E. Kant, F. Daube, E. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis", in: Michael R. Lowery and Robert D. McCartney (eds.), Automating Software Design, MIT Press, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....2) automating software design, and 3) automating software design methods. 3. 1 Automatic Programming Unlike automated, design assistants, which help a human analyst complete a single, if essential, transformation in the software development process, automatic programming systems [34] 35] [36], 37] 38] attempt to perform, without human intervention, every transformation required to generate a working implementation from an initial specification of user requirements. A different form of automatic programming, end user programming, enables a computer naive user to interact with an ....

E. Kant, F. Daube, W. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis," Automating Software Design, M.R. Lowry and R.D. McCartney eds., pp. 141-168, 1991.

....no domain specific component built in. In this system, an informal problem description written in a domain specific language is trans lated into mathematical formalisms that require no domain knowledge. All the domain information is used only in this translation process. In the SINAPSE system [Kant et al. 1991], an environment for scientific computation, a system is described in terms of application domain key words. This description is guided through menus provided by a graphical user interface. Based on domain model and symbolic manipulation, this description is then translated into a more complete ....

E. Kant, F. Daube, W. MacGregor, and J. Wald. Scientific programming by automated synthesis. In M. Lowry and R. McCartney, editors, Automating Software Design, pages 169 205. AAAI Press/The MIT Press, 1991.

....of interest. Like SIGMA and ECO, f 0 makes extensive use of scientific domain knowledge to aid in the model building process. However, SIGMA s equation representation appears to incorporate more domain knowledge and constraints than the representation used in the f 0 prototype. In a similar vein, (Kant, Daube, MacGregor, Wald, 1991) describes the SINAPSE system under development at Schlumberger. SINAPSE helps scientists build mathematical modeling software used in the context of data interpretation tasks such as seismic interpretation. In particular, the system synthesizes finite difference programs that implement partial ....

E. Kant, F. Daube, W. MacGregor, and J. Wald (1991). Scientific Programming by Automated Synthesis. In M. R. Lowry & R. D. McCartney (Eds.), Automating Software Design Menlo Park: AAAI Press.

....Then, for each module, data structures and algorithms are selected. A detailed description of the approach to 43 selecting data structures and algorithms is available in another paper. SETL91] Finally, source code is generated. Another automatic system for synthesizing software is SINAPSE. [KANT91] SINAPSE aims: 1) to reduce the time needed for scientists and engineers to implement mathematical models, 2) to allow natural language specification of requirements for such models, 3) to reuse existing implementations, and 4) to avoid the introduction of careless errors into the ....

E.Kant, F. Daube, W. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis," in Automating Software Design, M.R. Lowry and R.D. McCartney (eds.), AAAI Press, Menlo Park, California, 1991, pp. 169-206..

....Transformation Systems Transformation systems repeatedly replace parts of an abstract algorithm specification with code that is closer to an implementation, until executable code is reached. Our views specify transformations from features of abstract types to their implementations. Kant et al. [33] describe Sinapse, which generates programs to simulate spatial differential equations, e.g. for seismic analysis. Sinapse transforms a small specification into a much larger program in Fortran or C; it is written using Mathematica [75] and appears to work 31 well within its domain. Setliff s ....

E. Kant, F. Daube, W. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis," in [39], pp. 169-205.

....specification; they repeatedly apply transformations that replace parts of the abstract algorithm with code that is closer to an implementation, until executable code is finally reached. Our views specify transformations from features of abstract types to their implementations. Kant et al. [23] describe the Sinapse system for generating scientific programs involving simulation of differential equations over large spatial grids for applications such as seismic analysis. Sinapse accepts a relatively small program specification and generates from it a 24 much larger program in Fortran or ....

E. Kant, F. Daube, W. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis," in [28], pp. 169-205.

....involvement thus reappeared for each new technological level. As a step beyond this, a fair amount of work has been applied in the 1990 s towards automated synthesis of computer software, in some cases already optimized for specific compilers and hardware architectures [26] Systems like SINAPSE [27] and ELLPACK [28] generate source code implementing numerical techniques (finite differences and finite elements) for solving systems of PDE s. Custom technology of this type is what has been used at NRL in composite materials research for automated model generation. 4.2 Material Behavior ....

E. Kant, F. Daube, W. MacGregor, J. Wald, Scientific Programming by Automated Synthesis, Chapter 8 in Automatic Software Design, M. Lowry and R. McCartney, eds, (AAAI Press/MIT Press, Menlo Park, CA, 169205, 1991).

....on the system every day. DMS supports large scale applications by several means. DMS is implemented in a parallel processing language, Parlanse, Bax96] running on easily obtained Windows NT multiprocessor workstations. Our experience with transformational code synthesis systems such as SINAPSE [KAN91], BAX93] implemented on symbolic manipulation systems (e.g. Mathematica) showed that code synthesis alone sometimes required hours to generate some 5,000 lines, well short of our target ceiling of 10 million lines. Symbolic manipulation is expensive. Parlanse provides the DMS implementers with ....

E. Kant, F. Daube, E. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis," Automating Software Design, Michael R. Lowery and Robert D. McCartney, eds., MIT Press, 1991.

