Table 3. Effectiveness of conducting 100,000 tests on a population of 1 million; full statistics over 10 simulations each.

	Method
	Individual testing	2-level pooling	Binary splitting	Recursive bin. splitting	Purim	Sobel-R1
No. individuals tested	100,000.0	490,791.9	804,985.4	976,797.2	720,729.5	1,083,181.0(*))
Cases found (tp)	990.0	4,810.7	7,578.5	9,353.0	6,993.6	10,442.2(*))
False-positives (fp)	990.3	460.5	76.6	459.5	393.6	850.4(*))
Cases missed (fn)	10.0	97.2	471.3	415.0	213.7	389.6(*))
Cleared (tn)	98,009.7	485,423.4	796,858.9	966,569.8	713,128.6	1,071,498.7(*))
Infected individuals not even tested due to limited test capacity	9,000.0	5,092.1	1,950.1	232.0	2,792.7	0.0 (-831.8(*))

Simulating 100,000 tests on a total population of 1 million with an infection rate of 1%. We assume a sensitivity of 99% and a false negative rate of 1% for a single PCR, and no dilution due to pooling. We report how many individuals are tested with 100,000 (pooled) tests as well as true positives (tp), false positives (fp), false negatives (fn) and true negatives (tn). The reported numbers are averaged over 10 runs testing the whole population, measuring a relative standard deviation of <0.3% each.

(*) Note that Sobel-R1 is able to test the whole population 1.08 times; this methods identifies 10, 442.2 > 10, 000 cases, identifying 832 cases in the second run leading to a negative value in the last entry.