The fixed logarithmic smoothing schedule is negligible
ProvedInfoLossLAR.WorstCase.RegevSketch.smoothingError_negligibleasymptotic-analysislar-regev-bridgelattices
The one fixed function is nonnegative at every natural number and negligible in the published security sense: for every positive real constant , there is a threshold such that for every . Here size is the natural-number binary length. The schedule is fixed independently of any distinguisher, source-basis encoding length, sample polynomial or accuracy exponent. This statement concerns the explicitly defined numerical function only; it does not claim existence of a DGS sampler or a lattice-solving algorithm.
Preamble
import Definitions.Def_InfoLossLAR_RegevSketch_DGS set_option autoImplicit false open QAlgorithms.Complexity
Formal statement
namespace InfoLossLAR.WorstCase.RegevSketch theorem smoothingError_negligible : InfoLossLAR.WorstCase.Negligible smoothingError := by sorry end InfoLossLAR.WorstCase.RegevSketch
Source
Existing locally checked proof: Solutions/RegevBridgeNegligibleSmoothingError.lean (fixedSmoothingError_negligible) and Solutions/RegevBridgeGaussianVerification.lean (platform_negligible_of_native), combined in Solutions/RegevBridgeFixedLogSmoothing.lean (fixedSmoothingError_platform_negligible). Reused under the definitionally identical schedule in the published InfoLossLAR_RegevSketch_DGS definition; no new mathematical assumption.