The ADPfusion package

[Tags: bsd3, library, program]

ADPfusion combines stream-fusion (using the stream interface provided by the vector library) and type-level programming to provide highly efficient dynamic programming combinators.

This library is described in:

Christian Hoener zu Siederdissen
Sneaking Around ConcatMap: Efficient Combinators for Dynamic Programming
2012. Proceedings of the 17th ACM SIGPLAN international conference on Functional programming preprint:

From the programmers' viewpoint, ADPfusion behaves very much like the original ADP implementation developed by Robert Giegerich and colleagues, though both combinator semantics and backtracking are different.

The library internals, however, are designed not only to speed up ADP by a large margin (which this library does), but also to provide further runtime improvements by allowing the programmer to switch over to other kinds of data structures with better time and space behaviour. Most importantly, dynamic programming tables can be strict, removing indirections present in lazy, boxed tables.

As a simple benchmark, consider the Nussinov78 algorithm which translates to three nested for loops (for C). In the figure, four different approaches are compared using inputs with size 100 characters to 1000 characters in increments of 100 characters. C is an implementation (.C directory) in C using gcc -O3. ADP is the original ADP approach (see link above), while GAPC uses the GAP language (

Performance comparison figure:

Please note that actual performance will depend much on table layout and data structures accessed during calculations, but in general performance is very good: close to C and better than other high-level approaches (that I know of).

Even complex ADP code tends to be completely optimized to loops that use only unboxed variables (Int# and others, indexIntArray# and others).

Completely novel (compared to ADP), is the idea of allowing efficient monadic combinators. This facilitates writing code that performs backtracking, or samples structures stochastically, among others things.

Two algorithms from the realm of computational biology are provided as examples on how to write dynamic programming algorithms using this library: and


Dependenciesbase (==4.*), deepseq (>=1.3), ghc-prim, primitive (>=0.5), PrimitiveArray (>=0.5.3), QuickCheck (>=2.5), repa (>=3.2), strict (>=0.3.2), template-haskell, transformers, vector (>=0.10)
CopyrightChristian Hoener zu Siederdissen, 2011-2013
AuthorChristian Hoener zu Siederdissen, 2011-2013
CategoryAlgorithms, Data Structures, Bioinformatics
Home page
Source repositoryhead: git clone git://
UploadedMon Feb 3 21:22:56 UTC 2014 by ChristianHoener
Downloads1784 total (161 in last 30 days)
StatusDocs available [build log]
Successful builds reported [all 1 reports]




Maintainers' corner

For package maintainers and hackage trustees