < t) (ht_sin : Real.sin t = 0) :
Real.pi ≤ t
Helper for Proposition 4.7.8: the infimum defining firstPositiveSineZero equals π.
lemma helperForProposition_4_7_8_firstPositiveSineZero_eq_pi :
firstPositiveSineZero = Real.pi
Helper for Proposition 4.7.8: Real.tan is differentiable on (-π/2, π/2).
lemma helperForProposition_4_7_8_differentiableOn_tan_Ioo :
DifferentiableOn ℝ Real.tan (Set.Ioo (-(Real.pi / 2)) (Real.pi / 2))
Helper for Proposition 4.7.8: the derivative formula tan' = 1 + tan^2 on (-π/2, π/2).
lemma helperForProposition_4_7_8_deriv_tan_eq_one_add_tan_sq_on_Ioo :
∀ x ∈ Set.Ioo (-(Real.pi / 2)) (Real.pi / 2),
deriv Real.tan x = 1 + Real.tan x ^ 2
Helper for Proposition 4.7.8: Real.tan is bijective from (-π/2, π/2) to ℝ.
lemma helperForProposition_4_7_8_bijective_tan_subtype :
Function.Bijective (fun x : Set.Ioo (-(Real.pi / 2)) (Real.pi / 2) => Real.tan x.1)
Helper for Proposition 4.7.9: Real.arctan is differentiable on all real numbers.
lemma helperForProposition_4_7_9_differentiable_arctan_global :
Differentiable ℝ Real.arctan
Helper for Proposition 4.7.9: pointwise derivative formula for Real.arctan.
lemma helperForProposition_4_7_9_deriv_arctan_pointwise (x : ℝ) :
deriv Real.arctan x = 1 / (1 + x ^ 2)
Helper for Proposition 4.7.9: alternate pointwise derivative formula from Real.deriv_arctan.
lemma helperForProposition_4_7_9_deriv_arctan_pointwise_alt (x : ℝ) :
deriv Real.arctan x = 1 / (1 + x ^ 2)
Helper for Proposition 4.7.4: the positive-zero set used in the definition of firstPositiveSineZero is nonempty and bounded below.
lemma helperForProposition_4_7_4_positiveSineZeroSet_nonempty_bddBelow :
Set.Nonempty {t : ℝ | t ∈ Set.Ioi (0 : ℝ) ∧ Real.sin t = 0} ∧
BddBelow {t : ℝ | t ∈ Set.Ioi (0 : ℝ) ∧ Real.sin t = 0}
Helper for Proposition 4.7.4: every positive real zero of Real.sin is at least π.
lemma helperForProposition_4_7_4_pi_le_of_positive_sine_zero
(t : ℝ) (ht_pos : 0 < t) (ht_sin : Real.sin t = 0) :
Real.pi ≤ t
Helper for Proposition 4.7.4: the definition of firstPositiveSineZero agrees with the standard constant Real.pi.
lemma helperForProposition_4_7_4_firstPositiveSineZero_eq_pi :
firstPositiveSineZero = Real.pi
Helper for Proposition 4.7.4: existence of an angle in (-π, π] with prescribed sin and cos coordinates on the unit circle.
lemma helperForProposition_4_7_4_exists_angle_in_Ioc_neg_pi_pi
(x y : ℝ) (hxy : x ^ 2 + y ^ 2 = 1) :
∃ θ : ℝ, θ ∈ Set.Ioc (-Real.pi) Real.pi ∧ Real.sin θ = x ∧ Real.cos θ = y
Helper for Proposition 4.7.4: uniqueness of the angle in (-π, π] with fixed sin and cos values.
lemma helperForProposition_4_7_4_angle_unique_in_Ioc_neg_pi_pi
(x y θ₁ θ₂ : ℝ)
(hθ₁ : θ₁ ∈ Set.Ioc (-Real.pi) Real.pi)
(hθ₂ : θ₂ ∈ Set.Ioc (-Real.pi) Real.pi)
(hsin₁ : Real.sin θ₁ = x)
(hcos₁ : Real.cos θ₁ = y)
(hsin₂ : Real.sin θ₂ = x)
(hcos₂ : Real.cos θ₂ = y) :
θ₁ = θ₂
Helper for Proposition 4.7.5: the norm of exp(i x) is 1 for real x.
