DSApr 15

Lawler-Moore Speedups via Additive Combinatorics

arXiv:2604.1364236.6h-index: 30
Predicted impact top 34% in DS · last 90 daysOriginality Incremental advance
AI Analysis

This work provides a significant theoretical speedup for fundamental scheduling problems, improving upon classical algorithms that are optimal under SETH but leaving room for improvement in terms of maximum processing time.

The paper presents the first major speedup of the Lawler-Moore dynamic programming framework for scheduling on parallel machines, reducing the dependence on total processing time P to a dependence on the square of the maximum processing time p_max. For objectives Pm||∑w_jC_j and Pm||L_max, they achieve O(p_max^{2m-2}n) time, and for Pm||∑w_jU_j, O(p_max^{2m-1}n^2) time, yielding near-linear-time algorithms when processing times are polylogarithmic in n.

The Lawler-Moore dynamic programming framework is a classical tool in scheduling on parallel machines. It applies when the objective is regular, i.e. monotone in job completion times, and each machine follows a fixed priority order such as Smith's Rule or Jackson's Rule. For the basic objectives $Pm||\sum w_jC_j$, $Pm||L_{\max}$, and $Pm||\sum w_jU_j$, it gives running times $O(P^{m-1}n)$, $O(P^{m-1}n)$, and $O(P^mn)$, respectively, where $P$ is the total processing time. Recent SETH-based lower bounds indicate that the dependence on $P$ is essentially optimal, but they do not rule out improved dependence on the maximum processing time $p_{\max}$. We give the first major speedup of the Lawler-Moore recurrence. Our main ingredients are a new state-pruning method and a swapping argument based on an additive-combinatorial lemma. We prove that, whenever this swap does not increase the objective value, there exists an optimal schedule in which, for every prefix of jobs, the load difference between any two machines is at most $4p_{\max}^2$. This lets us prune redundant states throughout the dynamic program, replacing the dependence on $P$ by a dependence on $p_{\max}^2$. We show that the swap is non-increasing for all three objectives above. Hence $Pm||\sum w_jC_j$ and $Pm||L_{\max}$ admit algorithms with running time $O(p_{\max}^{2m-2}n)$, while $Pm||\sum w_jU_j$ can be solved in time $O(p_{\max}^{2m-2}Pn)\le O(p_{\max}^{2m-1}n^2)$. These bounds strictly improve the original Lawler-Moore runtimes whenever $p_{\max}=o(\sqrt{P})$. In particular, for $Pm||\sum w_jC_j$ and $Pm||L_{\max}$, we obtain the first near-linear-time algorithms when processing times are polylogarithmic in $n$.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes