Describes a n u m b e r o f currently available devices for the input and o u t p u t of graphical information to and from a computer, and illustrates some typical architectural applications.
Renfrow, N. and Bogdanski, J
76.84 'ARK-2'. Bull. Comput. Aid. Arcbit. Des. No 20, (May 1976) pp 1 6 - 2 5 Describes and illustrates the c o m p u t e r aided design system developed by the American architectural practice Perry-Dean Partners Inc. Westlake, B . J . 76.85 'Talking to computers in the 70's - the structural engineer'. Arcbit. Sci. Rev. Vol 19 No 1 (March 1976) pp 14--16 Discusses how the advances in c o m p u t e r and terminal hardware, particularly the advent of cheap graphics facilities, has eased the 'communication problem' for the nonspecialiast user.
Electrical and electronic engineering Adams, R . L . 76.86 'An optimization algorithm suitable for computer-assisted network tuning'. Proc. 1EEE (1975) ISCAS, pp 2 1 0 - 2 1 2 The linearity of the roots of network immittance affords more general application of least squares optimization to computerassisted tuning of IC filters. The m e t h o d described in this paper is s h o w n to be viable b y Monte-Carlo simulation on several high degree bandpass filters.
Breen, R. H. Jr., Temes. G . C .
76.87 'Application o f Golub's algorithm in circuit optimization and analysis'. IEEE Trans. CT-20, No 6, (Nov 1973) pp 687--690 The recently developed algorithm o f Golub for solving an overdetermined s y s t e m of linear equations is supplied to the least p-th optimization o f circuits and to the nodal and mesh analysis of passive resistive networks. The results show that Golub's technique is well-suited for t h e solution of such problems, even in the presence o f severe illconditioning. Bryant, P . R . 76.88 'Tracking sensitivity: an alternative algorit h m for linear non-reciprocal circuits'. Electron. Lett. Vol 11, No 5, (March 1975), pp 1 1 4 - 1 1 6 An alternative algorithm is presented that is always applicable and which is believed to be computationally more attractive than that previously proposed by Leung and Spence for circuits containing a b o u t 20 or few, er nodes. The similarities between these techniques and some earlier work done by Branin for c o m p u t i n g the steadystate a.c. analysis o f a circuit are also pointed out. Dembart, B., Erisman, A . M . 76.89 'Hybrid sparse-matrix methods'. IEEE Trans. CT-20, No 6, (Nov 1973), pp 641--649 In c.a.d, for large systems, efficient solution o f large sparse s y s t e m s o f nonlinear equations is important. Two parts o f t h e solution viz., ordering and factorization are critically examined. Markowitz ordering criterion is found to be m o s t practical ordering algorithm. A combination o f solution processes provides a very efficient hybrid
algorithm for factoring the ordered sparse matrix. Gadenz, R. N., Rezai-Fakfr, M. G, and Temes, G . C . 76.90 'A m e t h o d for the computation of large tolerance effects'. IEEE Trans. CT-20, No 6, (Nov 1973), pp 704--708 A m e t h o d is described for the tolerance analysis of linear circuits with large element variations. It uses the adjoint network concept to set up a reduced system of equations, solvable by either Gaussian elimination or iteration. The process appears to be considerably more economical than simple repeated analysis of the circuit, provided that not all the circuit elements are toleranced.
Jessel, G . P . 76.91 'Network statistics for computer-aided network analysis'. IEEE Trans. CT-20, No 6, (Nov 1973) pp 6 3 5 - 6 4 1 This paper derives b o u n d s and presents empirical values for: the n u m b e r of elements pernode; the sparsity of the cutset matrix; the density of nonlinear nodes; the relative percentage of capacitors and inductors; the relationship between the n u m b e r of elements of each element type, and the total n u m b e r of elements in an electrical circuit. This information is essential for comparative studies of computer-aided network analysis program performance,
Laker, K. R. and Ghausi, M.S. 76.92 'Statistical multiparameter sensitivity - a valuable measure for c.a.d.'. IEEE (1975), ISCAS, pp 3 3 3 - 3 3 6 Statistical multiparameter sensitivity measures suitable for computation and optimization in active filters are presented. The physical meanings of the measures are assessed and a comparison with Monte-Carlo analysis is provided. The measures are shown to be useful in c.a.d, and to provide insight to circuit performance. Effects of c o m p o n e n t correlation on sensitivity performance are evaluated.
