Просмотр исходного текста страницы Honest Specifics Regarding The Q-VD-Oph Accomplishments

if (the actual physical fitness associated with temp_solution is larger than that of Px)10.           Px Equates to temp_solution;}}} 5.6. Local Optimal Avoidance Another drawback of SFLA applied to e-RTSP is that it is easy to fall into a local optimal solution. A local optimal solution is a scheme in which tasks are assigned to processors extremely unevenly. Namely the fitness of the local optimal solution is quite high, but the assigned computing capacity of the processor exceeds its maximal computing capacity. Construction of a sub-population in each sub-population can prevent the occurrence of such [http://en.wikipedia.org/wiki/Thymidine_kinase Thymidine kinase] a situation to a certain extent. To prevent local optimal solutions, several frogs in a sub-population will be selected into the sub-population. As the convergence speed of SFLA cannot be slowed down, the frogs with higher fitness should be assigned a higher selected probability. As in [13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32], the selected probability of each frog is defined as follows: p(j)=2(n+1?j)n(n+1),j=?1,2,?��,n (8) where n is the number of the frogs in a sub-population. As the frogs in a sub-population are sorted in descending order according to their fitness, [http://www.selleckchem.com/products/q-vd-oph.html Q-VD-Oph research buy] so Equation (8) gives higher selected probability to the frogs with higher fitness. 5.7. Convergence Acceleration As the original population is generated randomly when applying SFLA��s whole process to e-RTSP, the overall qualities of these scheduling schemes are low. We apply the neighborhood search to the original population to improve the fitness of each frog. It consists of two parts: (1) migrate a task from a processor to a different one if the energy consumption becomes less without lowering the fitness; and (2) exchange two tasks between two processors if the energy consumption becomes less without lowering the fitness. A frog becomes a better solution in its neighborhood by applying the neighborhood search. The overall quality of the population is improved with this preprocessing and so SFLA can find a feasible scheduling scheme more quickly. [http://www.selleckchem.com/products/sch58261.html click here] 5.8. Summary of the SFLA Applied to e-RTSP In this section, Figure 3 concludes the above procedures. Figure 3 Flow chart of SFLA for e-RTSP. 6. Experiment Extensive experiments are conducted to estimate SFLA��s performance for the energy-aware real-time task-scheduling problem. Our task and processor sets are not only generated from synthetic data for simulation but also from real data of benchmarks. The synthetic data are generated following the same setting in [11]. The real data are from the Embedded System Synthesis Benchmarks Suite (E3S) [37]. First, we generate the synthetic data sets for simulation. Then, the parameter tests are conducted to determine the optimal parameters for SFLA. After that, SFLA is compared with several familiar scheduling algorithms with synthetic and real data of benchmarks. Their performance is analyzed in detail. 6.1.