Licensed under ISC license.
A binary tree with smart constructors that ensure the resulting tree is balanced.
This data structure can be used as a primitive on top of which one can easily build balanced data structures, including but not limited to binary search trees.
For instance, implementing stdlib-like Set/Map is trivial and suffers only a ~5 % overhead (and one gains a O(1) length/cardinal operation).
These two modules can be used as a drop-in replacement for Map and Set.
The performance characteristics are slightly different: cardinal is now O(1),
some operations use that as a shortcut (compare, subset, ...).
In addition, the representation is exposed (the internal structure of the tree
can be pattern matched). It is protected by a private modifier, such that
invariants cannot be broken. However, custom operations are much easier to
implement (e.g. rank to access the n'th element, which enables uniform
sampling in O(log n)).
A generic algorithm that turns a directed graph intro a tree. It finds where binding nodes should be introduced to make the resulting tree readable. The idea is described in this blog post.
For instance, this is useful to print cyclic values (see Cmon).
Dbseq is a small data structure that offers operations halfway between a list and an immutable array. Most operations have a logarithmic cost. In practice, it is a log with base 4 and small constant factors.
The name comes from the fact that the data structure is particularly suitable to associate metadata to variables in De-Bruijn notation when traversing terms.
This data structure allows efficient implementation of text markers for text editors (see Emacs Markers).
More generally it allows to track the movement of objects on a line where chunks are added and removed, with queries O(log n) amortized time.
Finally, it is persistent so you easily compare markers movement between different revisions.
See Order-maintenance problem for a detailed description of what this intent to solve.
Main algorithm follows the amortized solution from "Two Simplified Algorithms for Maintaining Order in a List", Michael A. Bender, Richard Cole, Erik D. Demaine, Martín Farach-Colton, and Jack Zito.
A managed implementation provide finer integration with OCaml GC to collect items that are no longer reachable via the public API.
An implementation of Maxrects packing algorithm in 2D. This algorithm try to pack a maximum number of 2d boxes inside a 2d rectangle.
See Even More Rectangle Bin Packing
Useful for generating spritesheets, texture atlases, etc.
An implementation of double-double arithmetic.
Code is translated from DD by Martin Davis. See tsusiatsoftware for more information.
An implementation of the HyperLogLog probabilistic cardinality estimator. See HyperLogLog.
An implementation of "A Fast, Minimal Memory, Consistent Hash Algorithm" from John Lamping and Eric Veach.
Hashtables indexing OCaml values by their physical indentities. A proof-of-concept, playing with the GC in tricky ways.
Its main purpose is to efficiently observe sharing, detect cycles, etc, in arbitrary OCaml values without having to stop and stay out of the OCaml runtime.
Can be used to experiment and learn about the GC but do expect bugs and don't expect any kind of compatibility with future OCaml versions. (Would be nice to have proper upstream support for such feature though!)
This library converts e-NFA (including NFA and DFA) to regular expressions.
Unfortunately the regular expression is often of exponential size, unless you extend the language to allow sharing sub-expressions (for instance with let binders).
This library defines a few strongly typed idioms that are sometimes useful in OCaml codebase:
- type-level equality and ordering
- unhabitated type
- an encoding of type-level naturals
- finite sets (the set of numbers less than a given constant)
An implementation of the algorithm desribed in Fast brief practical DFA minimization by Valmari et al.
The tests and some fixes come from WalkerCodeRanger/dfaMinimizationComparison, thanks!
An implementation of A Simple, Fast Dominance Algorithm by Keith D. Cooper, Timothy J. Harvey, and Ken Kennedy.
A congruence closure algorithm, inspired by Fast congruence closure and extensions by Robert Nieuwenhuis and Albert Oliveras. Support backtracking and interpretation of equivalence classes to OCaml value.