Dynamic programming solves optimization problems by recursively decomposing them into subproblems, storing optimal subproblem values and reassembling them efficiently.

In particular, a RAM machine \(A\) consists of a finite sequence of instructions (or program) \(\langle \pi_1,\ldots,\pi_n \rangle\) expressing how numerical operations (typically addition and subtraction) are to be applied to a sequence of registers \(r_1,r_2, \dots\) in which values may be stored and retrieved directly by their index. Showing that one of these models \(\mathfrak{M}_1\) determines the same class of functions as some reference model \(\mathfrak{M}_2\) (such as \(\mathfrak{T}\))