Real Time Relativity: exploration learning of special relativity C. M. Savage, A. Searle, and L. McCalman Department of Physics, Australian National University, ACT 0200, Australia∗ “RealTimeRelativity”isacomputerprogramthatletsstudentsflyatrelativisticspeedsthougha simulatedworldpopulatedwithplanets,clocks,andbuildings. Thecounterintuitiveandspectacular opticaleffectsof relativity areprominent,while systematicexploration of thesimulation allows the 7 usertodiscoverrelativistic effectssuch aslength contraction andtherelativityof simultaneity. We 0 report on the physics and technology underpinning the simulation, and our experience using it for 0 teaching special relativity tofirst yearuniversity students. 2 PACSnumbers: n a J I. INTRODUCTION space and time. To overcome their non-relativistic pre- 8 conceptionsstudentsmustfirstrecognizethem,andthen 1 “RealTime Relativity”1 is a firstpersonpointof view confrontthem.3 TheRealTimeRelativitysimulationcan ] game-like computer simulation of a special relativistic aid this process. h world,whichallowstheusertomoveinthreedimensions In the next section we discuss some relevant physics p - amongst familiar objects, Fig. 1. In a first year univer- education research. Section III outlines the relativistic d sityphysicscourseithasprovedcomplementarytoother opticsrequiredtounderstandthesimulation. SectionIV e relativity instruction. briefly overviews the computer technology that is mak- . s Sincethereislittleopportunityforstudentstodirectly ing interactive simulations of realistic physics, such as c experience relativity, it is often perceived as abstract, Real Time Relativity, increasingly practical. Section V i s and they may find it hard to form an integrated rela- describes students’ experience of the simulation. Section y tivistic world-view. They find relativity interesting and VI shows how it provides fresh perspectives on physics h p exciting,butmaybe leftbemusedbythe chasmbetween suchastherelativityofsimultaneity. SectionVIIreports [ the theoryandtheir everydayexperience.2,3,4 RealTime ourevaluationofitsuseinafirstyearphysicslaboratory. Relativity can help bridge this chasm by making visual 1 observationsthe basis fromwhichthe theory is deduced. v 0 In his original relativity paper Einstein discarded the II. EDUCATIONAL BACKGROUND 0 personal observer, who collects information from what 2 he sees, in favor of more abstract inertial observers who There is substantial evidence for the value of ac- 1 use distributed arrays of rulers and conventionally syn- 0 chronizedclocks.5HoweverKomar6andothers7,8showed tive learning.9,10 Effective learning is stimulated by stu- 7 dentsparticipatingintheconstructionandapplicationof thatspecialrelativitymaybeformulatedintermsofpos- 0 physics based world-views.11 A common factor in active tulates about a personal observer’s visual observations. / learning is a cycle of developing, testing, and correcting s ThisapproachtorelativityunderpinsouruseoftheReal c understanding in a collaborative environment. Peer in- Time Relativity simulation. i struction is one way for this to occur in the classroom,12 s Studieshaveshownthatstudentsmayfailtolearnfun- y while in the laboratory inquiry based approaches are damentalconcepts,suchastherelativityofsimultaneity, h known to be effective.11 p evenafter advancedinstruction.2,3,4 This is because spe- Computer simulations may promote active learning in : cial relativity contradicts some deeply held ideas about v physics, especially where real laboratories are difficult i to provide. However, research has shown that a testing X and development cycle is required to ensure good learn- r a ing outcomes.13,14 The effectiveness of simulations is re- duced by poor interfaces,13 and by students’ lack of the skillsrequiredtolearnfromthem.15 Theyalsoloseeffec- tiveness if the exploration is not conducted according to the scientific method.16 However, when such issues are addressed the results can be spectacular.14 Computer simulations are most effective when di- rectedtowardscleargoals,withanunderstandingoftheir strengths and limitations.14,16 They can provide an ad- FIG. 1: Screenshot from Real Time Relativity. The speed ditional active learning mode, and address broad goals of the camera relative to the objects is v = 0.9682c. The such as “thinking like a physicist”.17 However, learning Doppler and headlight effects havebeen turned off. to use software increases cognitive load, lessening capac- 2 ity to learn other new material.15 The value for physics unit vector in the propagation direction. From