In complexity theory, ZPL (Zero-error Probabilistic Logarithmic space) is the set of problems solvable by a probabilistic Turing machine which always yields the correct answer and uses logarithmic space on average. Probabilistic algorithms that always give the correct answer are called Las Vegas algorithms.
Unlike its deterministic counterpart L, a ZPL machine can potentially use exponential time by exploiting randomness. If ZPL is restricted to polynomial time, we get the more interesting class ZPLP.
A surprising result is that ZPL is equal to both RL and NL; thus, if a problem can be solved in logarithmic space with nondeterminism or with one-sided error, it can be solved with no error and logarithmic space on average. See the articles on RL and NL for more information about ZPL.