serie of sqrt

[laurent couraud]

Hello,



First, sorry for my bad English, I am French.



I think that it have a way in that the calculus of mpz_sqrt will can be more
good than the method used actually in gmp

In fact it is the case when the user want the square root of a series of
integer that differ by a small value

Example:



Suppose that we search a value of an integer n that verify:

            a + n*h = b^2

Where a, n, h and b are all integer and h is very less than a.

I think that in this case it will be good to calculate the square root of a

mpz_sqrt(a0, a);

Then for the next value of h, we can call a function like:

            mpz_sqrt_init(rop, op, initialvalue);

Then in the example, we will call this function like this:

            mpz_sqrt_init(b, a+n*h, a0);

            a0 = b;

Because the entire part of the square root of a + n*h and the entire part
 of the square root of a + (n + 1)*h can be equal or very near, if we inject
the initial value in the Newton algorithm we probably gain the square root
with a very small number of iteration.

In addition, may be we can use in the same idea a function like:

            mpz_perfect_square_init