Computer experiments and uncertainty quantification: an overview
主 题: Computer experiments and uncertainty quantification: an overview
报告人: Jeff Wu (Georgia Institute of Technology)
时 间: 2016-06-22 16:00-17:00
地 点: 理科一号楼 1114
Because of the advances in complex mathematical models and fast computer codes, experiments on a computer have become popular in engineering and scientific investigations. Computer simulations can be much faster or less costly than running physical experiments. Research on computer experiments has progressed rapidly in the last 10 years. In this talk I will give an overview of the field. Difference from traditional physical experiments will be described. Some major development will be outlined, including multi-fidelity modeling, calibration, spatial-temporal model, and experimental design. In parallel to and independent of the work in statistics, applied mathematicians have approached the same problem under the banner of uncertainty quantification (UQ). Thus UQ can be viewed as an interface between applied math and statistics. A cursory introduction to UQ will be made. Some real examples will be given for illustrations.