Appendix LEMMA: the projections of and onto agree for
ProvedIncentivesInTeams.Conglomerate.appendix_lemmaFix a conglomerate model, a joint strategy , a subunit and a strategy . Write for the head's information when is played and for his information when is played. For a state define (A.3)
and let , be their projections onto the component state space . If is nonempty, then
The lemma says that, for every component other than the deviating subunit, the head's information under the deviation and under restricts that component's state in the same way. It is the device by which the paper compares the two sides of (A.2).
Formalization Note. The hypothesis that is nonempty is added. The paper states the lemma for all , but its proof reads off the single-coordinate condition (A.4), which presumes . Without it the statement is false: if sends a message that never sends, while . The hypothesis is kept as in the paper's setting. The one-exchange protocol of the model file is used for the information functions.
import Mathlib import Definitions.Def_IncentivesInTeams_Conglomerate_Model
namespace IncentivesInTeams.Conglomerate
theorem appendix_lemma {ι : Type*} [Fintype ι] [DecidableEq ι] {S₀ : Type*} [Fintype S₀] {S : ι → Type*}
[∀ i, Fintype (S i)] {Z₀ : Type*} {Z M₀ M : ι → Type*} {D₀ : Type*} {D : ι → Type*}
(T : Model S₀ S Z₀ Z M₀ M D₀ D)
(βs : JointStrategy S₀ S Z₀ Z M₀ M D₀ D) (i : ι)
(b : SubStrategy (S i) (Z i) (M₀ i) (M i) (D i)) (hb : b ∈ T.B i) (s : S₀ × ∀ k, S k)
(hB : {s' | βs.headInfo s' = (βs.update i b).headInfo s}.Nonempty) :
Prod.fst '' {s' | (βs.update i b).headInfo s' = (βs.update i b).headInfo s} =
Prod.fst '' {s' | βs.headInfo s' = (βs.update i b).headInfo s} ∧
∀ j, j ≠ i →
(fun s' : S₀ × ∀ k, S k => s'.2 j) ''
{s' | (βs.update i b).headInfo s' = (βs.update i b).headInfo s} =
(fun s' : S₀ × ∀ k, S k => s'.2 j) ''
{s' | βs.headInfo s' = (βs.update i b).headInfo s} := by sorry
end IncentivesInTeams.Conglomerate
Read-back
What the Lean code literally says, in plain math · claude-opus-5-5
Setting. is a finite set of subunits with decidable equality. and each are finite state sets, and are arbitrary types. A state is .
A joint strategy consists of:
- head functions , and ;
- for each , subunit functions and , plus a decision rule .
The head's information under is
is with subunit 's strategy replaced by ; the head and the other subunits are unchanged.
Theorem. Take:
- a model , from which only the strategy set is used here;
- a joint strategy ;
- a subunit ;
- a subunit- strategy ;
- a state .
Set and define
Assume is nonempty. Then:
- The sets of head-component values coincide:
- For every subunit , the sets of -component values coincide:
Nothing is asserted about the -th component. No assumption is made on weights or payoffs, and is not required to lie in any strategy set.
Degenerate cases. always contains . If or some is empty, there is no state and the statement is vacuous. If , part 2 is vacuous and only part 1 is asserted. The hypothesis plays no role in the conclusion beyond restricting .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.