OAI/Combinatorics/LowSpaceSubsetSum/Counting/DeficiencyVariance.lean
7 declarations, 0 contain sorry, 0 declared as axioms
Deficiency under sampling without replacement
The variance estimate in the paper's deficiency lemma is a specialization
of the finite completion-coupling theorem
Balanced.WithoutReplacement.finiteVariance_powersetCard_le.
Its probability space is exactly the uniform law on all size-v subsets
of the supplied domain U, including the cases v = 0 and v = |U|.
- def
OAI.LowSpaceSubsetSum.meanDeficiencyline 24Source documentation
Mean deficiency of a uniform size-
vsubset ofU.noncomputable def meanDeficiency (a : ι → ℤ) (U : Finset ι) (v : ℕ) : ℝ := …
- def
OAI.LowSpaceSubsetSum.deficiencyVarianceline 28Source documentation
Variance for uniform sampling of exactly
vindices, without replacement.noncomputable def deficiencyVariance (a : ι → ℤ) (U : Finset ι) (v : ℕ) : ℝ := …
- theorem
OAI.LowSpaceSubsetSum.deficiency_exchange_boundline 31 - theorem
OAI.LowSpaceSubsetSum.deficiencyVariance_leline 39Source documentation
The paper's exact variance bound
Var(d(V)) ≤ v.theorem deficiencyVariance_le (a : ι → ℤ) (U : Finset ι) (v : ℕ) (hv : v ≤ U.card) : deficiencyVariance a U v ≤ v := by …Mentions: OAI.LowSpaceSubsetSum.deficiencyVariance
- theorem
OAI.LowSpaceSubsetSum.deficiency_centered_secondMoment_leline 47 - theorem
OAI.LowSpaceSubsetSum.deficiency_tail_fraction_leline 54Source documentation
A direct finite Chebyshev bound for deviations from the mean deficiency. The left side is the exact probability under uniform size-
vsampling.theorem deficiency_tail_fraction_le (a : ι → ℤ) (U : Finset ι) (v : ℕ) (hv : v ≤ U.card) {t : ℝ} (ht : 0 < t) : (((U.powersetCard v).filter (fun V => t ≤ |deficiency a V - meanDeficiency a U v|)).card : ℝ) / (U.powersetCard v).card ≤ (v : ℝ) / t ^ 2 := by …Mentions: OAI.LowSpaceSubsetSum.deficiency, OAI.LowSpaceSubsetSum.meanDeficiency
- theorem
OAI.LowSpaceSubsetSum.deficiency_uniform_variance_leline 89Source documentation
Measure-theoretic form for the actual uniform fixed-cardinality law.
theorem deficiency_uniform_variance_le [MeasurableSpace (Finset ι)] [MeasurableSingletonClass (Finset ι)] (a : ι → ℤ) (U : Finset ι) (v : ℕ) (hv : v ≤ U.card) : ProbabilityTheory.variance (deficiency a) (PMF.uniformOfFinset (U.powersetCard v) (Finset.powersetCard_nonempty.mpr hv)).toMeasure ≤ v := by …Mentions: OAI.LowSpaceSubsetSum.deficiency