lemma helperForProposition_4_7_5_norm_exp_I_mul_real (x : ℝ) :
‖Complex.exp (Complex.I * (x : ℂ))‖ = 1
Helper for Proposition 4.7.5: equality of scaled polar exponentials with positive radii forces equality of radii.
lemma helperForProposition_4_7_5_radius_eq_of_scaled_exp_eq
(r s θ α : ℝ)
(hr : 0 < r)
(hs : 0 < s)
(hEq : (r : ℂ) * Complex.exp (Complex.I * (θ : ℂ)) =
(s : ℂ) * Complex.exp (Complex.I * (α : ℂ))) :
r = s
Helper for Proposition 4.7.5: after identifying radii, one can cancel the nonzero real scalar and deduce equality of the exponential factors.
lemma helperForProposition_4_7_5_exp_eq_of_radius_eq
(r s θ α : ℝ)
(hr : 0 < r)
(hrs : r = s)
(hEq : (r : ℂ) * Complex.exp (Complex.I * (θ : ℂ)) =
(s : ℂ) * Complex.exp (Complex.I * (α : ℂ))) :
Complex.exp (Complex.I * (θ : ℂ)) = Complex.exp (Complex.I * (α : ℂ))
Helper for Proposition 4.7.5: equality of exp(iθ) and exp(iα) forces an integer 2π angle shift between θ and α.
lemma helperForProposition_4_7_5_angle_shift_of_exp_I_eq
(θ α : ℝ)
(hExpEq : Complex.exp (Complex.I * (θ : ℂ)) = Complex.exp (Complex.I * (α : ℂ))) :
∃ k : ℤ, θ = α + 2 * Real.pi * (k : ℝ)
Helper for Proposition 4.7.6: the canonical witness (‖z‖, arg z) satisfies the required positivity, interval, and reconstruction properties.
lemma helperForProposition_4_7_6_exists_polar_witness
(z : ℂ) (hz : z ≠ 0) :
∃ p : ℝ × ℝ,
0 < p.1 ∧
p.2 ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero ∧
z = (p.1 : ℂ) * Complex.exp (Complex.I * (p.2 : ℂ))
Helper for Proposition 4.7.6: if two angles lie in (-π, π] and differ by an integer multiple of 2π, then the integer shift is zero.
lemma helperForProposition_4_7_6_int_shift_zero_of_interval_bounds
{θ α : ℝ} {k : ℤ}
(hθ : θ ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero)
(hα : α ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero)
(hEq : θ = α + 2 * Real.pi * (k : ℝ)) :
k = 0
Helper for Proposition 4.7.6: two admissible polar representations of the same nonzero complex number have identical radius and angle.
lemma helperForProposition_4_7_6_radius_angle_unique
{z : ℂ} {r s θ α : ℝ}
(hr : 0 < r)
(hs : 0 < s)
(hθ : θ ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero)
(hα : α ∈ Set.Ioc (-firstPositiveSineZero) firstPositiveSineZero)
(hEqR : z = (r : ℂ) * Complex.exp (Complex.I * (θ : ℂ)))
(hEqS : z = (s : ℂ) * Complex.exp (Complex.I * (α : ℂ))) :
r = s ∧ θ = α
Helper for Theorem 4.7.6: Euler's formula in the exp(i x) direction for real x.
lemma helperForTheorem_4_7_6_exp_I_mul_ofReal (x : ℝ) :
Complex.exp (Complex.I * (x : ℂ)) = (Real.cos x : ℂ) + Complex.I * (Real.sin x : ℂ)
Helper for Theorem 4.7.6: Euler's formula in the exp(-i x) direction for real x.
lemma helperForTheorem_4_7_6_exp_neg_I_mul_ofReal (x : ℝ) :
Complex.exp (-Complex.I * (x : ℂ)) = (Real.cos x : ℂ) - Complex.I * (Real.sin x : ℂ)
Helper for Theorem 4.7.6: real and imaginary parts of exp(i x) for real x.
