Same issue here, worked on my machine and failed for 450010, and thanks to you I changed :math.pow(n,3) |> round to nnn
I guess this is part of the difficulty : you have to know every nook and cranny of the language you use :)

It is - I think - test 109:

Test 109
  1) test sum_fracts (SumfractsTest)
     fixture:10
     Assertion with == failed
     code: Sumfracts.sum_fracts(u) == ans
     lhs:  {239033687673439875, 44847850151521494}
     rhs:  {79677895891146625, 14949283383840498}

You can see that lhs and rhs are different. Can't you print the input and your ouput?

Isn'it test 109 instead of 111 where you are stuck?

As you could have seen I slightly modified the description and give a hint in each language where it is helpful. It was one of my first kata, now I'm using some kind of "fuzzy assert" where I test if abs(actual - expected) is less than a given tolerance.
Thanks for your post.

When rounding doesn't fit try truncating since it seems that your value is a bit too high.
I added something about that in "Your solution" for Elixir and you could have a look at it.
Thanks for your post.
I answer late but I have just been notified, CW has problems!

Right, forgot it, fixed. Thanks!

Good work!

Maybe you don't need big decimals?

Did you solve the kata in Elixir? As often as possible avoid math.pow in many languages which is not precise enough for big numbers.