- Email: [email protected]
FE systems survey
I
Electronic cir Sensitivity and optimization, R Brayton and R Spence, Elsevier Scientific Publishing Company (1980) pp 368 $63.50
Sensitivity and optimization is the second book in the series on the subject of computer-aided design of electronic circuits. In comparison with the first book in the series, its overall appearance is a considerable improvement; the binding is stronger and the typeface more pleasing in appearance. It contains a clear table of contents with a full subject index which on random selection seems to be correct. (Some mistakes do, however, appear in the author index.) As the title suggests, the book is split into two areas of computer-aided circuit design. The first section, on sensitivity, covers a topic which, sadly, has lacked the attention of book publishers and in part, summarizes work which had only previously appeared in journals. It concentrates on the numerical methods used in a number of procedures for computing sensitivity measures without being side-tracked into explaining Gaussian elimination and LU factorization, additions which would only clutter the general flow of the material. Step-by-step algorithms for computing these measures are included. It also states some of the major theoretical sensitivity relationships which have been established. Methods for computing differential sensitivity (up to second order), multiple parameter sensitivities, large change sensitivity and tracking sensitivity are covered together with such side issues as model simplification, systematic exploration and Tellegen's Theorem. It is pleasing to note that quiescent and time domain sensitivities have not been overlooked and a whole chapter is devoted to this, and to the effect which nonlinear devices have on the results of a frequency domain differential sensitivity analysis. The second section of this book complements the first by concentrating on optimization as an application of sensitivity analysis. Although not an
120
I
teets exhaustive coverage of optimization procedures, it does include the more successful techniques. Unconditional optimization is presented first, emphasizing the quasi-Newton methods with some mention of non-gradient techniques. Of course most optimization problems in computer-aided circuit design are of a constrained nature and methods for transforming the problem into an unconstrained form by adding penalty functions, or by using transformation variables are discussed. The more recent Lagrange multiplier methods developed to handle constrained problems are also covered together with the underlying principles of linear, nonlinear and quadratic programming. In situations where there are multiple objectives, optimal trade-off curves provide a means by which the circuit's best capabilities can be revealed to the designer. The concept of Pareto points and their relation to the optimal trade-off curves is intro-
FE
duced. One chapter is largely devoted to minimax methods of handling multiple objectives and their use in statistical design. The authors deal with the use of optimization to maximize yield, either by means of design centring or by suitable tolerance assignment. With the inclusion of design possibilities at the beginning of both sensitivity and optimization sections, the reader can gain some overall feeling as to the importance and usage of such procedures, but the book seems to be aimed at either students studying computeraided circuit design where the course material emphasizes the analysis of analogue circuits, or aimed at those whose task it is to provide engineers with sensitivity and optimization tools. Sensitivity and optimization is a readable text and is recommended to those involved in computer-aided circuit design.
/an Benest, Rutherford Appteton Laboratory, UK
su,-vey
Finite element systems, C A Brebbia, CML Publications (1981) pp 480, £28.00 This book is intended to be a guide to finite element systems for practising engineers. It is a collection of papers by different authors presented at a seminar held in September 1981 ; together with a number of classification tables and a 'contact' list added by the editor. In the tables, programs have been classified purely on availability of type of elements, material properties, etc without any references to the basic distinctions between them, namely that some of the programs are complete systems capable of performing the analysis with pre- and post-processor facilities of varying magnitudes while others are only pre- and postprocessors. This major difference in
the type of programs becomes evident only from reading through the text, thus defeating the purpose of the tables. In addition to this variation one finds that some systems deal with only specific types of elements or theory, and this too should have been reflected in the tables. The content of the papers varies in nature. Some appear to be based on extracts from their respective sales brochures, some describe the structure of the programs and their internal organizations, and others deal with the theory to varying degrees. Some of the papers appear to be motivated by the marketing interests behind the relevant programs while others are probably intended to be more general and consequently deal with the theory and function. These differences in the papers are probably due to the differ-
computer-aided design
I
I
I
ent interpretations authors would have with respect to aim and purpose of the seminar. To suggest that these papers be used as a guide to the selection of a finite element system would be to do an injustice to the authors. A number of other systems exist and they too should be included in a book of this nature, even though no papers were presented at that seminar. Further, although the topic is finite element systems, programs dealing with boundary elements have been included. The title of the conference, like the title of this book, may have excluded a number of other boundary element
programs thus making the list rather incomplete in this respect. Individuals or organizations interested in selecting a finite element/ boundary element system or for that matter even a pre- or post-processor to finite element systems would do better to start with a more complete list of programs and to contact the suppliers for information about the latest versions on offer. Most of these programs are in constant development and a publication such as this is outof-date within a very short time. Those interested in special finite element systems, which incorporate
recent research work, would normally keep abreast of the journals dealing with those aspects and therefore know what programs they require and where to get them. When considering programs it is necessary to establish what verification and testing has been carried out. This aspect has not been covered by the papers. In conclusion the book has little to recommend it as either a guide to selection, or as a handbook to understanding finite element systems.
