Time-Space Tradeoffs for Computing Functions, Using Connectivity Properties of Their Circuits

Abstract Recent research has investigated time-space tradeoffs for register allocation strategies of certain fixed sets of expressions. This paper is concerned with the time-space tradeoff for register allocation strategies of any set of expressions which compute given functions. Time-space tradeoffs for pebbling superconcentrators and grates are developed. Corollaries which follow include tradeoffs for any straight-line program which computes polynomial multiplication, polynomial convolution, the discrete Fourier transform, straight-line merging, and the computation of “most” sets of n linear functions in n indeterminates.