theorem
proved
tactic proof
log_aczel_data_of_laws
show as:
view Lean formalization →
formal statement (Lean)
158theorem log_aczel_data_of_laws
159 (C : ComparisonOperator) (h : SatisfiesLawsOfLogicContinuous C) :
160 ∃ P : ℝ → ℝ → ℝ,
161 LogAczelData (fun t : ℝ => derivedCost C (Real.exp t)) P := by
proof body
Tactic-mode proof.
162 obtain ⟨P, hPcont, hPsym, hCons⟩ := h.route_independence
163 refine ⟨P, ?_⟩
164 have hFcont : ContinuousOn (derivedCost C) (Set.Ioi (0 : ℝ)) :=
165 excluded_middle_implies_continuous C h.excluded_middle
166 have hNorm : derivedCost C 1 = 0 :=
167 identity_implies_normalized C h.identity
168 have hSymm : IsSymmetric (derivedCost C) :=
169 non_contradiction_and_scale_imply_reciprocal C h.non_contradiction h.scale_invariant
170 refine
171 { continuous_G := continuous_log_cost_of_continuousOn_positive (derivedCost C) hFcont
172 zero_G := by simpa [derivedCost] using hNorm
173 even_G := ?_
174 continuous_P := hPcont
175 symmetric_P := hPsym
176 aczel_eq := ?_ }
177 · exact IndisputableMonolith.Cost.FunctionalEquation.G_even_of_reciprocal_symmetry
178 (derivedCost C) (by intro x hx; exact hSymm x hx)
179 · intro t u
180 have htu_pos : 0 < Real.exp t := Real.exp_pos t
181 have huu_pos : 0 < Real.exp u := Real.exp_pos u
182 have h := hCons (Real.exp t) (Real.exp u) htu_pos huu_pos
183 simpa [Real.exp_add, Real.exp_sub] using h
184
185/-! ## 2b. Piece 1 mollifier scaffold
186
187The former residual `continuous_combiner_log_smoothness_bootstrap` says
188`G(t) = F(exp t)` is `C^∞`. The classical argument is mollification:
189convolve `G` with a `ContDiffBump` kernel to produce a smooth approximant
190`G_φ` and then push smoothness back to `G` through the Aczél equation.
191
192The scaffold below packages the parts Mathlib hands us for free and
193isolates the genuine analytic content for later work.
194
195In this section we prove:
196
197* `mollified G φ` exists as a measurable function and is continuous.
198* `mollified_pointwise_tendsto`: as the bump support shrinks, the
199 mollification converges to `G` pointwise (via Mathlib's
200 `ContDiffBump.convolution_tendsto_right_of_continuous`).
201
202Smoothness of `G` itself still needs the equation-side argument, but
203`G_φ` already inherits arbitrary regularity from the bump kernel.
204-/
205
206open scoped Convolution
207open MeasureTheory
208
209/-- Mollification of a continuous function on `ℝ` by a normalized
210`ContDiffBump` kernel, written as a left convolution to align with
211`ContDiffBump.convolution_tendsto_right_of_continuous`. -/
depends on (40)
-
of -
isolates -
G -
G -
kernel -
smooth -
smooth -
G -
G_even_of_reciprocal_symmetry -
via -
back -
IsSymmetric -
of -
continuous_combiner_log_smoothness_bootstrap -
continuous_log_cost_of_continuousOn_positive -
LogAczelData -
mollified -
mollified_pointwise_tendsto -
SatisfiesLawsOfLogicContinuous -
ComparisonOperator -
derivedCost -
excluded_middle_implies_continuous -
identity_implies_normalized -
non_contradiction_and_scale_imply_reciprocal -
identity -
is -
of -
from -
as -
is