The paper uses the discrete approximate iteration method and bi-convergent method to solve the multidimensional
continuing convex dynamic programming model. Firstly, the state value of one of state equations is set to be unknown and the
others be known. Then, discrete approximate iteration method is used to find the optimal value of the unknown state values
until all state equations have found optimal values. If the objective function is non-concave and non-convex, the algorithm is
proved convergent. If the objective function is convex, the algorithm is proved linear convergent. Finally, an example shows
the effectiveness of the formation and the algorithm.