Making automatic differentiation truly automatic: coupling PETSc with ADIC

Despite its name, automatic differentiation (AD) is often far from an automatic process. Often one must specify independent and dependent variables, indicate the derivative quantities to be computed, and perhaps even provide information about the structure of the Jacobians or Hessians being computed. When AD is used in conjunction with a toolkit with well-defined interfaces, however, many of these issues can be dealt with automatically. We describe recent research into coupling the ADIC automatic differentiation tool with PETSc, a toolkit for the parallel numerical solution of PDEs. This research leverages the interfaces and objects of PETSc to make the AD process very nearly invisible.     

Pseudo-timestepping and verification for automatic differentiation derived CFD codes

The derivation of adjoint CFD codes through automatic differentiation is discussed with focus on the verification of the resulting code. Two alternative time-stepping formulations are discussed and compared, and the inclusion of multi-grid acceleration in the AD-derived code is shown. The adjoint code is embedded in a one-shot design loop and results for an inverse design case are presented.     

On a computer program for investigating automatic scanning of bubble chamber photographs

In a previous paper, the authors have described an automatic scanning system, and a computer program (ASINI 1) to test the system on simulated data. A second version of the program has been written (ASINI 2) which embodies more sophisticated tracking techniques. Furthermore, simulated photographs in ASINI 2 have dimensions approaching those of real photographs. Experimental results are discussed, and the limitations of the present program in automatic scanning of real photographs are pointed out. This work has been supported by Impresa Cibernetica, CNR. Ricevuto il 16-7-1964. CNR, Laboratori CISE. Laboratori CISE.     

A non-iterative shooting method for a non-linear diffusion problem using automatic differentiation

We present a non-iterative shooting method for the solution of a non-linear two-point boundary value problem that models the steady-state temperature distribution in a cylinder of unit radius. By non-iterative it is meant that there is no need to solve one or more initial value problems repeatedly. The method of this article avoids the need for such repetition by numerically obtaining a single algebraic non-linear equation involving only the initial condition. Thus, the appropriate initial condition for the final solution is available just after one solution of the initial value problem corresponding to an arbitrary initial value. In addition, the solution of the initial value problem is obtained as a Taylor series expansion of arbitrary order, using a technique known as automatic differentiation, which is the process of obtaining the coefficients in the Taylor series expansion using recursive formulas. Thus, the method does not face the need to deal with step size issues or the need to carry out lengthy algebraic manipulations. The method successfully reproduces the solutions obtained previously by other researchers.     

Looking for narrow interfaces in automatic differentiation using graph drawing

Automatic differentiation is a powerful technique for evaluating derivatives of functions given in the form of a high-level programming language such as Fortran, C, or C++. This technique is superior, in terms of accuracy, to numerical differentiation because it avoids the truncation error involved in divided difference approximations. In automatic differentiation, the program is treated as a potentially very long composition of elementary functions to which the chain rule of differential calculus is applied over and over again. Because of the associativity of the chain rule, there is room for different strategies computing the same numerical results but whose computational cost may vary significantly. Several strategies exploiting high-level structure of the underlying computer code are known to reduce computational cost as opposed to blindly applying automatic differentiation. An example includes “interface contraction” where one takes advantage of the fact that the number of variables passed between subroutines is small compared with the number of propagated directional derivatives. Unfortunately, these so-called narrow interfaces are not immediately available. The present study investigates the use of the VCG graph drawing tool to recognize narrow interfaces in the computational graph, a certain directed acyclic graph used to represent data dependences of variables in the underlying computer code.     

Computation of state and input trajectories for flat systems using automatic differentiation

For path planning and trajectory tracking it is necessary to find an input signal for a given output signal, i.e., it is useful to invert the plants model (system inversion). Up to now, system inversion was practical only for simple systems due to the amount of symbolical computations involved. The inversion of flat state-space systems using automatic differentiation will be discussed.     

An algebraic foundation for automatic feature-based program synthesis

Feature-Oriented Software Development provides a multitude of formalisms, methods, languages, and tools for building variable, customizable, and extensible software. Along different lines of research, different notions of a feature have been developed. Although these notions have similar goals, no common basis for evaluation, comparison, and integration exists. We present a feature algebra that captures the key ideas of feature orientation and that provides a common ground for current and future research in this field, on which also alternative options can be explored. Furthermore, our algebraic framework is meant to serve as a basis for the development of the technology of automatic feature-based program synthesis and architectural metaprogramming.     

Nonlinear system identification for predictive control using continuous time recurrent neural networks and automatic differentiation

In this paper, a continuous time recurrent neural network (CTRNN) is developed to be used in nonlinear model predictive control (NMPC) context. The neural network represented in a general nonlinear state-space form is used to predict the future dynamic behavior of the nonlinear process in real time. An efficient training algorithm for the proposed network is developed using automatic differentiation (AD) techniques. By automatically generating Taylor coefficients, the algorithm not only solves the differentiation equations of the network but also produces the sensitivity for the training problem. The same approach is also used to solve the online optimization problem in the predictive controller. The proposed neural network and the nonlinear predictive controller were tested on an evaporation case study. A good model fitting for the nonlinear plant is obtained using the new method. A comparison with other approaches shows that the new algorithm can considerably reduce network training time and improve solution accuracy. The CTRNN trained is used as an internal model in a predictive controller and results in good performance under different operating conditions.     

carlomat, version 2 of the program for automatic computation of lowest order cross sections

Version 2 of carlomat, a program for automatic computation of the lowest order cross sections of multiparticle reactions, is described. The substantial modifications with respect to version 1 of the program include: generation of a single phase space parameterization for the Feynman diagrams of the same topology, an interface to parton density functions, improvement of the color matrix computation, the Cabibbo–Kobayashi–Maskawa mixing in the quark sector, the effective models including scalar electrodynamics, the Wtb$Wtb$ interaction with operators of dimension up to 5 and a general top–Higgs coupling. Moreover, some minor modifications have been made and several bugs in the program have been corrected.     

