Suitability of Elixir to math-oriented apps

[peerreynders]

Update: the ets code is using a bulk insert (a list of tuples) - giving it the apparent edge over the map code. Replacing individual Map.put/3 calls with a consolidated maps:merge/2 call improved matters for the map as a cache case.

With the Fib example ets outperformed a map significantly - but still didn’t quite catch up to the last call optimized version. The results seemed to imply that ets tables as a BEAM artifact are optimized enough to have better lookup and write performance - though that may be a function of size.

For the process dictionary approach see:
Improving a slow solution for Longest Collatz Sequence