Nisan--Wigderson overlap hardwiring #
When one NW block is replaced by a challenge, each predecessor block depends on
only the challenge coordinates in their intersection. Its observed Boolean
value is therefore recoverable from a canonical table with exactly
2^|S_i ∩ S_j| entries. Summing these entry counts gives precisely the
predecessor term in the design's overlap cost.
Mapping overlap coordinates through the current block embedding produces exactly the intersection of the two block supports.
Challenge overlap has exactly the design's recorded intersection size.
Inserting a challenge makes the current block read exactly that challenge.
Challenge insertion leaves every coordinate outside the current support unchanged.
The challenged current block is definitionally faithful to the challenge.
A predecessor block depends on the challenge only through the two blocks' intersection coordinates.
Any Boolean observation of a predecessor block has the same overlap-only dependency.
Looking up the actual overlap restriction in the canonical predecessor table recovers the observed predecessor value.
The assignment space indexing one predecessor table has exactly one entry for every Boolean assignment to the intersection.
Total predecessor-table entries equal the sum of the standard exponential intersection costs.
The exact design overlap cost is the predecessor-table cost plus one unit for every later output coordinate.
Predecessor tables fit within the exact overlap cost.
A weak-design overlap budget bounds all predecessor hardwiring tables.