Publications

This paper presents an automatic fusion system, stream fusion, based on equational transformations, that fuses a wider range of functions than existing short-cut fusion systems. In particular, stream fusion is able to fuse zips, left folds and functions over nested lists, including list comprehensions. A distinguishing feature of the stream fusion framework is its simplicity: by transforming list functions to expose their structure, intermediate values are eliminated by general purpose compiler optimisations.

We have reimplemented the entire Haskell standard list library on top of our framework, providing stream fusion for Haskell lists. By allowing a wider range of functions to fuse, we see an increase in the number of occurrences of fusion in typical Haskell programs. We present benchmarks documenting time and space improvements.

We address the tension between software generality and performance in the domain of simulations based on Monte-Carlo methods. We simultaneously achieve generality and high performance by a novel development methodology and software architecture centred around the concept of a specialising simulator generator. Our approach combines and extends methods from functional programming, generative programming, partial evaluation, and runtime code generation. We also show how to generate parallelised simulators.

We evaluated our approach by implementing a simulator for advanced forms of polymerisation kinetics. We achieved unprecedented performance, making Monte-Carlo methods practically useful in an area that was previously dominated by deterministic PDE solvers. This is of high practical relevance, as Monte-Carlo simulations can provide detailed microscopic information that cannot be obtained with deterministic solvers.

A novel, parallelised approach to Monte Carlo simulations for the computation of full molecular weight distributions arising from complex polymerisation reactions is presented. The methods developed enable rapid simulation of the outcomes of reactions that previously required many hours or even days. Some systems (e.g. those arising in the authors’ own research into star-polymer formation processes) which once required circa 8 hours of simulation time (using the commercial program package PREDICI) can now be simulated in a few minutes on a parallel computer.

In addition to this significant speed improvement, new insights have been developed with regard to the Monte Carlo process in at least four key areas: (1) an insufficient system size has been shown to create inaccuracies via poor representation of the most improbable events and least numerous species; (2) from a speed point of view, the method has been found to under-perform for some simple systems when compared to PREDICI but is vastly superior when applied to complex systems; (3) advanced algorithmic principles and compiler technology known to computer scientists have been used to provide speed improvements and (4) the parallelisability of the algorithm has been explored and excellent scalability been demonstrated.

For the first time, specific implementation strategies and their suitabilities are discussed. Further, the code is available, allowing other researchers to use or examine in detail the methods put forward.

We address the tension between software generality and performance in the domain of scientific and financial simulations based on Monte-Carlo methods. To this end, we present a novel software architecture, centred around the concept of a specialising simulator generator, that combines and extends methods from generative programming, partial evaluation, runtime code generation, and dynamic code loading. The core tenet is that, given a fixed simulator configuration, a generator in a functional language can produce low-level code that is more highly optimised than a manually implemented generic simulator. We also introduce a skeleton, or template, capturing a wide range of Monte-Carlo methods and use it to explain how to design specialising simulator generators and how to generate parallelised simulators for multi-core and distributed-memory multiprocessors.

We evaluated the practical benefits and limitations of our approach by applying it to a highly relevant problem in computational chemistry. More precisely, we used a Markov-chain Monte-Carlo method for the study of advanced forms of polymerisation kinetics. The resulting implementation executes faster than all competing software products, while at the same time also being more general. The generative architecture allows us to cover a wider range of chemical reactions and to target a wider range of high-performance architectures (such as PC clusters and SMP multiprocessors).

We show that it is possible to outperform low-level languages with functional programming in domains with very stringent performance requirements if the domain also demands generality.

The Haskell String type is notoriously inefficient. We introduce a new data type, ByteString, based on lazy lists of byte arrays, combining the speed benefits of strict arrays with lazy evaluation. Equational transformations based on term rewriting are used to deforest intermediate ByteStrings automatically. We describe novel fusion combinators with improved expressivity and performance over previous functional array fusion strategies. A library for ByteStrings is implemented, providing a purely functional interface, and approaches the speed of low-level mutable arrays in C.

Some Lisp programs, such as Emacs, but also the Linux kernel (when fully modularised) are mostly dynamic; i.e., apart from a small static core, the significant functionality is dynamically loaded. In this paper, we explore fully dynamic applications in Haskell where the static core is minimal and code is hot swappable. We demonstrate the feasibility of this architecture by two applications: Yi, an extensible editor, and Lambdabot, a plugin-based IRC robot. Benefits of the approach include hot swappable code and sophisticated application configuration and extension via embedded DSLs. We discuss both benefits in detail at the example of a novel embedded DSL for editor interfaces.

Extension languages enable users to expand the functionality of an application without touching its source code. Commonly, these languages are dynamically typed languages, such as Lisp, Python, or domain-specific languages, which support runtime plugins via dynamic loading of components. We show that Haskell can be comfortably used as a statically typed extension language, and that it can support type-safe dynamic loading of plugins using dynamic types. Moreover, we discuss how plugin support is especially useful to applications where Haskell is used as an embedded domain-specific language (EDSL). We explain how to realise type-safe plugins using dynamic types, runtime compilation, and dynamic linking, exploiting infrastructure provided by the Glasgow Haskell Compiler. We demonstrate the practicability of our approach with several applications that serve as running examples.

The quality of a programming language itself is only one component in the ability of application writers to get the job done. Programming languages can succeed or fail based on the breadth and quality of their library collection. Over the last few years, the Haskell community has risen to the task of building the library infrastructure necessary for Haskell to succeed as a programming language suitable for writing real-world applications. This on-going work, the Cabal and Hackage effort, is built on the open source model of distributed development, and have resulted in a flowering of development in the language with more code produced and reused now than at any point in the community’s history. It is easier to obtain and use Haskell code, in a wider range of environments, than ever before. This demonstration describes the infrastructure and process of Haskell development inside the Cabal/Hackage framework, including the build system, library dependency resolution, centralised publication, documentation and distribution, and how the code escapes outward into the wider software community. We survey the benefits and trade-offs in a distributed, collaborative development ecosystem and look at a proposed Haskell Platform that envisages a complete Haskell development environment, batteries included.

xmonad is a tiling window manager for the X Window system, implemented, configured and dynamically extensible in Haskell. This demonstration presents the case that software dominated by side effects can be developed with the precision and efficiency we expect from Haskell by utilising purely functional data structures, an expressive type system, extended static checking and property-based testing. In addition, we describe the use of Haskell as an application configuration and extension language.