Finding Solutions to Different Problems Simultaneously in a Multi-molecule Simulated Reactor

In recent years, the chemical metaphor has emerged as a computational paradigm based on the observation of different researchers that the chemical systems of living organisms possess inherent computational properties. In this metaphor, artificial molecules are considered as data or solutions, while the interactions among molecules are defined by an algorithm. In recent studies, the chemical metaphor was used as a distributed stochastic algorithm that simulates an abstract reactor to solve the traveling salesperson problem (TSP). Here, the artificial molecules represent Hamiltonian cycles, while the reactor is governed by reactions that can re-order Hamiltonian cycles. In this paper, a multi-molecule reactor (MMR-n) that simulates chemical catalysis is introduced. The MMR-n solves in parallel three NP-hard computational problems namely, the optimization of the genetic parameters of a plant growth simulation model, the solution to large instances of symmetric and asymmetric TSP, and the static aircraft landing scheduling problems (ALSP). The MMR-n was shown as a computational metaphor capable of optimizing the cultivar coefficients of CERES-Rice model, and at the same time, able to find solutions to TSP and ALSP. The MMR-n as a computational paradigm has a better computational wall clock time compared to when these three problems are solved individually by a single-molecule reactor (MMR-1).

Keywords – Artificial chemistry, combinatorial optimization, traveling salesperson problem, TSP