Figure 1
From: Reverse engineering a gene network using an asynchronous parallel evolution strategy
Behaviour of the serial island-ES. (A) The effect of the number of islands on serial algorithm performance. We plot the inverse of the time needed to reach 'good-enough' and 'good' solutions (values of the objective function less than 550000 (blue) and 350000 (red) respectively) against the number islands used. (B) Prediction of the maximum achievable absolute speed-up of a piES algorithm across different numbers of islands, calculated by assuming that each island is running on a separate processor, and there is no communication overhead. To achieve this we divide the time needed for an N-island ES to converge by the number of islands Nisl. This measure of the optimal parallel performance is then divided with that of the fastest serial algorithm (running on 1 island). The resulting ratios are shown for 'good-enough' and 'good' solutions. The black line indicates perfect absolute speed-up. Note that no run on 50 islands was able to reach the 'good' solution value within the time limit of 36 hours for jobs on the Darwin cluster. Error bars in (A) and (B) represent 95% confidence intervals on the mean.