Suboptimal dynamic programming with error bounds
This paper presents a method to relax Dynamic Programming. The methodmakes it possible to findsuboptimal solutions with known error bounds to hard problems.The bounds are chosen by the user, who can then effectively trade-offbetween solution time and accuracy. Several examples from differentdomains where the method is highly useful are presented.