Calcium

_images/ca2.svg

Calcium (pronounced “kalkium”) is a C library for exact computation with real and complex numbers (both algebraic and transcendental), presently in early development.

Calcium is free software (LGPL). It depends on GMP, MPFR, Flint, Antic and Arb.

Source code: https://github.com/fredrik-johansson/calcium

FAQ

I thought exact calculation with real numbers isn’t possible. Isn’t x = 0 undecidable?

In general, yes. In practice, much of calculus and number theory only depends on numbers that are simple combinations of well-known elementary and special functions, and there are heuristics that work quite well for deciding predicates about such numbers. Calcium will be able to give a definitive answer at least in simple cases (for example, proving \(16 \operatorname{atan}(\tfrac{1}{5}) - 4 \operatorname{atan}(\tfrac{1}{239}) = \pi\) or \(\sqrt{5+2\sqrt{6}} = \sqrt{2}+\sqrt{3}\)), and will simply answer “Unknown” when its heuristics are not powerful enough.

Isn’t it going to be horribly slow?

It will definitely be too slow to replace floating-point numbers for 99.9% of scientific computing. The target is symbolic and algebraic computation. A big factor in making this kind of library practical is that Flint now has extremely fast multivariate polynomial arithmetic which can be used to implement multivariate transcendental number fields with adequate performance.

Calcium will generally be much slower than arbitrary-precision ball arithmetic. It will often make sense for users to first try a numerical evaluation with Arb, and fall back on an exact calculation with Calcium only if that fails (typically because an exact comparison is needed).

Why this C library? Why not use a computer algebra system?

Calculating with constant values is only a small part of what a computer algebra system has to do, but it is actually one of the most complex parts. Calcium is intended to take some of the complexity out of the task of building a computer algebra system by offering a black-box solution to the “constant problem”. This solution will not be perfect, but it will be “good enough” for many applications. C is a pain, but it is the best choice for building a library that will not be tied to a particular computer algebra system or programming language ecosystem. Initially, the goal of Calcium will be to implement high-performance basic data structures in C; this will enable future experimentation with algorithms in both C and in high-level languages using wrappers.

Credits and references

Indices and tables