Variations on Constants: Flow Analysis of Sequential and by Markus Müller-Olm (auth.)

By Markus Müller-Olm (auth.)

Program research is anxious with thoughts that immediately ascertain run-time homes of given courses sooner than run-time. it's used for validation on the way to make sure that courses serve their meant goal and in extra processing for effective execution comparable to in optimizing compilers. optimum application research offers a warrantly concerning the precision of the computed results.

This monograph, a revised model of the author's habilitation thesis, focusses on optimum circulate research of sequential and parallel courses. It experiences algorithmic houses of assorted types of the well known constant-propagation challenge. so one can come to grips with the variations thought of, it combines recommendations from assorted components akin to linear algebra, computable ring idea, summary interpretation, application verification, complexity thought, and so on. mix of thoughts is the main to extra growth in computerized research and constant-propagation permits us to demonstrate this element in a theoretical study.

After a normal evaluation, the monograph comprises 3 primarily self-contained elements that may be learn independently of one another. those components learn: a hierarchy of constants in sequential courses, inherent limits of movement research of parallel courses, and the way to beat those limits via forsaking a vintage atomic execution assumption.

Show description

Read or Download Variations on Constants: Flow Analysis of Sequential and Parallel Programs PDF

Best computers books

Fire in the Valley: The Birth and Death of the Personal Computer

Within the Seventies, whereas their contemporaries have been protesting the pc as a device of dehumanization and oppression, a motley selection of university dropouts, hippies, and electronics fans have been engaged in whatever even more subversive. captivated with the belief of having laptop strength into their very own arms, they introduced from their garages a hobbyist flow that grew into an undefined, and eventually a social and technological revolution.

STACS 2006: 23rd Annual Symposium on Theoretical Aspects of Computer Science, Marseille, France, February 23-25, 2006. Proceedings

This e-book constitutes the refereed lawsuits of the twenty third Annual Symposium on Theoretical points of computing device technology, STACS 2006, held in Marseille, France, in February 2006. The fifty four revised complete papers offered including 3 invited papers have been rigorously reviewed and chosen from 283 submissions.

Compel (Vol. 23, 2004): Special Issue

Targeted factor: chosen papers from PMAPS 2002 -Conference on Probabilistic equipment utilized to strength structures, Naples 2002

Extra resources for Variations on Constants: Flow Analysis of Sequential and Parallel Programs

Example text

0) and b = (1). In the concrete algorithm the elements of D are represented in this way by a matrix A and a vector b but this further representation step is suppressed in this section. We show, however, that all the needed operations on affine spaces can efficiently be performed on their representation by a matrix and a vector. Conceptually, it is simpler to consider the affine spaces themselves as representations because they are ordered. On pairs (A, b) we have only the pre-order induced by their interpretation as affine spaces: (A, b) ≤ (A , b ) :⇔ {x | Ax = b} ⊆ {x | A x = b } .

The most important reason for this assumption is that it ensures that we can validly compute on representations without losing precision: if we precisely mirror the equations characterizing weakest preconditions on representations, the largest solution of the resulting equation system on representations characterizes the representation of the weakest precondition by the following wellknown lemma. It appears in the literature (for the dual situation of least fixpoints) under the name Transfer Lemma [4] or μ-Fusion Rule [49].

Xn ] by standard fixpoint iteration. 2 Zeros As mentioned, we represent assertions by the zeros of ideals in our algorithm. A state σ is called a zero of polynomial p if pσ = 0; we denote the set of zeros of polynomial p by Z(p). More generally, for a subset B ⊆ Z[x1 , . . , xn ], Z(B) = {σ | ∀p ∈ B : pσ = 0}. For later use some facts concerning zeros are collected in the following lemma, in particular of the relationship of ideal operations with operations on their zeros. 1. Suppose B, B are sets of polynomials, q is a polynomial, I, I are ideals, and I is a set of ideals in Z[x1 , .

Download PDF sample

Rated 4.71 of 5 – based on 31 votes

Related posts