Data.IntMap vs Data.Sequence in Haskell (Haskell Collections)

Question

What is the main difference between Data.Sequence / Seq a and Data.IntMap / IntMap a in Haskell? From everything I've read, they both offer a fast cons function, fast random access, and fast random update. What's the difference?

For context, I need all of these (fast cons, fast random access, and fast random update).

Which should I choose?

What implementation do you have for cons for IntMap? IntMap indexes don't "shift over." — Louis Wasserman
– Louis Wasserman, Commented Aug 7, 2021 at 13:36

dfeuer · Accepted Answer · 2021-08-08 17:46:14Z

3

A Seq stores something for every index between 0 and some number. An IntMap stores something for some but not all Int values. This is a big conceptual difference with obvious consequences for space usage—if your data are sparse, you probably don't want a dense representation. Cons and snoc are somewhat cheaper for Seq than for IntMap, I'd expect, though they have different meanings, while random access is significantly slower. It sounds like you might be interested in what Okasaki calls a "random access list", but between Seq and IntMap, you'd best analyze your specific algorithm and benchmark.

edited Aug 8, 2021 at 17:46

answered Aug 7, 2021 at 5:26

dfeuer

48.8k5 gold badges69 silver badges171 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Carl Over a year ago

Random access into Seq is going to be somewhat slower than IntMap on most occasions, but it's still sub-linear. That's not nothing.

dfeuer Over a year ago

@Carl, for sure. Louis Wasserman's point that cons and snoc do different things is probably more important in the end.

Collectives™ on Stack Overflow

Data.IntMap vs Data.Sequence in Haskell (Haskell Collections)

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related