--Winter 1988 California Institute of Technology In This Issue
Light Fantastic But Not Escher or Bach A little over three years ago the Olga Taussky-Todd met Kurt American Physical Society G6del in 1925 at the University appointed a study group of of Vienna where they were both scientists to look into the tech students. G6del went on to nological aspects (not the ethical prove the existence of undecid or philosophical ones) of the able mathematical statements Strategic Defense Initiative and and to become one of the giants to present a report to the of 20th-century mathematics. society's membership. That Although she hasn't starred report was released last April. in any bestsellers, Taussky-Todd One of the 17 members of the also went on to become a study group was Cal tech's famous mathematician (algebraic Amnon Yariv, the Thomas G. number theory, topological alge Myers Professor of Electrical bra, matrix theory). After earn Engineering and professor of ing her doctorate in 1930, she applied physics. In "Star Wars taught at Bryn Mawr College, Technology: Will It Work?" and Girton College, Cambridge, beginning on page 29, Yariv and worked for the English Min describes some basics of lasers istry of Aircraft Production dur Sun Run and missiles and then sums up ing the war. She and her hus On the cover-the solar-powered the study group's conclusions. band Jack then worked at the GM Sunraycer, which outdis The article was adapted from a National Bureau of Standards in tanced all competition in a Watson Lecture. Washington, D.C. for 10 years recent race across Australia, rests Yariv came to Caltech as (with a year off at the Institute on the grounds of Pasadena's associate professor of electrical for Advanced Study) before Tournament House before engineering in 1964. His BS coming to Caltech in 1957. undertaking the less strenuous (1954), MS (1956), and PhD Taussky-Todd has been profes chore of leading the 1988 Rose (1958) are all from Berkeley. He sor emeritus (a term she doesn't Parade as the "pace car of the has been a pioneer in integrated care for) since 1977. In 1980 future." With Sunraycer are optoelectronics-marrying lasers she was awarded a Golden Doc (from left) Peter Lissaman, Alec and electronic circuits on a sin torate from the University of Brooks, and Paul MacCready, gle semiconductor chip-and in Vienna-50 years after the who played leading roles in the phase conjugate optics-a tech original. Taussky-Todd recalls vehicle's development; all hold nique for correcting atmospheric the earlier times in "Remem PhDs in aeronautics from Cal distortion, which could provide a brances of Kurt G6del," begin tech. In addition to these three, solution for one of the problems ning on page 24. The article there was a strong Caltech com of beaming lasers over great was adapted from a talk given in ponent in the whole project distances. Salzburg in July 1983. with alumni at GM and Hughes, as well as at Aero Vironment and among its consultants. MacCready (MS 1948, PhD David Harper - President of the Alumni Association 1952) is president of AeroViron Theodore P. Hurwitz - Vice President for Institute Relations ment, Inc., a firm whose staff Robert L. O'Rourke - Director of Public Relations specializes in environmental and alternative energy projects but is STAFF: Editor - Jane Dietrich perhaps best known for creating Writer - Douglas Smith unusual aeronautical vehicles Production Artist - Barbara Wirick powered by muscle, batteries, or Business Manager - Marilee Wood solar cells. So the small Monro Circulation Manager - Susan Lee via, California, company was a Photographer - Robert paz natural place for GM to turn to for meeting the Pentax World PICTURE CREDITS: Cover, 7, 43 - Robert Paz; 2, 10, 12, 13, back cover Solar Challenge. - Paul MacCready; 7 - Alec Brooks; 4,5,7, 10, 12 - GM; 13 - Mar- MacCready's story of how tyn Cowley; 13 - DuPont; 14-21 - Photo CERN; 23 - R. Zhu; 25- the car was conceived, designed, Caltech Archives; 25-27 - Olga Taussky-Todd; 29, 30, 34 - Physics Today; and built (beginning last March) 32, 33 - Scientific American; back cover - Walt Mancini/Pasadena Star News in time for the November 1 race appears on page 2. "Sunraycer Engineering & Science (ISSN 0013-7812) is published quarterly, Fall, Winter, Spring, Odyssey" also gives an overview and Summer, at the California Institute of Technology, l201 East California Boulevard, of some of the vehicle's ad Pasadena, California 91125. Annual Subscription $8.00 domestic, $20.00 foreign air mail, single copies, $2.00. Third class postage paid at Pasadena, California. All rights vanced technology as well as an reserved. Reproduction of material contained herein forbidden without authorization. account of the race itself and the " 1988 Alumni Association California Institute of Technology. Published by the strategy that contributed to the California Institute of Technology and the Alumni Association. Telephone: 818-356- victory. 3630. Postmaster: Send change of address to Caltech, 1-71, Pasadena, CA 91125.
ENGNE~~~~
CAliFORNIA INSTITUTE OF TECHNOLOGY WINTER 1988 VOLUME LI, NUMBER 2
.sunraycer Odyssey - by Paul B. lvlacCready Page 2 An innovative solar-powered car was created and built in just eight months; one of Sunraycer's creators describes its genesis and its victory in a race across Australia.
Quark Quest Page 14 Scientists may find the top quark at LEP, an accelerator under construction at CERN.
Remembrances of Kurt Giidel - by Olga Taussky-Todd Page 24 A Caltech mathematician recalls the late 1920s and early 1930s at the University of Vienna, when her friend and fellow student proved undecidability.
Star Wars Technology: Will It Work? - by Amnon Yariv Page 29 Conclusions of the American Physical Society's study group on directed-energy weapons are explained by a member of that committee.
Departments
Research in Progress Page 37 Photographic Memory Computer's- Eye View
Books Page 41
Random Walk Page 42 Sunraycer Odyssey Winning the Solar-Powered Car Race Across Australia
by Paul B. M acCready
~R DAYS WE HAl) BEEN TRA VEUNG south on the desk of General Motors Chairman Roger r Australia's deserted Stuart Highway, but Smith. Smith found the concept intriguing now crowds of spectators lined the road for and sent the invitation on to GM's subsidi the final few kilometers to watch the GM ary, Hughes Aircraft, for consideration of the Sunraycer, powered by sunbeams, win the project's feasibility. From GM's standpoint, 3,005-kilometer (l,867-mile) race from developing and racing such a vehicle would Darwin to Adelaide. Sunraycer completed serve to focus technological developments the Pentax World Solar Challenge course within the whole company, would make across the continent in 44.9 hours of running GM's technological capabilities more evident time during 51/4 days; the car's speed averaged to the public, would fit GM's racing philoso 66.9 kilometers per hour (41.6 mph), 50 per phy, and would attract students to engineer cent faster than the runner-up; the average ing as an exciting and rewarding career. electric power to the motor was just a bit over This article gives an overview of the 1,000 watts (11/3 horsepower). Of the 24 solar GM-Sunraycer story and my personal view of powered vehicles that started out from some of the main issues. More detailed treat Darwin on November 1, 1987 (nine from ments of the technology and of the race are Australia, four from Japan, four from the available elsewhere. Many people contributed United States, three from Germany, two from significantly to the success of Sunraycer, al Switzerland, and one each from Denmark though space allows me to cite only a few of and Pakistan), 13 completed the course. The them here. runner-up from Ford of Australia finished 23 My company's involvement began last Besides creaLing interesting hours behind Sunraycer, and only six had February when Edmund Ellion of Hughes photographic opportunities, the gold plating on arrived by the time of the banquet and prize (whom I knew 35 years ago at Caltech) con Sunraycer's canopy (leji) keeps ceremony November 13. The last one finally tacted me to explore whether Aero Vironment out 98 percent q{ the infrared reached Adelaide December 2. (A V), with its reputation for developing and 90 percent of the visible The story of the race began in 1982 when unusual low-powered vehicles, could help. radiation. As a result of this the Australian visionary and adventurer Hans In early March Hughes and A V began an shielding, plus a bit of ventila tion, the driver remains com Tholstrup drove a pioneering solar-powered intense three-week program-planning effort. jimable even on the hottest car slowly from Perth to Sydney. Soon after The basic overall design had to be esta days. From inside, the visibil ward he had the idea that a dramatic com blished during this three-week phase. Engi ity is surprisingly good; some petition would stimulate global interest in this neers always wish for adequate time to deli times a driver would even inexhaustible and nonpolluting energy re berate and explore alternatives before irrevo wear dark glasses. Such a windscreen wouldn'l be suil source, and he acted on the idea. In 1986 he cable decisions must be made, but the real able jl)r nighl driving, but sent out invitations to the 1987 World Solar world rarely grants that wish. We quickly thaL's of small concern for a Challenge. One of these invitations reached considered the trade-offs involved in a dozen solar-powered car.
3 different configurations: different shapes and orientations of solar panels, various body shapes, and the location and structural sup port of the wheels (and whether to have three or four of them). We explored virtually every vehicle configuration that finally showed up at Darwin. Our study supported the advan tages of what was to become the Sunraycer concept, which emphasizes low aerodynamic drag and invulnerability to crosswinds even at the sacrifice of some solar power from addi tional or tilting panels. Almost all the other cars in the race had solar panels distinct from the car bodies. Although their frontal areas could be small, the vehicles suffered from the drag associated with large wetted areas, as well as the interference drag of separate Alan Cocconi (left) and Alec wheels and connecting structures exposed to Brooks check a power the wind. Such vehicles also had to cope electronics module prototype. THE CONNECTION with some structural complexities. For tunately, our concept happened to result in a design that was not only efficient but also looked dramatically super-streamlined-like something James Bond might drive. Essential to our design was the peak power tracker, a prototype of which was designed, built, and demonstrated during that PhD '81) was Sun three-week stage by AV's electronics consul raycer manager and also one of the tant, Alan Cocconi. The peak power tracker six drivers in the race; and Bart Hibbs (BS continually adjusts the voltage of an array of '77) served as lead ael·oayn: 4 ENGINEERING & SCIENCE I WINTER 1988 refined by wi~d~tunnel tests-and then built. The crosswind safety problem loomed large SOLAR CHALLENGE RULES during the configuration-planning stage and subsequently during the refinement of the shape by computer, wind tunnel, and field tests. For a lightweight vehicle operating occasiOlially in the 70- to lOO-km/h regime and encountering real-world winds, our pre liminary calculations indicated that a tilted flat solar panel would be out of the question. A number of vehicles of such configuration did finish the Australian race without mishap, but they were slower and mostly heavier and sometimes frightening to watch as they jittered along in a crosswind. Even a non tilted flat panel above the body was con sidered dangerous because of the lift that could be generated at high speeds from momentary gusts (especially the up-gusts from the vortices in the wakes of the enor mous trucks that ply Australia's outback region). Sunraycer's basic body shape is rela tively impervious to crosswinds; but the four wheels, especially if covered with steamlined pants, tend to straighten the crossflow under the vehicle, creating a sidt- force and also dis turbing the balance between the vertical force on the upper and lower rear half. The side force is associated with some beneficial thrust, as with a sailboat, but for the wind speeds expected in Australia and the anticipated vehicle speeds the net speed bene fit would be insignificant. And there are drawbacks to consider. A side force adds a This quarter-scale model of bit to tire drag and wear. Much more trou Sunraycer scored the lowest blesome is the possibility that the side force drag coefficient ever recorded will be enough to destabilize the vehicle and for a land vehicle in Caltech's even blow it off the road. lO-foot wind tunnel. From left Also, at high vehicle speeds the vertical to right Bart Hibbs, Max Schenkel, and Kent Kelly view force from strong crosswinds could lighten the the tests. load on the rear tires enough to make the vehicle yaw. The two strakes, or fins, on Sunraycer's top are there to help solve this problem. They serve somewhat as flow straighteners for the top to balance partially the flow straightening produced by the wheels on the underside. They may also serve as vortex generators to modify boundary-layer separation. Tests in Caltech's lO-foot wind tunnel showed that the strakes yielded the desired result, while the many other con figurations of add-on devices we tried-fins and spoilers that seemed more logical-did not do as well. There was no time to explore the matter more deeply; the empirical solu tion was adopted. Of course, like all engineers, we hope someday to be able to 5 treat the subject in a more fundamental way. tion against back voltage when a portion of These strakes also offered several benefits the array is shadowed. Both solar arrays were beyond extending the envelope of safe speed designed and built by Hughes. in crosswinds. They added height, which The battery, also designed by Hughes, is allowed Sunraycer to meet the one-meter silver zinc, has 68 cells, can store three minimum height requirement while maintain kilowatt hours and weighs 27 kilograms (60 ing a slender silhouette for low drag and a lbs). It is the electrical equivalent of a lead low center of gravity for stability. They pro acid battery weighing four times as much. A vided a convenient place for tum signals. silver zinc battery tires rapidly as it is cycled a And they helped solve the problem of giving number of times, however, which lowers its the driver a rear view without the drag of an capacity, but the effect was small for the few exterior rear-view mirror. Hughes concocted cycles of the Australian race. a fiber optics remote-viewing system whereby The motor, developed by GM's Research the driver could see on a tiny screen what was Laboratories, uses Magnequench magnets observed through a lens pointed rearward from GM's Delco Remy Division. It weighs from the top of the right strake. And finally, 5 kg (11 lbs), reaches an efficiency of 92 per we think the strakes look rather jazzy, and we cent, and can deliver 4 horsepower continu would not be surprised to see their descen ously and more than 10 horsepower briefly. dants sprouting on cars (and caps), whose A cogged belt conveys its output to the left owners appreciate Sunraycer's style. rear wheel. Because of the tight schedule, many inter Twelve separate solar panel arrays are related aspects of the design had to be pur connected to the battery by peak power track sued in parallel. While the shape was being ers. The battery drives the motor with a refined, the chassis and suspension were motor drive inverter that generates three under development, the electronic system was phase AC power. The drive inverter includes being designed, new tires were being control based on constant current, constant researched, wheels and axles were being built, speed (rpm), or constant torque; its efficiency the solar cells were being ordered, and an is greater than 97 percent. expedition was mounted to Australia to The human driver has instruments to gather meteorological data and to survey the monitor the operating conditions and tem entire Stuart Highway for road hazards and peratures of all the electronic devices, and road-surface details. there is a 90-channel telemetry system de A prototype. vehicle was built containing livering all the information to the nearby 7,200 single-crystal silicon 6- by LS-cm cells. observer vehicle. Manufactured by Hughes' subsidiary Spectro Sunraycer weighs 175 kg and carries a lab (the same brand that six years earlier driver ballasted up to 85 kg for a gross weight powered our Solar Challenger on its 163-mile of 260 kg (573 lbs). The space-frame chassis, flight from Paris to England), these cells con which weighs less than 7 kg, is constructed of vert approximately 16.5 percent of the aluminum tubing. The body's exterior is incident radiation to electricity. The final made of a Kevlar-Nomex-Kevlar sandwich. race vehicle had 20 percent of its panel area The portion supporting the solar cells is built covered with these silicon cells, but the with high-temperature epoxy to maintain remaining 80 percent held 4- by 2-cm gallium structural integrity when heated strongly by arsenide cells. Half of these were obtained the sun. To keep the driver cool, the canopy from Applied Solar Energy Corporation and is gold plated. While every aspect of the Sun the other half from Mitsubishi International raycer development had its pressures, con Corporation. Use of these sp~ce-quality gal struction of the body was especially trying lium arsenide cells resulted in a solar array because of the need to let panel layers cure, having about 25 percent more power than the sometimes for several days, before the next prototype array. In the best steady-sun condi step could be undertaken. tions in Australia this 8-metei array would The technological phase of the Sunraycer deliver more than 1,550 watts. In com program was essentially finished by mid parison to silicon the gallium arsenide cells August-all the major engineering challenges have higher output and a lower negative tem were solved just four and a half months after perature coefficient of power, but they cost the April 1 start. In the remaining two and a more, are smaller, more delicate, and heavier, half months we placed a strong emphasis on and require more diodes to provide protec- reliability testing, somewhat to the neglect of 6 F.NGI~EERING & SCIENCE / WINTER 1988 Above: Alec Brooks, who was one of the six drivers in the race, observes Paul Mac Cready through a fiber optics remOle viewing system con nected (0 the lOp of the slrake Barl Hibbs (left) and Graham Gyalt check the inpm dala at left. This served as 10 the VSAERO program llsing a graphic depiction of the Sunra)'cer's rear-view body shape. "mirror Above: Nomex honeycomb, the sandwich fifling, is laid onto the Kevlar skin in the mold for the solar-array panel. Left: the lightness of the ciIassi.