Abstract. Cost analysis of Java bytecode is complicated by its unstructured control flow, the use of an operand stack and its object-oriented programming features (like dynamic dispatching). This paper addresses these problems and develops a generic framework for the automatic cost analysis of sequential Java bytecode. Our method generates cost relations which define at compile-time the cost of programs as a function of their input data size. To the best of our knowledge, this is the first approach to the automatic cost analysis of Java bytecode. 1

Abstract. The classical approach to automatic cost analysis consists of two phases. Given a program and some measure of cost, we first produce recurrence relations (RRs) which capture the cost of our program in terms of the size of its input data. Second, we convert such RRs into closed form (i.e., without recurrences). Whereas the first phase has received considerable attention, with a number of cost analyses available for a variety of programming languages, the second phase has received comparatively little attention. In this paper we first study the features of RRs generated by automatic cost analysis and discuss why existing computer algebra systems are not appropriate for automatically obtaining closed form solutions nor upper bounds of them. Then we present, to our knowledge, the first practical framework for the fully automatic generation of reasonably accurate upper bounds of RRs originating from cost analysis of a wide range of programs. It is based on the inference of ranking functions and loop invariants and on partial evaluation. 1

Machine The semantics presented in this section is essentially Sestoft's \mark 1" abstract machine for laziness [Sestoft 1997]. In that paper, he proves his abstract machine 6 A. K. Moran and D. Sands h fx = Mg; x; S i ! h ; M; #x : S i (Lookup) h ; V; #x : S i ! h fx = V g; V; S i (Update) h ; M x; S i ! h ; M; x : S i (Unwind) h ; x:M; y : S i ! h ; M [ y = x ]; S i (Subst) h ; case M of alts ; S i ! h ; M; alts : S i (Case) h ; c j ~y; fc i ~x i N i g : S i ! h ; N j [ ~y = ~x j ]; S i (Branch) h ; let f~x = ~ Mg in N; S i ! h f~x = ~ Mg; N; S i ~x dom(;S) (Letrec) Fig. 1. The abstract machine semantics for call-by-need. semantics sound and complete with respect to Launchbury's natural semantics, and we will not repeat those proofs here. Transitions are over congurations consisting of a heap, containing bindings, the expression currently being evaluated, and a stack. The heap is a partial function from variables to terms, and denoted in an identical manner to a coll...

In this thesis we present and analyse a set of automatic source-to-source program transformations that are suitable for incorporation in optimising compilers for lazy functional languages. These transformations improve the quality of code in many different respects, such as execution time and memory usage. The transformations presented are divided in two sets: global transformations, which are performed once (or sometimes twice) during the compilation process; and a set of local transformations, which are performed before and after each of the global transformations, so that they can simplify the code before applying the global transformations and also take advantage of them afterwards. Many of the local transformations are simple, well known, and do not have major effects on their own. They become important as they interact with each other and with global transformations, sometimes in non-obvious ways. We present how and why they improve the code, and perform extensive experiments wit...

Depth-first search is the key to a wide variety of graph algorithms. In this paper we express depth-first search in a lazy functional language, obtaining a linear-time implementation. Unlike traditional imperative presentations, we use the structuring methods of functional languages to construct algorithms from individual reusable components. This style of algorithm construction turns out to be quite amenable to formal proof, which we exemplify through a calculationalstyle proof of a far from obvious strongly-connected components algorithm. Classifications: Computing Paradigms (functional programming) ; Environments, Implementations, and Experience (programming, graph algorithms). 1 Introduction The importance of depth-first search (DFS) for graph algorithms was established twenty years ago by Tarjan (1972) and Hopcroft and Tarjan (1973) in their seminal work. They demonstrated how depth-first search could be used to construct a variety of efficient graph algorithms. In practice, this...

The goal of program transformation is to improve efficiency while preserving meaning. One of the best known transformation techniques is Burstall and Darlington's unfold-fold method. Unfortunately the unfold-fold method itself guarantees neither improvement in efficiency nor total-correctness. The correctness problem for unfold-fold is an instance of a strictly more general problem: transformation by locally equivalence-preserving steps does not necessarily preserve (global) equivalence. This paper presents a condition for the total correctness of transformations on recursive programs, which, for the first time, deals with higher-order functional languages (both strict and non-strict) including lazy data structures. The main technical result is an improvement theorem which says that if the local transformation steps are guided by certain optimisation concerns (a fairly natural condition for a transformation), then correctness of the transformation follows. The improvement theorem make...

We introduce a framework for assessing the effectiveness of partial evaluators in functional logic languages. Our framework is based on properties of the rewrite system that models a functional logic program. Consequently, our assessment is independent of any specific language implementation or computing environment. We define several criteria for measuring the cost of a computation: number of steps, number of function applications, and pattern matching effort. Most importantly, we express the cost of each criterion by means of recurrence equations over algebraic data types, which can be automatically inferred from the partial evaluation process itself. In some cases, the equations can be solved by transforming their arguments from arbitrary data types to natural numbers. In other cases, it is possible to estimate the improvement of a partial evaluation by analyzing the associated cost recurrence equations.

The powerful abstraction mechanisms of functional programming languages provide the means to develop domain-specific programming languages within the language itself. Typically, this is realised by designing a set of combinators (higher-order reusable programs) for an application area, and by constructing individual applications by combining and coordinating individual combinators. This paper is concerned with a successful example of such an embedded programming language, namely Fudgets, a library of combinators for building graphical user interfaces in the lazy functional language Haskell. The Fudget library has been used to build a number of substantial applications, including a web browser and a proof editor interface to a proof checker for constructive type theory. This paper develops a semantic theory for the non-deterministic stream processors that are at the heart of the Fudget concept. The interaction of two features of stream processors makes the development of such a semantic theory problematic: (i) the sharing of computation provided by the lazy evaluation mechanism of the underlying host language, and (ii) the addition of non-deterministic choice needed to handle the natural concurrency that reactive applications entail We demonstrate that this combination of features in a higher-order functional language can be tamed to provide a tractable semantic theory and induction principles suitable for reasoning about contextual equivalence of Fudgets.

Abstract not found

interpretation based cost and termination analyzer for Java bytecode. The system receives as input a bytecode program, (a choice of) a resource of interest and tries to obtain an upper bound of the resource consumption of the program. costa provides several non-trivial notions of cost, as the consumption of the heap, the number of bytecode instructions executed and the number of calls to a specific method. Additionally, costa tries to prove termination of the bytecode program which implies the boundedness of any resource consumption. Having cost and termination together is interesting, as both analyses share most of the machinery to, respectively, infer cost upper bounds and to prove that the execution length is always finite (i.e., the program terminates). We report on experimental results which show that costa can deal with programs of realistic size and complexity, including programs which use Java libraries. To the best of our knowledge, this system provides for the first time evidence that resource usage analysis can be applied to a realistic object-oriented, bytecode programming language. 1