APPROXIMATING CLOSED FORM SOLUTIONS TO CONVERGING BRANCH DYNAMIC-PROGRAMMING PROBLEMS Academic Article

abstract

• This paper concerns a method for obtaining closed form approximate solutions to converging branch dynamic programming problems with quadratic returns and linear transitions. For quadratic returns, not necessarily convex, the optimum solutions for each stage generally contain multiple segments. A procedure for dealing with these multiple segments is presented. Problems with nonlinear returns at the converging branch have not been effectively dealt with in the literature. Utilizing a piecewise linear approximation to these nonlinear returns and the multiple segment procedure, closed form solutions to the resulting problems can be obtained. This method facilitates the computerization of this class of converging branch problems. 1987.

authors

• Curry, Guy

published proceedings

• JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS

author list (cited authors)

• GIM, B. J., & CURRY, G. L.

citation count

• 0

complete list of authors

• GIM, BJ||CURRY, GL

publication date

• February 1987

publisher

• Elsevier  Publisher

published in

• Journal of Mathematical Analysis and Applications  Journal

Digital Object Identifier (DOI)

• 10.1016/0022-247X(87)90344-1

start page

• 58

end page

• 69

volume

• 122

issue

• 1

URL

• http://dx.doi.org/10.1016/0022-247x(87)90344-1