In this paper, some algorithms for searching representative points (RPs) of two-dimensional mixed normal distribution are given. Some theoretical properties are also given. The performance of three different types of RPs respectively based on Monte Carlo method, number-theoretical method and mean square error (MSE) method, are compared in sense of distribution discrepancy and mean square error criterion. It is shown that RP based on MSE method performs better than RP based on Monte Carlo method. Moreover, the bootstrap technique is used to resample the RPs, which also verifies the superiority of RPs based on the MSE method.
