There are many, but which one to use is strongly-coupled to what operations you need to be efficient:

• if you want sparse data (most indexes don’t have anything in them), you could use a Map of index → value. This wastes space with keys, but has O(log(n)) access for more than 32 elements.

• if you have arrays of fixed size (not too large), then tuples are very fast (elem is constant-time) to read. Updating is expensive, and resizing is as well - both copy every pointer in the structure.

• if you’re converting an algorithm that uses zero-based indexing into a block of memory to simulate some other data structure (ie, a queue / a stack / a circular list), a better data structure may help. For instance, the Erlang :queue module allows amortized-constant-time pushing and popping from either end.