“Binary Test” is a name I gave to a theoretical testing method to identify the infected samples. I got the idea from this tweet
If you have a limited number of tests but unlimited ability to take individual samples what would be the ideal way to use them? If you expect a single positive result per 128 tests should you mix samples and binary search? Is there a better algorithm for other distributions?— Sam Pullara (@sampullara) March 13, 2020
It is a similar core idea as the Binary Search algorithm, hence the
name, but instead of looking for one specific sample, we are trying to
identify all samples that are infected. The basic idea is, you mix all
samples, perform one single test on the batch – if it is negative then
none of the samples are infected, if it is positive then you split the
batch into two and recurse.