\· and emptiness of the inte rior are apparelll when the canopy is removed and the solar-array panel lilled up. An AeroVironment crew checks oUI/he Sunraycer pro totype during tests in the GM wind tunnel. There were no surprises, bllt the tests Quallli/led the effects 0/ many small cleanups, making if pos sible 10 decide which ones justified being incorporated into fhe race vehicle. 7 In constructing Sunraycer's some obvious improvements in aerodynamic idealized performance curve, details. We put the prototype vehicle through the rolling-friction drag (tires, a rigorous 3,000-km field-test program while bearings) portion is assumed constant and so consumes the actual race vehicle was built. Also time power directly proportional to was spent planning the logistics of the race speed; the aerodynamic drag and developing race strategy. Then the race is proportional to speed vehicle also received extensive field testing, squared and so consumes much of it in Australia under the guidance of power proportional to speed cubed; and the conversion of Ray Borrett (of GM's Holden subsidiary in electrical power to mechanical Australia), who put the team through its prac power of the drive wheel is tice for every conceivable contingency. taken as 88 percent efficient. Race strategy would be simple if we didn't For a smooth road surface have batteries to store energy: except for and no wind or hills at the temperature and altitude typi speed limits imposed by safety or traffic laws, cal of the Australia race always go as fast as you can. The course is course, the curve was con prescribed; you just follow the Stuart High sistent with observations in the way (and drive on the left side). But a bat 50-80 km/h range. Note that tery complicates the situation. It serves as a the graph can also be inter preted as energy (in watt bank in which you can store your charge of hours) versus distance (in energy "money" in sunny times for with kilometers) for one drawal at a time when you can make better hour of running. use of it. This is a great convenience, but one for which you pay the price of having to use certainty. If the last day of the race is ,going your brains without certain necessary inputs. to be cloudy instead of sunny, you should go On a windless day in nonhilly terrain, a more slowly during the early days so as to reasonable initial strategy is to maintain con have more energy in the bank to spend on stant speed independent of the sun situation maintaining speed at the end. If all forecasts -the speed that just consumes the total through the end of the race could be counted energy available from sun and/or battery in on to be perfect, there would be a specific the available time. Referring to the perfor optimal speed for any moment, and it can be mance curve above, if in one hour you go 70 calculated. The complete calculations depend km/h you will "spend" 1,080 watt hours. If on many factors besides the amount of instead you drive a half hour at 50 km/h and cloudiness. For example, a headwind or a a half hour at 90 km/h, your 70-km journey rougher road surface in the future will then in one hour will "cost" more, namely 1,250 affect the Sunraycer performance curve (watts watt hours, because of the nonlinearity of the vs. speed), and this affects the present optimal performance curve. speed. The solar array power depends on Selection of the proper constant speed Sunraycer's heading relative to the sun's posi requires that you assume how much energy tion, a function of the highway orientation will be available for the time interval in ques where the vehicle will be at a particular time. tion. Unfortunately, this assumption depends Motor efficiency varies with both the power on future events that can't be forecast with and the rpm (vehicle speed). Battery 8 ENGINEERING & SCIENCE / WINTER 1988 efficiency, the charge you get out compared to Another rule-of-thumb was to maintain what you put in, yaries.with the charge/dis constant speed during cumulus cloud condi charge rate, the total charge in the battery, tions when small cloud shadows alternated and the battery's prior history of charge/dis with sunny spots. But when cloud shadow charge cycles. Because of the motor and bat areas were large (say, five kilometers or tery inefficiencies, you don't save as much more), the driver should speed up when in energy going down 100 m of altitude as you shadow. It may seem counterintuitive to go spend going up. faster when nature is giving you less power, A further complication is that the battery but since you get through the shadow more storage is not infinite. The charge is just quickly you actually benefit. In a headwind three kilowatt hours at most, and only the (or tailwind), you should slow down (or speed region between 20 percent and 80 percent of up) from the zero-wind optimum by an full charge should be used for best efficiency amount of about half of the wind speed. A and cell longevity. This 1.8 kilowatt hours of constant battery current mode is convenient preferred battery range is only about a fifth of for handling varying winds as long as the ter the total energy available to Sunraycer on a rain is rather flat. sunny day. The right speed at one moment Fortunately, operating a bit off optimum depends on events throughout subsequent does not hurt much. If you operate 10 per days, but only insofar as these narrow battery cent faster or slower than the optimum speed limits are not exceeded each day. In practice, for 100 seconds, you lose only about 1 the battery operating limits for race strategy second; for a 20-percent difference the net solution are even more restrictive. The target loss is under 4 seconds. Going too slowly discharge condition for the end of a day had costs you distance in a given time but con to be pre-selected within a narrow range of sumes less energy, and so the battery ends up only 0.5 kilowatt hours, a conservative target containing more energy. This excess can then for preserving "fuel" to get through a cloud be spent efficiently over a long period of time cover the next morning. The target would be to increase your subsequent speed. The net revised as the end of the day neared, and the consequence on time lost is surprisingly forecast of evening and morning radiation small. became more certain. Sunraycer managed to maintain a rather The complete race strategy was modeled steady average pace from Darwin to Adelaide. by Mike Cassidy and Joe Gurley of Hughes Aircraft Co. Their program was used to r S File Edit explore various scenarios and thereby develop t SunRaycer Telemetry Printer o 50 100 150 200 0 50 100 150 200 250 ~ guidelines or rules-of-thumb to guide deci ArrOll: output: sions during the race. For the race itself the 1528 W Tracker program was exercised before each day began, output: 13.9 A using the weather forecast, to yield optimal SPEED/DIST: :--: 1510 W 62.2 km/hr 99 ~ 38.6 mph speeds and battery charge conditions Rrrall 174.2 km temp: :--: 54°C throughout that day. As the day progressed, Battery Cell Uoltages: if forecast conditions changed significantly, 0.0 0.5 0.0 0.5 1.0 Amps Battert.l: (chQ 9 Sunraycer leads its entourage of observer vehicle (rhe moror home immediarely behind), supply trllck, and various cars containing helpers and media. The entollrage also included a lead car 100 merers ahead, a SCOllt car that roved some· limes many miles ahead, a communications car, a sareflire-system vehicle, tfllcks carrying rhe camping gear, and a media blls. Sunraycer may save global fuel in the long run bur definirely did nor do so during the race. The crew changes one of three jlar rires during rlie race. Sropping, changing rhe wheel, and reacceieraring rook only aboUl rwo minllles. Paul MacCready checks out sky conditions as Marher Nalllre recharges rlie ballery. Sunraycer's solar panel is rilled/or recharging at 5 p.m. Ar 7 p.m. of the third day Sunraycer is snuggled IInder its blanket/or the night. 10 ENGI NEERING & SCIENCE/ WINTER 1988 For the first five days the daily average speeds in the bank to handle possible poor sunlight ranged fn;>m62.8 to'69.5km/h~ The dif conditions in the evening and next morning. ference arose mostly from headwinds and The evening charging period indeed yielded tailwinds .and only secondarily from altitude almost no charge, but on the morning of changes, sections of unpaved road, or the November 2 the sun filled the battery even amount of battery charge consumed or above the top of its efficient range. replaced: We traveled 521 km on the shorter On the second day I decided to maintain (8-hour) initial day and on the subsequent speed late in the afternoon because I was four days, 605, 564, 563,and 600 km. For optimistic about the upcoming evening the last 152 km on the sixth day, since there morning radiation situation. But Nature was was no need to preserve battery charge for again uncooperative during the final running later, Sunraycer's average speed increased period of the day (perversely agreeing well dramatically but was held down to 77.1 km/h with the official forecast). The battery was by the heavier traffic. reduced to 25 percent of full charge by the It was not as easy as it may have looked. 5:00 p.m. stopping time, and large storms The steady pace belies the stress on the made the evening charge negligible. If the weather forecasters and strategists for the first next morning were overcast, we might have a several days, until improving weather and the serious problem. I spent a sleepless night, lack of any close competition took some of promising Nature I would be more respectful the pressure off. At the very start of the race in the future if she would just give us a fair we wanted to drive fast and escape the traffic shake. Nature didn't listen. It rained that problems associated with all the other solar night, and clouds kept the morning charge vehicles and their large entourages. This small. At 8:00 a.m. the battery was recharged required borrowing heavily from the battery to only 30 percent, but we started out coura bank to cope with low sun angle, clouds, and geously and spent charge rapidly to maintain hills. Operating Sunraycer in a constant a fair speed under the overcast to try to reach speed mode was appropriate for the hilly the sunny area far to the south. We did topography and cumulus-cloud weather, but reach sunshine as planned, and hindsight this mode makes the prediction of battery showed that we had used almost exactly the charge difficult. Our anxiety was com proper strategy for the previous 24 hours. It pounded by the fact that we were unsure of is sometimes better to be lucky (or to have a the battery charge condition; this could only special relationship with Nature) than to be be determined from an amp-hour meter, and skillful. the accuracy of ours was uncertain. Only On subsequent days we were usually a bit when the battery was recharged back up to more cautious, but the weather was improv the effective top at the official end of the race ing and the last half of the course proved to at Seppeltsfield were we able to ascertain that be almost a slam dunk, with only a bit of the amp-hour meter had been right all along. high cloud. Still, every day the strategy deci The strategy discussions continued sion process was dicey after 4:00 p.m., as throughout the first day, as we watched the solar radiation diminished and Sunraycer battery condition and modified 'our view of the weather forecast. Watching the weather unfold in our region sedU:ced us into upgrad TIRES ing the forecast that had been provided by the Australian Meteorological Service (and inter The race did demonstrate one preted for our purposes by AV's George "flaw" in We three flat tires Ettenheim). When we saw that the clouds have had more. If we had were developing unexpectedly slowly, we thinner tread, the lessened friction could not help but presume that radiation would have saved a hour over the 45- would be better than predicted for late in the hour race. more than for the day. But Nature disdained our revised expec extra 15 minutcs or so that would been tations. As the 5:00 p.m. stopping time lost another dozen flats. U""'1';lU'J<,a and neared that first day, the radiation was weak about two minutes because of cloud shadows, and the forecasts while the extra of the charging potential for the evening and from the sun the interval could later morning grew pessimistic. We drove slowly be used make up some the time lost. to conserve battery charge and leave enough 11 began feeding hungrily on an emptying bat cent faster than the runner-up and would tery, while we hunted for a stopping location probably still have been 30 percent faster if in the 5:00-5: 10 p.m. time slot that would the runner-up had had our weather. Com afford both a comfortable camping spot and a parisons become less clear as we explore "ifs" chance for a good evening charge unhindered further. The 30-percent figure would prob by shadows of trees or clouds. The scout ably have been below 20 percent if we had vehicle would be probing ahead for likely used the all-silicon solar array and the proto spots, and we would be coordinating by radio type Sunraycer, but we would have won even and trying to locate ourselves on featureless with a total power system equivalent to those terrain while using uncoordinated odometers. of the next two finishers. But if we had opted . The scout car always requested that we use for a less advanced power system, we would constant speed the last half hour so as to have saved development time, which could make the coordinated selection of a roosting have been invested in aerodynamic cleanup. location somewhat easier. We always agreed, And with a slower vehicle we would have but then worries about the battery would been less concerned with vehicle dynamics cause us to change speed anyway. After mak and could have used our alternate tires, which ing bum decisions about our stopping point featured substantially lower drag but less for several days, I was delighted to leave it to comfortable stability and control. Sunraycer's others thereafter. superiority and value lay not in one single Sunraycer cruised to the finish line on the advantage but in pushing the frontiers in all sixth day with a huge margin of victory. the technological areas encountered. The Most of this was due to Sunraycer's basic race was the design focus; technological speed, but other factors helped. There were advancement on a broad front was the goal. some jokes about GM's influence extending This success and the fact that the whole At right Paul MacCready holds the World Solar Chal to very high levels, and we did indeed have project was conceived and accomplished in so lenge Perpetual Trophy, which some luck with the weather. Our early speed short a time were partly the result of a re was donated by Broken Hill helped us escape the meteorological misery markably effective organizational setup, Associated Smelters Pty. Ltd. that moved in on those farther back in the which harnessed diverse talents. General It will reside in Detroit until race, including gully-washing precipitation Motors had the resources- GM's Magne (and after?) the next race in 1990. that flooded roads and a storm with I-inch quench motor and car-suspension design; hailstones in a spot where reportedly no rain Below: GM's President Bob had been recorded in three years. Stempel talks design with Was Sunraycer's edge due to the solar Sunraycer driver Molly array, power electronics, motor, battery, aero Brennan (also from GM) at dynamics, tires, race strategy, reliability, or Seppeltsfield, the ultimate luck? Every factor helped, with the effec finish of the race. The- car behind was the first to cross tiveness of the total power system from the A ustralia from south to north, solar cells to the motor output representing a afeat it accomplished in 1908. substantial advantage. Sunraycer was 50 per- 12 ENGINEERING & SCIENCE / WINTER 1988 Gossamer Albatross Solar Challenger Quetzalcoatlus northropi Flying Fish hydrofoil Hughes' unique abilities in solar arrays, silver cumstances. One should recognize the Some of the previous projects zinc batteries, electromechanical testing, and uniqueness of the confluence of challenge, ofAero Vironment's staff are strategy theory; the GM-Holden strengths in interest, resources, and capability that this shown above. race management in Australia; and the abili project represented. ties of another 10 GM groups that were in The second most frequently asked ques volved more peripherally. GM also had the tion (after "How much did it cost?") has vision and daring to enter the competition been, "What good is it?" or, phrased dif and, with priority commitment from the top ferently, "Will solar-powered cars ever be levels, to take it seriously. practical?" The answer seems clear. Cars Aero Vironment, which was given the powered solely by sunlight are unlikely to be charter to playa substantial role in both pro practical enough to justify wide usage. The gram management and technological areas, maximum solar power intercepted by a car had resources of another sort to contribute. sized object is small in comparison to the The company's background has included a power required by our present safe and com number of vehicle developments featuring the fortable automobiles, anp, in any case, sun emphasis on low power and efficiency that is light is not always available. But Sunraycer the essence of a solar-powered race car. and the W orId Solar Challenge are valuable Members of AV's staff or teams at the com when viewed more broadly. The attention pany itself had created human-powered air that a solar car race focuses on doing big jobs craft (the Gossamer Condor and Gossamer with little power expands our insights into, Albatross, and the battery-assisted Bionic expectations of, and demands for getting Bat), the solar-powered Gossamer Penguin better fuel economy with gasoline and edging and Solar Challenger, a flying replica of a battery-powered and hybrid cars toward prac giant pterodactyl (E&S, November 1985),and ticality. The project moves us toward han various human-powered land and water vehi dling the transportation needs of the future cles, such as the flying Fish hydrofoil. All of while making fewer demands on the earth's these projects had to operate with power in resources and environment. And, as a special the 0.25- to 2-horsepower range. These pro bonus, as Sunraycer goes on an extended jects, in addition to a number of specialized tour, it can serve as a stimulus to students to aircraft developments A V has conducted for appreciate that engineering is fun and that government customers, made it uniquely . nonpolluting transportation is achievable. 0 qualified to take on Sunraycer. Also, AV, W- 4 the size of GM, with its short chain of command and versatile staff, has the capacity for quick response toa problem and rapid Stories about Sunraycer and the World Solar development of solutions. Together-the Challenge are also currently appearing in the huge company and the tiny one-we formed February issues of Smithsonian magazine and a team of unusual strength-one especially Popular Science. A book on Sunraycer is appropriate for this challenge. expected out by the middle of 1988, and a Can this sort of organizational system and National Geographic television documentary the excitement of a race situation (which on the race is scheduled for the summer. A demands absolute deadlines and produces an series of papers is in preparation for technical unequivocal winner) be applied to handle symposia. Sunraycer will also be touring other developments as quickly and effec museums and schools as well as auto shows tively? The answer is probably "yes" in prin and will be demonstrated in action on the ciple, but perhaps only in rather special cir- highways of many states. 