A Gaussian limit for the composition-gap part-count model
ProvedAnalyticCombinatorics.Ch9.LimitLaws.compositionParts_tendstoInDistribution_gaussiananalytic-combinatoricslean4limit-laws
At size , the source probability space is the real sign-gap product model with coordinates and law compositionGapP n; natural subtraction is used. On the full real-coordinate carrier define , where the natural number is cast to the reals. On the sign-supported product law this equals one plus the number of positive gaps. Then
Finite mass identities relate this gap model to composition counts for positive and positive part counts; this headline is stated on the gap model, without an extra random-variable transport to uniform composition objects. The source carriers and measures vary with , and the limit is the identity random variable on with the standard Gaussian measure.
Preamble
import Definitions.Def_AnalyticCombinatorics_Ch9_LimitLaws_BinaryWordCLT import Definitions.Def_AnalyticCombinatorics_Compatibility_TendstoInDistribution import Definitions.Def_AnalyticCombinatorics_Ch9_LimitLaws_CompositionParts import Mathlib import Mathlib.MeasureTheory.Function.ConvergenceInDistribution set_option autoImplicit true set_option relaxedAutoImplicit false set_option maxSynthPendingDepth 3 /-! # Number of parts in a random composition For `n ≥ 1`, a composition of `n` with `k` parts is equivalently a choice of `k - 1` of the `n - 1` interior gaps. Thus the part count has the same law as `1 + Binomial(n - 1, 1 / 2)`. The central limit theorem is proved by reusing the local quasi-powers instantiation for binary words from `BinaryWordCLT.lean`, shifted from `n` to `n - 1` and then translated by the deterministic `+1`. -/ open Complex Filter MeasureTheory ProbabilityTheory open scoped Topology open AnalyticCombinatorics open AnalyticCombinatorics.Ch9 open AnalyticCombinatorics.Ch9.LimitLaws
Formal statement
theorem AnalyticCombinatorics.Ch9.LimitLaws.compositionParts_tendstoInDistribution_gaussian :
AnalyticCombinatorics.Compatibility.TendstoInDistribution
(fun n ω =>
(compositionParts n ω - ((n : ℝ) + 1) / 2) /
Real.sqrt (((n - 1 : ℕ) : ℝ) / 4)) atTop
(fun x : ℝ => x) compositionGapP (gaussianReal 0 1) := by sorrySource
Human review
Confirmed by the mission captain (proposal self-audit).