lemma helperForTheorem_4_7_6_re_im_of_exp_I_mul (x : ℝ) :
Real.cos x = Complex.re (Complex.exp (Complex.I * (x : ℂ))) ∧
Real.sin x = Complex.im (Complex.exp (Complex.I * (x : ℂ)))
Helper for Proposition 4.7.7: rewrite cos θ + i sin θ as exp(iθ).
lemma helperForProposition_4_7_7_base_as_exp (θ : ℝ) :
((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) =
Complex.exp (Complex.I * (θ : ℂ))
Helper for Proposition 4.7.7: normalize the scaled exponential argument for integer powers.
lemma helperForProposition_4_7_7_exp_int_argument_rewrite (θ : ℝ) (n : ℤ) :
(n : ℂ) * (Complex.I * (θ : ℂ)) = Complex.I * (((n : ℝ) * θ : ℝ) : ℂ)
Helper for Proposition 4.7.7: convert (cos θ + i sin θ)^n to one exponential at argument (n : ℝ) * θ.
lemma helperForProposition_4_7_7_power_as_exp_scaled (θ : ℝ) (n : ℤ) :
(((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) ^ n) =
Complex.exp (Complex.I * (((n : ℝ) * θ : ℝ) : ℂ))
Helper for Proposition 4.7.7: identify cos(nθ) and sin(nθ) as the real and imaginary parts of (cos θ + i sin θ)^n.
lemma helperForProposition_4_7_7_re_im_from_power (θ : ℝ) (n : ℤ) :
Real.cos ((n : ℝ) * θ) =
Complex.re (((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) ^ n) ∧
Real.sin ((n : ℝ) * θ) =
Complex.im (((Real.cos θ : ℂ) + Complex.I * (Real.sin θ : ℂ)) ^ n)
Helper for Theorem 4.7.7: the positive sine-zero set is nonempty and bounded below.
lemma helperForTheorem_4_7_7_positiveSineZeroSet_nonempty_bddBelow :
Set.Nonempty {x : ℝ | x ∈ Set.Ioi (0 : ℝ) ∧ Real.sin x = 0} ∧
BddBelow {x : ℝ | x ∈ Set.Ioi (0 : ℝ) ∧ Real.sin x = 0}
Helper for Theorem 4.7.7: every positive real zero of sine is at least π.
lemma helperForTheorem_4_7_7_pi_is_lower_bound_of_positive_sine_zeros
(y : ℝ) (hy_pos : 0 < y) (hy_sin : Real.sin y = 0) :
Real.pi ≤ y
Helper for Theorem 4.7.7: the definition of firstPositiveSineZero agrees with Real.pi.
lemma helperForTheorem_4_7_7_firstPositiveSineZero_eq_pi :
firstPositiveSineZero = Real.pi
Helper for Theorem 4.7.7: transport π-identities to firstPositiveSineZero.
lemma helperForTheorem_4_7_7_transport_antiperiodic_periodic
(hpi : firstPositiveSineZero = Real.pi) :
(∀ x : ℝ,
Real.cos (x + firstPositiveSineZero) = -Real.cos x ∧
Real.sin (x + firstPositiveSineZero) = -Real.sin x) ∧
(∀ x : ℝ,
Real.cos (x + 2 * firstPositiveSineZero) = Real.cos x ∧
Real.sin (x + 2 * firstPositiveSineZero) = Real.sin x) ∧
Function.Periodic Real.cos (2 * firstPositiveSineZero) ∧
Function.Periodic Real.sin (2 * firstPositiveSineZero)
Helper for Theorem 4.7.8: firstPositiveSineZero is nonzero.
lemma helperForTheorem_4_7_8_firstPositiveSineZero_ne_zero :
firstPositiveSineZero ≠ 0
Helper for Theorem 4.7.8: divide-by-firstPositiveSineZero form equals integer-multiple-of-π form.
lemma helperForTheorem_4_7_8_div_eq_int_iff_mul_eq_int_pi (x : ℝ) (n : ℤ) :
x / firstPositiveSineZero = (n : ℝ) ↔ x = (n : ℝ) * Real.pi
Helper for Theorem 4.7.8: the integer-existence formulations using division by firstPositiveSineZero and multiplication by π are equivalent.
