Abstract | Monte-Carlo method is actually widely used in science and engineering. The theoretical foundation of this method is the law of large number in which a standard condition for data sequences is the well-known i.i.d. condition. But for most real world data sequence, this i.i.d. requirement is certainly too strict. In this talk we present a new Monte-Carlo algorithm which can be applied to situations with much higher degree of uncertainty, known as probability-distribution uncertainty. This new algorithm are based on the author’s law of large number and central limit theorem in a framework of nonlinear expectation theory. We also provide concrete explanation through some challenging real world problems in financial risk controls and in machine learning. |