Abstract: An improvement of constituting Markov partition is proposed for calculating measure theoretic entropy by computing invariant distribution of nonlinear mapping instead assuming equal likelihood for all points over a interval; and the formulations and the iteration forms of invariant distribution of logistic map are derived.It is shown that the better result can be obtained through several iterations and the amount of computation is not larger.

Key words: metric entropy, invariant distribution, generating partition, Markov partition, logistic map