An Analysis of Algorithms and Heuristics for Combinatorial Optimization

Loading...
Thumbnail Image

Date

2023-05

Journal Title

Journal ISSN

Volume Title

Publisher

The Ohio State University

Research Projects

Organizational Units

Journal Issue

Abstract

Combinatorial optimization problems involve a search in a finite search space of multiple parameters that minimize a set metric. In this work, the practical implementation of a classical combinatorial optimization problem and applicable heuristics are explored in a survey of several models. The combinatorial problem explored is combining the possibilities of flights and hotels across a search window proportionally longer than the number of travel days where optimality is defined solely on cost of hotels and flights. For the trivial case, only all of the departing flights, hotels, and return flights for a single destination are considered. Then the trivial problem is expanded to include a second destination so the problem's dynamics necessitate finding the cheapest itinerary for visiting both locations given the total trip time. For this expanded problem, the matrix data structure fails to capture the magnitude of the data and so the data is queried from two databases but aggregated into a graph data structure modeling the problem as a classic Shortest Path Problem in the field of Combinatorial Optimization. The work compares the reduction in time complexity leveraging the problem dynamics to preprocess the graph and prune branches and then finding the optimal solution with Dijkstra's Algorithm. The work then explores the effect of assuming uncertainty in the prices transforming the problem into an NP-Hard problem in the scenario-represented uncertainty. By modeling the uncertainty as arc intervals, the problem can be solved in polynomial-time. Finally the work explores the applicability of heuristic algorithms to the multi-destination graph with certainty in the arc weights.

Description

Keywords

combinatorial optimization

Citation