G Ganesalingham, Ore Arup Partnership, UK
CAN WE KEEPYOU ON COURSE Our 3 new CAD courses cover.
DURING
1 982?
CAD T E C H N O L O G Y TODAY • current hardware -- micro and larger systems -- as standalone and satellite systems; new communication services • display technology -- vector/rester/DVST including colour and plxels • man-system interface -- keyboards vs. digitizing pads vs. interactive devices • CAD software and data management systems Course G1 -- 1 day • London 19 April 1982 • North-West 18 October 1982
E you need computer talk to Benson. Benson offer more choice. From our ranse of drum and flatbed plotters and electrostatic printer/ plotters, Benson will ensure you have the plotterto suityour application, whatever your computer confisuration. Superb accurate drawinss in a choice of sizes and speeds, at a price to suit. So m a n y ......... realised Benson has MORE s or write for details of our nge of products.
~
b~n~n
TheComputerDraftingProfesskamls Benson ElectronicsLtd. TechnoHouse, RedcliffeWay, BristolBS16NH Telephone:(0272)211501(5 lines) Telex:444597Benson
volume 14 number 2 march 1982
W H Y MICROCOMPUTER-AIDED DESIGN? • technology -- hardware and peripherals overview • software, including firmware • economic justification -- costs vs. benefits • abstracted case histories Course M8 -- 1 day • North-West 22 March 1982 • London 28 September 1982 I M P L E M E N T I N G SMALL-SCALE CAD SYSTEMS • choice of correct system components, to match needs • methods of storing drawings and coding components • typical drawing software functions and dialogue • abstracted case histories Course M 9 - - 3 days • London 29-31 March 1982 • North-West 11-13 October 1982
Sytech Edtmation Contributors include Simon Bensasson of the Bensasson-Chalmers Partnership (author of the DOC report 'Micros in construction') and Frank Taylor of Sytech-formerly of the NCC. FEES -ALL COURSES-- £63 + VAT/day for bookings at least 4 weeks in advance, thereafter £70 + VAT/day. Reservations and/or further details -- please contact Course Regiatmr~ PO Box 8, Knutsford, Cheshire WA16 9DU, UK. Tel: Knutsford (0565) 52911 Telex: 868371 Attn. SYTECH
121
Electronic cir Sensitivity and optimization, R Brayton and R Spence, Elsevier Scientific Publishing Company (1980) pp 368 $63.50
Sensitivity and optimization is the second book in the series on the subject of computer-aided design of electronic circuits. In comparison with the first book in the series, its overall appearance is a considerable improvement; the binding is stronger and the typeface more pleasing in appearance. It contains a clear table of contents with a full subject index which on random selection seems to be correct. (Some mistakes do, however, appear in the author index.) As the title suggests, the book is split into two areas of computer-aided circuit design. The first section, on sensitivity, covers a topic which, sadly, has lacked the attention of book publishers and in part, summarizes work which had only previously appeared in journals. It concentrates on the numerical methods used in a number of procedures for computing sensitivity measures without being side-tracked into explaining Gaussian elimination and LU factorization, additions which would only clutter the general flow of the material. Step-by-step algorithms for computing these measures are included. It also states some of the major theoretical sensitivity relationships which have been established. Methods for computing differential sensitivity (up to second order), multiple parameter sensitivities, large change sensitivity and tracking sensitivity are covered together with such side issues as model simplification, systematic exploration and Tellegen's Theorem. It is pleasing to note that quiescent and time domain sensitivities have not been overlooked and a whole chapter is devoted to this, and to the effect which nonlinear devices have on the results of a frequency domain differential sensitivity analysis. The second section of this book complements the first by concentrating on optimization as an application of sensitivity analysis. Although not an
120
I
teets exhaustive coverage of optimization procedures, it does include the more successful techniques. Unconditional optimization is presented first, emphasizing the quasi-Newton methods with some mention of non-gradient techniques. Of course most optimization problems in computer-aided circuit design are of a constrained nature and methods for transforming the problem into an unconstrained form by adding penalty functions, or by using transformation variables are discussed. The more recent Lagrange multiplier methods developed to handle constrained problems are also covered together with the underlying principles of linear, nonlinear and quadratic programming. In situations where there are multiple objectives, optimal trade-off curves provide a means by which the circuit's best capabilities can be revealed to the designer. The concept of Pareto points and their relation to the optimal trade-off curves is intro-
FE
duced. One chapter is largely devoted to minimax methods of handling multiple objectives and their use in statistical design. The authors deal with the use of optimization to maximize yield, either by means of design centring or by suitable tolerance assignment. With the inclusion of design possibilities at the beginning of both sensitivity and optimization sections, the reader can gain some overall feeling as to the importance and usage of such procedures, but the book seems to be aimed at either students studying computeraided circuit design where the course material emphasizes the analysis of analogue circuits, or aimed at those whose task it is to provide engineers with sensitivity and optimization tools. Sensitivity and optimization is a readable text and is recommended to those involved in computer-aided circuit design.
