Subject Grouping: Technology

Computing in high energy physics has changed over the years from being something one did on a slide-rule, through early computers, then a necessary evil to the position today where computers permeate all aspects of the subject from control of the apparatus to theoretical lattice gauge calculations.

The state of the art, as well as new trends and hopes, were reflected in this year’s ‘Computing In High Energy Physics’ conference held in the dreamy setting of Oxford’s spires. The 260 delegates came mainly from Europe, the US, Japan and the USSR. Accommodation and meals in the unique surroundings of New College provided a special atmosphere, with many delegates being amused at the idea of a 500-year-old college still meriting the adjective ‘new’.

The conference aimed to give a comprehensive overview, entailing a heavy schedule of 35 plenary talks plus 48 contributed papers in two afternoons of parallel sessions. In addition to high energy physics computing, a number of papers were given by experts in computing science, in line with the conference’s aim – ‘to bring together high energy physicists and computer scientists’.

The complexity and volume of data generated in particle physics experiments is the reason why the associated computing problems are of interest to computer science.These ideas were covered by David Williams (CERN) and Louis Hertzberger (Amsterdam) in their keynote addresses.

The task facing the experiments preparing to embark on CERN’s new LEP electron–positron collider is enormous by any standards but a lot of thought has gone into their preparation. Getting enough computer power is no longer thought to be a problem but the issue of storage of the seven Terabytes of data per experiment per year makes computer managers nervous even with the recent installation of IBM 3480 cartridge systems.

With the high interaction rates already achieved at the CERN and Fermilab proton–antiproton colliders and orders of magnitude more to come at proposed proton colliders, there are exciting areas where particle physics and computer science could profitably collaborate. A key area is pattern recognition and parallel processing for triggering. With ‘smart’ detector electronics this ultimately will produce summary information already reduced and fully reconstructed for events of interest.

Is the age of the large central processing facility based on mainframes past? In a provocative talk Richard Mount (Caltech) claimed that the best performance/cost solution was to use powerful workstations based on reduced instruction set computer (RISC) technology and special purpose computer-servers connected to a modest mainframe, giving maybe a saving of a factor of five over a conventional computer centre.

Networks bind physics collaborations together, but they are often complicated to use and there is seldom enough capacity. This was well demonstrated by the continuous use of the conference electronic mail service provided by Olivetti-Lee Data. The global village nature of the community was evident when the news broke of the first Z particle at Stanford’s new SLC linear collider.

The problems and potential of networks were explored by François Fluckiger (CERN), Harvey Newman (Caltech) and James Hutton (RARE). Both Fluckiger and Hutton explained that the problems are both technical and political, but progress is being made and users should be patient. Network managers have to provide the best service at the lowest cost. HEPNET is one of the richest structures in networking, and while OSI seems constantly to remain just around the corner, a more flexible and user-friendly system will emerge eventually.

Harvey Newman (Caltech) is not patient. He protested that physics could be jeopardized by not moving to high speed networks fast enough. Megabit per second capacity is needed now and Gigabit rates in ten years. Imagination should not be constrained by present technology.

This was underlined in presentations by W. Runge (Freiburg) and R. Ruehle (Stuttgart) of the work going on in Germany on high bandwidth networks. Ruehle concentrated on the Stuttgart system providing graphics workstation access through 140 Mbps links to local supercomputers. His talk was illustrated by slides and a video showing what can be done with a local workstation online to 2 Crays and a Convex simultaneously! He also showed the importance of graphics in conceptualization as well as testing theory against experiment, comparing a computer simulation of air flow over the sunroof of a Porsche with a film of actual performance.

Computer graphics for particle physics attracted a lot of interest. David Myers (CERN) discussed the conflicting requirements of software portability versus performance and summarized with ‘Myers’ Law of Graphics’ – ‘you can’t have your performance and port it’. Rene Brun (CERN) gave a concise presentation of PAW (Physics Analysis Workstation). Demonstrations of PAW and other graphics packages such as event displays were available at both the Apollo and DEC exhibits during the conference week. Other exhibitors included Sun, Meiko, Caplin and IBM. IBM demonstrated the power of relational database technology using the online Oxford English Dictionary.