Fortran IV program to compute apparent resistivity of a perfectly conducting sphere buried in a half-space

A FORTRAN IV program is presented that computes apparent resistivity of a perfectly conducting sphere buried in a half-space at any depth and for any electrode configuration. The program is based on solution of the problem by the method of images.     

An automatic tuning methodology for a unified dead-time compensator

In this paper, an automatic tuning methodology for a modified Smith predictor control scheme is proposed. The main feature of the procedure is that it is applied in closed-loop (by either evaluating a set-point or a load disturbance step response) and it is suitable for self-regulating, integral and unstable processes. Further, the process parameter estimation technique is based on the evaluation of the integral of signals, thus making it inherently robust to measurement noise. Simulation and experimental results demonstrate the effectiveness of the methodology.     

HERSIM: a microcomputer program designed to compute the limits of conversion for real homogeneous isothermal enzymic reactors

HERSIM is a program written in BASIC designed to aid the investigator interested in determining the substrate conversion in a real homogeneous isothermal enzymic reactor, for various kinetic equations. The program runs after tracer data relative to a Dirac impulse to the reactor have been entered, and computes the two limits of real conversion: total segregation and maximum mixedness. The kinetic constants of the reacting system are input as data, and the variation of conversion with reactor temperature between given limits is computed as accurately as requested.     

State of the art on automatic road extraction for GIS update: a novel classification

This paper surveys the state of the art on automatic road extraction for GIS update from aerial and satellite imagery. It presents a bibliography of nearly 250 references related to this topic. The work includes main approaches on general methods of road network extraction and reconstruction, road tracking methods, morphological analysis, dynamic programming and snakes, methods multi-scale and multi-resolution, stereoscopic and multi-temporal analysis, hyper-spectral experiments, and other techniques for road extraction. Likewise, other approaches related in any way with the road extraction topic are also considered. Between them different papers on segmentation, vectorization, optimization, evaluation, semantic nets and neural networks, fusion techniques, fuzzy logic, and other methods are discussed. A novel classification of road extraction methods according to our criteria is included in order to provide a significant contribution to research in this topic.     

FTSyn: a framework for automatic synthesis of fault-tolerance

In this paper, we present a software framework for adding fault-tolerance to existing finite-state programs. The input to our framework is a fault-intolerant program and a class of faults that perturbs the program. The output of our framework is a fault-tolerant version of the input program. Our framework provides (1) the first automated tool for the synthesis of fault-tolerant distributed programs, and (2) an extensible platform for researchers to develop a repository of heuristics that deal with the complexity of adding fault-tolerance to distributed programs. We also present a set of heuristics for polynomial-time addition of fault-tolerance to distributed programs. We have used this framework for automated synthesis of several fault-tolerant programs including a simplified version of an aircraft altitude switch, token ring, Byzantine agreement, and agreement in the presence of Byzantine and fail-stop faults. These examples illustrate that our framework can be used for synthesizing programs that tolerate different types of faults (process restarts, Byzantine and fail-stop) and programs that are subject to multiple faults (Byzantine and fail-stop) simultaneously. We have found our framework to be highly useful for pedagogical purposes, especially for teaching concepts of fault-tolerance, automatic program transformation, and the effect of heuristics.     

Modified automatic teller machine prototype for older adults: a case study of participative approach to inclusive design

The goal of this study was to enhance an existing automated teller machine (ATM) human-machine interface in order to accommodate the needs of older adults. Older adults were involved in the design and field test of the modified ATM prototype. The design of the user interface and functionality took the cognitive and physical abilities of older adults into account. The modified ATM system included only “cash withdrawal” and “transfer” functions based on the task demands and needs for services of older adults. One hundred and forty-one older adults (aged 60 or above) participated in the field test by operating modified or existing ATM systems. Those who operated the modified system were found to have significantly higher success rates than those who operated the existing system. The enhancement was most significant among older adults who had lower ATM-related abilities, a lower level of education, and no prior experience of using ATMs. This study demonstrates the usefulness of using a universal design and participatory approach to modify the existing ATM system for use by older adults. However, it also leads to a reduction in functionality of the enhanced system. Future studies should explore ways to develop a universal design ATM system which can satisfy the abilities and needs of all users in the entire population.     

A comparison of techniques for automatic detection of stiction: simulation and application to industrial data

This paper illustrates the effect of stiction in actuators on closed loop performance and the importance of techniques, to be incorporated in the monitoring system, for automatic detection of the onset of this phenomenon. Three different techniques are compared on the basis of simulation by means of stiction models and it is shown how loop variables are affected by stiction parameters and by process characteristics. A major finding is that every technique has an uncertainty region where no decision can be taken in the absence of further information about the process. The application on industrial data, recorded on loops affected by stiction during routine plant operation, allows to confirm this result and to assess relative efficiency of the techniques. A simple test to be performed directly on the plant is also proposed to solve the remaining uncertain cases.     

Make — a program for maintaining computer programs

Good programmers break their projects into a number of pieces, each to be processed or compiled by a different chain of programs. After a set of changes is made, the series of actions that must be taken can be quite complex, and costly errors are frequently made. This paper describes a program that can keep track of the relationships between parts of a program, and issue the commands needed to make the parts consistent after changes are made. Make has been in use on UNIX

1 UNIX is a trademark of Bell Laboratories.

systems since 1975. The underlying idea is quite simple and can be adapted to many other environments.