/an Benest, Rutherford Appteton Laboratory, UK
su,-vey
Finite element systems, C A Brebbia, CML Publications (1981) pp 480, £28.00 This book is intended to be a guide to finite element systems for practising engineers. It is a collection of papers by different authors presented at a seminar held in September 1981 ; together with a number of classification tables and a 'contact' list added by the editor. In the tables, programs have been classified purely on availability of type of elements, material properties, etc without any references to the basic distinctions between them, namely that some of the programs are complete systems capable of performing the analysis with pre- and post-processor facilities of varying magnitudes while others are only pre- and postprocessors. This major difference in
the type of programs becomes evident only from reading through the text, thus defeating the purpose of the tables. In addition to this variation one finds that some systems deal with only specific types of elements or theory, and this too should have been reflected in the tables. The content of the papers varies in nature. Some appear to be based on extracts from their respective sales brochures, some describe the structure of the programs and their internal organizations, and others deal with the theory to varying degrees. Some of the papers appear to be motivated by the marketing interests behind the relevant programs while others are probably intended to be more general and consequently deal with the theory and function. These differences in the papers are probably due to the differ-
computer-aided design
I
I
I
ent interpretations authors would have with respect to aim and purpose of the seminar. To suggest that these papers be used as a guide to the selection of a finite element system would be to do an injustice to the authors. A number of other systems exist and they too should be included in a book of this nature, even though no papers were presented at that seminar. Further, although the topic is finite element systems, programs dealing with boundary elements have been included. The title of the conference, like the title of this book, may have excluded a number of other boundary element
programs thus making the list rather incomplete in this respect. Individuals or organizations interested in selecting a finite element/ boundary element system or for that matter even a pre- or post-processor to finite element systems would do better to start with a more complete list of programs and to contact the suppliers for information about the latest versions on offer. Most of these programs are in constant development and a publication such as this is outof-date within a very short time. Those interested in special finite element systems, which incorporate
recent research work, would normally keep abreast of the journals dealing with those aspects and therefore know what programs they require and where to get them. When considering programs it is necessary to establish what verification and testing has been carried out. This aspect has not been covered by the papers. In conclusion the book has little to recommend it as either a guide to selection, or as a handbook to understanding finite element systems.
G Ganesalingham, Ore Arup Partnership, UK
CAN WE KEEPYOU ON COURSE Our 3 new CAD courses cover.
DURING
1 982?
CAD T E C H N O L O G Y TODAY • current hardware -- micro and larger systems -- as standalone and satellite systems; new communication services • display technology -- vector/rester/DVST including colour and plxels • man-system interface -- keyboards vs. digitizing pads vs. interactive devices • CAD software and data management systems Course G1 -- 1 day • London 19 April 1982 • North-West 18 October 1982
E you need computer talk to Benson. Benson offer more choice. From our ranse of drum and flatbed plotters and electrostatic printer/ plotters, Benson will ensure you have the plotterto suityour application, whatever your computer confisuration. Superb accurate drawinss in a choice of sizes and speeds, at a price to suit. So m a n y ......... realised Benson has MORE s or write for details of our nge of products.
~
b~n~n
TheComputerDraftingProfesskamls Benson ElectronicsLtd. TechnoHouse, RedcliffeWay, BristolBS16NH Telephone:(0272)211501(5 lines) Telex:444597Benson
volume 14 number 2 march 1982
W H Y MICROCOMPUTER-AIDED DESIGN? • technology -- hardware and peripherals overview • software, including firmware • economic justification -- costs vs. benefits • abstracted case histories Course M8 -- 1 day • North-West 22 March 1982 • London 28 September 1982 I M P L E M E N T I N G SMALL-SCALE CAD SYSTEMS • choice of correct system components, to match needs • methods of storing drawings and coding components • typical drawing software functions and dialogue • abstracted case histories Course M 9 - - 3 days • London 29-31 March 1982 • North-West 11-13 October 1982
Sytech Edtmation Contributors include Simon Bensasson of the Bensasson-Chalmers Partnership (author of the DOC report 'Micros in construction') and Frank Taylor of Sytech-formerly of the NCC. FEES -ALL COURSES-- £63 + VAT/day for bookings at least 4 weeks in advance, thereafter £70 + VAT/day. Reservations and/or further details -- please contact Course Regiatmr~ PO Box 8, Knutsford, Cheshire WA16 9DU, UK. Tel: Knutsford (0565) 52911 Telex: 868371 Attn. SYTECH
121