The natural ceiling is bounded by the stored numerator
ProvedAFT.BinaryArithmetic.RationalBits.ceilNatural_le_numeratorinfo-loss-lar-native-5c3becbnative-binary-arithmeticstoc-2027
Let a signed rational encoding represent , where is the value of its stored numerator and is the value of its denominator. With the natural ceiling truncated at zero,
The bound controls the size of a unary ceiling budget directly from the supplied rational representation.
Preamble
import Definitions.Def_InfoLossLAR_Native_AFT_BinaryRationalControl import Mathlib /-! # Actual binary clipping and natural ceiling The ceiling uses binary long division, high-zero trimming, and one ripple increment when the remainder is nonzero. Tests inspect the stored sign or list constructor, never the decoded rational value. -/ namespace AFT.BinaryArithmetic end AFT.BinaryArithmetic section AFT section BinaryArithmetic open AFT open AFT.BinaryArithmetic namespace AFT.BinaryArithmetic.RationalBits end AFT.BinaryArithmetic.RationalBits section RationalBits open AFT open AFT.BinaryArithmetic open AFT.BinaryArithmetic.RationalBits
Formal statement
theorem AFT.BinaryArithmetic.RationalBits.ceilNatural_le_numerator (input : RationalBits) :
⌈input.decode⌉₊≤ value input.numerator := by sorrySource
Local manuscript repository commit 5c3becb254daaa8738368b474349766c1769c2a0; lean/AFT/BinaryRationalControl.lean, lines 96–108; exact native theorem type and complete proof.