Interactive data analysis on workstations is well established and can be applied to all stages of program development and design. Richard Mount likened interactive graphics to the ‘oscilloscope’ of software development and analysis.

Establishing a good data structure is essential if flexible and easily maintainable code is to be written. Paulo Palazzi (CERN) showed how interactive graphics would enhance the already considerable power of the entity-relation model as realized in ADAMO. His presentation tied in very well with a fascinating account by David Nagel (Apple) of work going on at Cupertino to extend the well-known Macintosh human/computer interface, with tables accessed by the mouse, data highlighted and the corresponding graphical output shown in an adjacent window.

The importance of the interface between graphics and relational databases was also emphasized in the talk by Brian Read (RAL) on the special problems faced by scientists using database packages, illustrated by comparing the information content of atmospheric ozone concentration from satellite measurements in tabular and graphical form – ‘one picture is worth a thousand numbers’.

The insatiable number-crunching appetite of both experimental and theoretical particle physicists has led to many new computer architectures being explored. Many of them exploit the powerful new (and relatively cheap) RISC chips on the market. Vector supercomputers are very appropriate for calculations like lattice gauge theories but it has yet to be demonstrated that they will have a big impact on ‘standard’ codes.

An indication of the improvements to be expected – perhaps a factor of five – came in the talk by Bill Martin (Michigan) on the vectorization of reactor physics codes. Perhaps more relevant to experimental particle physics are the processor ‘farms’ now well into the second generation.

Paul Mackenzie (Fermilab) and Bill McCall (Oxford) showed how important it is to match the architecture to the natural parallelism of a problem and how devices like the transputer enable this to be done. On a more speculative note Bruce Denby (Fermilab) showed the potential of neural networks for pattern recognition. They may also provide the ability to enable computers to learn. Such futuristic possibilities were surveyed in an evening session by Phil Treleavan (London). Research into this form of computing could reveal more about the brain as well as help with the new computing needs of future physics.

The importance of UNIX

With powerful new workstations appearing almost daily and with novel architecture in vogue, a machine-independent operating system is obviously attractive. The main contender is UNIX. Although nominally machine independent, UNIX comes in many implementations and its style is very much that of the early 7Os with cryptic command acronyms – few concessions to user-friendliness! However it does have many powerful features and physicists will have to come to terms with it to exploit modern hardware.

Dietrich Wiegandt (CERN) summarized the development of UNIX and its important features – notably the directory tree structure and the ‘shells’ of command levels. An ensuing panel discussion chaired by Walter Hoogland (NIKHEF) included physicists using UNIX and representatives from DEC, IBM and Sun. Both DEC and IBM support the development of UNIX systems.

David McKenzie of IBM believed that the operating system should be transparent and looked forward to the day when operating systems are ‘as boring as a mains wall plug’. (This was pointed out to be rather an unfortunate example since plugs are not standard!)

The insatiable number-crunching appetite of both experimental and theoretical particle physicists has led to many new computer architectures being explored

Two flavours of UNIX are available – Open Software Foundation version one (OSF1), and UNIX International release four (SVR4) developed at Berkeley. The two implementations overlap considerably and a standard version will emerge through user pressure. Panel member W. Van Leeuwen (NIKHEF) announced that a particle physics UNIX group had been set up (contact HEPNIX at CERNVM).

Software engineering became a heated talking point as the conference progressed. Two approaches were suggested: Carlo Mazza, head of the Data Processing Division of the European Space Operations Centre, argued for vigorous management in software design, requiring discipline and professionalism, while Richard Bornat (‘SASD- All Bubbles and No Code’) advocated an artistic approach, likening programming to architectural design rather than production engineering. A. Putzer (Heidelberg) replied that experimenters who have used software engineering tools such as SASD would use them again.

Software crisis

Paul Kunz (SLAC) gave a thoughtful critique of the so-called ‘software crisis’, arguing that code does not scale with the size of the detector or collaboration. Most detectors are modular and so is the code associated with them. With proper management and quality control good code can and will be written. The conclusion is that both inspiration and discipline go hand in hand.

