Finite bounds for the explicit sparse parameter choice #
Ceiling division covers the source bits; rounding to a power of three costs at most another factor of three. The logarithmic slack then pays for the extension-degree and entropy factors in the normalized degree-loss bound.
theorem
Algebraic.Cutwidth.Extractor.Internal.explicitCondenserPowerBits_pos
(n k e : ℕ)
{u : ℕ}
(rate : 0 < u)
:
theorem
Algebraic.Cutwidth.Extractor.Internal.explicitCondenserExtensionExponent_pos
(n k e u : ℕ)
:
theorem
Algebraic.Cutwidth.Extractor.Internal.explicitCondenser_entropy_capacity
(n k e : ℕ)
{u : ℕ}
(rate : 0 < u)
:
k ≤ sparsePowerBits u (explicitCondenserBudget n k e) * condenserCoordinates k (sparsePowerBits u (explicitCondenserBudget n k e))
theorem
Algebraic.Cutwidth.Extractor.Internal.explicitCondenser_output_rate
(n k e u : ℕ)
:
u * (condenserCoordinates k (sparsePowerBits u (explicitCondenserBudget n k e)) * sparseFieldBits u (explicitCondenserBudget n k e)) ≤ (u + 1) * k + u * sparseFieldBits u (explicitCondenserBudget n k e)