AFTD Auto-Formalizing Theoretical Domains

random_switching_channel

def

Given channels W_G and W_H and switching parameter p ∈ (0, 1), the random switching channel has input pairs (a, b) and tagged outputs in Y ⊕ Z with transition probabilities p * W_G(y|a) and (1-p) * W_H(z|b).

Statement

def random_switching_channel {A B Y Z : Type*} [Fintype Y] [Fintype Z]
    (W_G : DiscreteMemorylessChannel A Y) (W_H : DiscreteMemorylessChannel B Z)
    (p : ℝ) (hp0 : 0 < p) (hp1 : p < 1) :
    DiscreteMemorylessChannel (A × B) (Sum Y Z)
Defined in
arXiv:2610.41977, Section 5
Verified
07 Nov 2026
Axioms
none listed
Built on 1
  • DiscreteMemorylessChannelA discrete memoryless channel from input alphabet A to finite output alphabet Y is defined by non-negative transition probabilities…
Used by 2

Read back from the Lean

Given types $A, B$, finite types $Y, Z$, discrete memoryless channels $W_G$ (from $A$ to $Y$) and $W_H$ (from $B$ to $Z$), a real number $p$, and proofs that $0 < p$ and $p < 1$, random_switching_channel defines a DiscreteMemorylessChannel with input type $A \times B$ and output type $Y \oplus Z$ (the sum type Sum Y Z). Its transition probability function sends an input pair $(a, b) \in A \times B$ to: - $p \cdot W_G.\mathrm{prob}(a, y)$ for output $\mathrm{inl}(y)$, where $y \in Y$ - $(1 - p) \cdot W_H.\mathrm{prob}(b, z)$ for output $\mathrm{inr}(z)$, where $z \in Z$. The definition includes the proofs that these probabilities are non-negative and sum to $1$ over all outputs in $Y \oplus Z$ for every input pair $(a, b)$.

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

Lean source view module on GitHub

/-- The random switching channel between two discrete memoryless channels with switching parameter p in (0, 1). -/
def random_switching_channel {A B Y Z : Type*} [Fintype Y] [Fintype Z]
    (W_G : DiscreteMemorylessChannel A Y) (W_H : DiscreteMemorylessChannel B Z)
    (p : ℝ) (hp0 : 0 < p) (hp1 : p < 1) :
    DiscreteMemorylessChannel (A × B) (Sum Y Z) where
  prob := fun ab => fun
    | Sum.inl y => p * W_G.prob ab.1 y
    | Sum.inr z => (1 - p) * W_H.prob ab.2 z
  prob_nonneg := by
    rintro ⟨a, b⟩ (y | z)
    · exact mul_nonneg hp0.le (W_G.prob_nonneg a y)
    · exact mul_nonneg (sub_nonneg.mpr hp1.le) (W_H.prob_nonneg b z)
  prob_sum := by
    rintro ⟨a, b⟩
    rw [Fintype.sum_sum_type]
    simp_rw [← Finset.mul_sum]
    rw [W_G.prob_sum a, W_H.prob_sum b, mul_one, mul_one]
    linarith
See it in the graphDiscuss this resultChallenge it