A closely related issue is that of verifiable code – being able to prove that the code will do what is intended before any executable version is written. The subject has not yet had much impact on the physics community and was tackled by Tony Hoare and Bernard Sufrin (Oxford) at a pedagogical level. Sufrin, adopting a missionary approach, showed how useful a mathematical model of a simple text editor could be. Hoare demonstrated predicate calculus applied to the design of a tricky circuit.

Less high technology was apparent at the conference dinner in the candle-lit New College dining hall. Guest speaker Geoff Manning, former high energy physicist, one-time Director of the UK Rutherford Appleton Laboratory and now Chairman of Active Memory Technology, believed that physicists have a lot to learn from advances in computer science but that fruitful collaboration with the industry is possible. Replying for the physicists, Tom Nash (Fermilab) thanked the industry for their continuing interest and support through joint projects and looked forward to a collaboration close enough for industry members to work physics shifts and for physicists to share in profits!

Summarizing the meeting, Rudy Bock (CERN) highlighted novel architectures as the major area where significant advances have been made and will continue to be made for the next few years. Standards are also important provided they are used intelligently and not just as a straitjacket to impede progress. His dreams for the future included neural networks and the wider use of formal methods in software design. Some familiar topics of the past, including code and memory management and the choice of programming language, could be ‘put to rest’.

During the past few years, electronic circuitry techniques have been developed which enable complex logic units to be produced as tiny elements or ‘chips’. These units are now mass-produced, and are available relatively cheaply to anyone building data processing equipment.

Just a few years ago, the first complete processing unit on a single chip was produced. Now micro logic elements can be combined together to provide micro data processing systems whose capabilities in certain respects can rival those of more conventional computers. Commercially- available microcomputers are used widely in many fields.

Where an application requires special capabilities, it is preferable to take the individual micro logic units and wire them together on a printed circuit board to provide a tailor-made processing unit: If there is sufficient demand for the perfected design, the printed circuit board stage subsequently can be dispensed with and the processor can be mass-produced by large-scale integration (LSI) techniques as a single microprocessor.

With these processing units, there generally is a trade-off between speed and flexibility, the ultimate in speed being a hard-wired unit which is only capable of doing one thing. Flexibility can be achieved through programmable logic, but this affects the overall speed.

Programming micros is difficult, but one way of sidestepping these problems would be to design a unit which emulates a subset of an accessible mainframe computer. With such an emulator, programs could be developed on the main computer, and transferred to the micro after they have reached the required level of reliability. This could result in substantial savings in program development time. In addition, restricting the design to a subset of the mainframe architecture results in a dramatic reduction in cost.

High energy physics, which has already amply demonstrated its voracious appetite for computer power, could also soon cash in on this microcomputer boom and produce its own ‘brand’ of custom-built microprocessors.

According to Paolo Zanella, Head of CERN’s Data Handling Division, now is the time to explore in depth the uses of microprocessors in high energy physics experiments. If initial projects now under way prove to be successful, the early 1980s could see microprocessors come into their own.

One of the biggest data processing tasks in any physics experiment is to sift through the collected signals from the various detecting units to reject spurious information and separate out events of interest. Therefore to increase the richness of the collected data, triggering techniques are used to activate the data collection system of an experiment only when certain criteria are met.

Even with the help of this ‘hardwired’ selection, a large proportion of the accumulated data has to be thrown away, often after laborious calculations. With experiments reaching for higher energies where many more particles are produced, and at the same time searching for rarer types of interaction, physicists continually require more and more computing power.

Up till now, this demand has had to be met by bringing in more and bigger computers, both on-line at the experiments and off-line at Laboratory computer centres. With the advent of microprocessors, a solution to this problem could be in sight. Micros could be incorporated into experimental set-ups to carry out a second level of data selection after the initial hard-wired triggering – an example of the so called ‘distributed processing’ approach where computing power is placed as far upstream as possible in the data handling process. In this way the demand on the downstream central computer would be reduced, and the richness of the data sample increased.

