Abstract. Haskell is a functional language, with features such as higher order functions and lazy evaluation, which allow succinct programs. These high-level features present many challenges for optimising compilers. We report practical experiments using novel variants of supercompilation, with special attention to let bindings and the generalisation technique. 1

This thesis describes techniques for transforming and analysing functional programs. We operate on a core language, to which Haskell programs can be reduced. We present a range of techniques, all of which have been implemented and evaluated. We make programs shorter by defining a library which abstracts over common data traversal patterns, removing boilerplate code. This library only supports traversals having value-specific behaviour for one type, allowing a simpler programming model. Our library allows concise expression of traversals with competitive performance. We make programs faster by applying a variant of supercompilation. As a result of practical experiments, we have identified modifications to the standard supercompilation techniques – particularly with respect to let bindings and the generalisation technique. We make programs safer by automatically checking for potential patternmatch errors. We define a transformation that takes a higher-order program

Abstract. Haskell is a functional language, with features such as higher order functions and lazy evaluation, which allow succinct programs. These high-level features are difficult for fast execution, but GHC is a mature and widely used optimising compiler. This paper presents a wholeprogram approach to optimisation, which produces speed improvements of between 10 % and 60 % when used with GHC, on eight benchmarks. 1