[SciPy-Dev] Proposal - optimize callback given a state keyword

[Jaime Fernández del Río]

On Fri, Feb 9, 2018 at 2:35 AM Andrew Nelson <andyfaff at gmail.com> wrote:

> The `differential_evolution` function is backed by the (as yet) **non
> public** DifferentialEvolution solver object.
>
> This has different modes of operation (by virtue of implementing __next__
> and __iter__):
>
> ```
> def ackley(x):
>     arg1 = -0.2 * np.sqrt(0.5 * (x[0] ** 2 + x[1] ** 2))
>     arg2 = 0.5 * (np.cos(2. * np.pi * x[0]) + np.cos(2. * np.pi * x[1]))
>     return -20. * np.exp(arg1) - np.exp(arg2) + 20. + np.e
>
> bounds = [(-5, 5), (-5, 5)]
>
> solver = DifferentialEvolutionSolver(ackley, bounds)
>
> # run's through to completion
> res = solver.solve()
>
> # control how many iterations you want
> for idx in range(10):
>     x, fun = next(solver)
>     # amend internal state of solver if you know what you're doing
>
> # keep on going until you decide to stop
> for x, fun in solver:
>     print(x, fun)
>     intol = (np.std(solver.population_energies) <=
>              solver.atol +
>              solver.tol * np.abs(np.mean(solver.population_energies)))
>     if intol:
>         break
> ```
>
> It seems like this could be a useful design pattern going forwards, as I
> think it meets Jaime's points. However, this is obviously a large package
> of work.
>

It looks very nice indeed. I will try to find some time to look into the
approach in differential evolution and see if I can find inspiration there
to turn my thoughts into a more articulate proposal. It certainly looks
like it would be a lot of work, but I think there is a chance I could
convince my manager to work on some of this on his dime.

Jaime

-- 
(\__/)
( O.o)
( > <) Este es Conejo. Copia a Conejo en tu firma y ayúdale en sus planes
de dominación mundial.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scipy-dev/attachments/20180209/f16c6064/attachment-0001.html>