A new advance on a method known as a randomness extractor makes it easier for machines generate truly random numbers by harvesting randomness from the environment.
The new randomness extractor combines two independent sources of weakly random numbers into one set that is nearly random, with only minor deviations. Then the researchers use a “resilient function,” a method of combining information, to turn the string of numbers into one truly random bit — a 1 or 0.
Compared with the previous state-of-the-art randomness extractors, which required input that was already very close to random, the new method can mine sources that are “much, much, much, much weaker,” says computer scientist Avi Wigderson of the Institute for Advanced Study in Princeton, N.J. The new extractor is a “substantial improvement over the previous results, and it’s very close to the best you can hope for.”