ULSAlgorithms 1.1.0-g3e5595996d
High-performance exact and heuristic algorithms for uncapacitated lot sizing
Loading...
Searching...
No Matches
Jacobs-Khumawala simplified optimal lot sizing

Jacobs-Khumawala simplified optimal lot sizing

Public class:

JacobsKhumawalaBranchAndBoundSolver

Reference:

F. R. Jacobs and B. M. Khumawala (1987), A Simplified Procedure for Optimal Single-Level Lot Sizing, Production and Inventory Management 28(3), 39-43.

Publication identity

The published abstract describes:

  • a simple branch-and-bound procedure;
  • exact single-item, single-level lot sizing;
  • computational equivalence to Wagner-Whitin;
  • a graphical representation intended to make the method easier to apply;
  • decomposition into subproblems;
  • elimination of subproblems that cannot lead to an optimum.

C# reconstruction

A node is a zero-inventory boundary.

A branch from boundary i to boundary j corresponds to one replenishment in period i satisfying demand through j-1.

For each boundary the algorithm stores only its cheapest partial-plan label. Any more expensive branch reaching the same boundary is dominated because the remaining subproblem is identical.

An explicit Lot-for-Lot solution provides an initial upper bound. Branches whose accumulated cost already exceeds that incumbent are fathomed.

Because boundary subproblems are processed in topological order, labels are final when expanded.

Complexity

  • O(T²) branch evaluations;
  • O(T) auxiliary memory;
  • exact result.

This class is a modern programmatic reconstruction of the publication's branch/subproblem representation. It does not claim to reproduce the original paper's graphical worksheet line by line.