The micros would filter the readout in the few microseconds before the data is transferred to the experimental data collection system. Zanella is convinced that this could significantly improve the quality of the data and reduce the subsequent off-line processing effort to eliminate bad triggers.

As well as being used in the data collection system, micros would also be useful for control and monitoring functions. The use of off the-shelf microcomputers in accelerator control systems, for example, is already relatively widespread. Some limited applications outside the control area are already being made in experiments, a notable example being the CERN/Copenhagen/Lund/Rutherford experiment now being assembled at the CERN Intersecting Storage Rings.

Microcomputer projects are now being tackled at several Laboratories. At CERN three projects are under way in the Data Handling Division. Two of these are programmable emulators (one being based on the IBM 370/168 and the other on the Digital Equipment PDP-11), while the third is a very fast microprogrammable unit called ESOP.

High energy physics still has a lot to learn about microprocessor applications, and there is some way to go before their feasibility is demonstrated and practical problems, such as programming, are overcome.

However this year could see some of these initial projects come to fruition, and the early 1980s could live up to Zanella’s expectations as the time when the microprocessor becomes a routine part of the high energy physicists’ toolkit. 

CERN is the favourite showpiece of international co-operation in advanced scientific research. The public at large is, by now, quite used to the paradox of CERN’s outstandingly large-scale electromagnetic machines (accelerators) being needed to investigate outstandingly small-scale physical phenomena. A visitor finds it natural that this, largest-in-Europe, centre of particle research should possess the largest, most complex and costly accelerating apparatus.

But when told that CERN is also the home of the biggest European collection of computers, the layman may wonder: why is it precisely in this branch of knowledge that there is so much to compute? Some sciences such as meteorology and demography appear to rely quite naturally on enormously vast sets of numerical data, on their collection and manipulation. But high energy physics, not so long ago, was chiefly concerned with its zoo of ‘strange particles’ which were hunted and photographed like so many rare animals. This kind of preoccupation seems hardly consistent with the need for the most powerful ‘number crunchers’.

Perplexities of this sort may arise if we pay too much attention to the (still quite recent) beginnings of the modern computer and to its very name. Electronic digital computers did originate in direct descent from mechanical arithmetic calculators; yet their main function today is far more significant and universal than that suggested by the word ‘computer’. The French term ‘ordinateur’ or the Italian ‘elaboratore’ are better suited to the present situation and this requires some explanation.

What is a computer?

When, some forty years ago, the first attempts were made to replace number-bearing cogwheels and electro-mechanical relays by electronic circuits, it was quickly noticed that, not only were the numbers easier to handle if expressed in binary notation (as strings of zeros and ones) but also that the familiar arithmetical operations could be presented as combinations of two-way (yes or no) logical alternatives. It took some time to realize that a machine capable of accepting an array of binary-coded numbers, together with binary-coded instructions of what to do with them (stored program) and of producing a similarly coded result, would also be ready to take in any kind of coded information, to process it through a prescribed chain of logical operations and to produce a structured set of yes-or-no conclusions. Today a digital computer is no longer a machine primarily intended for performing numerical calculations; it is more often used for non-numerical operations such as sorting, matching, retrieval, construction of patterns and making decisions which it can implement even without any human intervention if it is directly connected to a correspondingly structured set of open-or-closed switches.

Automatic ‘black boxes’ capable of producing a limited choice of responses to a limited variety of input (for example, vending machines or dial telephones) were known before; their discriminating and logical capabilities had to be embodied in their rigid internal hardware. In comparison, the computer can be seen as a universally versatile black box, whose hardware responds to any sequence of coded instructions. The complication and the ingenuity are then largely transferred into the writing of the program.

The new black box became virtually as versatile as the human brain; at the same time it offered the advantages of enormously greater speed, freedom from error and the ability to handle, in a single operation, any desired volume of incoming data and of ramified logical chains of instructions. In this latter respect the limit appears to be set only by the size (and therefore the cost) of the computer, i.e. by the total number of circuit elements which are brought together and interconnected in the same computing assembly.