lemma helperForTheorem_4_7_8_exists_div_eq_int_iff_exists_mul_eq_int_pi (x : ℝ) :
(∃ n : ℤ, x / firstPositiveSineZero = (n : ℝ)) ↔
(∃ n : ℤ, x = (n : ℝ) * Real.pi)4.7 — Trigonometric Functions — Part 2
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Theorem 4.7.9: For every real number x (with π given by Definition 4.7.2), Real.cos x = 0 if and only if x / π is an integer plus (1/2).
theorem real_cos_eq_zero_iff_div_firstPositiveSineZero_half_integer (x : ℝ) :
Real.cos x = 0 ↔ ∃ n : ℤ, x / firstPositiveSineZero = (n : ℝ) + (1 / 2 : ℝ)
Proposition 4.7.11: the Leibniz series identity π = 4 * (1 - 1/3 + 1/5 - 1/7 + ⋯) holds in Lean via the conditional summation filter on ℕ; in particular, 4 - 4/3 < π < 4.
theorem firstPositiveSineZero_eq_four_mul_tsum_alternating_inv_odd_and_bounds :
firstPositiveSineZero =
4 * ∑'[SummationFilter.conditional ℕ] n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ)) ∧
4 - 4 / 3 < firstPositiveSineZero ∧ firstPositiveSineZero < 4
Proposition 4.7.12: for f(x) = ∑_{n=1}^∞ 4^{-n} cos(32^n π x) (with π = firstPositiveSineZero from Definition 4.7.2), the series converges uniformly on ℝ; in particular, f is continuous on ℝ.
theorem scaledCosineSeries_uniformly_convergent_and_continuous :
TendstoUniformly
(fun N => fun x =>
Finset.sum (Finset.range N) (fun n =>
(1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x)))
scaledCosineSeriesFunction Filter.atTop ∧
Continuous scaledCosineSeriesFunction
Proposition 4.7.13: for every j : ℤ and every integer m ≥ 1, for f(x) = ∑_{n=1}^∞ 4^{-n} cos(32^n π x) (with π = firstPositiveSineZero), the increment across consecutive points of mesh size 32^{-m} satisfies |f((j+1)/32^m) - f(j/32^m)| ≥ 4^{-m}.
theorem scaledCosineSeries_increment_lower_bound
(j : ℤ) (m : ℕ) (hm : 1 ≤ m) :
Helper for Theorem 4.7.9: shifting by firstPositiveSineZero / 2 converts Real.sin to Real.cos.
lemma helperForTheorem_4_7_9_sin_shift_half_firstPositiveSineZero_eq_cos (x : ℝ) :
Real.sin (x + firstPositiveSineZero / 2) = Real.cos x
Helper for Theorem 4.7.9: division by firstPositiveSineZero after the half-period shift simplifies to adding 1/2.
lemma helperForTheorem_4_7_9_shifted_division_formula (x : ℝ) :
(x + firstPositiveSineZero / 2) / firstPositiveSineZero =
x / firstPositiveSineZero + (1 / 2 : ℝ)
Helper for Theorem 4.7.9: integer-valuedness of the shifted quotient is equivalent to half-integer-valuedness of the original quotient.
lemma helperForTheorem_4_7_9_exists_integer_shifted_iff_half_integer (x : ℝ) :
(∃ n : ℤ, (x + firstPositiveSineZero / 2) / firstPositiveSineZero = (n : ℝ)) ↔
(∃ n : ℤ, x / firstPositiveSineZero = (n : ℝ) + (1 / 2 : ℝ))
Helper for Theorem 4.7.9: Real.cos x = 0 is equivalent to vanishing of the shifted sine value.
lemma helperForTheorem_4_7_9_cos_zero_iff_shifted_sin_zero (x : ℝ) :
Real.cos x = 0 ↔ Real.sin (x + firstPositiveSineZero / 2) = 0
Helper for Proposition 4.7.11: the Leibniz series gives π / 4 for the conditional summation filter on ℕ.
lemma helperForProposition_4_7_11_hasSum_alternating_inv_odd_pi_div_four_conditional :
HasSum (fun n : ℕ => (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ))) (Real.pi / 4)
(SummationFilter.conditional ℕ)
Helper for Proposition 4.7.11: the conditionally summed Leibniz identity for firstPositiveSineZero (hence for π).
