Re: [boost] [Review Request] Multiprecision Arithmetic Library

Subject: Re: [boost] [Review Request] Multiprecision Arithmetic Library
From: Christopher Kormanyos (e_float_at_[hidden])
Date: 2012-04-02 16:49:19

>> All the floating point types, have full std lib support (cos sin exp, pow

>> etc), as well as full interoperability with Boost.Math.
>> 
>> What is the precision of the following operations: *, /, +, -, sqrt
>> for the cpp_dec_float type?

> Good question, we don't really guarantee any at present. 
> In practice they're *usually* accurate to 1eps,
> but they're not intended to compete with MPFR's guarantees. 
> But Chris would know more?

My basic operations (add, sub, mul, div) are on-par or faster
for a variety of reasons. For example, I don't allocate---just use
std::array "on the stack".

Those GMP and MPFR guys can not program in a professionally
adherent fashion. But they sure do have efficient algorithms!
Their square root, exponential, logarithmic and trigonometric
functions beat mine hands down. I'd like to get a hold of their
algorithms and program them in a readable, portable fashion.
I think they use, for example, binary splitting for trigonometric functions,
whereas I use Taylor.

>> Any limitations on the maximum/minimum value of exponent?

> Must fit in a long long.  That still gives some pretty big numbers ;-)

In other words, approx. 10^(10^19) for 64-bit long long.
The Python language MP stuff has unlimited exponents.
I forgot what that thing is called, but it's another cool tool.

Best regards, Chris.