bernoulli_energy_moment_from_count_moment_and_pointwise_domination
ProvedRole. It controls sampled row/column counts or energies, which feed the moment bounds for random sampled matrices.
Problem and notation. Exact matrix completion asks when an unknown low-rank real matrix can be recovered from a random subset of its entries. Here has rank , entries are observed, and . Recovery means nuclear-norm minimization: minimize among matrices agreeing with on the observed entries. Probability notation. is the fixed-cardinality success probability: is chosen uniformly among all subsets of entries with , and the event is that the convex program uniquely returns . In Bernoulli nodes, or means each entry is sampled independently with probability , usually . Coherence notation. The object records SVD/singular-vector data for . The hypotheses and are the Candes-Recht incoherence assumptions: measures how spread out the singular vector spaces are, and measures the largest entry of the sign matrix . The parameter controls polynomial failure probabilities such as . For sampled row/column nodes, counts observed entries in row , counts observed entries in column , and the corresponding energies sum over sampled entries. These estimates feed the noncommutative Khintchine and spectral-norm concentration bounds.
Claim. Generic moment comparison for sampled energies. If a nonnegative energy statistic is pointwise dominated by times a nonnegative count statistic, then a moment bound for the count statistic implies the corresponding energy moment bound.
Lecture-note formulation:
The constants in this node are universal existential constants; the theorem asserts that some positive constants with these roles exist.
Decomposition status. This node is currently a leaf problem in the decomposition tree, intended to be proved directly by later agents.
import Definitions.Def_matrix_completion_sampled_counts open MatrixCompletion
theorem bernoulli_energy_moment_from_count_moment_and_pointwise_domination
(Ccount : ℝ) :
0 < Ccount →
∃ Cenergy : ℝ, 0 < Cenergy ∧
∀ (β : ℝ), 2 < β →
∀ (n₁ n₂ m q : ℕ) (X : Matrix (Fin n₁) (Fin n₂) ℝ),
0 < n₁ → 0 < n₂ → m ≤ n₁ * n₂ →
1 ≤ q →
(q : ℝ) ≥ β * Real.log (↑(max n₁ n₂)) →
(q : ℝ) ≤
((m : ℝ) / ((n₁ : ℝ) * (n₂ : ℝ))) * (↑(max n₁ n₂)) →
∀ (Energy Count : Finset (Fin n₁ × Fin n₂) → ℝ),
(∀ Omega : Finset (Fin n₁ × Fin n₂), 0 ≤ Energy Omega) →
(∀ Omega : Finset (Fin n₁ × Fin n₂), 0 ≤ Count Omega) →
(∀ Omega : Finset (Fin n₁ × Fin n₂),
Energy Omega ≤ entrySupNorm X ^ 2 * Count Omega) →
bernoulliExpectation ((m : ℝ) / ((n₁ : ℝ) * (n₂ : ℝ)))
(fun Omega : Finset (Fin n₁ × Fin n₂) => Count Omega ^ q) ≤
(Ccount * ((m : ℝ) / ((n₁ : ℝ) * (n₂ : ℝ))) *
(↑(max n₁ n₂))) ^ q →
bernoulliExpectation ((m : ℝ) / ((n₁ : ℝ) * (n₂ : ℝ)))
(fun Omega : Finset (Fin n₁ × Fin n₂) => Energy Omega ^ q) ≤
(Cenergy * ((m : ℝ) / ((n₁ : ℝ) * (n₂ : ℝ))) *
(↑(max n₁ n₂)) * entrySupNorm X ^ 2) ^ q := by
sorry