# Output distribution of rand(Float32) and rand(Float64), thread 2

**URL:** <https://discourse.julialang.org/t/output-distribution-of-rand-float32-and-rand-float64-thread-2/105184>\
**Category:** Internals & Design\
**Tags:** float, random\
**Created:** [October 19, 2023, 1:10pm UTC](https://discourse.julialang.org/t/output-distribution-of-rand-float32-and-rand-float64-thread-2/105184 "2023-10-19T13:10:07Z")\
**Posts on this page:** 1\
**Showing post:** 45

<div class="post-metadata">

**Author:** ![danielwe](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/danielwe/32/35657_2.png) [@danielwe](https://discourse.julialang.org/u/danielwe)\
**Post date:** [October 23, 2023, 10:35pm UTC](https://discourse.julialang.org/t/output-distribution-of-rand-float32-and-rand-float64-thread-2/105184/45 "2023-10-23T22:35:06Z")

</div>

> [@Sukera](#):
>
> don’t you get the same “overweighing” on the other end of the interval

No, that’s not the case now and won’t be the case in an improved implementation:

```julia-repl
julia> function test_rand(b::Bernoulli, N)
           hits = 0
           misses = 0
           for i in 1:N
              sample(b) ? (hits += 1) : (misses += 1)
           end
           @show N, hits, hits / N, b.p
           @show N, misses, misses / N, 1 - b.p
           return nothing
       end
test_rand (generic function with 1 method)

julia> test_rand(Bernoulli(prevfloat(1f0)), typemax(Int32))
(N, hits, hits / N, b.p) = (2147483647, 2147483529, 0.9999999450519681, 0.99999994f0)
(N, misses, misses / N, 1 - b.p) = (2147483647, 118, 5.4948031927900404e-8, 5.9604645f-8)

```

The current issue on the low end is that the sample space of `rand` is much too sparse to reflect the precision with which one can specify a small frequency `p::Float32`. On the upper end, the available precision for `p` is lower but perfectly matched by the sample space, so the Bernoulli random variable is weighted according to specification.

---

_[View the full topic](https://discourse.julialang.org/t/output-distribution-of-rand-float32-and-rand-float64-thread-2/105184)._