Boost C++ Libraries

...one of the most highly regarded and expertly designed C++ library projects in the world. Herb Sutter and Andrei Alexandrescu, C++ Coding Standards

This is the documentation for an old version of Boost. Click here to view this page for the latest version.
PrevUpHomeNext

log1p

#include <boost/math/special_functions/log1p.hpp>
namespace boost{ namespace math{

template <class T>
calculated-result-type log1p(T x);

template <class T, class Policy>
calculated-result-type log1p(T x, const Policy&);

}} // namespaces

Returns the natural logarithm of x+1.

The return type of this function is computed using the result type calculation rules: the return is double when x is an integer type and T otherwise.

The final Policy argument is optional and can be used to control the behaviour of the function: how it handles errors, what level of precision to use etc. Refer to the policy documentation for more details.

There are many situations where it is desirable to compute log(x+1). However, for small x then x+1 suffers from catastrophic cancellation errors so that x+1 == 1 and log(x+1) == 0, when in fact for very small x, the best approximation to log(x+1) would be x. log1p calculates the best approximation to log(1+x) using a Taylor series expansion for accuracy (less than 2ɛ). Alternatively note that there are faster methods available, for example using the equivalence:

log(1+x) == (log(1+x) * x) / ((1+x) - 1)

However, experience has shown that these methods tend to fail quite spectacularly once the compiler's optimizations are turned on, consequently they are used only when known not to break with a particular compiler. In contrast, the series expansion method seems to be reasonably immune to optimizer-induced errors.

Finally when macro BOOST_HAS_LOG1P is defined then the float/double/long double specializations of this template simply forward to the platform's native (POSIX) implementation of this function.

The following graph illustrates the behaviour of log1p:

Accuracy

For built in floating point types log1p should have approximately 1 machine epsilon accuracy.

Table 8.81. Error rates for log1p

GNU C++ version 7.1.0
linux
long double

GNU C++ version 7.1.0
linux
double

Sun compiler version 0x5150
Sun Solaris
long double

Microsoft Visual C++ version 14.1
Win32
double

Random test data

Max = 0.818ε (Mean = 0.227ε)

(<cmath>: Max = 0.818ε (Mean = 0.227ε))
(<math.h>: Max = 0.818ε (Mean = 0.227ε))

Max = 0.846ε (Mean = 0.153ε)

(Rmath 3.2.3: Max = 0.846ε (Mean = 0.153ε))

Max = 2.3ε (Mean = 0.66ε)

(<math.h>: Max = 0.818ε (Mean = 0.249ε))

Max = 0.509ε (Mean = 0.057ε)

(<math.h>: Max = 0.509ε (Mean = 0.057ε))


Testing

A mixture of spot test sanity checks, and random high precision test values calculated using NTL::RR at 1000-bit precision.


PrevUpHomeNext