Cryptography Reference
In-Depth Information
holds but also the successive square roots obtained starting with this element follow
a very specific pattern. Thus if an integer n satisfies the assertion of the theorem,
then it also passes the Fermat test and one might hope that this gives a necessary and
sufficient condition for n to be prime. But it is easy to see that this is not the case. For
example, if n
=
2047
=
23
·
89 and b
=
2 then, with the same notation as above,
t
=
1023 and we have that
> Power(2, 1023) mod 2047;
1
so that the composite number 2047 satisfies the assertion of Theorem 6.6 for b
2,
and hence 2047 is an odd composite that behaves like a prime as far as this theo-
rem is concerned. This leads to the following definition, similar to that of (Fermat)
pseudoprime above:
=
2 s t , with t odd, and b
an integer. Then n is a strong pseudoprime to the base b if either b t
Definition 6.3 Let n be an odd composite number, n
1
=
1
(
mod n
)
or
b 2 i t
≡−
1
(
mod n
)
for some i with 0
i
s
1.
Example 6.3 It is clear that a strong pseudoprime to the base b is also a
b -pseudoprime, but strong pseudoprimes are less abundant than pseudoprimes: for
example, the Carmichael number 561 is not a strong pseudoprime to the base 2 for
we have that 560
2 4
=
·
=
=
35 in the above notation. Then, if we
compute the sequence of values b 2 i t mod n ,for0
35, i.e., s
4 and t
i
3, we obtain:
> Power(2, 35) mod 561;
263
> Power(263, 2) mod 561;
166
> Power(166, 2) mod 561;
67
> Power(67, 2) mod 561;
1
Thus we see that 561 is not a strong pseudoprime to the base 2 (and this happens
because 67 is a square root of 1 modulo 561).
(
,
) =
Remark 6.4 Observe that if gcd
b
n
1 and p is a prime that divides both n and
and b 2 i t
b , then b t
1, and this
clearly implies that n cannot be a strong pseudoprime to the base b . Thus, although
not explicitly required in the definition, a necessary condition for n to be a strong
pseudoprime to the base b is that gcd
0
(
mod p
)
0
(
mod p
)
for all i with 0
i
s
(
b
,
n
) =
1.
An odd integer n that satisfies the congruences in the definition of strong pseudo-
prime for some b is said to pass the strong probable prime test to the base b .This
test is given in Algorithm 6.1.
According to the previous discussion, if n passes the strong probable prime test,
then n is either prime or a strong pseudoprime to the base b . If, on the contrary, n fails
the test to base b , then we know from Theorem 6.6 that n is composite and, in this
case, b is said to be a witness (also called a strong witness or a Miller-Rabin witness )
to the compositeness of n , meaning that b serves to guarantee the compositeness of
Search WWH ::




Custom Search