Leaf language is a method in computational complexity theory for characterizing a complexity class by formalizing what it means for a machine to "accept" an input.[1]
Complexity classes are typically defined in terms of a polynomial-time nondeterministic Turing machine (NTM). These machines possess multiple computational paths whose outcomes determine whether an input is accepted or rejected. Traditional NTMs accept the input if at least one path accepts it, and only rejects it if all paths reject it. Conversely, a co-NTM accepts the input only if all paths accept it and rejects it if any path rejects it. Similarly, different and more sophisticated notions of acceptance can be defined.
The characterization of a complexity class can be formalized by examining the formal language associated with each acceptance condition. This involves assuming an ordered tree, and reading the accept/rejects strings from the leaves of the computation tree. NTMs will accept if the leaf string is in the language, and will reject if the leaf string is in the language . [2]