1Blurb:: 2Construct a surrogate from multiple existing training points 3Description:: 4Multipoint approximations use data from previous design points to 5improve the accuracy of local approximations. The data often 6comes from the current and previous iterates of a minimization algorithm. 7 8Currently, only the 9Two-point Adaptive Nonlinearity %Approximation (TANA-3) method of 10\cite Xu98 is supported with the \c tana keyword. 11 12The 13truth model to be used to generate the value/gradient data used in the 14approximation is identified through the required \c 15actual_model_pointer specification. 16 17 18Topics:: 19Examples:: 20Theory:: 21Faq:: 22See_Also:: model-surrogate-local, model-surrogate-global, model-surrogate-hierarchical 23