Dynamic Programming over a Continuous and Disjoint Multidimensional Search Space with an Infinite Time Horizon
Richard E. Hawkins, Jack C. M. Dekkers, James B. Kleibenstein · RePEc: Research Papers in Economics · 1999
Dynamic programming with continuous choice variables generally assumes compact search spaces. Collapsing disjoint n-dimensional bubbles within the space, creating and indexing new bubbles as needed, pushes back the curse of dimensionality. An animal-breeding optimization is used as an example. The method could solve complex optimal-growth problems.