Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 1, case (ii), pp. 7–8 — ı̂ = i∗ ∉ M: i∗ is hired with probability ≥ 1/(m+1) + τ Σ (m k)(−1)^(k−1)(1−τ^k)/k

Proved
SecretaryPred.LearnedDynkin.case_ii

by mikedeng1 · Oct 24, 2026 · Mathlib 0df444a (Lean v4.33.1)

competitive-ratiolearning-augmented-algorithmsoptimal-stoppingp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-strengthenedp2o-v1secretary-problem

Let nnn candidates have values v(i)≥0v(i)\ge 0v(i)≥0 and predictions v^(i)≥0\hat v(i)\ge 0v^(i)≥0, let ı^\hat\imath^ be any maximizer of v^\hat vv^, and let i∗i^*i∗ be the best candidate (largest value, ties broken by the smaller index). Run Algorithm 1 (learned Dynkin) with threshold θ≥0\theta\ge 0θ≥0 and time τ∈(0,1)\tau\in(0,1)τ∈(0,1) on independent Uniform[0,1][0,1][0,1] arrival times. MMM is the set of candidates with ∣1−v^(i)/v(i)∣>θ|1-\hat v(i)/v(i)|>\theta∣1−v^(i)/v(i)∣>θ and m=∣M∣m=|M|m=∣M∣.

Case (ii). Suppose M≠∅M\neq\emptysetM=∅, ı^=i∗\hat\imath=i^*^=i∗ and i∗∉Mi^*\notin Mi∗∈/M. Then

Pr⁡[the algorithm hires i∗]  ≥  1m+1+τ∑k=1m(mk)(−1)k−11−τkk.\Pr[\text{the algorithm hires } i^*]\;\ge\;\frac{1}{m+1}+\tau\sum_{k=1}^{m}\binom{m}{k}(-1)^{k-1}\frac{1-\tau^k}{k}.Pr[the algorithm hires i∗]≥m+11​+τk=1∑m​(km​)(−1)k−1k1−τk​.

The first term is the probability that ı^\hat\imath^ precedes every candidate of MMM (and is hired in Prediction mode); the second is the closed form of ∫τ1{1−(1−t∗)m} τ/t∗ dt∗\int_\tau^1\{1-(1-t^*)^m\}\,\tau/t^*\,dt^*∫τ1​{1−(1−t∗)m}τ/t∗dt∗, the Secretary-mode contribution.

Formalization Note Candidates are Fin n (0-based). The probability that i∗i^*i∗ is hired is (arrivalLaw n).real {t | Hired θ τ v v̂ ı̂ t i∗}, the measure of the set of arrival vectors on which Algorithm 1 hires i∗i^*i∗, under the uniform law on [0,1]n[0,1]^n[0,1]n. The case hypotheses are stated with the division-free deviation predicate Deviates, i∗i^*i∗ is the top candidate of the tie-broken order Beats, and ı^\hat\imath^ is any maximizer of v^\hat vv^.

Differences from the source:

  • generalization: θ = 0.646, τ = 0.313 (Theorem 1) → any θ ≥ 0 and τ with 0 < τ < 1; reason: the case derivations never use the values of θ and τ (the paper evaluates the six case bounds on a grid of (θ, τ), p. 10); at θ = 0.646, τ = 0.313 the item is the source's step
  • reading: line 8 of Algorithm 1: 'i is the best so far', with ties not addressed; i∗ ∈ argmax v → i beats every earlier arrival in the strict order 'larger value, then smaller index' (Beats); istar is the top candidate of that order; reason: both untied readings make Theorem 1 false: strictly larger value fails for n ≥ 5 candidates all with v = 1 and exact predictions except v̂(ı̂) = 10 (P(hire) ≤ 1/n < 0.215); weakly larger value fails for i∗ with v = v̂ = 1, ı̂ with v = δ, v̂ = 10 and n − 2 fillers with v = v̂ = δ (ratio → δ as n → ∞); for distinct values the order is the source's
  • encoding: |1 − v̂(i)/v(i)| > θ (line 4 and the set M, p. 6) → θ·v(i) < |v(i) − v̂(i)| (Deviates); reason: identical when v(i) > 0; avoids Lean's x/0 = 0, and for v(i) = 0 a candidate deviates iff v̂(i) ≠ 0
  • encoding: candidates N = [n] interviewed in random order, arrival times iid Uniform[0, 1] (§2.1), exactly one hire → candidates Fin n (0-based), t uniform on the cube [0, 1]^n (arrivalLaw n), the mode at i is Secretary iff some deviating j has t_j ≤ t_i, and algValue sums the values of all candidates that are first to stop; reason: the run differs from the source's single-hire run only on the null set of arrival vectors with two equal times
Preamble
import Mathlib
import Definitions.Def_SecretaryPred_LearnedDynkin_Setting
Formal statement
namespace SecretaryPred.LearnedDynkin

theorem case_ii (n : ℕ) (theta tau : ℝ) (v vhat : Fin n → ℝ) (ihat istar : Fin n)
    (hv : ∀ i, 0 ≤ v i) (hvhat : ∀ i, 0 ≤ vhat i) (hihat : ∀ j, vhat j ≤ vhat ihat)
    (histar : ∀ j, j ≠ istar → Beats v istar j)
    (htheta : 0 ≤ theta) (htau0 : 0 < tau) (htau1 : tau < 1)
    (hcase : ihat = istar) (hstarM : ¬ Deviates theta v vhat istar)
    (hMne : (Mset theta v vhat).Nonempty) (m : ℕ) (hm : (Mset theta v vhat).card = m) :
    1 / ((m : ℝ) + 1) +
        tau * ∑ k ∈ Finset.Icc 1 m,
          (m.choose k : ℝ) * (-1 : ℝ) ^ (k - 1) * (1 - tau ^ k) / (k : ℝ)
      ≤ (arrivalLaw n).real {t | Hired theta tau v vhat ihat t istar} := by sorry

end SecretaryPred.LearnedDynkin
Source
Fujii, Yoshida, The Secretary Problem with Predictions, arXiv:2306.08340v1, pp. 7–8, proof of Theorem 1, case (ii)
Human review
  • Endorsed by Shuze Chen · Oct 24, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 24, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me