Deterministic strategies embed conservatively into behaviour strategies
ProvedAJGames.ofStrategy_playProbLet be a strategy on a game , read as a behaviour strategy. For every position of , the probability of under it is
So behaviour strategies extend deterministic ones without changing them.
import Mathlib import Definitions.Def_AJGames_Behaviour
namespace AJGames
open scoped ENNReal
universe u
open Classical in
/-- **Behaviour strategies extend deterministic ones conservatively.** Read as a behaviour strategy,
a deterministic strategy `σ` gives every position probability `1` if it lies in `σ` and `0`
otherwise. AJ axioms used: G2 (prefix closure) and G3 (alternation, to see that Player is to move
before a Player move); the strategy axioms first-O, determinism and receptiveness. No tensor axiom. -/
theorem ofStrategy_playProb {A : Game.{u}} (σ : Strategy A) (s : List A.M) (hs : s ∈ A.pos) :
playProb (ofStrategy σ) s = if s ∈ σ.carrier then 1 else 0 := by sorry
end AJGamesRead-back
What the Lean code literally says, in plain math · claude-opus-5-5
Games. Players are and . A game has a move type , a labelling , and a set of finite move sequences that contains the empty sequence, is prefix-closed, and whose members alternate (consecutive moves have different labels).
Strategies. A (deterministic) strategy on is a set of positions such that: (empty sequence); is prefix-closed; every nonempty member of begins with a move labelled ; determinism: if and with then ; receptivity: if , and , then .
Behaviour strategies. A behaviour strategy on assigns to every finite sequence (legal or not) a probability distribution on ( = "no move"), subject to: if then is nonempty with last move labelled , , and .
The play probability of under is the extended nonnegative real
i.e. the product, over the -labelled moves of , of the probability gives to that move after the preceding prefix; -labelled moves contribute the factor , and the empty product (in particular ) is .
The induced behaviour strategy. For a strategy , is defined by: if there is some with and , then is the point mass at one such , chosen by the axiom of choice; otherwise is the point mass at .
Statement. For every game , every strategy on , and every finite sequence of moves with ,
The hypothesis is required; nothing is asserted for illegal sequences. The case is included (then and the value is ). The membership test "" is decided classically.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.