13 Quark Quest HE BESTIARY OF subatomic physics draws its T denizens from a world where energy and mass are one, where time slows down as speed increases, and where particles conjure new particles from the void or transmute themselves through interactions with other particles-by tossing still more particles back and forth at each other. In this world where energy transforms into mass, a particle can only appear when the available energy is more than its rest mass, with a little some thing left over to give the newborn particle some velocity (kinetic energy). In the search for these particles, physicists continue to build bigger and more powerful particle colliders. The particle hunters, armed with computer ized detectors, hope to find ever more mas sive particles in a generous pool of free energy. Research teams from Caltech are partici pating in several experiments. One group, headed by Associate Professor of Physics Har vey B. Newman, hopes to find new particles in the wreckage of electron-positron colli sions. Newman's group is helping design an remained just beyond the view of today's experiment to be installed at the LEP (Large accelerators. Electron Positron) collider, currently under Physicists want the top quark for aesthetic construction at CERN, the European Labora reasons, as its discovery would bolster the tory for Particle Physics, in Geneva, fundamental theory's internal consistency. Switzerland. Quarks bind together to form protons, neu Newman hopes to find the top quark and trons, and most other particles, just as pro the Higgs particle. Both particles are believed tons and neutrons form atomic nuclei. As to lurk in the 150 - 200 GeV range. (GeV things currently stand, five quarks have been stands for giga [read billion] electron volts. found. The fifth, the bottom quark, An electron volt is the kinetic energy gained represents a loose end. Quarks and another by an electron passing through a one-volt set of particles, the leptons, appear to be fun electric field in a vacuum.) Both particles damental: they behave as points, with no have tantalized physicists for years, and a signs of internal structure. There are six false sighting of the top quark has been known leptons, not counting antiparticles: the reported at least once. But, in fact, they have electron, the electron neutrino, the muon, the 14 ENGINEERING & SCIENCE / WINTER 1988 muon neutrino, the tau, and the tau neutrino. The Higgs particle would confirm the Aerial view of CERN with The leptons divide neatly into three pairs or standard theory of electroweak interactions underground rings drawn in. "generations," each containing a particle and the Weinberg-Glashow-Salam (WGS) theory. LEP is the largest ring. (CERN's aboveground com its neutrino. Each generation is more mas The electroweak theory unifies electromagne plex is at left. surrounding the sive than its predecessor, but resembles it in tism and the weak nuclear force, and is a big smallest ring.) The Jura all other properties. Quarks display a similar step toward a Theory of Everything describing Mountains are in the back grouping, pairing "up" and "down," the four forces of nature-gravity, elec ground. "charmed" and "strange," but the bottom tromagnetism, and the weak and strong quark sits in splendid isolation in the heavy nuclear forces-as different aspects of one weight division. The top quark would round fundamental force. Just as quanta of elec out the roster, making the quark and lepton tromagnetic force are carried by photons, the families symmetric. This would cause several weak force is carried by three particles, the intractable processes predicted by the theory W+' W-, and Zoo But while photons are mass to cancel each other out, leaving a self less, these other particles have been found to consistent theory of quark and lepton have masses 90 to 100 times the mass of a interactions. proton. The WGS theory meets the challenge 15 of describing how two forces can be aspects of computing that there are only a handful of "one force when the particles that carry them high-energy particle accelerators in the world. appear so different. The mathematics is The accelerator complex at CERN, LEP's quite complex; but once the smoke clears future home, is one of the world's largest, several sets of infinite terms cancel each other most sophisticated collections. out. The residue is a solvable set of equations LEP should be operational by 1989. It with finite terms, but when all the terms will operate in the 44-200 GeV range, where describing the photon, W+, W-, and ZO have the top quark and Higgs particles are been sorted out, a set of terms is left over. expected to appear. It will be the largest These terms describe a new particle, electron-positron accelerator in the world, christened the Higgs after the physicist who and will incorporate many novel design developed the math that creates it. features. The LEP ring, 27 kilometers in cir The particle collider is to high-energy cumference (approximately 9 km in diame physics what the telescope is to astronomers. ter), is buried under the French-Swiss border Just as many astronomers use one telescope between Lake Geneva and the Jura for different purposes, several experiments Mountains. can be run on the same accelerator. In order LEP will accelerate electrons and positrons to understand Caltech's role in the LEP proj in opposite directions around a circular track. ect, therefore, we shall look first at the LEP Electron-positron pairs were chosen because itself, then at the experiment, and finish with they have no internal structure, unlike pro a closer look at one of Caltech's tons, which are made up of quarks. The col contributions. lisions are therefore easier to reconstruct, as The idea behind a particle collider is quite there are no secondary reactions among the simple. Take equal measures of electrons (e-) component quarks to confuse the picture. and positrons (e +), or protons (p) and antipro The price to be paid, and the reason LEP is tons (p), or protons and protons. Put them so big, is that any particle radiates photons in a high-vacuum tube (two tubes for proton (called synchrotron radiation) when proton collisions), goose them with a power accelerated. The electrons and positrons in ful electromagnetic field to rev them up to a LEP radiate vast amounts of high-energy pho very high kinetic energy-to a velocity close tons, requiring many megawatts of power to to lightspeed-and then send them headlong make up for the energy losses. If the ring at each other. The collisions will have twice were smaller, forcing the beam around tighter the kinetic energy of either particle alone, corners, the power loss to synchrotron radia plus all the energy released in the matter tion would be much higher. antimatter annihilation of e+e- or pp colli The electrons and positrons are confined sions. This energy creates new particles, in a high-vacuum pipeline only inches in which fly off in directions whose distribution diameter. The pipe passes through a series of may be determined by their charges, masses, radiofrequency (RF) cavities and magn.ets. and the initial collision parameters. An array Since electrons and positrons are oppositely of detectors around the collision zone deter charged, a single RF pulse will push them in mines the trajectories and charge/mass ratios opposite directions. Similarly, a single mag of the departing particles. The interesting netic field will bend the two beams in equal particles often don't reach the detectors, but but opposite directions. The beams collide decay into something else almost immedi head-on at the four experimental halls. ately. The physicists match their quarry's Each beam is 500 microns ().t, 10-6 meters) predicted decay patterns with observed parti wide and 25 ).t high at the interaction points, cle distributions and energies. less than the thickness of a human hair. That Putting this simple scheme into practice is space contains approximately 90 percent of another matter. Each component has the beam's particles. Since the number of numerous parts and subassemblies of its own. collisions is proportional to the beam's den Institutions around the world provide the sity, or "luminosity," the accelerator physi parts, which must fit more tightly than. the cists are working to make the beam even pieces of any jigsaw puzzle. The whole is run tighter. Initially, LEP will circulate two e by a vast array (a hierarchy, really) of com and two e+ bunches, each 1.6 mm long. puters whose software was assembled the Like all particle accelerators, LEP uses the same way. The entire operation requires so behavior of point charges in electromagnetic much complex equipment and sophisticated fields to accelerate subatomic particles. LEP 16 ENGINEERING & SCIENCE / WINTER 1988 uses resonance cavities-hollow, copper-lined cylinders divided into chambers by a series of plates. The plates are perpendicular to the beam line, which passes through their center. An oscillating RF field in the cavity, syn chronized with the particles' passage, gives them ali energy boost with every plate due to [ ~t, I .t.l_] interactions with the field's electric com .'4",( I ~ i -rr-] ponent. The oscillations are timed so that the RF field is decreasing while the particles are t I ,t'I.LJ passing through it. This helps keep the .. ,.ct Tr't-r-'] bunches together, as particles arriving early -having slightly less energy and thus follow ing a slightly smaller orbit-see a stronger field and get a bigger push. Overenergetic latecomers get a smaller push, slowing them down. The LEP has 128 RF cavities of five chambers each. They produce an accelera tion of 1.5 MeV(million electron volts)/meter using an RF field of 352 megahertz (MHz) at 16 megawatts of power. The total ring will be steer the beam around the ring. Quadrupole A particle moving through an able to produce up to 110 GeV initially. It magnets (magnets with two opposing sets of RF cavity (left) and its RF can be upgraded by adding 64 niobium super north and south poles, and hence two oppos field (right). Thefield gives the particle a boost with each conducting cavities producing 7 MeV/m. ing fields) keep the beam tightly focused. chamber. LEP II will eventually replace all the RF cavi Otherwise, the repulsion between like charges ties with superconductors and the ring will within a beam would cause it to spread out as operate at 186 GeV with no additional power it travels. Fine adjustments to the beam's consumption. According to Newman, "that's shape often require sextupole magnets. The actually less than at PETRA [a 46.8-Ge V LEP ring has 3,304 dipole magnets, 776 machine in Hamburg, Germany 1. When I quadrupoles, and 504 sextupoles. was there we could draw 50 megawatts from - LEP's steering magnet design is another RF cavity assemblies with the local net. This machine is much bigger, first. The magnets don't have to be very storage spheres. but it will use less power." The LEP ring broke new ground in its RF cavity design. Conventional RF cavities keep the field at constant strength between bunches, even though the cylindrical design does not store the field efficiently. The time between succeeding bunches is much too small (44 microseconds for LEP) to turn the power supplies feeding the field on and off. Unfortunately, power losses from the RF cav ities add a significant surcharge to the operat ing cost. The LEP designers added a spheri cal chamber atop each resonance cavity, con necting the two with a magnetic "gate." The sphere stores the RF field very efficiently. The magnetic gate opens to let the RF field down into the resonance cavity as each bunch passes, and closes again in its wake. Newman estimates the new design will save 40 percent on LEP's electric bill. Powerful electromagnets guide-the particle beams between RF cavities. Magnets exert a force perpendicular to both the field's orienta tion and the particle's path. Dipole magnets 17 , powerful, hut they must be absolutely rigid to In the meantime, the experimental halls , keep the particle beam properly aligned. The have been excavated, and work proceeds on designers substituted concrete for the metal the magnets, detectors, and computers needed core used in conventional electromagnets. to make it all work. All the magnets have The result is an electromagnet with thin been built and are on-site. Detector com metal stampings embedded in concrete, with ponents are being built around the world. the concrete making up most of the structure. Some portions are already in the final assem "The concrete magnets turned out to be very bly and testing stage on-site. LEP3NET, an rigid, with very good geometric integrity," international computer network, is up and Newman said. "We saved over 50 percent of running. Administered by Caltech, the cost of normal magnets." LEP3NET connects American and European research centers with the first transatlantic data communications link dedicated to high energy physics. Physicists worldwide are con tributing control and analysis programming. The LEP ring will host four experiments, ALEPH, DELPHI, L3, and OPAL. The Cal tech group is part of the L3 project. While the other three experiments' names are acro nyms generated by applying intense heat and pressure to a phrase describing the experi ment, "L3" simply means this experiment was the third proposal submitted to run on the LEP ring. "The L3 experiment is designed to emphasize very-high-resolution energy and momentum measurements of photons, elec trons, and muons," Newman said. "We are aiming for resolutions of 1 percent at 50 GeV. "We will make very precise measurements of the electroweak theory. LEP collides elec trons and positrons in a very small energy spread, so in just a few days' running you can get a very precise mass for the Zoo The exper iment is so precise, it's a bit of a theoretical challenge. You have to do third-order calcu The IUnneler's cutting head. Most of the ring's site has already been lations in order to obtain theoretical predic The machine bores a hole 4 excavated by a tunneling machine. The tions as good as the experimental precision. meters in diameter. machine clawed a 4-meter borehole through The ZO mass is the most precise quantity that the local sandstone at up to 60 meters per can be measured. It will serve as the refer- day. Preliminary designs had a good portion of the ring under the mountains, but test bor :\11 ;01'l.i PAIR ASYMMETRY ings soon revealed that the limestone there was more than a match for the tunneler. The designers were reluctant to make the ring smaller because of the synchrotron radiation problem. They couldn't put the ring under downtown Geneva or the lake, either. They wound up setting the ring on a 1.5 percent . grade in order to shoehorn it between the lake and the mountains. A portion of the ring still passes under the mountains' periphery. As of December, 1987, the ring has been com pletely excavated except for 80 meters in a dense limestone formation, where the work proceeds by pick, shovel, and explosives. 18 ENGINEERING & SCIENCE /WINTER 1988 ence point for further tests of the theory. a millimeter. Some components require an "We canthen riieasure something else 800-ton crane at the surface to lower them against that: the asymmetry of muon pair down the shaft. There are only two cranes production. We will run for 200 days and that big in all Europe, one of which will soon collect about 6,000,000 ZOs. Of these, visit CERN. These cranes are itinerant spe 200,000, produce muon pairs. Then we can cialists, always on the move from job to job. measure the asymmetry with a statistical The detector assembly contains four accuracy of 0.2 percent. different detectors nested around the collision "If the top quark has a mass of 180 GeV, point like Russian puzzle dolls. Each detec it will affect the asymmetry by 0.8 percent, so tor completely surrounds the beam line, so that's detectable at our experimental accuracy only particles leaving at very small angles, of 0.2 percent. As the particles get heavier, and the elusive neutrinos, will escape they have a greater effect on the symmetry. undetected. (All particles must be accounted But if the top quark gets too massive, when for to balance the mass-energy books. If the you go back to the low-energy experiments books are too far out of balance, it can indi that have been done and reinterpret them in cate that new, unpredicted particles are being order from low energy to high energy, putting formed.) The time expansion chamber (TEC) in corrections for the mass of the top quark, is innermost, surrounded by the electromag the experiments no longer agree with each netic calorimeter, the hadron calorimeter, and other. If the mass gets very large-like 400 the muon drift chamber. GeV, 450 times the mass of the proton-the A gargantuan electromagnet surrounds the whole self-consistent picture of the unified detectors. This electromagnet, the largest electroweak theory that occurs for a more rea ever built, provides the powerful magnetic sonable top quark mass falls apart. field needed to separate particles by their "After we study the Zo, we'll go on to charge/mass ratio. The electromagnet weighs higher energies. One of the best ways to see 7,600 tons. It creates a magnetic field 5,000 the Higgs is to look for certain rare events. gauss strong, constant over a volume of 1300 For example, at 165 GeV, electrons and posi cubic meters. trons can collide to make a ZO and a Higgs. The TEe, according to Newman, is "a The L3 experiment hal[, look The ZO goes to a muon pair, which you very-high-precision detector for tracking the ing toward the access shaft. detect, and by precise measurements of the trajectories of charged particles and measur Note the forklift in the fore muons you can measure the mass of the ing their momentum. It measures all the ground for scale. Higgs recoiling against that pair. This prob ably will require two years of continuous run ning at 4,000 hours per year, from which we may get twenty events. (This depends on the mass of the Higgs, and-of course-on its existence!)" The cavernous L3 experiment hall lies 50 meters below the French countryside. That hall is the shallowest of the four; the hall under the foothills of the Jura Mountains is a vertiginous 150 meters (approximately 500 feet) below ground. ("It doesn't sound like much," Newman said,"but when you look down this hole 500 feet deep, it's a long way down. And people had to go up and down stairs. The elevators have only recently come into service. ") L3's massive detector assembly, bigger than a two-story house, almost fills the hall's 21.4 m diameter by 30.5 m length. The detector components must be lowered piece meal down an access shaft; to be reassembled below. A crane traveling on rails high under the vaulted roof wrestles components weigh ing