Leung, K . H. a n d Spence, R. 76.93 'Efficient statistical circuit analysis'. Electron. Lett, Vol 10, No 17 (Aug 1974). pp 360--362 This letter reports the results o f a preliminary feasibility s t u d y of an efficient m e t h o d o f statistical analysis for linear nonreciprocal circuits. Proposed m e t h o d appears to offer the potential for a useful reduction in cost of a Monte-Carlo type o f statistical circuit analysis. Leung, K. H. and Spance, R. 76.94 'Tracking sensitivity: an efficient algorithm for linear non-reciprocal circuits'. Electron. Lett., Vol 10, No 18, (Sept 1974), pp 377-378 For the frequency-domain response of linear non-reciprocal circuits, an algorithm for c o m p u t i n g the c o m p o n e n t tracking sensitivity has been devised. Rapid exploration o f the effect is permitted b y the very limited calculation associated with each value o f the global variab1-, e.g. temperature, following an initial calculation c o m m o n to all values o f t h e global variable.
Rabbat, N. B., Ruehli, A. E. Mahoney, G. W., and Coleman J . J . 76.95 'A survey of macromodelling'. Proc. IEEE (1975 ISCAS, pp 1 3 9 - 1 4 2 This paper surveys a new area in circuit theory - macromodelling which bridges the gap between computer-aided circuit design and system design in a large scale integration environment. Macromodelling identifies a subnetwork by a model that reflects its terminal properties. Different macromodels are possible, depending on the accuracy, internal-parameter adjustment, analysis techniques required. This paper reviews the literature o n macromodelling, classifies techniques, and introduces new concepts where appropriate.
Rezai-Fakhr, M. G. and Temes, G . C . 76.96 'Statistical large-tolerance analysis of nonlinear circuits in the time domain'. IEEE Trans. CAS-22, No 1, (Jan 1975), pp 1 5 - 2 1 An algorithm is described for t h e MonteCarlo analysis of nonlinear circuits. In the proposed procedure, the n u m b e r o f nonlinear equations which m u s t be solved for every Monte-Carlo trial at each time point equals the total n u m b e r e + ~t o f t h e toleranced and nonlinear elements. The process is efficient provided that e + U is smaller than the n u m b e r of equations needed to analyse the circuit using the conventional techniques i.e. there are relatively few largetoleranced elements in the circuit. The process is also applicable, in a simplified form, to the transient analysis of linear circuits.
Seth, A. K. and Roe, P . H . 76.97 'Higher derivative network sensitivities using adjoint networks'. Int. J. Cir. Theor. Appl., Vol 1, No 3, (Sept 1973), pp 2 1 5 - 2 2 6 Higher derivative sensitivity coefficients for a lumped linear time-invariant n e t w o r k are described b y an extension of t h e m e t h o d used by Richards to calculate second derivative sensitivity coefficients. A computer algorithm to generate these coefficients efficiently is provided together with arrayspace and the n u m b e r of operations required at each stage. Extensions o f the above approach for networks containing distributed and nonlinear elements is also discussed. Finally, time domain sensitivity results are derived and the corresponding changes in the computer algorithm indicated.
Singhal, K., Vlach, J. and Bryant, P . R . 76.98 'Efficient c o m p u t a t i o n o f large change multiparameter sensitivity'. Int. J. Cir. Tbeor. Appl., Vol 1, No 3, (Sept 1973) pp 2 3 7 - 2 4 7 In m a n y applications, a network has to be solved ¢epeatedly while certain element values are changed. The c o m p u t a t i o n is reduced significantly if t h e network functions are generated in t e r m s o f these variable parameters explicitly. This paper presents a m e t h o d o f doing this which requires the solution of a n e t w o r k with at t h e m o s t m + 1 excitations when there are m variable parameters.
COMPUTER AIDED DESIGN