Question

I was reading a blog post on msdn about iterators which talks about about how Concat has O(m^2) performance where m is the length of the first IEnumerable. One of the comments, by richard_deeming on the second page, provides some sample code which he says is much faster. I don’t really understand why it’s faster and was hoping someone could explain it to me.

Thanks.

Answer 1

He’s simply saying that instead of using Concat to create an iterator which is actually equivalent to creating an iterator over:

...(((a+b)+c)+d)...

which is caused by:

for (int i = 0; i < length; ++i)
    ones = ones.Concat(list); 

create a list of iterators you need and return each of those iterators you created previously.

This way you’re not ending up with a lot of stacked iterators in the first collection of elements.

Also it’s worth mentioning that the claim about O(m^2) is not “really right”. It’s true in this specific case, but this is like saying + is O(m^2) when you’re calculating (((a+b)+c)+d)... case. It’s the specific usage pattern that makes it O(m^2).