The eight functions
AERRandFloatAERRandFloatRangeAERRandDoubleAERRandDoubleRangeAERRandGenFloatAERRandGenFloatRangeAERRandGenDoubleAERRandGenDoubleRange
provided by the aer/rand.h module use a method of obtaining floating-point values from integers that is known to introduce slight distribution-related bias. See this paper: https://hal.archives-ouvertes.fr/hal-02427338/file/fpnglib_iccs.pdf.
The eight functions
AERRandFloatAERRandFloatRangeAERRandDoubleAERRandDoubleRangeAERRandGenFloatAERRandGenFloatRangeAERRandGenDoubleAERRandGenDoubleRangeprovided by the
aer/rand.hmodule use a method of obtaining floating-point values from integers that is known to introduce slight distribution-related bias. See this paper: https://hal.archives-ouvertes.fr/hal-02427338/file/fpnglib_iccs.pdf.