Beyond massive parallelism: Numerical computation using associative tables (Q753471)

The authors discuss the basic problem to speed up algorithms by the use of available information from previous solutions or solution steps. This information should be stored in and delivered by an associative memory. In this context the solution of ordinary and partial differential equations and Karmarkar's algorithm are discussed. The authors write: Our results have a speculative character, since we deal with numerical computation on a connectionist (analog) parallel model and a putative class of hardware implementations.