Say I want to use scipy.optimize.differential_evolution to optimize a function, but I have a reasonable starting point where I believe the minimum is located.
Is there a way to feed at least one initial position to this algorithm?
I thought of restricting the bounds parameter to the near vicinity of my "solution". This has the drawback that it will affect the entire population (popsize) and I do want the algorithm to search the entire landscape of solutions.
Can this be done at all?
v1.7.0 or newer of scipy.optimize.differential_evolution supports x0 which is optionally your initial guess.