empiricalStrategy
defimported Lean
Given $T > 0$ pure actions $a_0, \dots, a_{T-1} \in \mathrm{Fin}\,n$, the empirical strategy is the mixed strategy assigning to each $i$ the fraction of rounds on which action $i$ was played: \[ q_i \;=\; \frac{1}{T}\#\{t : a_t = i\}. \]
Statement
set_option relaxedAutoImplicit false in set_option autoImplicit false in open Real Finset BigOperators in noncomputable def empiricalStrategy {n T : ℕ} [NeZero n] (hT : 0 < T) (actions : Fin T → Fin n) : MixedStrategy n
- Defined in
- TCSlib, `empiricalStrategy`
- Lean proof
- By Karim Abdel Sadek, Mark Bedaywi, from github.com/Shilun-Allan-Li/tcslib at
c8a02591b1e5(Copyright (c) 2026 Karim Abdel Sadek and Mark Bedaywi. All rights reserved, Apache-2.0). Compiled here against leanprover/lean4:v4.33.1, Mathlib v4.33.1; written for leanprover/lean4:v4.25.0.empiricalStrategy← empiricalStrategy (TCSlib/LearningTheory/Minimax/HedgeInteraction.lean:109): verbatim
- Verified
- 07 Oct 2026
- Axioms
- none listed
- Built on 1
- MixedStrategyA mixed strategy over $n$ actions is a record consisting of a weight function $w : \mathrm{Fin}\,n \to \mathbb{R}$ together with…
- Used by 2
- hedge_constructionHedge yields an approximate minimax pair. Let $G$ be a finite two-player zero-sum game with $M \ge 2$ row actions and $N \ge 1$ column…
- pureVsPayoff_empiricalStrategyPure-row payoff against the empirical column strategy. Let $G$ be a finite two-player zero-sum game with $M$ row actions and $N \ge 1$…
Lean source view module on GitHub
set_option relaxedAutoImplicit false in set_option autoImplicit false in open Real Finset BigOperators in /-- The empirical distribution of a sequence of `T > 0` pure actions, as a mixed strategy: the weight of action `i` is the fraction of rounds in which `i` was played. This is the column player's final mixed strategy `Q̄` in the Hedge-vs-best-response interaction. [FS99, §5]. -/ noncomputable def empiricalStrategy {n T : ℕ} [NeZero n] (hT : 0 < T) (actions : Fin T → Fin n) : MixedStrategy n where weights i := (∑ t : Fin T, if actions t = i then (1 : ℝ) else 0) / T nonneg i := by apply div_nonneg · exact Finset.sum_nonneg fun t _ => by split <;> positivity · exact Nat.cast_nonneg T sum_one := by rw [← Finset.sum_div, Finset.sum_comm] have hinner : ∀ t : Fin T, ∑ i : Fin n, (if actions t = i then (1 : ℝ) else 0) = 1 := by intro t rw [Finset.sum_eq_single (actions t)] · simp · intro b _ hb simp [hb.symm] · intro h exact False.elim (h (Finset.mem_univ _)) simp_rw [hinner, Finset.sum_const, Finset.card_fin, nsmul_eq_mul, mul_one] exact div_self (Nat.cast_ne_zero.mpr (Nat.pos_iff_ne_zero.mp hT))