High energy physics as a privileged user

We are only beginning to discover and explore the new ways of acquiring scientific knowledge which have been opened by the advent of computers. During the first two decades of this exploration, that is since 1950, particle physics happened to be the most richly endowed domain of basic research. Secure in their ability to pay, high energy physicists were not slow to recognize those features of their science which put them in the forefront among the potential users of the computing hardware and software in all of their numerical and non-numerical capabilities. The three most relevant characteristics are as follows:

1. Remoteness from the human scale of natural phenomena

Each individual ‘event’ involving sub-nuclear particles takes place on a scale of space, time and momentum so infinitesimal that it can be made perceptible to human senses only through a lengthy and distorting chain of larger-scale physical processes serving as amplifiers. The raw data supplied by a high energy experiment have hardly any direct physical meaning; they have to be sorted out, interpreted and re-calculated before the experimenter can see whether they make any sense at all – and this means that the processing has to be performed, if possible, in the ‘real time’ of the experiment in progress, or at any rate at a speed only a computer can supply.

2. The rate and mode of production of physical data

Accelerating and detecting equipment is very costly and often unique; there is a considerable pressure from the user community and from the governments who invest in this equipment that it should not be allowed to stand idle. As a result, events are produced at a rate far surpassing the ability of any human team to observe them on the spot. They have to be recorded (often with help from a computer) and the records have to be scanned and sifted – a task which, nowadays, is usually left to computers because of its sheer volume. In this way, experiments in which a prolonged ‘run’ produces a sequence of mostly trivial events, with relatively few significant ones mixed in at random, become possible without wasting the valuable time of competent human examiners.

3. High statistics experiments

As the high energy physics community became used to computer-aided processing of events, it became possible to perform experiments whose physical meaning resided in a whole population of events, rather than in each taken singly. In this case the need grew from an awareness of having the means to satisfy the need; a similar evolution may yet occur in other sciences (e.g. those dealing with the environment), following the currents of public attention and possibly de-throning our physics from pre-eminence in scientific computation.

Modes of application

In order to stress here our main point, which is the versatility of the modern computer and the diversity of its applications in a single branch of physical research, we shall classify all the ways in which the ‘universal black box’ can be put to use in CERN’s current work into eight ‘modes of application’ (roughly corresponding to the list of ‘methodologies’ adopted in 1968 by the U.S. Association for Computing Machinery):

1. Numerical mathematics

This mode is the classical domain of the ‘computer used as a computer’ either for purely arithmetic purposes or for more sophisticated tasks such as the calculation of less common functions or the numerical solution of differential and integral equations. Such uses are frequent in practically every phase of high energy physics work, from accelerator design to theoretical physics, including such contributions to experimentation as the kinematic analysis of particle tracks and statistical deductions from a multitude of observed events.

2. Data processing

Counting and measuring devices used for the detection of particles produce a flow of data which have to be recorded, sorted and otherwise handled according to appropriate procedures. Between the stage of the impact of a fast-moving particle on a sensing device and that of a numerical result available for a mathematical computation, data processing may be a complex operation requiring its own hardware, software and sometimes a separate computer.

3. Symbolic calculations

Elementary logical operations which underline the computers’ basic capabilities are applicable to all sorts of operands such as those occurring in algebra, calculus, graph theory, etc. High-level computer languages such as LISP are becoming available to tackle this category of problems which, at CERN, is encountered mostly in theoretical physics but, in the future, may become relevant in many other domains such as apparatus design, analysis of track configurations, etc.

4. Computer graphics

Computers may be made to present their output in a pictorial form, usually on a cathode-ray screen. Graphic output is particularly suitable for quick communication with a human observer and intervener. Main applications at present are the study of mathematical functions for the purposes of theoretical physics, the design of beam handling systems and Cherenkov counter optics and statistical analysis of experimental results.

5. Simulation

Mathematical models expressing ‘real world’ situations may be presented in a computer-accessible form, comprising the initial data and a set of equations and rules which the modelled system is supposed to follow in its evolution. Such ‘computer experiments’ are valuable for advance testing of experimental set-ups and in many theoretical problems. Situations involving statistical distributions may require, for their computer simulation, the use of computer-generated random numbers during the calculation. This kind of simulation, known as the Monte-Carlo method, is widely used at CERN.

