SOURCE-LINKED INTELLIGENCE

A Group-Based Resource Allocation Model for the Fractional Knapsack Problem

arXiv · AI, language, vision and robotics · article · Sep 6, 2026 · UTC

To solve the fractional knapsack problem, Dantzig's greedy rule orders items according to their value-to-cost ratio. This ordering introduces priority issues. An arbitrarily small perturbation to the input can change the allocation if the budget is exhausted between two items with very similar ratios. To mitigate that problem, we introduce a two-stage rule. We group items sharing attributes within a radius $δ$. These groups are then evaluated in descending order of ratio, and divide their group's budget share without further ranking. Consider a group featuring an aggregate capacity $U_G$, unit

Read original source ↗ Open in workspace

recordType

paper

region

Global