lemma helperForProposition_4_7_11_firstPositiveSineZero_eq_four_mul_conditional_tsum :
firstPositiveSineZero =
4 * ∑'[SummationFilter.conditional ℕ] n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ))
Helper for Proposition 4.7.11: the odd reciprocal series is not unconditionally summable.
lemma helperForProposition_4_7_11_not_summable_inv_odd :
¬ Summable (fun n : ℕ => (1 : ℝ) / (2 * n + 1 : ℝ))
Helper for Proposition 4.7.11: the alternating odd-reciprocal series is not unconditionally summable.
lemma helperForProposition_4_7_11_not_summable_alternating_inv_odd :
¬ Summable (fun n : ℕ => (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ)))
Helper for Proposition 4.7.11: under Lean's default unconditional tsum, the alternating odd-reciprocal series has value 0.
lemma helperForProposition_4_7_11_tsum_alternating_inv_odd_eq_zero :
(∑' n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ))) = 0
Helper for Proposition 4.7.11: consequently, 4 * ∑' n, (-1)^n / (2n+1) is 0 under unconditional tsum.
lemma helperForProposition_4_7_11_four_mul_tsum_alternating_inv_odd_eq_zero :
4 * ∑' n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ)) = 0
Helper for Proposition 4.7.11: the unconditional-series equality would force firstPositiveSineZero = 0.
lemma helperForProposition_4_7_11_eq_four_mul_tsum_implies_firstPositiveSineZero_eq_zero
(hmain : firstPositiveSineZero =
4 * ∑' n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ))) :
firstPositiveSineZero = 0
Helper for Proposition 4.7.11: lower bound 4 - 4/3 < firstPositiveSineZero.
lemma helperForProposition_4_7_11_lower_bound :
4 - 4 / 3 < firstPositiveSineZero
Helper for Proposition 4.7.11: upper bound firstPositiveSineZero < 4.
lemma helperForProposition_4_7_11_upper_bound :
firstPositiveSineZero < 4
Helper for Proposition 4.7.11: package the two stated numerical bounds into a single conjunction.
lemma helperForProposition_4_7_11_bounds_conjunction :
4 - 4 / 3 < firstPositiveSineZero ∧ firstPositiveSineZero < 4
Helper for Proposition 4.7.11: under unconditional tsum, the asserted Leibniz equality would force firstPositiveSineZero = 0, contradicting firstPositiveSineZero = Real.pi.
lemma helperForProposition_4_7_11_not_firstPositiveSineZero_eq_four_mul_tsum_alternating_inv_odd :
¬ firstPositiveSineZero = 4 * ∑' n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ))
Helper for Proposition 4.7.11: the full conjunction in the target statement is inconsistent with unconditional tsum for this series.
lemma helperForProposition_4_7_11_not_target_conjunction :
¬ (firstPositiveSineZero = 4 * ∑' n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ)) ∧
4 - 4 / 3 < firstPositiveSineZero ∧ firstPositiveSineZero < 4)
Helper for Proposition 4.7.11: combining any witness of the main Leibniz equality with the proven numerical bounds yields a contradiction under unconditional tsum.
lemma helperForProposition_4_7_11_false_of_main_conjunct_and_bounds
(hmain : firstPositiveSineZero =
4 * ∑' n : ℕ, (((-1 : ℝ) ^ n) / (2 * n + 1 : ℝ)))
(hbounds : 4 - 4 / 3 < firstPositiveSineZero ∧ firstPositiveSineZero < 4) : False
Helper for Proposition 4.7.12: each summand in the scaled cosine series is continuous.
lemma helperForProposition_4_7_12_termContinuous (n : ℕ) :
Continuous (fun x : ℝ =>
(1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x))
Helper for Proposition 4.7.12: the geometric majorant 4^{-(n+1)} is summable.
lemma helperForProposition_4_7_12_majorantSummable :
Summable (fun n : ℕ => 1 / (4 : ℝ) ^ (n + 1))
Helper for Proposition 4.7.12: each summand is bounded in norm by 4^{-(n+1)}.
lemma helperForProposition_4_7_12_termNormBound (n : ℕ) (x : ℝ) :
‖(1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x)‖ ≤
1 / (4 : ℝ) ^ (n + 1)
Helper for Proposition 4.7.12: partial sums converge uniformly to the infinite series.
