cmp.h File Reference

Floating-point comparison library for precise and robust comparisons. More...

#include <float.h>
#include <math.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdlib.h>

Include dependency graph for cmp.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	cmp_config_t
	Default configuration for each floating-point type. More...

Enumerations
enum	cmp_mode_t { CMP_ABSOLUTE , CMP_RELATIVE , CMP_ULPS , CMP_COMBINED }
	Comparison modes for floating-point numbers. More...

Detailed Description

Floating-point comparison library for precise and robust comparisons.

This library provides functions and macros for comparing floating-point numbers using different modes: relative epsilon, absolute epsilon, ULPS, and combined.

Comparison Mode Recommendations:

• CMP_ABSOLUTE: Use when you know the expected magnitude of differences (e.g., coordinates in a fixed grid). Epsilon should be based on your application's tolerance for absolute error.
• CMP_RELATIVE: Use for general scientific computing where values can vary widely in magnitude. Epsilon typically ranges from 1e-6 to 1e-15 depending on precision requirements.
• CMP_ULPS: Use for bit-level comparisons, testing numerical algorithms, or when you need to ensure results are identical to a reference implementation. ULPS of 1-4 is typical for most applications.
• CMP_COMBINED: Use when you need robustness across both very small and very large values. Combines the safety of absolute comparison near zero with the flexibility of relative comparison for larger values.

Typical Epsilon Values:

• Single precision (float): 1e-6 to 1e-7
• Double precision (double): 1e-12 to 1e-15
• Long double: 1e-15 to 1e-18

Author

[Your Name]

Date

September 2025

Definition in file cmp.h.

Enumeration Type Documentation

cmp_mode_t

enum cmp_mode_t

Comparison modes for floating-point numbers.

Enumerator
CMP_ABSOLUTE	Absolute epsilon: |a - b| ≤ ε. Best for fixed-scale data.
CMP_RELATIVE	Relative epsilon: |a - b| ≤ ε × max(|a|, |b|). Best for scientific computing.
CMP_ULPS	Units in Last Place: bit-wise difference ≤ N ULPS. Best for bit-exact comparisons.
CMP_COMBINED	Combined: |a - b| ≤ ε OR |a - b| ≤ ε × max(|a|, |b|). Most robust.

Definition at line 50 of file cmp.h.