Weakly Decisive Coalition
A coalition $C \subseteq N$ is weakly decisive for the ordered pair $(a, b)$ under a social welfare function $F$ ([[social_choice.voting.swf]]) if unanimous $a \succ b$ inside $C$, combined with unanimous $b \succ a$ outside $C$, forces society to rank $a \succ b$: $$ \forall P,\ \bigl(\forall i \in C,\ a \succ_{P_i} b\bigr) \wedge \bigl(\forall i \notin C,\ b \succ_{P_i} a\bigr) \Rightarrow a \succ_{F(P)} b. $$
In Lean: SocialChoice.Voting.IsWeaklyDecisiveFor F C a b.
Weak decisiveness is implied by decisiveness ([[social_choice.voting.is_decisive_for]])
— a witness profile satisfying the stronger constraint of total opposition
is in particular a witness for plain decisiveness:
SocialChoice.Voting.isWeaklyDecisiveFor_of_isDecisiveFor.
The opposite-direction-strengthening — turning weak decisiveness into full decisiveness — is the substantive content of the field expansion lemma [[social_choice.voting.decisive_spread]]. The split into "weak" vs full decisiveness lets the Arrow proof use only the simpler hypothesis when building witnesses.
References
- [MSZ 21.14ff] Maschler, Solan, and Zamir, Game Theory. Weak decisiveness in the field-expansion argument.