Futhark Library Documentation

Main libraries

lib/github.com/diku-dk/containers/array: Array related operations, like reduce_by_key.
lib/github.com/diku-dk/containers/arraymap: Array-based key-value maps.
lib/github.com/diku-dk/containers/bitset: Bitset module
lib/github.com/diku-dk/containers/hashkey: Key module.
lib/github.com/diku-dk/containers/hashmap: Static hashmaps.
lib/github.com/diku-dk/containers/hashset: Static hashsets with an unlifted type.
lib/github.com/diku-dk/containers/key: Definitions of modules that implement both the hashkey and ordkey module types for a variety of types.
lib/github.com/diku-dk/containers/list: Linked lists.
lib/github.com/diku-dk/containers/map: Abstract interface for working with key-value mappings.
lib/github.com/diku-dk/containers/opt: opt type.
lib/github.com/diku-dk/containers/ordkey: Ordered key type.
lib/github.com/diku-dk/containers/set: Abstract interface for working with sets.
lib/github.com/diku-dk/containers/slice: Types representing slices of arrays.

/prelude/ad: Definitions related to automatic differentiation.
/prelude/array: Utility functions for arrays.
/prelude/functional: Simple functional combinators.
/prelude/math: Basic mathematical modules and functions.
/prelude/prelude: The default prelude that is implicitly available in all Futhark files.
/prelude/soacs: Various Second-Order Array Combinators that are operationally parallel in a way that can be exploited by the compiler.
/prelude/zip: Transforming arrays of tuples into tuples of arrays and back again.
lib/github.com/diku-dk/cpprandom/random: Random number generation inspired by <random> in C++.
lib/github.com/diku-dk/segmented/segmented: Irregular segmented operations, like scans and reductions.
lib/github.com/diku-dk/sorts/merge_sort: Work-efficient parallel mergesort
lib/github.com/diku-dk/sorts/radix_sort: A non-comparison-based sort that sorts an array in O(k n) work and O(k log(n)) span, where k is the number of bits in each element. The library contains two variants of radix sort with different use cases. blocked_radix_sort should be used on large arrays, if the array is small then it may be the case that radix_sort is faster. radix_sort can also be much more suitable in cases where nested parallelism is utilized.