lca: O(log h) Online Lowest Common Ancestor Search
This package provides a reference implementation of my skew binary random access algorithm for performing an online lowest common ancestor in logarithmic time without preprocessing. This improves the previous known asymptotic bound for this problem from O(h) to O(log h), where h is the height of the tree. Mostly importantly this bound is completely independent of the width or overall size of the tree, enabling you to calculate lowest common ancestors in a distributed fashion with good locality.
While algorithms exist that that provide O(1) query time, they all require O(n) preprocessing, where n is the size of the entire tree, and so are less suitable for LCA search in areas such as revision control where the tree is constantly updated, or distributed computing where the tree may be too large to fit in any one computer’s memory.
Slides are available as Purely Functional Data Structures for On-Line LCA
Contributions and bug reports are welcome!
Please feel free to contact me through github or on the #haskell IRC channel on irc.freenode.net.
lcanow always exports
lengthfunctions that are always specialized to
length :: Path a -> Int null :: Path a -> Bool
Previously, these specialized versions of
lengthwere only exported with GHC 7.8 or older, and for more recent GHCs, the
Data.Foldableversions were exported instead. The exported API is now uniform across all supported versions of GHC.
The build-type has been changed from
Simple. To achieve this, the
docteststest suite has been removed in favor of using
cabal-docspecto run the doctests.
- Fix the build with GHC 8.4.
cabal-doctestfor the test suite.
- Updated to build without warnings in the wake of GHC 7.10.
- Use (and re-export) the new overloaded
lengthfrom Prelude on GHC 7.10+
mlcato parameterize them by monoid homomorphisms. This permits cheaper summaries to be calculated over the dropped path, when only a portion of the information in the path is required.
- Fixed a bug in path reconstruction
- Improved documentation to also note that this package also provides an improvement in the online version of the level ancestor problem.
- Added README
- Better haddock coverage
- Added links to documentation