Posterior Cumulants as Log-Partition Derivatives
ProvedFunctionalInequalities.Probability.Posterior.cumulant_compactPosterior_modularThis formalizes the derivative representation of posterior cumulants used in Lemma 4.3 of Bizeul–Klartag–Lehec (arXiv:2610.05474v1), cited there for cumulant dynamics.
Let be a complete finite-dimensional real inner-product space, let be a compactly supported probability law, let , let be continuous linear, and let . Then the th cumulant tensor of the compact posterior equals the th iterated derivative in the linear parameter of the log-partition function:
In Lean, the derivative is iteratedFDeriv ℝ m and the cumulant is Analysis.BakryEmery.cumulant. This identity is the deterministic bridge between posterior cumulants and smooth parameter calculus.
Superseded upload. This snapshot was replaced by FunctionalInequalities.Probability.Posterior.cumulant_compactPosterior_readable: open the readable replacement. The replacement preserves the frozen source and mathematical statement, with a compact declaration context. This historical node remains available for existing imports.
import Definitions.Def_KLSModular_FunctionalInequalities_Probability_Posterior_LogPartition
noncomputable section
open MeasureTheory ProbabilityTheory
open scoped RealInnerProductSpace ContDiff
section
namespace FunctionalInequalities
end FunctionalInequalities
open FunctionalInequalities
section
namespace FunctionalInequalities.Probability
end FunctionalInequalities.Probability
open FunctionalInequalities FunctionalInequalities.Probability
section
namespace FunctionalInequalities.Probability.Posterior
end FunctionalInequalities.Probability.Posterior
open FunctionalInequalities FunctionalInequalities.Probability FunctionalInequalities.Probability.Posterior
variable {E : Type*} [NormedAddCommGroup E] [InnerProductSpace ℝ E] [CompleteSpace E]
[FiniteDimensional ℝ E] [MeasurableSpace E] [BorelSpace E]
theorem FunctionalInequalities.Probability.Posterior.cumulant_compactPosterior_modular (μ : ProbabilityMeasure E)
(hc : IsCompact (μ : Measure E).support) (c : E) (Q : E →L[ℝ] E) {m : ℕ} (hm : 1 ≤ m) :
Analysis.BakryEmery.cumulant (compactPosterior μ hc c Q : Measure E) m =
iteratedFDeriv ℝ m (fun y => logPartition μ y Q) c := by sorry