These are names for the three groups of truth-table rows:
- ON-set: rows where F must be 1. This is the Σm list.
- OFF-set: rows where F must be 0. This is the ΠM list.
- DC-set: don't-care rows, listed in d(…).
For a completely specified function the DC-set is empty, and the ON-set and OFF-set are complements: together they cover all 2ⁿ rows with no overlap. That is why converting between Σm and ΠM is just "take the other rows"; see canonical form conversion.
With don't-cares, the rows split three ways. The sizes always add up to 2ⁿ.
The sets give a clean way to state what a correct implementation must do: be 1 on the whole ON-set, 0 on the whole OFF-set, and anything on the DC-set. An implicant is a product term that never touches the OFF-set.
The complement F′ simply swaps the ON-set and the OFF-set.