When writing a dynamic programming algorithm, a problem's solution is formulated in terms of its smaller subproblem solution. The first thing you do when developing a dynamic programming algorithm is to formulate the recursive optimality constraint among subproblems. For example, optimal matrix chain multiplication:

  m[i,i] = 0
  m[i,j] =  min  { m[i,k] + m[k+1,j] + p[i-1]*p[k]*p[j] }
           i<=k<j
[download]

Even the Cormen et al book does dynamic programming this way (starting with the recursive optimality). In general, this is enough to describe a memoized dynamic programming algorithm, since the details are always the same for these kinds of problems. I consider this more of a logic programming formulation. It doesn't say "do this, do that", it only gives a production rule for constructing optimal subproblems and some base cases.