Power decay on a dense subspace bounds a symmetric operator norm
ProvedMarkovChainCLT.opNorm_le_of_symmetric_power_decay_on_densedense-subspacefunctional-analysisself-adjointspectral-gap
Let be a bounded symmetric operator on a real inner-product space, and let be a dense linear subspace. Suppose and, for every , there is a finite constant such that
Then the operator norm is bounded by the same rate:
The constant may depend on the vector. The result upgrades vectorwise moment decay on a dense core to a uniform spectral bound by normalization, continuity of the quadratic form, and the Rayleigh-quotient characterization of the norm of a symmetric operator.
Preamble
import Theorems.Thm_MarkovChainCLT_quadratic_le_of_symmetric_power_decay_unit import Mathlib.Analysis.InnerProductSpace.Rayleigh open Filter Function Set open scoped NNReal Topology
Formal statement
theorem MarkovChainCLT.opNorm_le_of_symmetric_power_decay_on_dense
{E : Type*} [NormedAddCommGroup E] [InnerProductSpace ℝ E]
(T : E →L[ℝ] E) (hT : (T : E →ₗ[ℝ] E).IsSymmetric)
(D : Submodule ℝ E) (hD : Dense (D : Set E))
(t : ℝ) (ht : 0 ≤ t)
(hdecay : ∀ z : D, ∃ C : ℝ, 0 ≤ C ∧
∀ n : ℕ, |inner ℝ ((T ^ n) (z : E)) (z : E)| ≤ C * t ^ n) :
‖T‖ ≤ t := by sorrySource
G. O. Roberts and J. S. Rosenthal, Geometric Ergodicity and Hybrid Markov Chains, Electronic Communications in Probability 2 (1997), Theorem 2 and proof, pp. 7-9, https://www.probability.ca/jeff/ftpdir/hybrid.pdf. This packages the dense-subspace-to-spectral-gap step for symmetric operators.