lemma helperForProposition_4_7_12_uniformLimit_raw :
TendstoUniformly
(fun N => fun x =>
Finset.sum (Finset.range N) (fun n =>
(1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x)))
scaledCosineSeriesFunction
Filter.atTop
Helper for Proposition 4.7.12: each partial sum in the scaled cosine series is continuous.
lemma helperForProposition_4_7_12_partialSumsContinuous (N : ℕ) :
Continuous (fun x : ℝ =>
Finset.sum (Finset.range N) (fun n =>
(1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x)))
Helper for Proposition 4.7.13: the n-th increment term of scaledCosineSeriesFunction across mesh size 32^{-m}.
noncomputable def helperForProposition_4_7_13_diffTerm (j : ℤ) (m n : ℕ) : ℝ :=
((1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero *
(((j + 1 : ℤ) : ℝ) / ((32 : ℝ) ^ m)))) -
((1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero *
((j : ℝ) / ((32 : ℝ) ^ m))) )
Helper for Proposition 4.7.13: pointwise summability of the cosine series.
lemma helperForProposition_4_7_13_seriesSummableAtPoint (x : ℝ) :
Summable (fun n : ℕ =>
(1 / (4 : ℝ) ^ (n + 1)) *
Real.cos (((32 : ℝ) ^ (n + 1)) * firstPositiveSineZero * x))
Helper for Proposition 4.7.13: for n ≥ m, the phase shift 32^(n+1-m) * π is an integer multiple of 2π.
lemma helperForProposition_4_7_13_phaseShift_asTwoPiMultiple
(m n : ℕ) (hmn : m ≤ n) :
∃ k : ℕ,
((32 : ℝ) ^ (n + 1 - m)) * firstPositiveSineZero = (k : ℝ) * (2 * Real.pi)
Helper for Proposition 4.7.13: all modes n ≥ m vanish in the increment.
lemma helperForProposition_4_7_13_highModesVanish
(j : ℤ) (m n : ℕ) (hnm : m ≤ n) :
helperForProposition_4_7_13_diffTerm j m n = 0
Helper for Proposition 4.7.13: the increment equals a finite sum over modes < m.
lemma helperForProposition_4_7_13_increment_asFiniteSum
(j : ℤ) (m : ℕ) :
scaledCosineSeriesFunction (((j + 1 : ℤ) : ℝ) / ((32 : ℝ) ^ m)) -
scaledCosineSeriesFunction ((j : ℝ) / ((32 : ℝ) ^ m)) =
∑ n ∈ Finset.range m, helperForProposition_4_7_13_diffTerm j m n
Helper for Proposition 4.7.13: the main mode n = m-1 contributes exactly 2 / 4^m in absolute value.
lemma helperForProposition_4_7_13_mainMode_exact
(j : ℤ) (m : ℕ) (hm : 1 ≤ m) :
Helper for Proposition 4.7.13: finite geometric bound for ∑_{n < m-1} 8^{n+1}.
lemma helperForProposition_4_7_13_sumPowEight_bound
(m : ℕ) (hm : 1 ≤ m) :
(∑ n ∈ Finset.range (m - 1), (8 : ℝ) ^ (n + 1)) ≤ ((8 : ℝ) ^ m) / 7
Helper for Proposition 4.7.13: each increment mode is bounded by a Lipschitz estimate for cosine.
lemma helperForProposition_4_7_13_singleLowMode_absBound
(j : ℤ) (m n : ℕ) :
Helper for Proposition 4.7.13: the finite sum of absolute low-mode terms is bounded by the corresponding geometric series.
lemma helperForProposition_4_7_13_lowMode_sumAbs_bound
(j : ℤ) (m : ℕ) :
(∑ n ∈ Finset.range (m - 1), |helperForProposition_4_7_13_diffTerm j m n|) ≤
(Real.pi / (32 : ℝ) ^ m) *
(∑ n ∈ Finset.range (m - 1), (8 : ℝ) ^ (n + 1))
Helper for Proposition 4.7.13: the geometric-factor estimate implies the final 1 / 4^m bound.
