Version 0.3.0

Changes

QOL

Adds many quality of life overloads that were missing before. Including:

• pow with floating-point exponent
• floating-point <+,-,*,/> interval and vice versa
• integer <+,-,*,/> interval and vice versa
• added isnan, behaving equivalently to isnai
• improved overloading behavior when combined with other libraries (e.g., CuTangent, CuMcCormick)

Numeric limits

The std::numeric_limits may be overloaded for custom data types. For cu::interval, it should behave like the underlying type.
E.g.,

std::numeric_limits<double>::epsilon() == std::numeric_limits<cu::interval<double>>::epsilon()

As per the standard, overloading this class for our custom data type is valid.

No formatting by default

By default we do not include the formatting specializations (for std::ostream, std::format, etc.) to allow seamless use of custom
printing behavior. If the default behavior is wanted, simply include #include <cuinterval/format.h> to get it.

Full Changelog: v0.2.1...v0.3.0

Version 0.2.1

Changes

• Fix log implementations not using the intersected variables. This may have caused incorrect results when the interval contained out of domain values. In particular, below 0 as a lower bound but a valid upper bound.
• Added icpx preset in CMakePresets.json
• Documentation fixes.

Full Changelog: v0.2.0...v0.2.1

Version 0.2.0

Full Changelog: v0.1.0...v0.2.0

Added std::numbers and std::format overloads for cu::interval.

std::numbers support:

It is allowed by the standard to overload the numbers (pi, e, ln2, etc.) and is now possible to do for cu::interval.
Just use, e.g., std::numbers::pi_v<cu::interval<double>> to get the smallest interval that still contains the real
pi value. This will not just be a point interval with std::numbers::pi_v because pi (and all the other numbers
in std::numbers) cannot be represented in IEEE-754 floating-point numbers.

The same numbers are also accessible from within the cu:: namespace.

std::format support:

Now supports using std::format with the cu::interval type. For example,

std::format("interval: {}", cu::interval<double>{1.0, 2.0});

prints: interval: [1, 2].

All the format specifiers that the underlying type T of cu::interval supports are also supported for cu::interval.
For example,

std::format("interval: {:.4f}", cu::interval<double>{1.0, 2.0});

prints: interval: [1.0000, 2.0000].

Version 0.1.0

v0.1.0

update git tag in FetchContent example