4 ms·
Interesting, but I can't think of any situations where I'd take chances with probabilistic periods in exchange for a slight speed boost over something with a kn
by clarry 7y ago
Interesting, but I can't think of any situations where I'd take chances with probabilistic periods in exchange for a slight speed boost over something with a known period and good output (e.g. a large enough mcg/lcg where you drop the low bits on output, or pcg -- these are all plenty fast and can pass the same tests).
- FabHK 7y agoFor Monte Carlo eg in finance, you’d definitely take the speed improvement. This thing passes Big Crush and PractRand - it’s going to compute the approximate value of your three factor Power Reverse Dual note just fine.
- clarry 7y ago> This thing passes Big Crush and PractRand ... for the seeds that were tested. In theory, you can get unlucky and end up in a short cycle, and it will not pass these tests. That's why I'm not comfortable with it, when there are generators with a known, guaranteed period. They simply do not have this failure mode.
- FabHK 7y agoThe probability of the problem occurring is vanishingly small - much smaller than a bit flip due to a cosmic ray in the machine computing it, leading to a wrong answer or a crash. And even that doesn’t matter, because the computations are repeated so often. Again, the application determines what the most suitable PRNG is.
- MarkOverton 7y agoThat's why I pointed out that the probability of a too-short cycle or sequence-overlap is no higher than that of randomly selecting one snowflake out of all snowflakes in Earth's history. Also, the paper has a graph and accompanying discussion of what happens with the shorter cycles.