tons into positions accurate to fractions of 19 Y L3 DETECTOR i x ... MAGNET POLE MAGNET YOKE ACCESS S HAH MAGNET COIL MUON ORIFT CHAMBER F xp~ RIM~ N 11\\ IIf\U [ --. J. ) HADRON CALORIMETER 2 PHOTON TAGGING DETECTORS ELECTROMAGNETIC SHOWER COUNTER I BGO) TIME EXPANSION CHAMBER charges within 10 microseconds after an e>e heat, either. It intercepts protons, neutrons, collision, and its hardware integrates them mesons, and hyperons, all comparatively mas into particle tracks in real time." Further sive particles that interact with heavy nuclei. analysis determines how the magnetic field It also absorbs the small part of the elec has deflected the particle. The direction tromagnetic "cascade"-electrons, positrons, shows whether the particle has a positive or and especially photons-that may slip past negative charge, and the curvature gives the the BGO crystals. The densely packed nuclei charge-to-mass ratio. The combination in the calorimeter's 600 tons of depleted reveals the particle's identity. uranium-"'U from which all "'u has been The particles pass through the TEC and removed-are so many sitting ducks for the into the electromagnetic calorimeter. The hadrons from the collision zone. A particle "calorimeter" doesn't actually measure heat. will probably ricochet off a nucleus as it It's a scintillation counter on a grand scale. passes through the detector. The particle will A barrel-shaped array of 12,000 bismuth ger change direction and lose some kinetic energy manate (BGO) crystals detects photons, elec with each collision. It will also produce new trons, and positrons. A single photon particles, which may interact in turn, forming penetrating the crystal dissociates into an a cascade of particles. Approximately 60 con electron-positron pair. Each particle interacts centric sets of 5-millimeter-thick uranium with the crystal to produce photons, creating plates alternate with parallel tubes containing additional electron-positron pairs. The result a mixture of 80 percent argon and 20 percent ing avalanche of electrons and positrons pro carbon dioxide. As a particle zips through a duces a lot pf light that eventually reaches a tube, it ionizes the gas and generates a signal photodiode on the crystal's rear surface, on a detector wire. Computer analysis of sig where the photons signal their arrival with an nals from successive layers of tubes helps re electrical pulse. The amount of light pro create the particle's path. A particle's track duced is proportional to the energy deposited length is proportional to its initial energy, in that particular crystal. which can then be deduced. The modules are The hadron calorimeter doesn't detect being built in Michigan, Aachen, Bombay, 20 ENGI N EERING & SCI EN CE I W I NTER 1988 and Beijing, with production scheduled to be completedin December" 1987: Muons, having only about one-tenth the mass ofhadrons, and being immune to the blandishments of the strong nuclear force, whizz unscathed through the hadron calorim eter and into the muon drift chamber. This is actually three concentric sets of chambers. Each chamber is a long,. flat box made of thin aluminum sheets. The boxes contain a total of 116 kilograms ofargon and ethane. Again, as the muons pass through the gas, they leave ionized trails that are detected on a three dimensional wire array. The innermost box has 16 wire layers perpendicular to the particle's path. The center box has 24 layers, and the outer box has 16. Each particle thus leaves a trail of 56 data points. Optically flat glass sheets keep the wires in perfect align ment down to lO-micron tolerances. The entire drift chamber system is dynamically stabilized to the same tolerance. "With the very-high-energy particles we'll be looking at, high precision is critical," Newman explained. "For example, with a 45-GeV muon the devi ation from a straight line, by which you can sions must be made in real time while the Muon drift chamber "half measure the momentum, is only 4 millime detector is running. Soon after, LEPICS, an octant" assembly. The two ters in 6 meters. We want to reach 1 percent off-line computer at CERN, sifts through the upper rows have two chambers each. accuracy for the muon pair's mass reconstruc stored particle tracks, identifies the particles, tion, which means 1.5 percent for each and winnows out those combinations that muon. To do this, we need to measure the could not have been produced by the particles deviation from a straight line to the order of or decay modes of interest. The off-line 50 microns. The total systematic error, i. e., analysis can take many months to complete. the total error in the computer's knowledge of Thanks to LEP3NET, the analysis can be the location of the detector wires in space at coordinated among participating institutions, any instant, has to be less than 30 microns." and some computer work can be done The muon chambers will be assembled into locally. 16 "half-octants" of five chambers each. Most projects at CERN are international Most of the octants have already been built collaborations, and the L3 experiment is the and tested at CERN. largest and most widespread. Four hundred The data from the TEC, the BGO array, physicists from forty institutions in thirteen the hadron calorimeter, and the muon drift countries are contributing to the project. The chambers go through a massive on-line paral list of American participants includes Caltech, lel processing system. The computer has a Carnegie-Mellon, Harvard, the University of complete picture of the detector arrays in its Hawaii, Johns Hopkins, M.LT., the Univer memory. Each component's location and sity of Michigan, Northeastern, Ohio State, orientation in space is known to a precision the University of Oklahoma, Princeton, dictated by the physics, which in some cases Rutgers, and Yale. Other contributors is within 10 microns. The system takes all include groups from China, France, East and the signals from all the detectors, over West Germany, Hungary, India, Italy, the 100,000 channels of data, and integrates them Netherlands, Spain, Sweden, Switzerland, and into a three-dimensional "picture" of all the the Soviet Union. particle tracks. Some of the computational L3's Caltech contingent numbers only one power goes into deciding what particles came associate professor, one senior research fellow, from uninteresting events, and discarding two postdocs, and assorted undergraduates. them before too much computational time or Yet this small group has been responsible for storage space is wasted on them. These deci- key elements of the basic design of the muon 21 drift chamber; for developing LEPICS; for target with high-energy hydrogen atoms, pro d~veloping and operating LEP3NET; and for ducing a high-intensity flux of 17-MeV pho developing a calibration technique for the tons. The crystals' response can be calibrated BGO crystals used in the electromagnetic to the photon's known energy. The entire calorimeter. system, including the vacuum line carrying The BGO subproject preserves the inter the hydrogen beam, is completely indepen national flavor of the whole. The USSR sup dent of the LEP ring. plied some of the raw materials. The The system can operate between runs, or Shanghai Institute of Ceramics is making the even while the LEP is running. The RFQ's crystals. France provided the technology to electronics will be strobed to an external cut, grind,and polish the crystals to toler trigger. The RFQ will emit a high-intensity ances of 100 ~ in 24 cm. West Germany fur pulse, one to three microseconds long. The nished the photodiodes. An Italian firm intensity must be carefully controlled: high specializing in helicopter rotors built the sup enough that photons hit an appreciable porting barrel. Caltech is calibrating test number of crystals, yet not so high that two arrays of 20 and 49 crystals. photons hit the same crystal. The pulse Each BGO crystal is a long, thin, truncated length is no less important-it must be short pyramid. The crystal's front face is 2 x 2 cm. enough for the crystals' response to be meas The back face, where the photodiode is ured accurately, and for the detector's elec mounted, is 3 x 3 cm, and the crystals are tronics to work with a high ratio of signal to 24 cm long. When the crystals are assembled electronic noise. A single pulse calibrates into a barrel, their tapered shapes aim their only 200 to 300 of the 12,000 crystals, but at long axes at the barrel's center-the collision 200 pulses per second the entire array can be point. calibrated in an hour or two. Although RFQs Besides having the requisite transparency running at low pulse rates have been used as and scintillation properties, BGO is quite proton sources in the first, or preinjector, dense, almost as dense as iron. The heavier stages of high-energy proton accelerators, no the atoms in a material, the more likely one had ever attempted to run one at 200 Hz incoming particles are to interact with its within a physics experiment before. nuclei. The crystals are long enough to The Caltech group tested their idea on a absorb 99 percent of the energy from an elec 20-crystal array in the first week of November tromagnetic cascade produced by incoming 1987. They had originally planned to do it at high-energy electrons, positrons, or photons. the Los Alamos National Laboratory, but it Unfortunately, great density also means seems international cooperation only goes so great weight. Each crystal weighs a kilogram. far. Graduate students Hong Ma and But the supporting structure between crystals Renyuan Zhu, who designed the calibration must be as thin as possible to minimize the setup and built the readout system, were odds of any particle from a cascade missing denied security clearances. Both are from the crystals. The support must also be very mainland China. So they wound up at a rigid to keep the crystals aligned accurately. commercial RFQ manufacturer, AccSys, Inc., The final design uses a lightweight grid of of Pleasanton, California, instead. molded carbon fibers. The grid is only Since the calibration occurs while the 0.2 mm thick between crystals. The grid experiment is running, the lithium target can weighs a mere 400 kg, yet it supports ten tons not be set up at the collision pOInt. With an of crystals. offset target, though, some photons would hit The BGO crystal array is designed to crystals at angles of up to 55° from the long measure energies in the neighborhood of 100 axis. The test showed that good results can Ge V with a precision of 0.5 percent. The still be attained at 60°. Thus the entire barrel array will require frequent calibration while can be calibrated from a single location. in use to maintain such high resolution. The The experimenters also let the system Caltech group developed a calibration method pulse continuously for two days· at an inten using a Radiofrequency QuadrupOle (RFQ) sity equivalent to a couple of weeks' running proton accelerator. in the LEP ring. The system proved quite The RFQ would be installed underneath stable. the LEP ring. An angled beamline would Although the test was successful, the connect the RFQ to a lithium target in the experiment hit one not-unexpected glitch. collision zone .. The RFQ would bombard the The unshielded RF source generated a power- 22 ENGINEERING & SCIENCE / WINTER 1988 ful field that interfered with the detector's 1988, and production of the two end caps (of H. Ma. R, Mount. H. New sensitive electronics. (This shouldn't happen 2,000 crystals each) will then begin. Overall, man and an AccSys employee at LEP, where the entire RF system will be it will take another year and a half at current (I. to r) testing the BGO cali bration system (along right well shielded, and the RFQ will be much production rates to finish the assembly. hand wall. from left) RF farther away from the controls.) Unfazed, the And so it goes. There's still a lot to be source (in wood and wire team slapped an RF shield together from done before L3 is up and running. The chief enclosure). RFQ accelerator wood slats, chicken wire, and generous French mechanical engineer said it best. (on white cart-BGO array swatches of aluminum foil. This four-hour With elegant calligraphy, he painted an old [not shownl would go at left end). detector electronics, The exercise in low technology reduced the RF Chinese proverb on the finished half-barre1's RF conduit emerges near the interference tenfold. They gained another housing. Roughly translated, it says, "In a top of the enclosure and tenfold reduction by moving the RFQ away journey of 100 miles, 90 miles is halfway." passes down behind the RFQ, from the rest of the system. This was no Will the results be worth the wait? New mean feat-the RFQ weighs about 800 man is confident. "We've been searching for pounds. Six people and a one-ton hoist were the last ten years for the top quark. My last needed to push'it across the floor. Further experiment was actually built to find the top more, the vacuum system had to be opened quark, and in eight years of running we didn't to the air while the team inserted a ten-foot find it. With L3, we have achieved very extension into the RF conduit. The conduit ambitious technological goals in electron, itself is a two-inch copper pipe containing a muon, and photon resolution. We have to be smaller pipe held in alignment by teflon rather unlucky, or nature has to be almost spacers. The system acts as a high-power perverse, for us not to see the top quark. If coaxial cable, carrying the RF power to the the standard theory is working, the top quark RFQ accelerator. should be in the LEP range. And if we don't Even before the calibration test was see the top quark, the self-consistency of the arranged, BGO crystal production entered unified electroweak theory will be in jeop high gear. As these are the largest BGO crys ardy. Either way, we will have a new, excit tals ever made, early production runs were by ing situation for the experimenters and for trial and error. The system is up to speed the theorists."D-DS now, with more than 100 people working around the clock to produce 420 finished crystals every month. The first half-barrel of 4,000 crystals has been shipped to CERN, Newman will give the Caltech Associates where it was set up and precalibrated in July President's Circle a two-day tour of CERN through August, 1987. Crystal production for and L3 this May. The President's Circle will the second half-barrel will be finished by early tour Europe May 3 through May 12. 23 Remembrances of Kurt Codel by Olga Taussky-Todd , This article is adapted from a T WAS MORITZ SCHLICK'S seminar on the philoso tify to the fact that any argument could be talk given in Salzburg in July I phy of mathematics that put me in contact settled by citing his Tractatus. 1983, The remarks on the with Kurt GOdel in my first year (1925) at the In spite of my admiration for Schlick, I occasion were collected in a book, "Godel Remembered," University of Vienna. In this seminar we myself left his seminar and even his private published last summer by studied Bertrand Russell's book, Introduction circle, to which I had been admitted. I was Bibliopolis, Naples, Italy. The to Mathematical Philosophy, in German the youngest in age in the Vienna Circle, but American distributor is translation. In the first meeting of this sem I was disappointed that these gatherings could Humanities Press Interna inar the subject of axioms was discussed, in not give me guidance for my work in number tional, Inc., Atlantic High lands, New Jersey. particular the axioms for number theory and theory. Had I realized what G6del would the axioms of geometry. In the second meet achieve later, I would not have run away. ing I dared to ask a question, namely about a For G6del's results show that logic is not a connection between the two systems of subject that stands alone and is a basis for axioms. This led, unexpectedly, to a long dis mathematical thinking; it is in fact part of cussion between participants who enjoyed mathematics. making lengthy speeches. Schlick never said G6del and I were both born in Moravia, a word, and I myself was unable to follow then a part of the Austro-Hungarian empire much of it. At the end of the meeting and now in Czechoslovakia. G6del's birth Schlick asked who would like to report on the place was Brno (Brunn) and mine, Olomouc seminar. Nobody replied. Then Schlick (Olmiitz). GOdel's family had moved to asked: "Who will report on this seminar?" Vienna in 1923, while my family had already Then G6del volunteered. He started with the gone there in 1909, then later to Linz, then words: "Last week somebody asked the fol back to Vienna. G6del entered the Univer lowing question ...." He did not know my sity of Vienna in 1923. Among our teachers name, of course. However, from then on I were the number theoretician Philip Furt was at least "somebody" for G6del. wangler, Hans Hahn, Wilhelm Wirtinger (an Although G6del's home was in the mathe expert on algebraic functions), Karl Menger, matics seminar, although he was to become a Walter Mayer (who became Einstein's assis student of Hahn's, although he was not a tant), Lense, Helly, and Vietoris (who is con member of the Vienna Circle (Wiener Kreis), sidered the founder of algebraic topology), he was nevertheless an offspring of the Furtwangler, under whom I wrote my Vienna Circle to which Hahn, Menger, Car thesis, had earned recognition through his nap, Waissman, and others belonged. The early work in geodesy-I suppose during Vienna Circle was a group created by Schlick World War 1. Later he emerged as the man and concerned, if I understand it correctly, who proved, and disproved, David Hilbert's with the development of a language for sci conjectures in class field theory. He was self ence and mathematics. The time was ripe for taught and not, as is usually assumed, Hil such a creation in the late 1920s, because the bert's student; in fact, the two men never necessity of testing the foundation of mathe met. His first appointments were in his matical thinking, the methods of proofs, the native Germany. He finally had an offer axioms, the rules, had become pressing. It from Vienna, as successor to Mertens. It was a time of Sturm und Drang in mathemat was only through the Royal Society obituary ics. Wittgenstein was the idol of this group. notice by G. Kreisel that I learned that I never actually saw him there, but I can tes- Furtwangler's course on class field theory 24 ENGINEERING & SCIENCE I WINTER 1988 almost lured GOdel into this subject. How vised a PhD thesis for which he was the class theory could have profited from a man official supervisor. This thesis concerned like G6del! However, elementary number sequence spaces and used earlier papers of his theory was an essential ingredient in G6del's that were based on work by Helly. He made work. no secret of this and, in fact, he asked me to Hans Hahn was professor of geometry in be sure to cite HeIly when I gave a lecture the widest sense. He was also politically about the results of that thesis. It was active for the socialist party, he wrote many through Hahn, who recommended me to papers in mathematics, he was a member of Courant at G6ttingen, that I later captured the Vienna