What's the difference between lazy mean and immediate mean in Boost accumulator?

Question

my question is as in the title. By default it's lazy mean. It seems I got wrong result. However, before I switch to using immediate mean, I want to clarify the difference. Thanks.

Answer 1

http://en.wikipedia.org/wiki/Lazy_evaluation

In the particular context of accumulators it means that in the lazy case the calculation is performed in the extractor - when you actually request the value of the mean (as opposed to being performed in the accumulator, i.e. when you record the data from which the mean is going to be calculated).