6. File management and retrieval

As a big organization, CERN has its share of necessary paper-work including administration (personnel, payroll, budgets, etc.), documentation (library and publications) and the storage of experimental records and results. Filing and retrieval of information tend nowadays to be computerized in practically every field of organized human activity; at CERN, these pedestrian applications add up to a non-negligible fraction of the total amount of computer use.

7. Pattern recognition

Mainly of importance in spark-chamber and bubble-chamber experiments – the reconstruction of physically coherent and meaningful tracks out of computed coordinates and track elements is performed by the computer according to programmed rules.

8. Process control

Computers can be made to follow any flow of material objects through a processing system by means of sensing devices which, at any moment, supply information on what is happening within the system and what is emerging from it. Instant analysis of this information by the computer may produce a ‘recommendation of an adjustment’ (such as closing a valve, modifying an applied voltage, etc.) which the computer itself may be able to implement. Automation of this kind is valuable when the response must be very quick and the logical chain between the input and the output is too complicated to be entrusted to any rigidly constructed automatic device. At CERN the material flow to be controlled is usually that of charged particles (in accelerators and beam transport systems) but the same approach is applicable in many domains of engineering, such as vacuum and cryogenics.

Centralization versus autonomy

The numerous computers available at CERN are of a great variety of sizes and degrees of autonomy, which reflects the diversity of their uses. No user likes to share his computer with any other user; yet some of his problems may require a computing system so large and costly, that he cannot expect it to be reserved for his exclusive benefit nor to be kept idle when he does not need it. The biggest computers available at CERN must perforce belong to a central service, accessible to the Laboratory as a whole. In recent years, the main equipment of this service has consisted of a CDC 6600 and CDC 6500. The recent arrival of a 7600 (coupled with a 6400) will multiply the centrally available computing power by a factor of about five.

We are only beginning to discover and explore the new ways of acquiring scientific knowledge which have been opened by the advent of computers

For many applications, much smaller units of comp­uting power are quite adequate. CERN possesses some 80 other computers of various sizes, some of them for use in situations where autonomy is essential (for example, as an integral part of an experimental set-up using electronic techniques or for process control in accelerating systems). In some applications there is need for practically continuous access to a smaller computer together with intermittent access to a larger one. A data-conveying link between the two may then become necessary.

Conclusion

The foregoing remarks are meant to give some idea of how the essential nature of the digital computer and that of high energy physics have blended to produce the present prominence of CERN as a centre of computational physics. The detailed questions of ‘how’ and ‘what for’ are treated in the other articles of this [special March 1972 issue of CERN Courier devoted to computing] concretely enough to show the way for similar developments in other branches of science. In this respect, as in many others, CERN’s pioneering influence may transcend the Organization’s basic function as a centre of research in high energy physics.

The popular picture of a biologist shows him wearing a white coat, peering through a microscope. If it is at all possible to draw a popular picture of an experimental high-energy physicist, he will probably be sitting at his desk and looking at his output from the computer. For the computer is increasingly becoming the tool by which raw experimental results are made intelligible to the physicist.

By one of those apparent strokes of luck that occur so often in the history of science, electronic digital computers became available to scientists at just that stage in the development of fundamental physics when further progress would otherwise have been barred by the lack of means to perform large-scale calculations. Analogue computers, which had been in existence for a number of decades previously, have the disadvantage for this work of limited accuracy and the more serious drawback that a more complicated calculation needs a more complicated machine. What was needed was the equivalent of an organized team of people, all operating desk calculating machines, so that a large problem in computation could be completed and checked in a reasonable time. So much was this need felt, that in England, for example, before digital computers became generally available, there was a commercial organization supplying just such a service of hand computation.

