Neural Networks without Training for Optimization - Neural Networks: Methodology and Applications - page 455

Fig. 8.3. Example of 1D function to be minimized
where C 1 and C 2 are real positive numbers.
This cooling schedule is faster than the law given by Mitra et al. in 1986
for the simulated annealing [Mitra et al. 1986].
Pros and Cons
In practice, the qualities of the simulated annealing are preserved with the
rescaling of the energies. Moreover, when the allotted time is limited, that
Fig. 8.4. Deformation of the function of Fig. 8.3 as a function of E target