Sorting Can Exponentially Speed Up Pure Dynamic Programming

TL;DR

This paper demonstrates that incorporating sorting via min and max operations in pure dynamic programming algorithms can exponentially reduce the complexity of solving certain problems like the minimum spanning tree.

Contribution

It introduces a novel pure (min,max,+) DP algorithm that solves the minimum spanning tree problem in polynomial time, significantly improving over the exponential lower bound.

Findings

Pure (min,+) DP algorithms require exponential operations for MST.

The new (min,max,+) algorithm solves MST in polynomial time.

Sorting enabled by min and max operations is key to the speedup.

Abstract

Many discrete minimization problems, including various versions of the shortest path problem, can be efficiently solved by dynamic programming (DP) algorithms that are "pure" in that they only perform basic operations, as min, max, +, but no conditional branchings via if-then-else in their recursion equations. It is known that any pure (min,+) DP algorithm solving the minimum weight spanning tree problem on undirected n-vertex graphs must perform at least $2^{\Omega(\sqrt{n})}$ operations. We show that this problem can be solved by a pure (min,max,+) DP algorithm performing only $O(n^3)$ operations. The algorithm is essentially a (min,max) algorithm: addition operations are only used to output the final values. The presence of both min and max operations means that now DP algorithms can sort: this explains the title of the paper.