Olfa Mraihi

Invariant assertions, invariant relations, and invariant functions

Abstract Invariant assertions play an important role in the analysis and documentation of while loops of imperative programs. Invariant functions and invariant relations are alternative analysis tools that are distinct from invariant assertions but are related to them. In this paper we discuss these three concepts and analyze their relationships. The study of invariant functions and invariant relations is interesting not only because it provides alternative means to analyze loops, but also because it gives us insights into the structure of invariant assertions, hence it may help us enhance techniques for generating invariant assertions.

Computing preconditions and postconditions of while loops

Weakest preconditions were introduced by Dijkstra as a tool to define the semantics of programming constructs, and thereby as a means to prove the correctness of programs; the dual concept of strongest postcondition was introduced subsequently as an alternative means for the same ends. In this paper, we present and discuss a method to compute weakest preconditions and strongest postconditions of while loops in a C-like programming language; to this effect, we use the concept of invariant relation. Whereas the task of computing weakest preconditions and strongest postconditions of while loops is usually approached by limiting the number of iterations and applying successive sequential compositions, invariant relations afford us a crisper, closed form solution.

A versatile concept for the analysis of loops

Abstract Ever since their introduction by Hoare in 1969, invariant assertions have, justifiably, played a key role in the analysis of while loops. In this paper, we discuss a distinct but related concept, viz invariant relations, and show how these can be used to answer many questions pertaining to the analysis of loops, including: how to compute the function of the loop; how to compute an invariant assertion of the loop; how to compute a weakest precondition of the loop; how to compute a strongest postcondition of the loop; how to compute the termination condition of a loop; how to verify whether the loop computes a given function; how to verify whether the loop is correct with respect to a given specification; and finally how to compute an invariant function for the loop. Using a tool we have developed at the University of Tunis to derive invariant relations, we show how all these tasks can be automated by means of a computer algebra system, viz Mathematica (©Wolfram Research). Whenever applicable, we compare the performance of our tool against the performance of others.

Verifying while loops with invariant relations

Traditionally, invariant assertions are used to verify the partial correctness of while loops with respect to pre/post specifications. In this paper we discuss a related but distinct concept, namely invariant relations, and show how invariant relations are a more potent tool in the analysis of while loops: whereas invariant assertions can only be used to prove partial correctness, invariant relations can be used to prove total correctness; also, whereas invariant assertions can only be used to prove correctness, invariant relations can be used to prove correctness and can also be used to prove incorrectness; finally, where traditional studies of loop termination equate termination with iterating a finite number of times, we broaden the definition of termination to also capture the condition that each individual iteration proceeds without raising an exception.

Invariant functions and invariant relations: An alternative to invariant assertions

Whereas the analysis of loops in imperative programs is, justifiably, dominated by the concept of invariant assertion, we submit a related but different concept, of invariant relation, and show how it can be used to analyze diverse aspects of a while loop. We also introduce the concept of invariant function, which is used to generate a broad class of invariant relations.

Invariant relations, invariant functions, and loop functions

The increasing criticality of software applications, the increasing size and complexity of such applications, and the increasing reliance of software engineering paradigms on third party software assets combine to place a high premium on the ability to analyze software products to an arbitrary level of thoroughness and precision. Yet despite several decades of research, the goal of analyzing the functional properties of software products to an arbitrary level of thoroughness and precision remains unfulfilled. In this paper, we discuss the use of a relation-theoretic approach inspired from Mills’ logic to analyze while loops, and we support our approach by an operational prototype tool. The proposed method and tool have applications in program comprehension, reverse engineering, program verification, software maintenance, and programmer education.

Using invariant functions and invariant relations to compute loop functions

In this short paper we discuss the design, implementation and operation of an automated tool that computes the function of while loops written in C-like programming languages.

Towards an automated tool for computing loop functions

Despite several decades of on-and-off research efforts, the functional analysis of computer programs to an arbitrary level of thoroughness and precision remains an unfulfilled challenge. Also, despite several decades of advances in programming language design and implementation, most of the software that is written and maintained today is written in traditional imperative programs; in such programs, the most critical/ most complex control structure is the loop. In this paper, we discuss our ongoing effort to develop a tool that reviews a while loop written in an imperative language (C, C++, Java) and derives its function, by specifying the final values of all its program variables as a function of its initial values.