At CERN the requirement for computing facilities in the Theory Division was at first largely satisfied by employing a calculating prodigy, Willem Klein. Mr. Klein is one of those rare people who combine a prodigious memory with a love of numbers, and it was some time before computers in their ever-increasing development were able to catch up with him. He is still with us in the Theory Division, giving valuable help to those who need a quick check calculation. It is of interest to note, however, that he has now added a knowledge of computer programming to his armoury of weapons for problem solution!

The first computer calculations made at CERN were also done in the very earliest days of the Organization. Even before October 1958, when the Ferranti Mercury computer was installed, computer work had been sent out to an English Electric Deuce in Teddington, an IBM 704 in Paris and Mercury computers at Harwell and Manchester. Much of this work was concerned with orbit calculations for the proton synchrotron, then being built.

We have now reached a stage at which there is hardly a division at CERN not using its share of the available computer time. On each occasion that a new beam is set up from the accelerator, computer programmes perform the necessary calculations in particle optics as a matter of routine; beam parameters are kept in check by statistical methods; hundreds of thousands of photographs from track-chamber experiments are ‘digitized’ and have kinematic and statistical calculations performed on them; the new technique of sonic spark chambers, for the filmless detection of particle tracks, uses the computer more directly. In addition, more than 80 physicists and engineers use the computer on their own account, writing programmes to solve various computational problems that arise in their day-to-day work.

There are now two computers at CERN, the original Ferranti Mercury and the IBM 7090, a transistorized and more powerful replacement for its predecessor, the IBM 709. The 7090, in spite of its great speed (about 100 000 multiplications per second!), is rapidly becoming overloaded and is to be replaced towards the end of this year by a CDC 6600, which at present is the most powerful computing system available in the world.

It is hoped that this new machine will satisfy the computing needs of CERN for upwards of five years. The Mercury computer is now being used more and more as an experimental machine and there is, for instance, a direct connexion to it at present from a spark-chamber experiment at the proton synchrotron. Calculations are performed and results returned to the experimental area immediately, giving great flexibility.

Track-chamber photographs

Among the biggest users of computer time are the various devices for converting the information on bubble-chamber and spark-chamber photographs, usually on 35-mm film, into a form in which the tracks of the particles can be fitted with curves and the entire kinematics of an event subsequently worked out. To this end, from the earliest days of CERN, IEPs (instruments for the evaluation of photographs) [rumour once had it that IEP stood for ‘instrument for the elimination of physicists’!] have been built and put into use. These instruments enable accurately measured co-ordinates of points on a track, together with certain identifying information, to be recorded on punched paper tape. Their disadvantage is that measuring is done manually, requires skill and, even with the best operator, is slow and prone to errors. The paper tapes produced have to be copied on to a magnetic tape, checking for various possible errors on the way, and the magnetic tape is then further processed to provide in turn geometric, kinematic and statistical results.

It was recognized at an early stage, both in Europe and in the United States, that for experiments demanding the digitization of very large numbers of pictures, for example those requiring high statistical accuracy, some more-automatic picture-reading equipment would be needed. Two such devices are now coming into use at CERN. One, the Hough-Powell device (known as HPD), developed jointly by CERN, Brookhaven, Berkeley and the Rutherford Laboratory, is an electro-mechanical machine of high precision, which still requires a few pilot measurements to be made manually, on a measuring table named ‘Milady’, when used for bubble-chamber pictures. It has already been used in one experiment, for the direct processing of 200 000 spark-chamber photographs (for which the pilot measurements are unnecessary). The other device is called ‘Luciole’, a faster, purely electronic machine, although of lower precision, specially developed at CERN for digitizing spark-chamber photographs.

With the sonic spark chamber, the position of the spark between each pair of plates is deduced from the time intervals between its occurrence and the detection of the sound by each of four microphones. Arrays of such devices can be connected directly to the computer, thus dispensing with the taking, developing and examination of photographs.

