Mbt.Maketype +'a t = private | Leaf| Node of int * 'a t * 'a M.measurable * 'a t * M.measureNode(size, l, x, r) where:
size is the number of elements in the tree,l is the left sub-treex is a user-defined valuer is the right sub-tree.Trees are guaranteed balanced by construction, the depth of all branches is O(log size).
Type of balanced trees
val leaf : 'a tLeaf constructor, the empty tree
val node : 'a t -> 'a M.measurable -> 'a t -> 'a tSmart Node constructor, ensuring that the resulting tree is balanced and has the appropriate size.
Cost of node l x r is expected to be O(log |size l - size r|) amortized, i.e proportional to the logarithm of the disbalance. In particular, if l and r are similarly-sized, it operates in constant time on average. NOT PROVEN
User-values can be moved in different subtrees of the result, but the ordering is preserved (so data stay correct if the operation applied on values is associative or the relation expected between them is transitive).
val size : 'a t -> intAccessor to the size
Concatenate two trees. Cost of join l r is O(log (min size l size r)). NOT PROVEN
val rank : int -> 'a t -> 'a M.measurableReturn the n'th node in tree order