Short Simplex Paths in Lattice Polytopes
summary
Algorithms produce explicit simplex edge paths on lattice polytopes in a box, with a first constructive method giving a path length bounded roughly by n^4 · K · log K (independent of the number of inequalities and of the cost vector) using bit-scaling plus a Franke Tardos preprocessing, and a second iterative method that peels off one equality per iteration to reduce dimension and improves the bound to roughly n^2 (with the entry-size α appearing only inside logs). Both are implemented as oracle-based pivoting routines and provably terminate.
AI-written summary · about
chapters (11)
- 0:00Intro
- 1:48Diameter of lattice polytopes
- 5:21LP on lattice polytopes
- 7:38How "short" can a simplex path be?
- 9:45Short simplex paths in lattice polytopes
- 12:40How does it work?
- 14:39Basic algorithm
- 21:51Preprocessing & scaling algorithm
- 22:09Iterative algorithm
- 27:22Correctness - idea
- 29:11Short simplex paths - idea