The study of dynamics of particles in magnetic and electric fields gives rise to another important family of computer programmes. The electron storage ring, or beam-stacking model, required the writing of a programme that followed the motion of individual batches of particles during their acceleration and stacking in the ring. A previous study by the same group resulted in a series of programmes to examine the behaviour and stability of a proposed fixed-field, alternating-gradient stacking device. Various aspects of the performance of the linac (the linear accelerator that feeds the PS) have been studied and improved using the computer, and the later stages of the design of the PS itself involved a detailed computer simulation of the beam in the ring, including the various transverse or ‘betatron’ oscillations to which it is subjected.

There also exists a series of particle-optics programmes used for the design and setting up of particle beams, particularly the ‘separated’ beams producing particles of only one kind. These are ‘production’ calculations, in the sense that the programme is run with new parameters every time there is a major change in beam layout in any of the experimental halls.

Computer language

At first, a major bar to the use of computers for small, but important, calculations was the difficulty of programming them in their own special ‘language’ to solve the specific problem in hand. What is the use of a machine that can perform a particular calculation in a minute, the would-be-user asks, if it will take a month to provide the programme for the calculation? Given a hand calculating machine, a pencil and paper, and a quiet room, I can do it myself in three weeks! This valid argument limited the use of computers to two kinds of calculation: those too long to be performed by hand, and those that had to be carried out so often that the original effort of producing the programme was justified.

This situation was rectified by the use of ‘programming languages’, which make it possible to express one’s problem in a form closely resembling that of mathematics. Such languages, if defined rigorously enough, express the problem unambiguously, and they can be translated automatically (by the computer) into the instructions for a particular computer. Two such languages have been used at CERN: Mercury Autocode, and Fortran. The use of Mercury Autocode has recently been discontinued, but until a short time ago many physicists used both languages with great success to express their computational problems in a form directly comprehensible by a computer. Courses in the Fortran language, given both in English and French, are held regularly, and usually last about three weeks. Such ‘compiler languages’, as they are called, used to be considered a rather inferior method for programming computers, as the translations obtained from them often used the machine at a low efficiency, but it is now recognized that the advantage of writing programmes in a language that can be translated mechanically for a number of different computers far outweighs a small loss in programme efficiency.

In the Theory Division, computer programmes are often written by individual theorists to check various mathematical models. Having devised a formula based on a novel theory, the physicist computes theoretical curves for some function that can be compared with experimental results.

Operation

The Data Handling Division, which actually has CERN’s central computers under its charge, contains a number of professional programmers, mostly mathematicians by training. Some of these are responsible for the ‘systems programmes’, that is, the Fortran compiler and its associated supervisor programme. Some have the job of disseminating programming knowledge, helping individual users of the computer and writing programmes for people in special cases. Others are semi-permanently attached to various divisions, working on particular experiments as members of the team.

A small number of mathematicians are also engaged in what might be called ‘specialist computer research’, covering such things as list-programming languages and methods of translating from one programme language into another. Such work might be expected to yield long-term profits by giving increases in computing power and efficiency.

As at all large computing installations, computer programmers at CERN do not operate the machine themselves

As at all large computing installations, computer programmers at CERN do not operate the machine themselves. Data and programmes are submitted through a ‘reception office’ and the results are eventually available in a ‘computer output office’, leaving the handling and organizing of the computer work-load, and the operating of the machine, to specialist reception staff and computer operators in the Data Handling Division.

What of the future of computers at CERN? In a field as new as this, predictions are even more dangerous than in others, but it is clear that the arrival of the new computer at the end of the year will cause a great change in the way computers are used. Having ten peripheral processors, each of which is effectively an independent computer, the machine may have many pieces of equipment for data input and output attached to it ‘on-line’. The old concept that a computer waiting for the arrival of data is standing idle, and that this is therefore wasteful and expensive, need no longer be true. With the new system, a computer that is waiting for new data for one problem is never idle, but continues with calculations on others. Every moment of its working day is gainfully employed on one or other of the many problems it is solving in parallel. Even so, there will still be a need for a number of smaller computer installations forming part of particular experiments.

As M.G.N. Hine, CERN’s Directorate Member for Applied Physics, pointed out at a recent conference, even with the growth of such facilities, the amount of computing time available may one day dictate the amount of experimental physics research done at CERN, in much the same way as the amount of accelerator time available dictates it now.