Thinking About Probabilistic Filters

I've been a big fan of probabilistic filters such as Bloom, and (more so) Cuckoo filters for a while now. I've put them to great use in my personal projects, as well as in major data processing projects at work. They fall into a neat class of data structures that let you trade certainty for performance. They are wildly useful for membership checks without loading the entire set into memory in cases where occasional false positives are acceptable. This can make certain cases of checks far more...