Expect to underflow to 0 and overflow to biggest value, but results below seem opposite? I train ML and had nan from dividing by big # in gradients julia> 1f0/2^64 Inf32 julia> 1f0/(2^64) Inf32 julia> 1f0*(2^64) 0.0f0 julia> 1/(2^64) Inf julia> 1.0*(2^64) 0.0

Weird floating point under/overflow behavior

mbauman March 31, 2024, 1:26am 2

This isn’t about floating point — it’s the integers! Check the value of 2^64.

4 Likes

Topic		Replies	Views
Should large exponentials not yield Inf? General Usage question , numerics , integer-overflow	15	1355	May 23, 2022
Integer power of large Integer returning very wrong results General Usage base	7	528	May 15, 2023
Division of Powers of Ten is wrong. Why? New to Julia numbers	4	1966	June 28, 2019
Exponent operator failing for large exponents General Usage	11	373	October 5, 2022
Negative integer exponent General Usage question	4	118	October 3, 2024