Reducing the Memory Requirements of Dynamic Programming

Abstract
This paper presents a decomposition procedure for extending the size of problems that can be solved using dynamic programming. It essentially consists of decomposing the tabular arrays of data into blocks of data, and then performing the dynamic programming calculations over the whole tabular array by calculating on each block separately.