the 4- teaching of first person simulations, such as Real Time frequency components in a particular frame, the compo- Relativity, is largelyunexplored,as existing researchhas nents in any other frame may be found using a Lorentz concerned simulations of models, such as might be used transformation. The transformation between the usual byanexpertphysicist,ratherthanimmersivefirstperson standard configuration24 frames S and S′ are sufficient simulations.13,14,15,16 forour purposes. We willuse “world”(w) and“camera” ′ Real Time Relativity differs from other physics simu- (c) to refer to the frames S and S , lations in providing a realistic, explorable environment. f =γf (1−n v/c), In the context of a first year university physics class, we c w w,x areaskingthe question: Canaspects ofspecialrelativity fcnc,x =γfw(nw,x−v/c), be learntby explorationof the RealTime Relativity vir- f n =f n , f n =fn , c c,y w w,y c c,z w,z tual world? Many students are comfortable interactively γ =(1−v2/c2)−1/2, (2) discovering the rules of virtual worlds; perhaps they can use this experience for discovering the rules of physics? where v is the component of the relative velocity of the Successful learning from simulations is more likely frames along the positive xw axis of the world frame. if students are suitably prepared and guided.11,13,15 The first equation expresses the Doppler effect, and the Preparation should develop a basic understanding of remainingequationsexpressthe dependenceoftheprop- the physics which determines what is seen in the sim- agationdirectionontherelativeframevelocity: aneffect ulation. In our case this includes the finite speed of knownas“relativisticaberration”. IntheRealTimeRel- light, the Doppler effect, and relativistic optical aber- ativity simulation let the world frame be that in which ration. This preparationmight use conventionalinterac- the objects are at rest, and the camera frame be the tive multimedia.18 Preparation should also include how user’s instantaneous rest frame, as we represent the user thescientificmethodisusedtodevelopunderstandingof byacamera. Werequirethefrequenciesandpropagation novel phenomena. directions in the camera frame, fc and~nc. Scherr, Shaffer, and Vokos have found that students’ Since ~nc is a unit vector, its x-component, nc,x is the understanding of time in special relativity is poor.2,3 cosine of the angle θc between the light ray and the They conclude that “... many students who study spe- xc axis: if the ray is coming towards the observer nc,x cialrelativityattheundergraduatetograduatelevelsfail changessign. DividingthesecondandthirdofEqs.(2)by to develop a functional understanding” .3 They identify the first and using nc,x = −cosθc and nw,x = −cosθw, the reason for this as students misunderstanding funda- we get mental ideas such as: the “time of an event, simultane- cosθ = cosθw+v/c , ity, and reference frame”.3 They have developedinstruc- c 1+(v/c)cosθw tional materials to address these problems. Mermin19 sinθ = sinθw . (3) hasalsonotedthattraditionalrelativisticpedagogymay c γ(1+(v/c)cosθw) makeincorrectassumptionsaboutstudents’priorknowl- Relativistic aberration is analogous to non-relativistic edge. Real Time Relativity can address these problems, forms ofaberrationthat students may haveexperienced. as fundamental ideas, such as the time of an event, have For example, the dependence of the angle of falling rain intuitive operational meanings. on an observer’s velocity, or the difference between the visualpositionofahighflyingaircraftandthatindicated by its sound. This understanding may be made quanti- III. RELATIVISTIC OPTICS tative using the relativistic velocity addition formulae.23 Penrose20 showed that relativistic aberration implies thatstraight-linesareseenaseither linesorcirculararcs Some ofthe basicphysicsofrelativisticoptics,namely in other frames. He also showed that a sphere, which the Doppler effect andaberration,wasdiscussed by Ein- stein in his first relativity paper.5 However it was not always has a circular outline (unlike a circle which may until about 1960 that the pioneering work of Penrose,20 haveanellipticaloutline),willcontinuetohaveacircular Terrell,21 and Weisskopf22 showed that relativity gives a outline after aberration, and hence continue to look like a sphere. These effects are immediately apparent in the rich and unexpected visual environment. Real Time Relativity simulation, Fig. 2. Inthissectionwesummarizerelativisticopticsusing4- Thenon-relativisticDopplereffectmayalsobefamiliar vectors,becausethatishowitisimplementedintheReal