Circle, and he was very interested a prestigious temporary appointment there, in logic. But he was also an ardent follower of ESP. I myself attended one of his lectures on this subject when he fended off people who asked doubting questions. He was extremely averse to fakes who harmed the relevant research. Since G6del was Hahn's student and had apparently similar inclinations him self, there may have been conversations between the two men in this connection. But I don't know this for certain. Hahn had done a great deal of work in pure mathematics. He had started off as a young man in a university position in Poland and was attached to the topology school there, but his contributions were in other branches of mathematics too. I was Hahn's assistant in my last year in Vienna (1933/34), which was also Hahn's last year of life. He Left: Kurt G6del was already very ill, and I practically super- Far left: Olga Taussky in G6ltingen in the early 1930s. Below: Taussky, at left, as Hans Hahn's assistant, in a seminar in Vienna. Hahn is seated next to her. 25 papers. He saw immediately that there was only one case of the discussion that could lead to difficulties. He seemed a rather silent man, but he offered his help whenever it was needed. Austria was not a very prosperous country in those days, so the mathematical seminar enriched its library by publishing the journal Monatshefie. The seminar members thus had a lively exchange with the publishers of other mathematical journals in many countries. In addition, by including reviews of books, which the local reviewers had to return to the library, they acquired many books. Godel took on a number of these reviews. In addi tion, he was an editor of Menger's journal Ergebnisse eines Kolloquiums. I was working on my dissertation at this time. Shortly before I received my doctorate, my family moved back to Vienna. I was then allowed to invite colleagues to tea. Godel enjoyed these invitations, but he was always very silent. I have the impression that he enjoyed lively people but didn't like to contri bute to nonmathematical conversations him self. On the whole these tea parties were for colleagues of my own age group, or for visi tors to Vienna whom I had met and whose work was not unfamiliar to me. They were either class field experts such as Iyanaga from Japan (a student of Takagi) or attached to Menger's school. Menger himself came too. There was also Midutani, a statistician, but I do not know how he came to join Menger's At the Taussky home in helping with the edition of Hilbert's papers group. But, as usual, I was asked to check his Vienna: Above, GOdel; and in number theory. work. Most visitors came on sabbatical from below, a tea attended' by Karl I know nothing about Hahn's work in the U.S., or from Japan. Menger (left), next to GOdel. Taussky is second/rom right, logic, but he was GOdel's thesis man, and I I did not realize that GOdel, who clearly expect he was highly competent. I was seemed to enjoy himself, felt superior to this present at a lecture on logic that he gave to circle, as in fact he was. He somehow heard an audience of nonexperts. There he stressed that I had invited Hahn and the great Takagi, the fact that mathematical work was nothing on his visit to Vienna after the Zurich confer but a set of tautologies. This made me un ence, to meet my family, and he remarked happy, for I am a devoted number theoreti that I counted him among the minores gen cian. My feelings were expressed in one of tium. This is a medical term for doctors not my poems by the words, "Number theory is of the highest standing, who of course could greater than what comes later in the strict not command the highest fees. (In Austria athletics of mathematics" (Primary Press, doctors still conversed in some sort of 1979). medieval Latin slang, partly to prevent During our student years it slowly became patients from understanding them.) obvious that GOdel would stick with logic, There is no doubt that Godel had a liking that he was to be Hahn's student and not for members of the opposite sex, and he Schlick's, and that he was incredibly talented. made no secret about this fact. Let me tell a His help was much in demand. One day I little anecdote. I was working in the small told him of my new system of axioms for seminar room outside the library in the groups. He re-proved its validity immedi mathematical seminar. The door opened, ately. Menger asked him to study one of my and a very small, very young girl entered. 26 ENGINEERING & SCIENCE.! WINTER 1988 She was good-looking, with a slightly gloomy die in allen Fachkreisen grosstes Aufsehen face (maybe timidity), and wore a beautiful, erregte und-wie sich mit Sicherheit quite unusual, summer dress. Not much later voraussehen lasst-ihren Platz in der Kurt ent~red, and she got up and the two of Geschichte der Mathematik einnehmen wird." them left together. It seemed a clear show-off (" ...A solution of the top rank, which pro on Kurt's part. voked the greatest respect in all mathematical That same girl changed quite a bit later; circles and, as can be confidently predicted, maybe she became a student. She came to will take its place in the history of mathemat talk to me occasionally, and she complained ics.") This did in fact happen and shows about Kurt being so spoiled-having to sleep Hahn's correct appreciation. By 1931 he had long in the morning and similar items. already achieved worldwide fame, only a year Apparently she was interested in him and after receiving his doctorate. He had proved wanted him to give up his prima donna the existence of undecidable mathematical habits. Quite a bit later she handed a paper statements and had given a shock to the to Professor Menger, something on topologi world of logicians and mathematicians. But cal spaces. Early on in my life many people he had achieved even more: he had shown have handed chores over to me, chores of all something that any person with a minimum sorts. This is still the case, and I still do not of background can understand. This is a very know what to do about it. Hence, Menger important accomplishment in my opinion. asked me to check her work. It really was G6del achieved many other results, perhaps not in my line. The best I could do was to sit just as important, but this is the one mainly down with the girl and read it with her, mak associated with his name. ing her explain it to me. It appeared soon I saw GOdel in 1931 at the conference of that she was unable to do so but was truly the German Mathematicians Association in grateful to have somebody to talk to about it. Bad Elster. It was there that he met Ernst It also appeared that she wanted to show Zermelo. I think that perhaps no other per Kurt that she could do something. son is alive who remembers this event. I had Godel was well trained in all branches of good reason to know about it, for I worked mathematics, and you could talk to him then with the number theoretician Arnold about any problem and receive an excellent Scholz, who was a great class field theory response. If you had a particular problem in expelt. Both Scholz and Zermelo worked in mind, he would start by writing it down in Freiburg. Scholz was eager to help Zermelo symbols. He spoke slowly and very calmly, and thought a discussion with G6del would and his mind was very clear. But you could achieve this. But Zermelo was a very irasci talk to him about other things too and his ble person. He had suffered a nervous break clear mind ma<;le this a rare pleasure. I down and felt ill-treated, but had actually understand that Einstein had many conversa~ recovered at that time. He had no wish to tions with him. meet GOdel. A small group suggested lunch In due course I heard that Godel'sl929 at the top of a nearby hill, which involved a dissertation, "Uber die Vollstiindigkeit des mild climb, with the idea that Zermelo Ernst Zermelo at Bad Elster. LogikkalkUls (On the completeness of the cal culus oflogic)" was an extremely important achievement and that Schlick was very much impressed with Godel's philosophicum-the one-hour examination in philosophy, which forms part of the D.Phil. examination. He achieved the position of Privatdozent (so called Habilitation) at a very early date with a paper called "Formal unentscheidbare Siitze (Formally undecidable statements)." G6del gave me copies of his first two fundamental publications. Unfortunately, I lent them to a colleague, from whom I could not extract them later. From the well-known Austrian mathematical historian Auguste Dick I learned that Hahn had written in his recom mendation, ".. .Eine Losung ersten Ranges, 27 should talk to Godel. Zermelo did not want about this and lashed out against Hilbert's to do. so and made excuses: he did not like 1932 paper "Tertium non datur, " saying Godel's looks (he actually had not met him); something like, "How can he write such a the climb was too much for him; there would paper after what I have done?" (GOdel had not be enough food if GOdel came along too. already proven undecidability.) When Godel joined the group, however, the Godel's father died young, apparently two immediately started discussing logic, and leaving the family well provided for finan Zermelo never noticed that he had made the cially. But it was never clear to me whether climb. Kurt was given his own share of the fortune The peaceful meeting between Zermelo then, or whether his mother received all of it. and Godel in Bad Elster was not the start of lt is my guess that he did not earn a penny a scientific friendship between two logicians. until his first invitation to Princeton. And The trouble with Zermelo was that he felt he then after that he presumably had no income had already achieved Godel's most admired until the second invitation. He then stayed in result himself. Scholz seemed to think that Vienna as an unpaid Privatdozent until he this was in fact the case, but Zermelo had not left Austria for good. announced it and perhaps would never have On GOdel's first visit to Princeton in 1933, done so. It is not impossible that others too several colleagues, including myself, went to had some of Godel's results. However, it see him off at the train station, where he seems to me that it needs an energetic person boarded the Orient Express. Later he con to take over and, with full knowledge of the fided to me that this was not his actual depar work that is in the air, give creative guidance ture from Vienna. He was taken ill before for the future. I doubt that Zermelo was such reaching his boat; he took his temperature a person. and decided to return home. His family per Godel suffered not infrequently from suaded him to try again, however. I do not severe mental breakdowns. I do not know recall anything he told me about Princeton or whether they were caused by the overstrain about the United States apart from the fact he suffered through the creative processes he that the steaks he ate were very small. made his brain carry out or whether they During his second stay in Princeton his were just in his makeup. Auguste Dick has health broke down. I heard about this in supplied me with an amusing remark by Cambridge about 1936 via a student of Miss Furtwangler concerning Godel's result when Stebbing, a professor of logic in London, who the latter had one of his paranoia attacks: "Is had heard it from Schlick. Schlick was sup his illness a consequence of proving the posed to have been pessimistic about G6del nonprovability or is his illness necessary for surviving much longer. I was truly upset such an occupation?" I felt honored when his when this was reported. (Godellived, how older brother Rudolf, a medical doctor, dis ever, until 1978.) cussed his anxiety about Kurt with me when I saw GOdel and his wife again in 1948. I tried to telephone Kurt during one of these Although I had not seen him since before the attacks. In addition Kurt had some physical war, it seemed as though we had seen each ailments. But he did not spare himself-a other the day before. My husband, Jack, and trait already noticeable in his performance in I were staying at the Institute for Advanced the Schlick seminar when I first met him. He Study in Princeton, attached to the von Neu was convinced of the value of his ideas and mann project and living in one of their little wanted to make sure that they were known houses. (The war years had transformed us and appreciated. into numerical analysts, and Jack had close The famous mathematician David Hilbert connections with von Neumann from when had, among other tremendous achievements, the latter had visited Great Britain.) made a list in 1901 of the unsolved mathe In 1948 GOdel had not yet obtained the matical problems he considered the most title of professor. It is hard to believe. There important. By doing this he channeled was a time when Einstein's influence there mathematical research for a long time after was not strong, although I do not know ward. In particular, his first two problems whether Einstein tried to do something about concerned the foundations of mathematics. Godel's position. But Einstein had great Although Hilbert's problems became the basis friendship for Godel. for GOdel's work, Godel was critical of Hil In Princeton, as in the old times in bert's claims and aims. He spoke to me Vienna, I invited Kurt to tea. 0 28 ENGINEERING & SCIENCE / WINTER 1988 Star Wars Technology: Will It Work? N MARCH 23, 19~3 President Reagan called by Amnon Yariv Oupon the American people, and especially the technological community, to undertake a research and development effort with the goal of finding an alternative to the current policy of assuring national security through the threat of nuclear retaliation. The new defense system, which was meant to "render nuclear weapons obsolete," was organized under the control of the Strategic Defense Initiative Office (SDIO) whose mission was to acceler ate and expand research in antiballistic mis sile defense. This organization and the pro gram have come to be known popularly as Star Wars. Almost immediately following Reagan's speech an intense debate ensued, centered on ethical, political, and philosophi cal issues. What was missing, however, was an in-depth technological analysis of the feasi bility of the proposed defense system. So in November 1984 the American Physical Society constituted a committee 29 Of the three stages of an ICBM's trajectory, the missile is most vulnerable during the boost phase. Intercepting it during the post-boost phase may still be possible, but in mid-course it becomes immensely more difficult to stop. of engineers, physicists, and chemists, whose independent warheads and the decoys make charter was to study the technological aspects up the third and longest phase, the mid of Star Wars and report on them to the mem course stage, which lasts about 20-22 minutes. bership, The 17 members of the committee The terminal stage, when these warheads re came from industry, academia, and the na enter the earth's atmosphere over the intend tional government laboratories. Its funding ed target, is very quick. came from private foundations, and it en The missile is most vulnerable during the joyed the full cooperation of SDIO, as well as boost phase. During this phase the missile still access to classified information. I was a includes the booster rockets, so the whole member of that committee. But before I thing is very large-about 30 meters tall and summarize the committee's findings, I'd like close to 200 tons in weight. Because it's so to set forth a primer on intercontinental large, it cannot be armored and is therefore ballistic missile (ICBM) technology and the relatively soft. It's also very hot, making it defense strategies, mainly lasers, envisaged relatively easy to detect with infrared sensors. under Star Wars. Duringthe post-boost stage the bus is still A missile under way from, say, the Soviet relatively vulnerable but much less so than Union to the United States goes through during the initial phase. Because it's no three phases on its journey to re-entry. The longer firing, it's much colder. In the third first of these, the boost phase, takes the mis phase, however, intercepting the missile is sile from rocket blast-off to an altitude of immensely more difficult, because now in about 200 kilometers and lasts about three to addition to the warheads there might be as four minutes. The burned-out stages of the many as 100,000 decoys moving with the rockets have been dropped by the end of this speed of the warheads and, using currently phase, and what is left in the post-boost phase available means, indistinguishable from them. is a big container called the bus, which is now It is generally agreed that for a ballistic moving at close to the orbital velocity of ap missile defense system to be most effective, it proximately seven kilometers per second. has to operate on the vulnerable boost phase The bus contains up to 10 warheads and knock out the great majority of the mis (MIRVs-multiple independently targeted siles during that stage. Failing that, the mis re-entry vehicles), which it unloads during the sile could still be attacked with appreciably post-boost phase one at a time, aiming each greater difficulty during the bus phase, when one at its specific target. At the same time it is still a single target containing all the war the bus may also unload lightweight decoys. heads. After that we may have to handle as This stage lasts about five minutes. The many as 100,000 targets. 