NaN detection and debugging utilities. More...

#include <cmath>
#include <iostream>
#include <mpi.h>
#include <vector>

Include dependency graph for nancheck.hpp:

Macros
#define	CHECK_AND_ABORT_IF_NAN(value)

#define	CHECK_AND_ABORT_IF_NANS(vec)

Functions
template<typename T >
bool	pfc::utils::hasNaNs (const std::vector< T > &vec)

template<typename T >
void	pfc::utils::abortIfNaN (T value, const char *filename, int line)

template<typename T >
void	pfc::utils::abortIfNaNs (const std::vector< T > &vec, const char *filename, int line)

Detailed Description

NaN detection and debugging utilities.

This header provides macros and functions for detecting NaN (Not-a-Number) values in floating-point computations, useful for debugging numerical issues.

Key features:

CHECK_AND_ABORT_IF_NAN(value): Macro to check single values
CHECK_AND_ABORT_IF_VECTOR_HAS_NAN(vector): Macro to check entire vectors
Enabled with NAN_CHECK_ENABLED preprocessor flag
MPI-aware: Reports rank where NaN was detected

NaN checks are disabled by default for performance. Enable with: cmake -DNAN_CHECK_ENABLED=ON

#include <openpfc/utils/nancheck.hpp>
 
double result = compute_something();
CHECK_AND_ABORT_IF_NAN(result);  // Aborts if NaN detected

See also: utils.hpp for other utility functions

Author: OpenPFC Development Team

Date: 2025

Function Documentation

◆ abortIfNaNs()

template<typename T >

void pfc::utils::abortIfNaNs	(	const std::vector< T > &	vec,
		const char *	filename,
		int	line
	)

Checks if there are any NaNs in a vector of floats and aborts the MPI application if NaNs are detected. Prints an error message indicating the process rank where NaNs were found before aborting.

Parameters

vec	The vector of floats to check.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ hasNaNs()

template<typename T >

bool pfc::utils::hasNaNs ( const std::vector< T > & vec )

Checks if there are any NaNs in a vector of floats.