论文部分内容阅读
The use of high-fidelity simulations for design optimization and uncertainty quantification is often prohibitively expensive.For this reason,algebraic surrogates are often fit to a sampling of the simulations in parameter space,and these surrogates are used instead of the simulations.Multi-fidelity surrogates attempt to take one more step in the direction of economy by combining a small number of high-fidelity simulations with a large number of inexpensive low-fidelity simulations.The latter usually represent coarser discretization,simpler physical models,or idealized geometry.