30 ENGINEERING & SCIENCE/ WINTER 1988 Today the Soviet arsenal consists mostly tight wave and a single, pure color at a single of SS-18 and' SS-19 ICBMs. Altogether we wavelength, in contrast to a light bulb, which assume 'that the Soviets' have something like puts out light in all directions at different 1,400 missiles, each one capable of carrying angles and different wavelengths. It spreads six to eight MIRVs. So we are dealing with a out, whereas the laser beam stays almost total number of about 10,000 warheads parallel and doesn't spread. But it turns out potentially aimed at various targets in the that the laser beam does spread just a very lit United States or elsewhere. tle bit, and this spreading limits what you can Since we only have three or four minutes do with it in some cases. to destroy even the current Soviet missiles, How much laser energy do we actually which are relatively slow, during the boost need for damaging a missile? We know how phase, we need a weapon that can strike much power we can get from lasers, so the quickly. Nothing moves faster than light, so basic physics is very simple. The skin of a the main weapon proposed under Star Wars Soviet missile is usually made of aluminum, is the laser. and it might have some ablative surface on it Other interesting approaches considered that would take more energy to burn through. include charged and neutral atomic beams that move at close to the speed of light. The first faser consisted ofa helical/lash lamp wound These technologies are still very far behind around a synthetic ruby rod the laser and are not effective in the atmo containing chromium ions. sphere. We considered them less viable, Mirrors at the ends of the rod although we still looked at them in great bounce photons (from detail. The main reason for the choice of the the excited ions) back and forth, building up to an laser, in addition to the speed of light, is the intense beam of light. coherent nature of its light output, which in principle makes it possible to deliver the requisite level of light intensity across inter continental distances and in many cases to penetrate the atmosphere. The first laser was designed in 1961 by Ted Maiman at the Hughes Research Labs in Malibu and is shown schematically at right. It consisted of a helical flash lamp wound An ordinary flashlight pro around a synthetic ruby rod, which is just duces light at different aluminum oxide containing chromium ions. wavelengths and different The end faces. of the rod were polished and angles spreading in all direc served as mirrors. These chromium ions are tions. Laser light is a tight, coherent beam at a single pumped by the surrounding helical flash lamp wavelength .. to excited states. A photon of the proper ruby frequency starts traveling between the mirrors and stimulates an excited chromium ion to make a downward transition. Now you have two photons of the same frequency. Another transition and you have three pho tons, and so on.. So the light builds up and becomes more intense as it travels. When it INCOHERENT LIGHT FROM A FLASHLIGHT hits the mirror, it turns around and comes back. In the meantime the atoms have been re-excited back to the upper state and the process continues. You end up with a very intense light beam bouncing back and forth between the two mirrors, and that's all a laser really is. What makes laser light different from the light from an ordinary light bulb is that, as the photons are forced back and forth by the mirrors, they organize themselves like COHERENT LIGHT FROM A LASER soldiers marching in step. This gives a very 31 tance R. A is the wavelength of the laser. Telescope From the formula you can see that if we D try to compensate for the spread (diffraction) I of the beam by using more laser power to LASER I start with, the amount we need becomes exor bitantly large. But since the spread angle is Diffraction of the laser beam given by the ratio (AID) of the wavelength of can be decreased by a tele D o=~ D scope. Because the'spread the radiation to the diameter D, you can angle is given by the decrease the spread by increasing the diame wavelength over the diameter L,LMER ' ter. You do this with a telescope. Unfor of the telescope mirror, the tunately, to get the required laser power, we larger the mirror, will need a telescope with a mirror about 10 the smaller the spread. meters in diameter. This is pretty big for a telescope. In fact, it's the size of the Keck Telescope, which is currently under construc tion, and which, when finished, will be the largest in the world. So, in addition, we wj]] need a sizable structure to aim and control the laser. Like the Keck Telescope, Star Wars mirrors will probably have to be constructed out of segments that will be carefully adjusted relative to each other to achieve a perfect surface. And where are we going to put these huge, expensive, delicately adjusted mirrors? If we want to get a laser beam from, let's say, Wyoming to Irkutsk, we can't go in a straight line because of the curvature of the earth and the atmosphere. So the current scenario is to put these mirrors in geosynchronous orbits To decrease diffraction over All we have to do is go to the lab and find about 36,000 kilometers above the earth and long distances, a ground out how much laser energy it will take to stationary relative to a point on the earth. based excimer laser (which burn a hole in missile skin_ This has, of produces an intense beam of Somewhere over the Soviet Union other 10- ultraviolet radiation at a course, been done, and it turns out to be meter mirrors, called fighting mirrors, would wavelength of.3 micrometers) roughly 30,000 watts per square centimeter of be positioned at much lower orbits, say 1,000 could be beamed to a large the target during an exposure of one second. kilometers above the earth. The laser would mirror in geosynchronous But if you want to deliver concentrated light be beamed from the earth to one of about orbit around the earth, and energy over long distances, you can't just from there to a fighting mirror 100 geosynchronous-orbit mirrors al}d then at lower orbit, which would point the laser at the target. Although a laser relayed to the fighting mirror, which would focus it on the missile. beam spreads very little in comparison to aim and focus it down on the ICBMs as they ordinary light, over 40,000 miles it spreads climb during the boost phase. too much to be effective for burning a hole in From "Space-based Ballistic-Missile What kind of power do we need to do Defense" by H. Bethe, R. Garwin, K. anything. The requisite power of the laser for this? We know it will take 30 kilowatts for Gottfried, W. Kendall. © October 1984 damaging a missile is given by by Scientific American, Inc. All rights one second to burn through the skin of the reserved. missile. The formula tells us how light P""ff ~ [I"'h"' 1nR' [ ; r diffracts as it propagates through the atmo sphere, so it's just simple bookkeeping. If we In this formula llethal stands for the inten were to use radiation of one micron we will sity (watts/cm 2) necessary to burn through or need a laser of roughly 30 megawatts. If this seriously weaken the skin of the missile, laser is going to be based on earth, so that we assuming approximately a one-second expo have to propagate through the atmosphere to sure. R is the effective range from the laser the geosynchronous mirror and down to the to a relay mirror (to be discussed later), and fighting mirror, we will need as much as one D is the diameter of the laser telescope aper gigawatt of power to achieve the requisite ture, that is, the final (large) diameter of the power density on the Russian missile in order launched laser beam. The factor RAID is the to burn through its skin. One gigawatt is, for theoretical radius of the laser beam at a dis- example, roughly the power output from one 32 ENGINEERING & SCIENCE / WINTER 1988 of the nuclear units in the San Onofre power relative to the climbing enemy missile. Then plant. The difference, of course, is that this the nuclear explosion is triggered, and x-rays kind of power has to be available only during are beamed at the vicinity of the target. One the engagement period of one to three of the main problems with the system is that minutes. The table at right shows some of x-rays don't penetrate the atmosphere, so that the typical power levels that result from the you have to wait until the Soviet missile has above cbnsiderations. risen above most of the atmosphere. The x-ray laser, the most exotic entry in the laser weapon field, doesn't need mirrors Bouncing an effective ground for the simple reason that no mirror has yet based laser off a lO-meter been made that can withstand x-ray radiation, mirror in space to an orbiting fighting mirror demands vari and its operation is not based on feedback as ous power outputs depending in our example above but, rather, on one-pass on the laser's wavelength. amplification of spontaneously emitted x-rays. Since most of the information on x-ray lasers is classified, a small group from our commit tee who had the top Department of Energy clearance was able to go to Lawrence Liver more Labs to be briefed on them. I was not one of that group, so what I am going to say about x-ray lasers comes not from our com Because it's very light, an mittee deliberations but from the open litera x-ray laser could be launched ture. An x-ray laser is essentially an efficient and accelerated quickly from a nuclear bomb surrounded by a bunch of submarine to a line-ofsight position relative to the missile. metal or plastic rods. When the bomb is A nuclear explosion triggers exploded, most of its output consists of x the laser. rays. In the billionth of a second before all the rest of the junk reaches the rods, the x From "Space-based Ballistic-Missile rays get there first and excite the atoms in the Defense" by H. Bethe, R. Garwin, K. rods. Here you are just replacing the helical Gottfried, W. Kendall. ~ October 1984 by Scientific American, Inc. All rights flash lamp around the ruby rod in Ted reserved. Maiman's laser with a nuclear bomb. When the x-rays excite the atoms in the rods, they lase along the rod axis. It's a rather crude laser in the sense that it isn't very collimated, or tight, and you can't concentrate the power on one small spot, but that's just what you might want in this case. The x-ray laser doesn't have to be pointed very accurately because it splashes a great deal of energy all over the place. The amount of energy envisaged is large enough that it can burn and disable objects over quite a broad area, so x ray lasers just need to be pointed at the gen eral vicinity of the target. It would work like this: say the Soviets launch some ICBMs. Radar and other satellite-based sensors detect them, and the information is relayed to a U.S. submarine a few thousand kilometers off shore, which launches an extremely fast rocket. The x-ray laser is very light in cbmparison to any other laser considered. It does not have enormous mirrors (all it consists of is a nuclear bomb, which is a relatively light device, and a few rods), so you can achieve fast acceleration from the submarine to a line-of-sight position 33 probes the atmosphere, and when it arrives, its own distortion (spread) contains informa tion about the atmospheric turbulence. A wavefront sensor extracts this information, which is then used to deform a mirror. The actual weapon laser beam is now reflected from this mirror and in the process gets dis torted in exactly the complementary distor tion necessary to balance the distortion caused by the atmospheric turbulence. The two distortions cancel each other, producing a diffraction-limited (collimated) beam. A de formable mirror would have to have as many as 100,000 little actuators that can push from below and deform the mirror locally. Caltech has played an important role, con ceptually and experimentally, in demonstrat ing many of the basic ideas that are involved in this technology, which goes under the Using phase conjugate optics, The atmosphere also bothers optical lasers. name of phase conjugate optics. About 30 a weak laser sent from a mir An ideal laser beam stays very tight and will years ago Caltech astronomer Horace Bab ror in space would carry an propagate with a very small angle of disper cock used techniques of this sort to compen "image" of the atmospheric turbulence in its own sion. But an optical beam launched from an sate for scintillation in order to resolve astro diffraction. This could be earth-based laser has to move through the nomical objects better. In my own group we interpreted by a wavefront sen atmosphere, and air turbulence makes the have been working for the last 10 years on sor and used to deform the atmosphere an uncongenial medium for opti phase conjugate optics and have developed, mirror that launches the cal propagation. Instead of remaining narrow unrelated to Star Wars, nonlinear optical actual laser weapon, so that the distortion of the outgoing and tight, the beam breaks up and spreads techniques that are relevant to correcting laser would cancel the effect of over an angle that might be thousands of atmospheric distortions. The technique has the atmospheric turbulence. times larger than the ideal angle-with a con many other applications,including photo comitant loss in laser power on target. The graphing through clouds and transmitting intensity loss is so large that by the time the images through fibers. power reaches its target the intensity is too So we now know, at least theoretically, low to do any real damage. One of the major how to generate light, how to launch it, and thrusts of current Star Wars research is deal how to compensate for distortion. What hap ing with this problem. pens when the weapon beam is actually over In one of the proposed solutions the large the target? Two or three fighting mirrors at Because of the earth's rota mirror assembly in geosynchronous orbit an altitude of 1,000-2,000 kilometers can tion, the number offighting mirrors actually necessary to (36,000 kilometers above the earth) will house cover the whole of the Soviet Union. But cover a launch site has to be a small laser that shoots a beam down to the two or three mirrors don't have enough time exceeded by a factor offive. earth-based laser station. This weak beam during the one to four minutes or so of the .. _--_._------, boost phase to take care of all the Soviet mis siles that are climbing. The Soviets can launch as many as 1,500 missiles at one time, ICBM Launching site so for an anticipated worst-case boost phase of one minute we will need some 12 mirrors, assuming half a second per kill. But these 12 mirrors can't be planted over the U.S.S.R. enemy territory They will have to be orbiting all the time, and since we don't know when the missiles will be launched, because of the earth's rotation we ~ earth will have to cover the whole earth with a blanket of orbiting mirrors. The factor by which you have to exceed the 12 or so that you actually need during engagement is referred to as the "absentee ratio," which "area covered by one satellite turns out to be about five. So we will need 34 ENGINEERING & SCIENCE! WINTER 1988 about 60 orbiting fighting mirror plaJforms, rockets that will bum in one minute instead each one JOmeters,in diameter. We will also of four to five. This is possible even now and need one 'relay mirror in a geosynchronous means that the critical time during the boost orbit for each of the 12 or so fighting mirrors. phase would be shortened by a factor of five. Even if we figure out how to do all this Almost every requirement would then be and make it work, it will take 10 to 15 years upped by that factor; we would need five to before this system is deployed, and the six times more power and so on. In addition, Soviets, of course, are not going to be sitting the boost phase would be completed at 80 still during thattime. They will be improving kilometers or so, which is still within the their own technology, especially in view of atmosphere. This would reduce the effec the fact that we are developing a ballistic mis tiveness of the x-ray weapon considerably, sile defense system. They have many possible because it can't penetrate through very much responses: they can attack the 60 orbiting of the atmosphere. And because the nuclear platforms, each containing mirrors, electron bomb that triggers the x-rays has to be ics, radars, sensors, and power supplies. launched from a submarine and now would These platforms that we are going to hang up have much less time to get to its line-of-sight there are enormously complicated and expen position relative to the target, new problems sive things, and they are relatively soft targets; would be presented. The post-boost vehicle that is, it will be far easier, I'd say by a factor could also be redesigned to discharge its war of at least a thousand, to destroy them than heads and decoys faster, thereby spending less for them to destroy the missiles. The Soviets time in this vulnerable stage. would only need to explode a nuclear bomb -Decoys and penetration aids-packing above the atmosphere just before launching the post-boost vehicle, the bus, with 200-300 the missiles. This would disable everything lightweight decoys. The decoys would be we had within range and make it easier for launched from the bus above the atmosphere their blind missiles to sneak through. and would move with the same velocity and Our mirror platforms could also be des have the same trajectory as the re-entry vehi troyed by lasers, lasers much weaker than the cle. They could also be shaped to look alike ones I've discussed here, because our mirrors, and to have similar responses to radar. By electronics, solar cells, and antennas are a today's mostly passive means there would be softer target than missiles. If the Soviets do no way of discriminating between them, with undertake a Star Wars-type program but are the result that, instead of concentrating on unsuccessful in developing strong enough one warhead, you would have to deal with lasers to defend against our missiles, the ones 200-300 objects that look the same. So, even they develop will very likely possess sufficient if we were successful in destroying 90 percent power to disable our weapon platforms with of the launched Soviet missiles in the boost all our mirrors and electronics. We really phase, so that instead of 10,000 warheads you have no clear idea how we are going to pro would have only 1,000, we might still have tect our assets in space. It's like buying a some 100,000 decoys to deal with. Cadillac and parking it in a high-crime area But forgetting possible Soviet counter every night for 10 years. measures for a moment, where are we with Other fairly simple options for the Soviets Star Wars technologies right now and how far that will make things more difficult for us do we have to go? Demonstrations of include: correcting for atmospheric turbulence are -Proliferation-increasing the number of short by a factor of 100 of what needs to be targets by building more missiles. done. Chemical lasers' effectiveness has to be -Booster rotation-rotating the missile as increased by a factor of 100 to get to the it rises so that the laser beam can't dwell on range where they can even theoretically be the same spot long enough to bum through. used. To do the job, excimer lasers need to This simple move would increase the laser be four orders of magnitude, or 10,000 times, power requirement by a factor of approxi more powerful than what we have today. mately three-a huge factor when gigawatts X-ray lasers still have enormous problems. are already involved. Other exotic possibilities include free - Ablation-using protective material that electron lasers and particle beams. A free~ will require more power to bum a hole electron laser uses a beam of electrons that through. have been accelerated to as much as 100 mil -Past-bum boosters-making booster lion volts to change the polarity of a magnetic 35 achievement is demonstrated in the table at left. As one can see, we need improvements DEW TECHNOLOGIES of from two to six orders of magnitude in the basic technologies. STATUS Our committee came to the following con TECHNOLOGY clusion: "Although substantial progress has been made in many of the technologies of directed-energy weapons over the last two LASERS decades, the study group finds significant gaps in the scientific and engineering understand ing of many issues associated with the devel opment of these technologies. We estimate that even in the best of circumstances, a decade or more of intensive research would be required to provide the technical knowl edge needed to make an informed decision about the potential effectiveness and surviva bility of directed-energy weapons." It is important for us to have a sound field in a regular fashion. This forces the elec understanding of the performance limitations trons to