notes.txt

To do list, no date estimate

0 hyper-rectangle (not just cubic) domains

1 specify some variables periodic, others not

2 sample continuation - start with some prescribed set of input samples (could be from a prior run with a large radius, as in var-rad, or really just anything), then continue sampling with a new (smaller) radius.

3 non-uniform sampling radius - specify a radius function, e.g. the reciprocal of a probability distribution