(minor bug) learner.choose_points gives wrong number of points in one very particular case

[basnijholt]

(original issue on GitLab)

opened by Jorn Hoofwijk (@Jorn) at 2018-04-18T13:17:25.768Z

I think this would almost never cause issues and would be easy to solve, it's just a bit odd.

The case is if you first request 1 point, and then 2 points, it will give you one point back in the second case.
The case where this happens is so specific I do not believe it would cause any trouble in the real world. But it would also be easy to solve, so might be worth the time.

import adaptive


def f(x):
    return x


learner = adaptive.Learner1D(f, bounds=(-5, 5))

print(learner.choose_points(1))
print(learner.choose_points(2))

# results in: 
# ([-5], [inf])
# ([5], [inf, inf])
# while we expect:
# ([-5], [inf])
# ([5, 0], [inf, inf])

[basnijholt]

originally posted by Bas Nijholt (@basnijholt) at 2018-04-18T13:35:12.954Z on GitLab

Nice catch, probably has an easy solution.