Calculating for modern operations
Managers of large-scale operations have an especially difficult time computing the optimal solution due to the many decisions they have to make in combination, according to DeValve, Pekeč, and Wei. To address this problem, the researchers designed a computationally efficient algorithm that can generate better decisions for:
• SUPPLY: How much to order of each component
• DEMAND: How to prioritize product assembly when orders arrive
More precise than established methods
An approach that can scale up: The researchers’ technique involves balancing inventory and shortage costs to come as close as possible to the optimal cost. Their algorithm, which takes a new approach to analyzing this problem, does not degrade in performance as the problem grows in size, a crucial stability guarantee for the scale of modern applications that has been lacking in the existing algorithms.
Closer to the optimal solution: In numerical simulations, the researchers find that their algorithm outperforms existing methods—on average, it comes within 1 percent of the optimal solution compared with 5 percent to 13 percent for other methods.