undulate as they move along, and the of a directed-energy weapon defense system. wiggly motion forces the electrons to radiate, For one thing, the system will be very expen that is, give up energy in the form of photons. sive. Just shooting wildly from the hip I Electron-beam lasers would need to be im would guess thousands of billions of dollars, proved by six orders of magnitude, and much with tens of billions for annual upkeep and of the basic physics isn't understood yet. Par repair. Embarking on such an endeavor ticle beams-beams of electrons, protons, or before we possess the basic technology and an atoms, which can be accelerated to velocities understanding of the phenomena involved nearly that of light-are still only a suggestion will almost surely lead to failure. And a and are even further behind than lasers. failure of that magnitude will be a national We don't know how to discriminate catastrophe. among warheads and decoys; only prelim But there are already political pressures, inary ideas have been proposed, and much which are increasing, to proceed with these work needs to be done just to establish their untested ideas and technologies. This is prob feasibility. We haven't yet built lO-meter ably due to the reluctance of Congress to shell mirrors, much less figured out how to put out a few billion dollars every year for a basic them in space and point them fast and research program in developing technologies. accurately. The pressure is to show results, and that pres And system issues have yet to be sure can force us to commit ourselves to pre addressed. Weare going to have the most mature technologies that are certain to fail. complicated system ever designed operating We hope that our report will make it easier to in a hostile environment of laser beams and resist such pressures. exploding nuclear bombs. Vast amounts of I want to reiterate in closing that our com information will have to be processed, includ mittee was not "for" or "against" the Strategic ing locations, decoy discrimination data, and Defense Initiative. We concentrated on the kill verification. This information will have technological and scientific issues. I person to be transferred and shared as the defense ally would like one day to see an effective focus moves with the surviving missiles from missile defense system developed strictly for one phase to the next. All the components of defensive purposes, because the list of nations the system have to communicate perfectly, possessing nuclear weapons will increase and and we don't know how to do that. This some of these nations may not be quite "rea problem is so difficult that I don't think peo sonable." But our committee concluded ple even know where to start. We considered that, based on present-day technology and this the most difficult part of the Star Wars understanding, the job cannot be done. What challenge. we need is at least 10 years of intensive The degree of shortfall between other research and development simply in order to needed performance levels and present-day find out if it can be done. 0 36 ENCINEERING & SCIENCE / WINTER 1988 Res'earch in Progress Photographic Memory EURAL NETWORK research is one of squeeze a highly interconnected, computer, the pattern can be modified N the hottest topics in computing three-dimensional structure onto an at will. For example, the chip could nowadays, but the field is still in its essentially two-dimensional chip is a be "taught" by modifying the pattern infancy. The very features that give tough proposition. With the addi between successive runs until a desired neural nets their unique attributes tional condition that the connections output is produced. make them hard to simulate on "regu be made variable, the problem But how does the pattern get on lar" serial computers, and harder still becomes well-nigh impossible. the chip? That's the key to Agranat's to model as physical devices. A An alternative approach, using op off-the-shelf design. The computer's number of esoteric devices are being tical technology, shows promise. Since video drive is connected to one of researched that could someday revolu light beams pass through each other those tiny LCD (liquid crystal display) tionize computer design. In the mean unhindered, and beam intensity is TV sets-the kind that have recently time, Aharon Agranat, a research fel easily controlled, the connection prob become the gift of choice for consumer low in Amnon Yariv's applied physics lem vanishes. In fact, small neural electronics addicts. A system of lenses group, is developing a system to model nets have been built entirely in optics; reduces the image 20-fold and focuses the flexible complexity of neural nets but optical information processing it on a charge-coupled device, or CeD. using mature technologies. Caltech technology is not nearly as advanced A CCD is the image-sensing device in has several research groups working as its silicon analog. a television camera. The dot matrix with neural nets, and the faculty Agranat decided to put the best becomes puddles of electric charge on includes some of the field's leading developed features of silicon and opti the CCD, and the chip reads them off scientists. ' cal technology together. The result is just as a camera's chips would read a Neural nets are loosely modeled a system where silicon neurons process video image. "Basically, we can put after the brain's own structure, where data (connection strengths) from an any pattern of dots on the CCD we every brain cell, or neuron, is con optically loaded memory. He was need, using the LCD screen as a spatial nected to thousands of others. The assisted this past summer by SURF light modulator," Neugebauer said. connection strengths (or "weights") (Summer Undergraduate Research Fel "People are using LCD-TV s in all differ. Data (memories) and programs lowship) student Chuck Neugebauer, a sorts of optical computing setups," (such as how to process visual infor senior interested in VLSI physics. Agranat notes. "They are cheap, mation to recognize a face) are stored Agranat's system codes the connec about $200, and readily available. as connection patterns of varying tion weights for a chip with N neurons There are high-quality spatial light weights. When the brain learns, the as an N x N matrix. Each matrix ele modulators built specifically for patterns change. ment codes the weight from the neu scientific use, but they're very expen Building a brain, with its approxi ron in the element's row to the neuron sive. There's a mageneto-optical de mately 10 12 neurons, each with up to in its column. Transformed into opti vice with a 48 x 48 matrix. It's a very 10,000 connections, is still beyond cal data, the matrix becomes a pattern good device, but it costs about $5,000, anyone's wildest dreams. Scientists of bright and dark dots. The brighter and it can't make shades of gray. We today are working with a handful of the dot, the greater the weight. A per need shades of gray for the weights. neurons, modeling systems from a sonal computer from a home electron With the LCD-TV, we can make a 100 dozen or so up to a few hundred. ics store generates the pattern accord x 100 matrix having more informa Even these small nets present complex ing to any of several algorithms. ("It's tion. The optical setup is routine, but problems. not standard, but we got it really it's not trivial. The image has to be Although microprocessing technol cheap, and it's got good graphics capa placed accurately to within 10 microns ogy is pretty sophisticated, trying to bilities," Neugebauer said.) Once in the everywhere on the CCD, and the 37 brightness has to be uniform as well." The chip assembly is enclosed in a light-tight box with a shuttered aper ture, like a still camera. A single flash exposure suffices to load the pattern. "My background is in optics," Agranat said. "I had the idea, but I didn't have the expertise to build it. Then Chuck showed up. I built the optics, and he designed the chips. Usually this would be done as a col laboration of two professionals- J would find a postdoc to share the work. But here one of the profession als is an undergraduate, and that's unique. 1 gave him the architecture Computer's-Eye View the CCD, the integrators, and so forth, and he developed the designs com pletely on his own. Without him, this project would still be on paper." F A PROPOSED Mars rover can find its developed an algorithm to separate In fact, the group developed three I way through Congress, it may use a objects from background based on architectures over the summer. The Caltech-designed vision system to find their relative motion. Since a CCD system is a semi-parallel, syn its way around Mars. The vehicle computer's field of vision consists of a chronous architecture: Just as creating would have to recognize and avoid matrix of discrete points (pixels), a a video image requires a number of obstacles unaided, as signals from an velocity vector can be assigned to each sweeps equal to the number of rows earthbound driver would take as long pixel. The velocity vector field is gen on the screen, so processing the CCD's as twenty minutes to arrive. erated by comparing two successive contents takes a number of computa Assistant Professor of Computation images of the same scene and deter tional cycles equal to the number of and Neural Systems Christof Koch, mining how each point moved neurons. The other two architectures along with Brian Wilcox of the JPL between images. The system actually can calculate the network's output in vision group, and Carl Ruoff, a gradu traces the apparent motion of pixels of either a single cycle (fully parallel syn ate student in mechanical engineering, constant brightness. It works from chronous processing) or continuously are evaluating Koch's design for space two assumptions. Neighboring points without cycling (parallel asynchronous applications. The project is supported in the field of view move the same way processing). "Chuck dreamed up the by NASA funds. Current computer unless they are on different objects. the third one himself," Agranat said. vision systems are based on so-called When the points have different "He got the idea for it, and he went "expert system" programs. "Expert motions, they are separated by discon back and figured out exactly what systems are very brittle," Koch said. tinuities (sharp changes between adja engineering principles and what com "You have to have rules for every cent vectors) coinciding with the ponents it would take to make it work. thing. If you built a vision system that object's edges. It's actualiy the easiest of the three to way, it would be made up of if-then Koch's algorithm poses the prob fabricate, so it may be the first one rules like, 'If you have a red blob at a lem in terms of minimizing an energy we'II see working." one-meter height in an office, it's prob function (an optimization problem). All three designs have been sent ably a telephone.' But it could be a The problem resembles a landscape of out for fabrication. Some of the chips cup. So you add another rule: 'If it's hills and valleys. A boulder dropped have returned recently, and the testing round, it's a cup.' But maybe it's on one of the hillsides should eventu process is about to begin. Several vari really an apple. Our approach is much ally roll to the lowest point in the val ations on each architecture will be more flexible." ley. Computing the boulder's path examined, and one design each will be "You never think about vision," mathematically is time-consuming, as chosen for the next scale-up. The larg Koch said. "You open your eyes and thousands of iterations may be est chip in the current batch has 32 the world is there. But your eyes just required over the entire landscape. neurons, enough gray matter to per make a big array of numbers-they're However, the problem is directly form simple computations. actually voltages, but conceptually they analogous to the behavior of a network A 1,ODD-neuron chip could be built correspond to numbers. Out of that of simple resistors. The junctions, or with current technology, Agranat said, array you have to infer that one object nodes, correspond to the pixels. An but there aren't as yet spatial light is in front of another, that things are external voltage fed to each node modulators big enough to load them. moving, how they're moving. This is corresponds to the brightness change at In the meantime, the lOO-neuron called early vision-getting three that pixel. Once the system has range accessible through the LCD-TV dimensional information from a two reached a stable state (minimum is sufficient for much network research dimensional intensity array." power dissipation), the voltage at each that would otherwise be impossibly Koch heads a research group work node corresponds to the velocity. expensive or time-consuming to ing on early vision and related compu "You can exploit the physics of the run.o-DS tational problems. The group has network to get the solution, instead of 38 ENGINEERING & SCIENCE / WINTER 1988 going to a computer and using a set of between the nodes, creating a velocity the active line processes break their logical rules," Koch ~aid. field discontinuity-an edge. respective connections, and the current Other people' have proposed The result is a hybrid chip design flows until another field stabilizes. relative-motion algorithms that can . with analog and digital components The system re-evaluates the nodes, determine an object's direction and connecting each node. The analog turning on new line processes as velocity. Unfortunately, these components are variable resistors needed, and the cycle repeats. "In its methods.generate a velocity field that designed by Carver Mead, the Gordon final state, you can read off the veloci varies smoothly over the entire visual and Betty Moore Professor of Com ties and discontinuities directly," Koch field. The smoothness hides an puter Science (E&S June '87). The said. "This thing has almost solved object's edges, leaving an unrecogniz chip actually has two identical the problem of object segmentation able blob. Furthermore, if overlapping networks- one each for the x and y deciding what surfaces belong to what objects move in opposite directions, components of motion. Each pixel's x objects. It's a very difficult problem." the network averages their velocities, and y nodes are linked by a variable The group has simulated the sys generating a zero-velocity region where resistor. tem on JPL's 32-node Mark III Hyper the objects meet. The system operates in cycles. Ini cube parallel computer, and on smaller Koch's network design includes an tially all line processes are "off" and 4- and 16-node NCUBE computers on edge simulation process. This so current flows until the field stabilizes. campus. The most recent work used called "line process" has recently been Next, an external unit evaluates each 128 x 128 pixel "before" and "after" introduced in a variety of vision appli node. If a neighboring node has a images from a video camera. Given a cations. The line process is a binary significantly higher voltage, the line moving hand on a featureless back switch in each resistive link. When the process between them turns "on," ground, the system generated an easily line process is "off," the switch is creating an edge. Evaluation is fast, recognizable image, complete with closed and current flows-smoothing even for large networks, because fingers, in only seven cycles. (1,000 the velocity field. When it's "on," the system-wide information isn't needed. iterations of the resistive net alone gen switch is open. No current flows Once all the nodes have been checked, erated a smooth-flow image resembling "Seeing" a moving hand. Top row, from left: 128 x 128 pixel image from video camera; the computer compares it to a second image (not shown) in which the hand has moved as much as two pixels. Initial velocity data; "snow" in background also registers. Smooth-flow field after 1,000 iterations, with grayness proportional to velocity. Bottom row, from left: State of the hybrid network after one, three, and seven cycles respectively. Line processes grow toward each other along edges, and the velocity aura around the hand disappears. After thirteen cycles (not shown) it vanishes altogether. 39 " " " In a hallway. Top row, from left: 128 x 128 video image; person in foreground is moving right and person in background is standing still. Discontinuities superimposed on initial velocity data; note band of camera noise across bottom of both images. Smooth-flow field after 1,000 iterations; noise vanishes as random variations cancel. Middle row: Hybrid network after one, three, and five cycles respectively. Bottom row: After seven, ten, and thirteen cycles. Line processes not supported by the velocity field wither, while line processes coinciding with edges grow. a mitten.) A person walking down a A 30 x 30 chip built to Koch's around 220 x 220. And Musca hallway filled with stationary objects design has recently returned from the domestica is a very good flier-try to was cleanly separated from the back fabricator, and network testing is catch it. So we think with 128 x 128 ground in thirteen cycles. about to begin. If this is successful, a we can do all the navigation we want. "Even though it's a Hypercube, it 128 x 128 chip could be built in two It may not be able to read, but our still takes 20 minutes to simulate two to three years with existing technology. Mars rover doesn't really need to do frames which may bea small fraction "The common housefly, Musca that," Koch concluded. Even if the of a second apart. The chips, we domestica, has 24,000 photoreceptors Mars rover never gets off the ground, think, will be able to do it in real time in each eye, or about 50,000 photore this rugged, lightweight technology will or very close," Koch said. ceptors covering almost 360°. That's find plenty of earthbound uses.D-DS 40 ENGINEERING & SCIENCE / WINTER 1988 Books First Light The Search for the Edge of the Universe by Richard Preston The Atlantic Monthly Press, 263 pages, $18.95 FI:--ID RICHARD PRESTO~'S book exciting, the 200-inch. A large supporting cast wastebasket," Jim Gunn's home I easy to read and informative, with a includes Caltech scientists and designed, home-built electronics and style that is vivid, slangy, and funny, engineers, past and present. Many of his deep cosmological insight. It describes two aspects of contem them are my good friends, but my The central plot is about how porary astronomy and some of the tri enthusiastic recommendation for the Maarten Schmidt needed a statistically als and fun of people who work at book is not biased by this fact-nor by complete survey of the numbers and Palomar, But the real heroes of this my own appearance in a "walk-on" properties of quasars at the limit of book, which reads in part like a novel role from an older pre-electronic detectability with CCDs (charge but is mostly true, are two telescopes generation, coupled photosensitive diodes). The (plus attached electro-optical scientific The Shoemakers, husband and CCD field of view is small and quasars instruments, computers and programs). wife, revi vified the study of the