Handling with regions unreachable inside the `ConvexHull` in `LearnerND`

[lsyxiaopang]

This library is very cool and I've been using it for controlling measuring instruments. However, for measuring instruments, there would be some situations where one or more regions inside the range cannot reach, and the parameters of LearnerND cannot handle this problem.

So, is there any ways to solve this problem without affecting the library too much, for example, if I got point that I can't reach, I may just return 0.

Thank you!

[basnijholt]

If you know these unreachable regions before hand, you could write a WrappedFunction around the function that controls your instrument. Then, if you are within those regions, you will simply return NaN.

[lsyxiaopang]

If you know these unreachable regions before hand, you could write a WrappedFunction around the function that controls your instrument. Then, if you are within those regions, you will simply return NaN.

Thank you, I think these NaNs may not affect the predictions too much.