Given $m$ points $P=\{p_0, p_1, ..., p_m\}$ in high dimensions (e.g. 100), it is known that computing (or even representing) their convex hull $\text{conv}(P)$ is generally intractable due to the exponential dependence on dimensions. In this case, I wonder:
- Is there an efficient algorithm to check whether some other given point $\tilde{p}$ is an interior point in the convex hull (i.e. there is a small ball centered at $\tilde{p}$ and lies in $\text{conv}(P)$)? I know we can construct a linear programming problem to check if a point lies inside the convex hull, but my question here is to further check if the convex hull has "volume" and if $\tilde{p}$ lies in its interior.
- Following 1, can we compute or efficiently lower-bounding the largest enclosed (inscribed) ball centered at $\tilde{p}$ and lies in $\text{conv}(P)$?
I'm an amateur in geometry and I appreciate any potential solutions or suggestions. Thank you for your time!
The same question posted on MathSE.