You are on page 1of 3

Dr. Z.

s Number Theory Lecture 14 Handout: Pseudoprimes and Probablistic Primality Tests


By Doron Zeilberger
Recall that Fermats Little Theorem tells you that if p is prime, and a is such that 1 a p1
then we must have that
a
p1
1 (mod p) .
But the converse is not always true, it could happen, (very rarely!) that x is composite, and
nevertheless
a
x1
1 (mod x) .
If that happens x is called Fermat pseudoprime to base a, or base-a pseudoprime.
Important Fact: The smallest base-2 pseudoprime is the composite number 341 = 11 31.
A composite number x that is a Fermat pseudoprime for any base a between 1 and p 1 is called
a Charmaichael Number.
We have
Theorem (A. Korselt 1899): A positive composite integer n is a Carmichael number if and only if
n is square-free, and for all prime divisors p of n, it is true that p 1 is divisible by n 1
Problem 14.1: Use Korslets theorem to show that 561 is a Carmichael number.
Solution to 14.1: 561 = 3 11 17. Now (5611)/(31), (5611)/(111), and (5611)/(171)
are all integers.
Fermats Primality Test
This should more appropriately called compositeness test. If a
n1
1 (mod n) for some a then
we know for sure that n is not prime. a is called a Fermat witness for the compositeness of n.
Problem 14.2: Use the Fermat primality test to investigate wheter 33 is prime or composite.
Solution to 14.2: The easiest witness to use is 2, so we start with a = 2, and hope for the best.
We need to nd 2
32
(mod 33).
2
2
(mod 33) = 4
2
4
(mod 33) = 16
2
8
(mod 33) = 16
2
(mod 33) = 256 (mod 33) = 33 7 + 25 (mod 33) = 25 (mod 33)
2
16
(mod 33) = 25
2
(mod 33) = (8)
2
(mod 33) = 64 (mod 33) = 31
1
2
32
(mod 33) = 31
2
(mod 33) = (2)
2
(mod 33) = 4 .
Since 2
32
1 (mod 33) we know for sure that 33 is composite.
If we compute a
n1
(mod n) for quite a few as and we always get 1 then there is a very good
chance that n is a prime.
Problem 14.3: Use the Fermat primality test to investigate whether 7 is a prime or composite,
by picking two random as (if necessary) in the set {2, 3, 4, 5, 6}.
Solution to 14.3: Lets pick a = 2 and a = 4, and nd a
6
(mod 7).
2
6
(mod 7) = 64 (mod 7) = 7 9 + 1 (mod 7) = 1
4
6
(mod 7) = 4096 = 7 585 + 1 (mod 7) = 1
We deduce with high probability that 7 is a prime, in other words it is a probable prime.
Remark: Of course, it is much easier to check that 7 is a prime directly. The test is only useful
for larger integers.
The Miller-Rabin Primality Test
Inputs: A positive integer n, and an integer a between 2 and n 2
Output: The knowledge that n is composite for sure, or a is a witness for the probable primality
of n
Step 1: By repeated division-by-2 (easy), write n 1 = 2
s
d, where d is an odd integer, and s is
an integer 0.
Step 2:
For i = 0, 1, . . . , s 1:
2.1: Compute a
2
i
d
(mod n)
If for all of them, they are neither 1 nor 1 (alias n 1) output composite.
Otherwise, keep trying several (say, R) randomly chosen as, and if all these witnesses testify,
output
n is very probably prime, using R random witnesses.
Problem 14.4: Use the Miller-Rabin primality test to investigage whether 7 is a prime or composite
by picking two (if necessary) random as in the set {2, 3, 4, 5, 6}
Solution to 14.4: 7 1 = 6 = 2 3 so d = 3 and s = 1
2
With a = 2
2
3
(mod 7) = 1, so a = 2 is a witness to the primality of 7
With a = 3
3
3
(mod 7) = 1, so a = 3 is a witness to the primality of 7.
Hence 7 is a probable prime with two witnesses.
Problem 14.5: Use the Miller-Rabin primality test to decide whether 21 is prime or composite
by picking (if necessary!) two random as in the set {2, 3, 4, 5, 6, 7, 8, 9}
Solution to 14.5: 21 1 = 20 = 5 2
2
so d = 5 and s = 2
With a = 2
2
5
(mod 21) = 32 (mod 21) = 11, that is neither 1 nor 20
2
10
(mod 21) = 11
2
(mod 21) = 121 (mod 21) = 16, that is neither 1 nor 20
So we know, for sure, that 20 is composite.
Ans. to 14.5: 20 is a composite number, by the Miller-Rabin test.
3

You might also like