Quarter bias of conjunction messages #
A conjunction summary that tests two distinct selected primary inputs is true on at most one quarter of the selected assignments. The proof overwrites the two forced coordinates to inject four copies of the accepting assignments into all assignments; repetitions and contradictory literals need no separate assumptions.
Fixing two distinct Boolean coordinates leaves at most a quarter of all assignments.
Two selected literal slots force the corresponding assignment coordinates.
A selected conjunction with two distinct primary variables has quarter bias.
Equations
Multiple-primary conjunctions supply two distinct designated primary variables.
A selected pair belongs to an actual multiple-primary conjunction.
Selecting every input makes the pair condition exactly the geometric gate predicate.
The actual line message has quarter bias whenever the selected-pair witness holds.
The extra output-message coordinate does not change a gate's quarter bias.
The recursive geometric gate count equals the number of qualifying actual lines.