Closed nlangrene closed 3 years ago
I found this book chapter https://onlinelibrary.wiley.com/doi/abs/10.1002/9780470400531.eorms0704, which is a very good literature summary on random search, makes is clear that SAJO should be referred to as Simulated Annealing, and should help us to dig out the actual name (if any) of our basic random search.
In the latest version of the paper (and ferrn),
Creeping Random Search (CRS): search_better()
Simulated Annealing (SA): search_better_random()
Pseudo Derivative (PD): search_geodesic()
By definition, simulated annealing contains a probability to accept a worse basis (jump out). Cooling refers to the decrease of the probability of acceptance (not of the step size). So, what we currently call SAJO should be actually called Simulated Annealing, and we need to find the right name of what we currently call SA ("Basic random search"?)