Obtaining the number of mantissa bits in use by an mpf

John Westwood john at westwoodsolutions.com
Fri May 14 14:28:56 CEST 2010

Hello GMP gurus!

I was wondering if you could help me with my problem. I wish to be able to 
obtain the numbers of mantissa bits in use by an mpf variable. In the manual, 
near the beginning of chapter 7, it states:

"Each variable keeps a size for the mantissa data actually in use. This means 
that if a float is exactly represented in only a few bits then only those bits 
will be used in a calculation, even if the selected precision is high."

I would to retrieve this size, but I see no function for doing so. The mpz 
variable has an mpz_sizeinbase function, but I see no equivalent for mpf.

I have tried using the mpf_get_str function, but the returned char array is 
longer than I expect. This is what I tried:

        long exponent;
        char* string = mpf_get_str(NULL, &exponent, 2, 0, delta);
        precision =  strlen(string);

I wish to know the number of bits in use, so that I can determine when to 
increase the precision of my mpf variables. At low precisions I use doubles 
and I determine the precision being used with:

        precision = static_cast<unsigned long>( fabs(log(delta) / log(2.0)) );

I cannot use this method for arbitrary precision, because GMP has no log 

My program is a fractal explorer, hence why I only wish to use adequate 
precision at a particular magnification.



More information about the gmp-discuss mailing list