I am running some computer simulations that eventually fail due to accumulating rounding errors. I need to improve the precision and would like a decent free library that provides higher or arbitrary precision. There seem to be quite a few out there and since I need to re-work several thousand lines of code I need to be sure of getting one that's reliable, fast, has a transparent syntax and provides all the functions I need.
GMP seems popular but, after reading the manual, I unfortunately have to rule it out because it has no logarithm function and I do not like the look of the interface.
Can anyone help?