Tracking and Distinguishing Slime Mold Solutions to the Traveling Salesman Problem through Synchronized Amplification in the Non-Equilibrium Steady State

Abstract

The plasmodium of the true slime mold Physarum polycephalum serves as a platform to study information processing in non-equilibrium active matter, showing complex oscillations and computation despite its simple morphology. Past experiments used Physarum's shape changes and photoavoidance in a stellate chip to find approximate solutions to the traveling salesman problem (TSP) for up to eight cities in linear time. To solve the $N$-city TSP, the organism elongated and withdrew its $N^2$ branches within the chip lanes, where an optical feedback loop controlled by a modified Hopfield neural network selectively illuminated certain lanes to trigger retraction of $N(N-1)$ branches. When the modified Hopfield network stabilizes the illumination pattern, the organism reaches a non-equilibrium steady state (NESS), where $N$ extended branches form a valid TSP solution. The illumination pattern induces a clear split between two distinct lane groups. Fourier and power spectral density analyses reveal that in NESS, the solution lanes exhibit lower frequency, larger-amplitude oscillations, an enhancement of these signals compared to the higher-frequency, smaller-amplitude fluctuations in the non-solution lanes. This frequency downconversion and amplification in power density is a hallmark of a Fröhlich condensate. Synchronization indices for Physarum-selected solution lanes in the NESS exhibit maximum $S\sim1$, while non-solution lanes reach minimum $S$-a consistent trend across problem sizes and tour lengths. Amoeba-inspired algorithms that leverage noise to accelerate TSP convergence make trade-offs between solution quality and speed, showing up to $\sqrt N$ scaling in iterations, akin to quantum algorithms like Grover's search. By tuning power density, frequency shifts, and synchronization, it may be possible to improve the quality and efficiency of TSP solutions from Physarum-based biocomputers.