Covering zeros is the test step. If all zeros can be covered by fewer than n rows and columns, the current zeros do not yet contain a complete assignment. The diagram marks the cover on headers so the lines are visible even when the covered body cells are zeros, making the certificate readable.

highlighted = computed this step

Fewest cover lines

The fewest cover uses row W2 and column J2. Why: the covered row and column headers show which lines cover all current zeros. This is a structural test, not a cost calculation: it asks whether the zero pattern is rich enough to contain a complete assignment.

lines=2\text{lines}=2
cover zerosJ1J2J3W1102W2000W3101

Not enough lines

There are 2 lines, which is less than n=3. Why: fewer than n lines means these zeros do not yet contain a full assignment. If a perfect zero assignment existed, covering its independent zeros would require a full set of assignment positions, so the short cover is a certificate that adjustment is still needed.

2<32<3
cover zerosJ1J2J3W1102W2000W3101

Minimum uncovered

The minimum uncovered cells have value 1. Why: this is the amount used to create more zeros. It is the smallest safe push: subtracting more would skip over information, while subtracting this amount creates at least one new opportunity without changing the optimal assignment.

δ=1\delta=1
cover zerosJ1J2J3W1102W2000W3101

Diagram note

Legend: zero body cells, minimum-uncovered cells, and covering-line headers are marked by role. The cover count is the certificate of whether the current zeros are sufficient; here it says the matrix must be adjusted before selecting an assignment. Pixel positions are rounded for layout; every number shown is exact.

cover lines decide whether to adjust\text{cover lines decide whether to adjust}
cover zerosJ1J2J3W1102W2000W3101