AFTD Auto-Formalizing Theoretical Domains

two_box_risky_value_mean

theoremverified

The mean of the risky distribution is 1.

Statement

theorem two_box_risky_value_mean (M : ℕ) (hM : 1 < M) :
    ∑ v ∈ (two_box_risky_value_dist M hM).support, (two_box_risky_value_dist M hM).prob v * v = 1
Source
step towards two_box_counterexample_family
Verified
24 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 2
  • two_box_risky_value_distThe distribution taking the value M with probability 1/M and the value 0 with probability 1 - 1/M.
  • DiscreteDistA discrete probability distribution with finite support on non-negative real values, defined by a finite support set, non-negative…
Used by 3
  • two_box_dtd_le_oneUnder deferred-to-deadline commitment, every valid policy on the two-box instance gets expected utility at most 1.
  • two_box_dtd_after_box_zeroWith the history [(0, 1)], any valid policy has deferred-to-deadline expected utility at most 1.
  • two_box_counterexample_familyFor any integer M > 1, there exists a 2-box Pandora deadline instance such that 2 - 1 / M ≤ optDUS inst and every valid policy π…

The proof, explained

The expected value is computed by directly evaluating the sum over the support of the distribution. Because $1 < M$, the support consists of the two distinct real values $0$ and $M$. The value $0$ contributes nothing to the expectation, while the value $M$ occurs with probability $1/M$, yielding an expected value of $(1/M) \cdot M = 1$.

  1. In the real numbers, $0 \neq M$.

    From the hypothesis $1 < M$ for $M \in \mathbb{N}$, casting to $\mathbb{R}$ gives $1 < M$. Linear arithmetic then implies $0 \neq M$ because $M > 1 > 0$.

    lines 4–6

      have hM0 : (0 : ℝ) ≠ (M : ℝ) := by
        have : (1 : ℝ) < (M : ℝ) := by exact_mod_cast hM
        linarith
  2. The expected value expands to the sum of the summand evaluated at $0$ and at $M$.

    Unfolding the definition `two_box_risky_value_dist` shows that the support is the two-element set $\{0, M\}$. Since $0 \neq M$, `Finset.sum_pair` expands the sum over $\{0, M\}$ into the sum of the value at $0$ and the value at $M$.

    lines 7–8

    uses two_box_risky_value_dist, Finset.sum_pair

      simp only [two_box_risky_value_dist]
      rw [Finset.sum_pair hM0]
  3. The sum simplifies to $M^{-1} \cdot M = 1$.

    At $v = 0$, the summand is multiplied by $0$, which vanishes by `mul_zero`. At $v = M$, the condition $M = 0$ is false because $M \neq 0$ (`if_neg` using `Ne.symm`), so the probability is $M^{-1}$. Dropping the zero term with `zero_add` leaves $(M^{-1}) \cdot M = 1$.

    line 9

    uses if_true, if_neg, Ne.symm, mul_zero, zero_add

      simp only [if_true, if_neg (Ne.symm hM0), mul_zero, zero_add]
  4. The expected value equals $1$.

    By symmetry of inequality (`Ne.symm`), $M \neq 0$. With $M$ non-zero, algebraic simplification (`field_simp`) cancels $M^{-1} \cdot M$ to $1$, concluding the proof.

    lines 10–11

    uses Ne.symm

      have hMpos : (M : ℝ) ≠ 0 := Ne.symm hM0
      field_simp
two_box_risky_value_dist
The discrete probability distribution with support $\{0, M\}$, assigning probability $1 - 1/M$ to $0$ and $1/M$ to $M$.
Finset.sum_pair
For distinct elements $a$ and $b$, the sum of a function over the two-element set $\{a, b\}$ equals $f(a) + f(b)$.
if_true
A conditional expression whose condition is True evaluates to the 'then' branch.
if_neg
A conditional expression whose condition is false evaluates to the 'else' branch.
Ne.symm
Inequality is symmetric: if $a \neq b$, then $b \neq a$.
mul_zero
Multiplying any number by zero yields zero ($a \cdot 0 = 0$).
zero_add
Zero is an additive identity on the left ($0 + a = a$).

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

/-- The expected value of box 1's value distribution is exactly `1`. -/theorem two_box_risky_value_mean (M : ℕ) (hM : 1 < M) :    ∑ v ∈ (two_box_risky_value_dist M hM).support, (two_box_risky_value_dist M hM).prob v * v = 1 := by  have hM0 : (0 : ℝ) ≠ (M : ℝ) := by    have : (1 : ℝ) < (M : ℝ) := by exact_mod_cast hM    linarith  simp only [two_box_risky_value_dist]  rw [Finset.sum_pair hM0]  simp only [if_true, if_neg (Ne.symm hM0), mul_zero, zero_add]  have hMpos : (M : ℝ) ≠ 0 := Ne.symm hM0  field_simp
See it in the graphDiscuss this resultChallenge it