On using compressibility to detect when slime mould completed computation

Slime mouldPhysarum polycephalumis a single cell visible by an unaided eye. The slime mould optimizes its network of protoplasmic tubes in gradients of attractants and repellents. This behavior is interpreted as computation. Several prototypes of the slime mould computers were designed to solve problems of computation geometry, graphs, transport networks, and to implement universal computing circuits. Being a living substrate, the slime mould does not halt its behavior when a task is solved but often continues foraging the space thus masking the solution found. We propose to use temporal changes in compressibility of the slime mould patterns as indicators of the halting of the computation. Compressibility of a pattern characterizes the pattern's morphological diversity, that is, a number of different local configurations. At the beginning of computation the slime explores the space, thus generating less compressible patterns. After gradients of attractants and repellents are detected the slime spans data sites with its protoplasmic network and retracts scouting branches, thus generating more compressible patterns. We analyze the feasibility of the approach on results of laboratory experiments and computer modelling. © 2015 Wiley Periodicals, Inc. Complexity 21: 162–175, 2016