Alan Carle

ADIFOR-Generating Derivative Codes from Fortran Programs

The numerical methods employed in the solution of many scientific computing problems require the computation of derivatives of a function f

Adifor 2.0: automatic differentiation of Fortran 77 programs

R^N

FIAT: A Framework for Interprocedural Analysis and Transfomation

→

Constructing the procedure call multigraph

R^m

Preliminary Results from the Application of Automated Adjoint Code Generation to CFL3D

. Both the accuracy and the computational requirements of the derivative computation are usually of critical importance for the robustness and speed of the numerical solution. Automatic Differentiation of FORtran (ADIFOR) is a source transformation tool that accepts Fortran 77 code for the computation of a function and writes portable Fortran 77 code for the computation of the derivatives. In contrast to previous approaches, ADIFOR views automatic differentiation as a source transformation problem. ADIFOR employs the data analysis capabilities of the ParaScope Parallel Programming Environment, which enable us to handle arbitrary Fortran 77 codes and to exploit the computational context in the computation of derivatives. Experimental results show that ADIFOR can handle real-life codes and that ADIFOR-generated codes are competitive with divided-difference approximations of derivatives. In addition, studies suggest that the source transformation approach to automatic differentiation may improve the time to compute derivatives by orders of magnitude.

Computing large sparse Jacobian matrices using automatic differentiation

Numerical codes that calculate not only a result, but also the derivatives of the variables with respect to each other, facilitate sensitivity analysis, inverse problem solving, and optimization. The paper considers how Adifor 2.0, which won the 1995 Wilkinson Prize for Numerical Software, can automatically differentiate complicated Fortran code much faster than a programmer can do it by hand. The Adifor system has three main components: the AdiFor preprocessor, the ADIntrinsics exception-handling system, and the SparsLinC library.

ADIFOR 2.0 user`s guide (Revision B)

The fiat system is a compiler-building tool that enables rapid prototyping of interprocedural analysis and compilation systems. Fiat is a framework because it provides parameterized templates and common drivers to support interprocedural data-flow analysis and procedure cloning. Further, fiat provides the complex underlying support required to collect and manage information about the procedures in the program. Fiats reliance on system-independent abstractions makes it suitable for use in systems with distinct intermediate code representations and enables sharing of system software across research platforms. Demand-driven analysis maintains a clean separation between interprocedural analysis problems, enabling tools built upon fiat to solve only the data-flow problems of immediate interest. Fiat drives interprocedural optimization in the ParaScope programming tools at Rice University and the SUIF compiler at Stanford University. Fiat has proven to be a valuable aid in development of a large number of interprocedural tools, including a data race detection system, a static performance estimation tool, a distributed-memory compiler for Fortran D, an interactive parallelizing tool and an automatic parallelizer in the SUIF compiler.

Derivative convergence for iterative equation solvers

An algorithm for constructing a precise call multigraph for languages that permit procedure parameters, extending the method of B. Ryder (see ibid., vol.5, no.3, p.216-225 (1979)) for handling recursion, is presented. If it is assumed that there is a constant upper bound on the number of procedure parameters to any procedure in the program, then the algorithm is polynomial in the total number of procedures in the program. >

Requirements for DataParallel Programming Environments

This report describes preliminary results obtained using an automated adjoint code generator for Fortran to augment a widely-used computational fluid dynamics flow solver to compute derivatives. These preliminary results with this augmented code suggest that, even in its infancy, the automated adjoint code generator can accurately and efficiently deliver derivatives for use in transonic Euler-based aerodynamic shape optimization problems with hundreds to thousands of independent design variables.

Automatic Data Layout for Distributed-Memory Machines in the D Programming Environment

The computation of large sparse Jacobian matrices is required in many important large-scale scientific problems. Three approaches to computing such matrices are considered: hand-coding, difference approximations, and automatic differentiation using the ADIFOR (automatic differentiation in Fortran) tool. The authors compare the numerical reliability and computational efficiency of these approaches on applications from the MINPACK-2 test problem collection. The conclusion is that ADIFOR is the method of choice, leading to results that are as accurate as hand-coded derivatives, while at the same time outperforming difference approximations in both accuracy and speed.