TimeRelativityprogram(seeSectionIV). Rindler23pro- to some students. This, together with the analogy to non-relativistic aberration, emphasize the closer relation videsamorecompleteintroduction,bothwithandwith- of relativistic optics to direct experience than the usual out using 4-vectors. space-time approachto special relativity. A plane light wave is described by its 4-frequency F, which has components23 A convenient form of the Doppler effect follows from the first two of Eqs. (2) after eliminating n , w,x F =f[1,~n], (1) 1−v2/c2 f = p f =Df , (4) where f is the frequency and ~n = (nx,ny,nz) is the c 1−(v/c)cosθc w w 3 the photons, and the change in photon flux due to the combined effects of time dilation and the observer’s mo- tion,whichisanadditionalmanifestationoftheDoppler effect. IntermsoftheDopplerfactorinEq.(4)thesecon- tribute factors to the change in intensity of D2, D, and D respectively, for a combined intensity change factor of D4. However for common detectors, such as the eye or a CCD camera, it is the photon number flux P that is detected, and this changes by a factor of D3, since the energy change per photon is irrelevant, P =D3P . (6) B A IV. TECHNOLOGY Computers can generate images incorporating special,28 and general,29 relativistic optics. By the early 1990s it was possible to interactively render simple objects, such as cubes.30 The highest quality images were generated by the ray-tracing method, which is capable of producing photo-realistic images.31 However ray-tracing is currently too slow for interactive simula- FIG. 2: Screenshots from Real Time Relativity. Top frame: tions,althoughindividualimagescanbe strungtogether at rest in the world frame. Bottom frame: v = 0.9682c, to make movies.18 corresponding to γ = 4. The Doppler and headlight effects have been turned off. In the world frame the camera is in The development of the programmable graphics front of the position in thetop frame. processor32 has made it possible to render complex rel- ativistic scenes in real time. The first such systems ap- pear to have been developed by D. Weiskopf33 and M. where this equation defines the “Doppler factor” D. For Bochers34withinthephysicseducationgroupattheUni- v/c ≪ 1 the denominator is the familiar non-relativistic versity of Tu¨bingen.35 This group has focussed on us- wave compression or expansion. For waves incoming at ing relativity visualization for science communication.36 θ =π/2radianstotherelativemotion,thedenominator Our Real Time Relativity simulation is similar to these, c is one and the observed frequency is less than the world is freely available, and is being developed as an Open frequency, at which they were emitted. This means that SourceprojectundertheLesserGeneralPublicLicense.1 thetimebetweenwavecrests,theperiod,islonger;which The screen image displayed by Real Time Relativ- isexactlytheeffectoftimedilation,ifthewavecrestsare ity is created using the computer graphics technique regarded as a clock. knownasenvironmentmapping,whichrendersthethree- The effect of aberrationon small angles may be found dimensional virtual world onto a two-dimensional cube by taking differentialsof the inverseLorentztransforma- map. A cube map may be visualised as the 4π ster- tion of the 4-frequency.25 This yields radian view-field mapped onto the interior surface of a cube centered on a camera, representing the user’s field dθ =D−1dθ . (5) ofview. Infact,thecubemapisadatastructureinwhich C W the image pixels are addressed by line of sight direction, HencesmallanglestransformbytheinverseDopplerfac- rather than by spatial position. The relativistically cor- tor. In particular, for objects directly ahead, so that rect scene is produced by transforming the cube map. θ = 0, and for v/c ≪ 1, the inverse Doppler factor is Each camera image pixel is formed by light incident C approximately D ≈ 1−v/c, and objects angular sizes from a particular direction; that is, with a specific prop- are shrunk. While for objects behind, D ≈ 1+v/c and agationdirectionn~ inthecameraframe. Therelativistic c objects are expanded. physicsproblemistofindthecorrespondingpropagation Perhapsthemostsubtleoftherelativisticopticseffects vector n~ in the world frame in which the cube map is w istheheadlighteffect. Indeed,acompletediscussionwas constructed. This vector then addresses the pixel on the not given until 1979, by McKinley.26,27 It refers to the cube map that is mapped to the camera pixel. The re- increased intensity of light coming from objects we are sulting camera image is displayed on the