> Are we just trying to estimate the ratio? To what desired accuracy?
Exactly. Number of samples depends on volume ratio and accuracy. It depends on dimension only through these two numbers. You got there in the end, proud of you.
You continue to miss the point. In this specific context, volume ratio scales super-exponentially with dimension, and unless you're willing to have accuracy also drop super-exponentially (as in, go very very quickly to zero), then you are wasting your time by trying to perform rejection sampling in a context where you are statistically never going to hit the target set.
No, I didn't miss that point. I'm well familiar with hyperspheres. And yet, the thing you described is only incidentally related to dimension.