Math.log1p()

x

A number greater than or equal to -1.

The natural logarithm (base e) of x + 1. If x is -1, returns -Infinity. If x < -1, returns NaN.

For very small values of x, adding 1 can reduce or eliminate precision. The double floats used in JS give you about 15 digits of precision. 1 + 1e-15 = 1.000000000000001, but 1 + 1e-16 = 1.000000000000000 and therefore exactly 1.0 in that arithmetic, because digits past 15 are rounded off.

When you calculate log(1 + x), where x is a small positive number, you should get an answer very close to x because: $\lim_{x \to 0} \frac{\log(1+x)}{x} = 1$. If you calculate Math.log(1 + 1.1111111111e-15), you should get an answer close to 1.1111111111e-15. Instead, you will end up taking the logarithm of 1.00000000000000111022 (the roundoff is in binary, so sometimes it gets ugly), and get the answer 1.11022…e-15, with only 3 correct digits. If you calculate Math.log1p(1.1111111111e-15) instead, you will get a much more accurate answer, 1.1111111110999995e-15, with 15 correct digits of precision (actually 16 in this case).

If the value of x is less than -1, the return value is always NaN.

Because log1p() is a static method of Math, you always use it as Math.log1p(), rather than as a method of a Math object you created (Math is not a constructor).

• Polyfill of Math.log1p in core-js
• Math.exp()
• Math.log()
• Math.expm1()
• Math.log10()
• Math.log2()
• Math.pow()