screen. moving towards. The intensity decreases for objects we A plane light wave is represented by the relativis- are moving away from. Three things combine to pro- tic 4-frequency, Eq. (1). The propagation direction duce these intensity changes: the change in angular size in the world frame is found by the inverse Lorentz of the emitting region, the Doppler change in energy of transformation25 of this 4-vector from the camera frame 4 into the world frame. This is implemented as a four- tails of any particular graphics processing unit. Conse- dimensional matrix multiplication of the 4-frequency. quently, it is only available on Windows computer sys- The transformation matrix is calculated before each tems. DirectX 9 includes the High Level Shader Lan- frameisrendered,usingthecurrentcameravelocity,and guage in which the pixel shader controlling the graphics isthenappliedtoa4-frequencyconstructedforeachcam- processing unit is written. era pixel. This has a spatial component equal to the Graphics processing units have been increasing in pixel’s imaging direction and the time component set to processing power more rapidly than central processing one. The spatialpartofthe transformed4-frequencyad- units.32 This is driven by the demand for parallel com- dresses the cube map pixel that is then rendered to the puting from the gaming community. For example, the screen. Xbox 360 graphics processing unit has forty-eight 32- The Graphics Processing Units (GPUs) on computer bit processors running at 500 MHz, each capable of a video cards are designed to do 4-vector matrix algebra floating point 4-vectoroperationper cycle, giving nearly efficientlyandinparallel. Thismakesitpossibleforsim- 100GigaFlops,comparedtoperhapsafewGigaFlopsfor ulations such as Real Time Relativity, and that due to a central processing unit.39 The main limitation is that Bochers,34,36 to perform the Lorentz transformations in graphics processing units do data-parallel computing, in real time. The 4-vectors that graphics processing units whichthesameoperationisrepeatedoneachelementofa normallyworkwithspecifythex,yandzcoordinatesofa dataarray. Nevertheless,computationalscientistsarede- vertex and a fourth w component that facilitates certain veloping algorithms that harness their processing power non-linear transformations (such as translation and per- for tasks such as solving partial differential equations.32 spectiveprojection),orspecifythered,green,bluecolour The Folding@Home distributed computing project has and alpha (transparency) of a (texture) pixel. Since the a client available which runs their molecular dynamics processing of different vertices or pixels is usually inde- calculationongraphicsprocessingunits, increasingcom- pendent, the operations can be performed in parallel. putationalpowerbyabouttwentytimespercomputer.40 The Doppler shift factor D is givenby the ratio of the These developments may have an impact on the kinds time components of the 4-frequencies in the camera and of physics teaching simulations that are possible in the world frames, Eq. (4). However, to determine the effect future. of the Doppler shift on a general colour requires the en- tire intensity spectrum. But in current graphics systems the spectrum is specified at just three frequencies; red, V. THE REAL TIME RELATIVITY green, and blue. Hence interpolation is used to gener- SIMULATION ate the spectrum. This simple approach, together with the lack of any infrared or ultraviolet spectra, prevents In this section we introduce the Real Time Relativ- a true representation of Doppler shifted colors, and is a ity simulation as experienced by students in the first significantlimitationof the currentversionofRealTime year course for physics majors at The Australian Na- Relativity. In particular, stars do not maintain a black- tionalUniversity.41Itwasusedinathreehourlaboratory body spectrum.26,37,38 structuredtoencourageexploration,whilerequiringthat The headlight effect, Eq. (6), is implemented by mul- certain measurements be made and compared to theory. tiplying eachpixelcolorvectorby the third powerofthe Studentswereprovidedwithamanualgivingbackground Doppler shift factor D. There are significant limitations informationandaskingbothqualitativeandquantitative on how the resulting large intensity range is rendered to questions. Many students completed the laboratory be- the screen by current hardware. fore they attended the relativistic optics lecture. Stu- The graphics processing unit does the Lorentz trans- dents worked in groups of two or three, and discussion