Transducers.jl: Efficient transducers for Julia
Transducers.jl provides composable algorithms on "sequence" of inputs. They are called transducers, first introduced in Clojure language by Rich Hickey.
Using transducers is quite straightforward, especially if you already know similar concepts in iterator libraries:
using Transducers 1:40 |> Partition(7) |> Filter(x -> prod(x) % 11 == 0) |> Cat() |> Scan(+) |> sum
However, the protocol used for the transducers is quite different from
iterators and results in a better performance for complex
compositions. Furthermore, some transducers support parallel
execution. If a transducer is composed of such transducers, it can be
automatically re-used both in sequential (
foldl etc.) and parallel
reduce etc.) contexts.
See more in the documentation.
If you are interested in parallel programming in general, see also: A quick introduction to data parallelism in Julia
using Pkg Pkg.add("Transducers")
Following packages are supported by Transducers.jl. In particular, they rely on the Transducers.jl protocol to support multi-threading, multi-processing, and GPU-based parallelism.
- Folds.jl implements parallelized
Base-like API based on Transducers.jl. This package can be used without knowing anything about transducers.
- FLoops.jl provides
for-loop syntax for using the loop executed by the Transducers.jl protocol.
- BangBang.jl implements
mutate-or-widen API. This is the foundation of
collect-like functions. Functions such as
union!!, etc. are useful as a reducing function.
- InitialValues.jl provides a framework for initial/identity element of folds.
- MicroCollections.jl provides empty and singleton collections (arrays, dicts and sets). They are useful when writing transducers and reducing functions that construct a data collection.