"NaN" stands for "not a number". "Nan" is produced if a floating point operation has some input parameters that cause the operation to produce some undefined result. For example, 0.0 divided by 0.0 is arithmetically undefined. Taking the square root of a negative number is also undefined.