Distributed Resource Allocation for Large-Scale Knapsack Problems

The patent divides the large-scale knapsack problem into multiple sub-problems by partitioning the set of items into different groups. Each sub-problem involves a subset of items and can be solved independently or in parallel. This segmentation reduces the computational complexity from exponential in the total number of items to exponential in the number of items per group, significantly lowering resource consumption while maintaining solution quality.

The patent performs preliminary actions by pre-computing and storing the optimal solutions for smaller sub-problems involving subsets of items. These pre-computed results are then reused when solving larger problems, avoiding redundant computations. This approach significantly reduces solving time for large-scale problems while increasing the effective throughput of the resource allocation system.

The patent introduces a new dimension by organizing items into hierarchical groups and levels. Instead of treating all items at the same level, the solution creates a multi-dimensional structure where items are partitioned across different groups and sub-groups. This dimensional organization transforms the intractable single-layer optimization problem into a manageable hierarchical structure, enabling the system to handle billions of decision variables by distributing computation across multiple levels of the hierarchy.