lemma helperForProposition_4_7_13_lowMode_series_to_quarterPow
(m : ℕ) (hm : 1 ≤ m) :
(Real.pi / (32 : ℝ) ^ m) * (((8 : ℝ) ^ m) / 7) ≤ 1 / (4 : ℝ) ^ m
Helper for Proposition 4.7.13: the sum of low modes n < m-1 is bounded by 1 / 4^m in absolute value.
lemma helperForProposition_4_7_13_lowModes_bound
(j : ℤ) (m : ℕ) (hm : 1 ≤ m) :
Helper for Proposition 4.7.13: if the main term has size 2/4^m and the error is bounded by 1/4^m, then the total has size at least 1/4^m.
lemma helperForProposition_4_7_13_absLowerBound_from_main_plus_error
(m : ℕ) {main error : ℝ}
(hmain : |main| = 2 / (4 : ℝ) ^ m)
(herror : |error| ≤ 1 / (4 : ℝ) ^ m) :4.7 — Trigonometric Functions — Part 3
Lean authors: Zichen Wang, Wanli Ma, Zaiwen Wen.
Proposition 4.7.14: for f(x) = ∑_{n=1}^∞ 4^{-n} cos(32^n π x) (encoded here as scaledCosineSeriesFunction from Definition 4.7.4), f is nowhere differentiable on ℝ; i.e., for every x₀ : ℝ, f is not differentiable at x₀.
theorem scaledCosineSeries_nowhere_differentiable
(x₀ : ℝ) :
¬ DifferentiableAt ℝ scaledCosineSeriesFunction x₀
Helper for Proposition 4.7.14: floor-bracketing x₀ between consecutive 32^{-m}-mesh points.
lemma helperForProposition_4_7_14_gridBracket_by_floor
(x₀ : ℝ) (m : ℕ) :
let j : ℤ := Int.floor (x₀ * (32 : ℝ) ^ m)
let a : ℝ := (j : ℝ) / ((32 : ℝ) ^ m)
let b : ℝ := ((j + 1 : ℤ) : ℝ) / ((32 : ℝ) ^ m)
a ≤ x₀ ∧ x₀ < b ∧ b - a = 1 / ((32 : ℝ) ^ m)
Helper for Proposition 4.7.14: ratio identity (8^m)/(32^m) = 1/(4^m).
lemma helperForProposition_4_7_14_powRatio_eight_div_thirtyTwo
(m : ℕ) :
((8 : ℝ) ^ m) / ((32 : ℝ) ^ m) = 1 / ((4 : ℝ) ^ m)
Helper for Proposition 4.7.14: a lower bound on numerator and an upper bound on distance imply a large secant quotient.
lemma helperForProposition_4_7_14_secantLower_from_numerator_and_distance
(x₀ y : ℝ) (m : ℕ)
(hnum : 1 / (2 * (4 : ℝ) ^ m) ≤
|scaledCosineSeriesFunction y - scaledCosineSeriesFunction x₀|)
(hdist_pos : 0 < |y - x₀|)
(hdist_le : |y - x₀| ≤ 1 / ((32 : ℝ) ^ m)) :
(8 : ℝ) ^ m / 2 ≤
Helper for Proposition 4.7.14: for every mesh scale, one can find a nearby point with secant slope at least (8^m)/2.
lemma helperForProposition_4_7_14_exists_large_local_secant
(x₀ : ℝ) (m : ℕ) (hm : 1 ≤ m) :
∃ y : ℝ,
0 < |y - x₀| ∧
Helper for Proposition 4.7.14: differentiability at x₀ forces eventual boundedness of punctured-neighborhood secant quotients.
lemma helperForProposition_4_7_14_eventual_secant_bound_of_differentiableAt
(x₀ : ℝ) (hf : DifferentiableAt ℝ scaledCosineSeriesFunction x₀) :
∀ᶠ y in nhdsWithin x₀ (({x₀} : Set ℝ)ᶜ),
Helper for Proposition 4.7.14: choose a scale m making 32^{-m} tiny and (8^m)/2 larger than a prescribed bound.
lemma helperForProposition_4_7_14_choose_m_against_delta_and_C
(δ : ℝ) (hδ : 0 < δ) (C : ℝ) :
∃ m : ℕ, 1 ≤ m ∧ 1 / ((32 : ℝ) ^ m) < δ ∧ C < (8 : ℝ) ^ m / 2