Module raw 

Raw trait implementations for rug::Float and rug::Complex.

Arbitrary-Precision Floating-Point Raw Implementations

This module provides raw trait implementations for the rug backend, enabling arbitrary-precision floating-point arithmetic with configurable precision.

Purpose and Role

The primary role of this module is to implement the core raw traits (RawScalarTrait, RawRealTrait, etc.) for rug::Float and rug::Complex. These implementations provide the computational foundations that validated wrappers build upon.

Precision Model

Unlike the native f64 backend with fixed 53-bit precision, the rug backend uses const generic precision:

• Precision is specified in bits (e.g., 100, 200, 500)
• All operations preserve precision of the operands
• Precision mismatches are detected during validation

MPFR Constant Optimization

For mathematical constants, this module uses MPFR’s precomputed constants when available, providing significant performance improvements:

• rug::float::Constant::Pi - ~10x faster than computing acos(-1)
• rug::float::Constant::Log2 - ~10x faster than computing ln(2)

Memory and Performance Characteristics

• rug::Float and rug::Complex are heap-allocated, non-Copy types
• All operations support reference-based variants to minimize cloning
• Memory usage scales with precision (approximately precision/8 bytes per value)