....domain, which is rarely possible. Thus, most existing expert systems function as interactive assistants. This change of emphasis also is true for the application of AI to software design. The early focus was on automatic programming (Barstow, 1979) Despite some interesting successes (Smith, 1991; Kant et al. 1991), the field of knowledge based software engineering has shifted its emphasis to creating systems that assist people in designing software (see the annual Knowledge Based Software Engineering Conference; this trend also appears in AI approaches to design in general, e.g. the annual International ....

Kant, E., Daube, F., MacGregor, W., & Wald, J. (1991). Scientific programming by automated synthesis: In M. Lowry & R. McCartney (Eds.), Automating Software Design (pp. 169-205). Menlo Park, CA: AAAI Press/MIT Press.

....Analyst DMS Remember Modify Build Rqmts Figure 2. The Design Maintenance System Concept Components 2.1 Domains A significant number of different notations, transformations, methods, and software components are needed to construct and maintain large systems. The SINAPSE system (see [Kan91]) for the generating mathematical modeling codes had several thousand chunks of synthesis knowledge, used for a narrow domain and producing relatively small (5K source lines of code) programs. It is not practical to manage such complex information as an amorphous collection. In DMS this knowledge ....

E. Kant, F. Daube, E. MacGregor, and J. Wald. Scientific Programming by Automated Synthesis. In: M.R. Lowery and R.D. McCartney (eds.). Automating Software Design. MIT Press, 1991.

....as the result of precompiling and caching the effects of a series of transformations (schema as a macro transformation ) 7.1. 2 Domain problem specific systems On the (literally) opposite side of the just discussed approaches are systems like 8NIX by Barstow [4] SINAPSE by Kant and coworkers [76], or ELF by Setliff [118] These are software synthesis systems which have been designed to synthesize code that solves specific types of problems in specific domains. Domain and problem specificity has the advantage that special features of the domains can be exploited to design efficient and ....

E. Kant, F. Daube, W. MacGregor, and J. Wald. Scientific Programming by automated synthesis. In M.R. Lowry and R.D. McCartney, editors, Automating Software Design. Menlo Park: AAAI Press, 1991.

....specification language and the underlying translation techniques; the system employs a set of refinement steps to the problem by which the algorithms and data structures are refined from an abstract level into a concrete implementation. Examples are the alpal system [7] the sinapse system [17], the suspense transformation system [20] and the dpml Data Parallel scientific Modelling Language [10] The Ctadel system should also be considered as a system of the last type. However, a novel approach is taken in which the description of the problem is transformed into code by taking target ....

E. Kant, F. Daube, W. MacGregor, J. Wald, Scientific Programming by Automated Synthesis, Chapter 8 in Automating Software Design, M. Lowry and R. McCartney, (eds), AAAI Press/MIT Press, Menlo Park, CA, 1991, pp. 169--205.

.... specification languages such as Z and VDM counteract some of the difficulties in the strictly descriptive slant of algebraic specifications; however, there is not much machine support for such methodologies [13] Finally, we contrast our system with highlydomain specific synthesis systems such as [11]. The issue is similar to the issue of weak vs. strong methods in AI. Every synthesis system which is practical must have both domain specific and domain independent components. For example, in our system we have domain independent knowledge in the form of general inference techniques and general ....

Kant, E., Daube, F., MacGregor, W., and Wald, J. Scientific programming by automated synthesis. In Automating Software Design, M. L. Lowry and R. McCartney, Eds. AAAI Press, Menlo Park, CA, 1991.

....forecast systems. The current and future role of PSEs for PDE based problems is discussed in [6] Well known PSEs for solving PDEs are ELLPACK [10] and the ALPAL system [1] PSEs for solving PDEs in the field of parallel computing are the = ELLPACK [7] and XELLPACK systems, the SINAPSE system [9], the SUSPENSE transformation system [11] and the DPML Data Parallel scientific Modelling Language [5] In comparison, in the CTADEL system main efforts have been put in generating efficient codes for possibly large RHSs of sets of PDEs. With the implemented symbolic optimization techniques, the ....

E. Kant, F. Daube, W. MacGregor, J. Wald, Scientific Programming by Automated Synthesis, Chapter 8 in Automating Software Design, M. Lowry and R. McCartney, (eds), AAAI Press/MIT Press, Menlo Park, CA, 1991, pp. 169--205.

....higher space dimension, and non Cartesian coordinates. Sinapse also contains substantial information about other equations, including wave equations in multiple dimensions. 1 Introduction The goal of this paper is to illustrate how the program synthesis tool, Sinapse (see Kant, et al. [4, 5, 6, 7]) can be used by a person who is familiar with partial differential equations and finite difference methods to generate and analyze numerical modeling programs for a wide range of applications. We illustrate the synthesis process with an initial boundaryvalue problem (IBVP) for the time dependent ....

E. Kant, F. Daube, W. MacGregor, J. Wald, Scientific Programming by Automated Synthesis, Chapter 8 in Automating Software Design, M. Lowry and R. McCartney, (editors), AAAI Press/MIT Press, Menlo Park, CA, 1991, pages 169-205.

No context found.

E. Kant, F. Daube, E. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis", in: Michael R. Lowery and Robert D. McCartney (eds.), Automating Software Design, MIT Press, 1991.

No context found.

E. Kant, F. Daube, W. MacGregor, and J. Wald, "Scientific Programming by Automated Synthesis", in Automating Software Design, M. R. Lowry and R. D. McCartney (eds.), AAAI Press, Menlo Park, California, 1991, pp. 141-168.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC