Deterministic GSO

In this section the deterministic GSO algorithm, where a glowworm selects the neighbor toward which it moves deterministically, as opposed to the probabilistic mechanism is introduced. In the deterministic GSO algorithm, the neighbor selection strategy is given by

where, l_i(t) represents the luciferin level associated with Glowworm i at time t, N_i(t) is the neighborhood of Glowworm i at time t and the Glowworm i selects to move toward a Glowworm j belongs to N_i(t) with probability p_j(t). In other words, a glowworm always chooses the neighbor which has the highest luciferin value to move toward. This strategy is seen to faster while locating single sources as can be seen from Table 1. Simulations assumed the source to be located at (-2,0) and a source strength, N = 10. The mean and standard deviation are calculated over 50 runs. A probabilistic mechanism is present in the basic GSO algorithm as it is advantageous for multimodal function optimization. In multimodal optimization, a probabilistic mechanism ensures that a few glowworms get to local peaks which might be of lesser magnitude than the global optima.

Table: Comparison between the probabilistic and deterministic strategies for a single source with constant wind.