The iterated-clock schedule in the conditional MinKT reduction #
This definitions layer records the exact clock schedule in Proposition 6.2 of
Hirahara's Symmetry of Information from Meta-Complexity. For
t' = max {t, |x| + |y|}, the two ordinary estimator queries use source clocks
p(t') and p^3(t'), symmetry of information is invoked at p^2(t'), and the
final conditional gap clock is p^4(t + |x| + |y|).
The logarithmic losses are not hidden in asymptotic notation. LossBudget
states exactly what the correction must absorb and what must fit inside the
final logarithmic slack.
Repeated application of a primitive clock. clockIterate clock k time is
the paper's p^(k)(t).
Equations
- One or more equations did not get rendered due to their size.
- Complexity.GapMINCKT.DifferenceEstimator.Unconditional.Iterated.clockIterate clock 0 x✝ = x✝
Instances For
The SoI base input t' = max {t, |x| + |y|}.
Equations
Instances For
The ordinary gap estimator advances its source clock by one application
of clock; the output length is deliberately irrelevant.
Equations
- Complexity.GapMINCKT.DifferenceEstimator.Unconditional.Iterated.ordinaryParameters clock = { clock := fun (_outputLength time : ℕ) => clock time }
Instances For
The final conditional clock p^4(t + |x| + |y|).
Equations
- One or more equations did not get rendered due to their size.
Instances For
The logarithmic SoI loss log_2(p(t)) + additive.
Equations
- Complexity.GapMINCKT.DifferenceEstimator.Unconditional.Iterated.logarithmicSoILoss clock additive time = Nat.log 2 (clock time) + additive
Instances For
The exact two-query plan, with an explicit composition loss and centering correction.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The paper's half-slack centering correction. Its adequacy is deliberately
kept as a separate finite inequality in LossBudget.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Exact finite loss inequalities for the iterated-clock construction.
The first field pays for the lower guarantee on the condition estimator and the ordinary pair compiler. The second field fits the paired-estimator loss, the SoI loss, and the centering correction inside the final gap slack.
- upper (inst : MINCKT.Instance) : Nat.log 2 (clockIterate clock 4 (paddedTime inst)) + pairUpperLoss inst ≤ correction inst
Losses that must be removed to obtain the source-clock upper estimate.
- lower (inst : MINCKT.Instance) : Nat.log 2 (clockIterate clock 2 (paddedTime inst)) + soiLoss (clockIterate clock 2 (paddedTime inst)) + correction inst ≤ (conditionalParameters clock).logarithmicSlack inst
Losses that must fit inside the final transformed-clock error.
Instances For
Monotonicity and widening are the exact order properties used to align the four iterated clocks. Polynomial growth is established separately.
- monotone : Monotone clock
More source time cannot reduce the clock allowance.
One clock application never gives less time.
Instances For
A regular primitive clock with a uniform polynomial upper bound.
- polynomiallyBounded : ∃ (coefficient : ℕ) (exponent : ℕ), ∀ (time : ℕ), clock time ≤ coefficient * (time + 1) ^ exponent
One polynomial controls the primitive clock at every input.