Arbitrary Precision (AP)

Cross-platform and cross-standard header-only arbitrary precision arithmetic library. Currently it offers integer types that behave as if they are built-in int, but able to store arbitrarily huge values.

Usage

1. Clone the repo.
2. include ap.hpp header.
3. Use either ap_int<BITSIZE> or ap_uint<BITSIZE> template type, parametrized with size in bits.

Interface and features

Two types are available out of the box: </br> ap_int<BITSIZE> - signed integer of size BITSIZE with two's complement representation.</br> ap_uint<BITSIZE> - unsigned integer of size BITSIZE.</br> Both behave and offer interface of the basic integer type (without advanced algorithms).</br>

• AP integer can be initialized from any built-in integer type or any other AP integer type.</br>
• AP integer can be initialized from binary, octal, decimal and uppercase hexadecimal strings. Base determined automatically, if it is set to 0.</br>
• Custom digits may be used to represent integers, see parameters set() and str() function.</br>

Operators

All arithmetic, comparison and bitwise operators are available (as for int).</br> The behavior is as follows:

• BiTSIZE aligned to size of AP_WORD.
• comparison, increment, decrement, sign change, addition, subtraction, multiplication behave as expected and return the result of the corresponding operation.
• bitwise operations mock two's complement behavior.
• Division by zero triggers native system behavior.
• Right shift by count of bytes greater than BITSIZE yields 0.
• Left shift by count of bytes greater than BITSIZE yields -1 for negative values, 0 in the other case.
• Output to ostream is base-aware, so it checks for iso_base hex, dec, oct flags.
• Hexadecimal strings are always uppercase, be it initialization or output. You can specify string "0123456789abcdef" if you want lowercase.

Note: For those who are concerned, it is possible to switch compilation approach and compile .cpp files separately.</br> Simply define AP_USE_SOURCES before including ap.hpp, or via compiler options. However, you will have to add .cpp files to your project by yourself.</br> Another option is to use different types for computation and storage of long integer, define AP_WORD and AP_DWORD with unsigned types of your choice, e.g.:</br> -DAP_WORD=unsigned\ char</br> -DAP_DWORD=unsigned\ short</br> Note that AP_DWORD must be exactly twice as big as AP_WORD, both must be unsigned.</br>