Effect of feedback delays on solution quality in amoeba-inspired computing system that solves traveling salesman problem

Abstract

We investigate how feedback delays affect the quality of solutions from an amoeba-inspired analog electronic computing system that solves the “traveling salesman problem”. Delays in the feedback process induce the oscillation of state variables. With an appropriate delay length, the system converges to the stable state that corresponds to the solution after oscillation. We find that the solution quality is improved by increasing the delay length. Consequently, delays bring a trade-off between the solution search time and the solution quality. Delay scheduling can further improve solution quality.

Amoeba-inspired analog electronic computing system integrating resistance crossbar for solving the travelling salesman problem

“…One of our future subjects is to improve the quality of the solution found by the electronic amoeba. Possible approaches are to assign appropriate initial states to the amoeba core units using the genetic algorithm, to impose stochastic fluctuations using a hardware random number generator to forcibly escape local minimum solutions (see SI), and to introduce delays in the bounceback control to induce the oscillation of state variables 45 .…”