The individual of population in standard self-adaptive evolutionary programming (EP) is composed as a pair of objective variable and strategy parameter. Therefore, EP must evolve both objective variable and strategy parameter. In standard evolutionary programming (CEP), these evolutions are implemented by mutation based on only Gaussian random number. On the other hand, fast evolutionary programming (FEP) uses Cauchy random number as evolution of objective variable and exponential evolutionary programming (EEP) uses exponential random number as evolution of objective variable. However, all of these EP (CEP, FEP and EEP) commonly uses Gaussian random number as evolution of strategy parameter. In this paper, we propose new EEP algorithm (NEP) which uses double exponential random number for both evolution of objective variable and strategy parameter. The experimental results show that this new algorithm (NEP) outperforms the existing CEP and FEP.