are rare; to obtain a large enough area, The book covers the stories of asteroids, from rock to mountains in a scan is needed. The stationary tele two quests. One of them is the search size, remnants of the origin of the solar scope images a continuous strip of sky with the Palomar 200-inch Hale system and fascinating hazards for drifting by. For this, Jim Gunn reflector for quasars at the largest red interplanetary travel. They use (often invents and builds a "kludge" which shifts attainable-and the implication in a shower of sparks) the oldest tele transforms the "four shooter" (which for cosmology of looking back so far scope on Palomar, the 18-inch carries four CCD's for stationary imag in time. The other, an account of Schmidt, taking photographs to find ing) into a synchronized scanner, which appeared in The New Yorker the trails of nearby, rapidly moving recording a night-long view of sky magazine in 1987, is the search with asteroids. Gravitational perturbations onto 12 large reels of tape output. the Palomar 18-inch Schmidt telescope by the earth and other planets may The kludge is connected, fastened on for asteroids whose eccentric orbits cause eccentric, earth-orbit crossing with tape; the computer program fails cross the path of the earth. Preston asteroids to strike us, and Gene and then triumphantly works. Juan explains reasons for these searches: Shoemaker has become an expert on Carrasco points the 200-inch and says What fuels the enormous output of a the geological remains (impact craters) "we are there" -a memorable phrase. quasar? (A black hole swallowing mil of ones that did. Preston includes a The astronomers munch Oreo cookies, lions of suns?) How dangerous are capsule history of the solar system, a marvel at glimpses they see as the sky asteroid impacts? (Very?) description of the Schmidt telescope, turns, listen to classical music and dis In addition to the telescopes there and his terrors on loading a piece of cuss why there might be so few quasars are also vivid human heroes. For this film into a holder in the darkroom. at large redshift. Are they seeing back remarkably intimate description of real The reader will also meet some of too far in time to before galaxies scientists as human beings, working at astronomy's pioneers in anecdotal existed, or before they contained all their limits, Preston spent many nights biographies of Bernhard Schmidt, devouring black holes? Was there an at Palomar, and conducted many Walter Baade and Fritz Zwicky, opaque screen at large redshift? Is it interviews. A particular novelty is the The search for remote quasars at all only a statistical fluctuation? description of how it feels to conceive, the 200-inch similarly introduces a Twenty-five years ago we did not design, build, de-bug, and use a one number of great men; a short biogra know that quasars existed; now we of-a-kind instrument. He describes phy of George Ellery Hale depicts the search for them near the unattainable personal and emotional characteristics charm and psychological stress of that edge of the Universe, from where the that make the scientists real, if some founder of great institutions. The cen "first light" emitted billions of years what larger than life. They are a tral hero, the 200-inch Hale reflector, ago just now reaches Juan Carrasco's varied group, many currently or for is particularly appealing to me because 200-inch. Dawn arrives, stopping the merly connected with Caltech: Carolyn of my 40-year love affair with that quest. This is an excellent and true and Eugene Shoemaker, Maarten steel and Pyrex monster of 1930s (even if somewhat romanticized) book Schmidt, Jim Gunn, Don Schneider. design. What is amazing is that it still about what it feels like to be an Another important and charming char remains adaptable to the newest tech explorer. 0 acter in the story is Juan Carrasco, nologies. Preston describes the born on a small Texas farm, now gadgeteers behind this application of Jesse L. Greenstein senior night assistant at Palomar, the electronics to the 200-inch-Jim West The Lee A. DuBridge Professor of only one in this group trusted to move phal and his "wizards of the Astrophysics. Emeritus 41 Random Walk Four Professorships Experimental Economics Laboratory Awarded Established ISTORY PROFESSORS John Benton and LABORATORY OF Experimental ence. Plott helped found the field of H Eleanor Searle and economics A Economics and Political Science is experimental economics, and, with his professor Charles Plott have been being established at Caltech. The co-workers, developed many of the lab appointed to endowed chairs in the laboratory will use people's behavior in techniques now in use. Other Caltech Division of Humanities and Social various simulations to study the researchers at the lab will include Sciences. The Division of Chemistry dynamics of decision-making and the David Grether, John Ledyard, Richard and Chemical Engineering, meanwhile, marketplace. The method has been McKelvey, Peter Ordeshoak, Tom Pal has named Peter Dervan to be its first applied to environmental and frey, Dave Porter, and Louis Wilde. Bren Professor of Chemistry. common-property resource problems, The Lynde and Harry Bradley Benton will become the Doris and finance (including futures markets and Foundation of Milwaukee, Wisconsin Henry Dreyfuss Professor of History. speculative bubbles), product quality has donated $500,000 to help establish Benton is an expert on medieval and warranties, regulation, deregula the lab. Other major contributors Europe's social and cultural history, tion, and anti-trust litigation. include General Motors, the Alfred P. with a special interest in France. Ben The lab will be directed by Charles Sloan Foundation, the Pacific Telesis ton joined the faculty in 1965. Plott, the Edward S. Harkness Profes Group, the National Science Founda Plott has been named the Edward sor of Economics and Political Sci- tion, NASA/JPL, and Caltech. S. Harkness Professor of Economics and Political Science. Plott is a pioneer in the emerging field of experi mental political economy, and will direct the just-established Laboratory of Experimental Economics and Politi Tug 0' War cal Science. Plott has been a faculty member since 1971. Searle is the new Edie and Lew Wasserman Professor of History. Searle, a specialist in the social and economic history of medieval England and northern Europe, is the first woman to hold a named professorship at Caltech. She joined the faculty in 1979. Dervan, professor of chemistry and a faculty member since 1973, has been named the first Bren Professor of Chemistry. Dervan's research centers on developing chemical methods to study the mechanisms by which vari ous drugs and proteins bind to DNA, laying the groundwork for eventual synthesis of molecules designed to bind at specific sites on the DNA. The Bren Foundation recently gave Caltech $1.5 million with which to endow the chair. The foundation, based in Los Angeles, is headed by This fall's ME72a final featured a tug-of war. Assistant Professor ofMechanical Engineer Donald Bren, a Caltech trustee since ing Erik Antonsson gave each student an identical set of components and a challenge: build 1984. Bren is chairman of the Irvine a device to crawl, slide or fly over a bed ofplastic "sand." While the challenge varies yearly, Company of Newport Beach, student enthusiasm is a constant. From left: Leslie McCaffree, Saroj Manandhar (TA), California. Brian Patterson, Kenneth Lin, and John Wiltse. 42 ENGINEERING & SCIENCE / WINTER 1988 about seven months in 1086, give his the United States, in this case for con Caltech on the Air torians an invaluable picture of tributions to immunology and molecu English life at the time. The book is lar biology. HE FIRST BROADCAST of" AirTalk: The considered to be England's single most Edward B. Lewis, the Thomas T Caltech Edition" took to the important historical document. Hunt Morgan Professor of Biology, has airwaves on January 20. The program A limited edition facsimilie (250 received a 1987 Gairdner Foundation is heard from 6 to 7 PM on KPCC, copies), written in Latin on parchment International Award for his research 89.3 FM, the National Public Radio and bound by hand between oak on genetic "master regulators"-genes affiliate of Pasadena City College. boards cut from medieval timbers, was that turn other genes on and off as an "AirTalk," a nightly talk show hosted issued two years ago in honor of the embryo develops, transforming a single by Larry Mantle, will be devoted to book's 900th anniversary. Richard undifferentiated cell into a complex Caltech on the third Wednesday of Hayman (,36), a frequent and gen organism of millions of specialized every month. The show includes erous donor to the Institute, formally cells. The award includes a $20,000 interviews with Caltechers and presented a copy to President Everhart cash prize. JPLers-faculty, staff, students, and in a ceremony on January 25. The Hans W. Liepmann, the von alumni; as well as features such as Cal book is on display in Millikan Karman Professor of Aeronautics, tech Almanac-a look at the people Library's Rare Book Room. Emeritus, has been appointed an and events that have shaped CaJtech; External Scientist Member of the Max Caltech Calendar-a listing of upcom Planck Institute for fluid Dynamics ing public events on campus; and Cal Research in G6ttingen, West Ger tech in the News. many. This honor has been granted The first show's guests included only once before in the Planck Vice President for Institute Relations Institute's history. Ted Hurwitz on Caltech's outreach Honors and Awards Rudy Marcus, the Arthur Amos activities and on finance in higher edu Noyes Professor of Chemistry, has cation, Institute Archivist Judith SSISTANT PROFESSOR of Biology David been honored by several organizations. Goodstein on Caltech's history, and AAnderson has been chosen by the He was elected a Foreign Associate of ASCIT president Jeff Tekanic on stu Chicago Community Trust to be a the Royal Society in June, was named dent life. President Everhart, JPL 1987 Searle Scholar. He will recieve a to the International Academy of director Lew Allen, and Pasadena City $180,000 grant over the next three Quantum Molecular Science in July, College president Jack Scott were also years to support his research. received the 1988 Peter Debye Award heard. Don L. Anderson, professor of geo in Physical Chemistry from the ACS in KPCC has received FCC approval physics and director of the Seismology September, and went to Sweden in to relocate its transmitting tower from Laboratory, has been awarded the October to receive an honorary Doctor the PCC campus to the top of Mount Geological Society of America's Arthur of Science degree from the University Wilson, and hopes to have the move L. Day medal for 1987. Anderson of Gothenburg. completed this summer. The station's studies the structure and evolution of Professor of Biology Paul H. signal will then reach from Ventura to the earth, moon, and terrestrial Patterson has received an Investigator southern Orange County, making it plancts. Initiated Research Grant from the one of the most powerful NPR sta William Goddard, the Charles and Alzheimer's Disease and Related tions in Southern California. Mary Ferkel Professor of Chemistry Disorders Association (ADRDA) to and Applied Physics, won the 1988 study how brain cells using the neuro American Chemical Society (ACS) transmitter acetylcholine develop. Award for Computers in Chemistry in Barry Simon, the IBM Professor of recognition of a lifetime of contribu Mathematics and Theoretical Physics, tions to the field. The award is spon has been elected vice president of the sored by Digital Equipment Corpora American Mathematical Society. Domesday Book Gift tion (DEC), and includes a $3,000 G. J. Wasserburg, the MacArthur prize. Professor of Geology and Geophysics ILLIAM THE CONQUEROR was an or Chemistry professor Robert and chairman of the Division of Geo W derly chap who liked to know Grubbs has been selected for the ACS logical and Planetary Sciences, gave what he owned and how much reve 1988 Award in Organometallic Chem the 100th Anniversary Goldschmidt nue it brought in. Once he had con istry. Dow Chemical Company spon Memorial Lecture to the Norwegian quered England, therefore, it was only sors the award, which carries a $3,000 Academy of Sciences in Oslo. His natural to take inventory of his new pnze. topic was "Isotope Connections domain. The result is the Domesday Leroy Hood, the Ethel Wilson between the Solar System and the Book- a field-by-field record of land Bowles and Robcrt Bowles Professor of Interstellar Medium." owners, their wealth, and their hold Biology and chairman of the division, Amcrican Cancer Society Postdoc ings: peasants and villagers, pigs and has been given the 1987-88 Dickson toral Fellowships have been awarded chickens, plows and grindstones, and Prize by the University of Pittsburgh. to chemistry research fellows Marc every building down to the last The award is prescnted annually to the Greenberg, Calvin Iida, and John beehive. The data, compiled over foremost investigator in medicine in Termini. 43 Random Walk (continued) Obituaries HARLES D. BABCOCK, professor of engineering by 1947, retiring in 1962. Princeton University. He came to C. aeronautics, died on July 1. Bab Converse's pioneering research on the Caltech in 1925 as an instructor in cock earned a BS from Purdue Univer vibratory compaction of sands and English. He had become an associate sity before coming west to Caltech, cohesive soils made him an invaluable professor of English by 1941, when he where he received his MS (1958) and advisor to builders, architects, and was named registrar and director of PhD (1962). He stayed on as a engineers throughout southern admissions. He became dean of research fellow in aeronautics, rising to California. admissions and director of undergradu become a full professor by 1974. Edward W. Hughes, 83, senior ate scholarships in 1953, and retired in Babcock's research into how structures research associate in chemistry, died 1968. Jones was 87 years old. fail has borne fruit in many branches on December 24. Hughes earned George P. Mayhew, professor of of engineering. Babcock was 53. BChem and PhD degrees from Cornell English, emeritus, died on October IS Frederick J. Converse, professor of University before joining the CaItech in Massachusetts, following a long ill soil mechanics, emeritus, died October faculty in 1938. He devoted his ness. He was 68. An expert on 9, six days before his 96th birthday. research to X-ray crystallography, and Jonathan Swift, Mayhew was educated Converse received his BS from the served as president of the American at Harvard, where he earned his AB, University of Rochester, and taught Crystallographic Association. MA, and PhD. He joined Caltech as there before coming to Caltech in 1921 Louis Winchester ("Winch") Jones, an assistant professor in 1954, and as an instructor of civil engineering. dean of admissions, emeritus, died on became a full professor in 1968. He He had become a professor of civil January 6. Jones was educated at had been on medical leave since 1974. IFTT I I E 5 Did you know you can make a gift to Caltech that The Institute has recently published a booklet actually pays you something in return? In fact, you providing information on these and other ways to may end up with more spendable income after your benefit yourself and (altech. It also has experts to gift than before. work with you and your advisors to come up with the financial plan that is right for you. If you would like Making a gift to Caltech as a life income· trust or gift a copy of A Tradition of Trust, call or write: annuity may provide numerous benefits: III Acharitable income tax deduction Mr. J. Thomas Gelder to reduce taxable income; Director of Gift and Estate Planning II Acomplete bypass of income Califomia Institute of Technology tax on appreciated assets; Pasadena, California 91125 II Saving estate taxes and avoiding (818) 356-6349 expense and delay of probate; III Possible income increase through sale and reinvestment of low income producing assets; and II Free asset management. 44 ENGINEERING & SCIENCE I WINTER 1988 Pioneer the future with us. All of the technological advancements that have part-time while studying at a nearby university; Full been pioneered by Hughes Aircraft Company are Study Fellows attend classes full-time. merely an introduction to what will come. And Hughes has just announced a special MSEE As a Hughes Fellow, you could be pioneering your program in Microwave Engineering in conjunction with own future, studying for your Master's or Doctorate in two major Southern California universities. Microwave Engineering (Electrical, Mechanical, Manufacturing), Fellows will be allowed time off at full pay to attend Computer Science or Physics. classes taught by university faculty right at Hughes' You'd be receiving full tuition, books and fees, an facilities. educational stipend, full employee benefits, relocation Since Hughes is involved with more than 90 expenses, professional-level salary, summer employ technologies, a wide range of technical assignments ment, technical experience ... with a combined value is available. An Engineering Rotation Program is also of $25,000 to $50,000 a year. available for those interested in diversifying their work While you're completing your degree, you'll also experience. have the opportunity to gain valuable experience at Since 1949, more than 5,500 men and women have Hughes facilities in Southern California, Arizona or earned advanced degrees in engineering and science Colorado. with the help of Hughes fellowships. We hope you'll Hughes Fellows work full-time during the summer. join us in creating the next generation of technological During the academic year, Work-Study Fellows work wonders, by pioneering the future-yours and ours. 1------, I Hughes Aircraft Company, Corporate Fellowship Office ) Dept. MC-8788, Bldg. C1/B168, P.O. Box 45066, Los Angeles, CA 90045-0066 I Please consider me a candidate for a I Hughes Fellowship and send rile the necessary information and application materials. 1 I I PLEASE PRINT: Name Date Home Phone & Hours 1 1 I Address City State Zip I I I am interested in obtaining a Master's __ Doctorate __ in the field of: ______~ _____ I I Rotation Program Interest: Yes __ No __ Microwave Program Interest: Yes __ No -- I I DEGREES NOW HELD OR EXPECTED: I 1 1 Bachelor's: Field Date School GPA I I I Master's: Field Date School GPA . HUGHES I I Minimum GPA-3.0/4.0. Proof of U.S. Citizenship May Be Required. Equal Opportunity Employer. J"!RCRAFT COMPANY I L ___ ~ ______~ NON·PROFIT ORG. ENGINE~~~~ U.S. POSTAGE PAID California Institute of Technology PASADENA , CA Pasadena, California 9112 5 PERMIT NO. 583 ADDRESS CORR!:CTION REQUESTED The Sunraycer saga is ilfllsrrated by fhe empty bOllles 0/ victory champagne at !he end vlrhe World Solar Chal lenge and by Slinraycer's logo showing Palll MacCready's company, AeroVironment, positioned in the midst a/rile giant corporate entilies o/GM. Wearing a garland of yellow roses, Sunraycer feads rhe Pasadena Rose Parade on January I, 1988 as {he "Pace Car a/the FWllre."