Multiprecision Support

O2scl

Some O₂scl classes support floating-point types beyond double. Generally, these classes support both long double and the boost::multiprecision types. (Though some of the boost multiprecision types require additional libraries, such as GMP.)

AWS 3/7/20: I don’t use the boost-supplied std::abs() and std::isfinite() functions for multiprecision types because some HPC compilers don’t have recent versions of boost installed. I have currently defined new functions o2scl::o2abs(), o2scl::o2isfinite(), and o2scl::o2hypot() which work with both long double and the boost::multiprecision types.

Testing with multiprecision types is enabled by defining O2SCL_LD_TYPES at the time the configure script is run.

List of classes which support multiprecision types:

• Numerical differentiation: deriv_gsl

• Function approximation: cheb_approx_tl

• Root-finding: root_brent_gsl and root_toms748

• String to double conversion: o2scl::dtos()

• Integration: inte_gauss56_cern, inte_gauss_cern, inte_adapt_cern, inte_kronrod_boost, inte_tanh_sinh_boost, inte_exp_sinh_boost, and inte_sinh_sinh_boost.