John W. Baugh

Modeling and verifying active structural control systems

Abstract This paper presents the results of a case study involving the use of a formal graphical notation, Modechart, and an automatic verification tool, the Concurrency Workbench, in the analysis of the design of a fault-tolerant active structural control system. Such control systems must satisfy strict requirements on their timing behavior; we show how to use various equivalence-based features supported by the Workbench to examine the timing behavior of different design alternatives, one of which has in excess of 10 19 states. The central insight arising from the study involves the importance of compositionality for reasoning about large and complex systems; in particular, the success of the case study depends integrally on our notations and tools support of componentwise minimization.

Network-distributed finite element analysis

The widespread availability of local-area networks has made the combined processing power of workstations a viable approach for compute-intensive analyses. In this paper, we describe several distributed algorithms for structural analysis using finite element methods, and we assess their performance on a conventional Ethernet-connected workstation network. Direct, iterative and hybrid equation solvers are evaluated for their performance on plane-elasticity problems, and are contrasted with respect to overall solution time and efficiency in distributing computations over a network. Equations modeling the costs of network communication and structural analysis computations are derived, and are subsequently used to predict the performance of several variations on the implemented algorithms. Our results show that each of the methods performs well on network architectures, and in particular that, while direct methods usually minimize network communication, certain iterative and hybrid methods can often be used to minimize overall solution time.

Evaluation of distributed finite element algorithms on a workstation network

This paper discusses the design, implementation and evaluation of linear finite element programs that distribute their computations over a network of workstations. We consider five different algorithms based on direct, iterative and hybrid equation solvers, each of which partitions and maps the model domain across conventional network hardware. A software architecture based on the client-server model distributes the computations and, at the language level, Berkeley sockets enable communication between processes. We evaluate and describe the performance of these algorithms in terms of execution time and speed-up, and we conclude that distributed solvers, particularly those based on substructuring and static condensation, can be effective even on high-latency communication networks.

Verification of an active control system using temporal process algebra

In this paper we describe complementary approaches that can be used to ensure the reliability of real-time systems, such as those used in active structural control systems. These approaches include both model-checking and simulation, and are based on a temporal process algebra. We combine these formal methods with a high-level, graphical modeling technique, Modechart, to specify an active structural control system consisting of several processors. Timing requirements on the system are specified and verified with a combination of process algebraic models and modal logic, and various simulation concepts are described for debugging models and for gaining insight into system behavior.

Asynchronous genetic algorithms for heterogeneous networks using coarse-grained dataflow

Genetic algorithms (GAs) are an attractive class of techniques for solving a variety of complex search and optimization problems. Their implementation on a distributed platform can provide the necessary computing power to address large-scale problems of practical importance. On heterogeneous networks, however, the performance of a global parallel GA can be limited by synchronization points during the computation, particularly those between generations. We present a new approach for implementing asynchronous GAs based on the dataflow model of computation -- an approach that retains the functional properties of a global parallel GA. Experiments conducted with an air quality optimization problem and others show that the performance of GAs can be substantially improved through dataflow-based asynchrony.

Formal specification of concurrent systems

Abstract This paper presents a formal methodology for developing concurrent systems. We extend the Larch family of specification languages and tools with the CCS process algebra to support the specification and verification of concurrent systems. We present and follow a refinement strategy that relates an implementation in a programming language to a formal specification of such a system. We illustrate our methodology on an example that uses the preconditioned conjugate gradient method for solving a linear system of equations.

SMT: An interface for localized storm surge modeling

Abstract The devastation wrought by Hurricanes Katrina (2005), Ike (2008), and Sandy (2012) in recent years continues to underscore the need for better prediction and preparation in the face of storm surge and rising sea levels. Simulations of coastal flooding using physically based hydrodynamic codes like ADCIRC, while very accurate, are also computationally expensive, making them impractical for iterative design scenarios that seek to evaluate a range of countermeasures and possible failure points. We present a graphical user interface that supports local analysis of engineering design alternatives based on an exact reanalysis technique called subdomain modeling, an approach that substantially reduces the computational effort required. This interface, called the Subdomain Modeling Tool (SMT), streamlines the pre- and post-processing requirements of subdomain modeling by allowing modelers to extract regions of interest interactively and by organizing project data on the file system. Software design and implementation issues that make the approach practical, such as a novel range search algorithm, are presented. Descriptions of the overall methodology, software architecture, and performance results are given, along with a case study demonstrating its use.

Modeling a Discrete Wet-Dry Algorithm for Hurricane Storm Surge in Alloy

We describe an Alloy model that helps check the correctness of a discrete wet-dry algorithm used in a system for hurricane storm surge prediction. Derived from simplified physics and encoded with empirical rules, the algorithm operates on a finite element mesh to allow the propagation of overland flows. Our study is motivated by complex interactions between the algorithm and a recent performance enhancement to the system that involves mesh partitioning. We briefly outline our approach and describe safety properties of the extension, as well as directions for future work.

Applications of coarse-grained dataflow in computational mechanics

Dataflow models are free of side effects and have no notion of state or sequencing. Because these representations place a partial, as opposed to a total, ordering on the execution of their component operations, the concurrent aspects of computation are clearly revealed. The correspondence between dataflow graphs and purely functional programs allows computations to be expressed in a high-level functional language and subsequently transformed into a dataflow graph. This paper describes the use of dataflow models as an alternative control strategy for engineering analysis programs and contrasts them with traditional imperative approaches. The characteristics of functional languages are also described, as is their inherent parallelism, which may be realized by compilation into dataflow graphs. The application of functional languages to finite element programming is presented, which allows the alternating assembly and solution of system equations found in frontal solvers. Issues such as the incremental update of arrays and the simulation of state are also addressed.

Is Engineering Software Amenable to Formal Specification

Unwieldy, incomprehensible, spaghetti-like Fortran code is very likely the image conjured by the term “engineering and scientific software.” What could the purveyors of such monstrosities possibly want with formal methods? Written from the perspective of a civil engineer, the following is a discussion of the potential role of formal specification tools, such as Larch, in engineering applications. The discussion also includes some examples of specifying engineering software components, and some comments on the potential benefits of formal specifications in engineering program design.