formations as well as its usual graphics work. First, a was encouraged. Preparation included answering simple non-relativistic three-dimensional scene is rendered to a pre-lab questions which were assessed at the beginning cube map, then relativistic transformations are applied of the laboratory. to it. To generate a frame, the 4-frequency associated Aninitialproblemoforientationwithinrelativisticop- with eachcamerapixelis inverseLorentztransformedto tics simulations arises because the speed of light is very findthecorrespondingworldframecubemappixel. This large in everyday terms. This means that either the ob- is then Doppler and intensity shifted, also by the graph- jects in the simulationmust be very large,roughly light- ics processing unit. An 800 by 600 window has 480,000 seconds, or the speed of light must be artificially slow, pixels, so displaying 50 frames per second requires 24 as in Gamow’s Mr. Tompkins story.42 In the interest of million pixel transformations per second, which is well realism, we have taken the former view, which allows within the capabilities of inexpensive graphics process- us to include realistic astronomical objects such as the ing units. Consequently, it is the conventional graphics Earth,which is 0.042light-seconds in diameter. The top processinggeneratingthe cube mapthatlimits the over- frame of Fig. 2 shows a screen from Real Time Relativ- all performance, not the relativistic calculations. ity. TheEarthisvisible,asistheSunbehindit.43 These RealTime Relativity isprogrammedusing Microsoft’s objects set the scale of the simulated world. Other ob- DirectX 9 interface, so that it is independent of the de- jects, such as the columns, have been chosen for their 5 they see a rainbow effect, as for directions towards that of motion, θ < θ , the color is blue shifted, while for 0 directionsawayfromthe directionofmotion,θ >θ , the 0 color is red shifted. Fig. 3 shows the red shifting of a bluecylinderthroughgreenandredastheviewingangle increases. If students already understand the non-relativistic Doppler effect, they may be guided to discover the rel- ativistic version. In particular, it is possible to deduce timedilationfromtheobservationthatthereisreddening when viewing perpendicular to the direction of motion, Eq. (4). FIG. 3: Screenshot from Real Time Relativity showing how the Doppler effect depends on the view angle. The camera The bottom frame of Fig. 2 shows the scene with the is looking perpendicular to the direction of motion, which is camera travelling down the row of columns with a speed fromrighttoleftatv=0.5c,atastippledbluecylinder. The corresponding to γ = 4. The Doppler and headlight ef- Doppler factor is approximately one, D ≈1, in the direction fects have been turned off. The circular curvature of the of the left edge of the image. The rest of the cylinder is red nearest columns is due to relativistic aberration, as dis- shifted through green and red. The headlight effect has been cussed in section III. The curvature of the more distant turned off. columns is barely noticeable. However, they are shrunk by approximately the inverse Doppler factor D−1, ac- familiar shapes, although they would be absurdly large cording to Eq. (5). The camera field of view covers a if they existed in the real world. Familiar objects aid in wide field in the world frame: the hoops and cubes on the recognition of the distortions caused by relativistic theedgesoftheimagearebehindthecameraintheworld aberration. frame. Students start by accelerating from rest down the row Aberrationmaybeunderstoodasaconsequenceofthe of columns shown in Fig. 2. At first it seems that they finite speed of light. The key idea is that the light that aremovingbackwards.44 Thisis completelycounterintu- reaches the camera at a particular instant was reflected itive and prompts them to question what they see: the by objects at different times. The light from closer ob- exploration has begun! The effect is due to relativistic jects wasreflectedlater thanthat fromfar awayobjects. aberration. An important way that motion is sensed is This is irrelevant when the camera is at rest relative to bythechangeinangularsizeasourdistancetotheobject theobjects,butwhenitismoving,thepositionoftheob- changes. Normally,as we approachanobjectits angular jects in the camera frame depends on time. For a large size increases, roughly proportionate with the distance. object this means that the parts nearer the camera re- In contrast, the decrease in angular size due to relativis- flected the received light later than the further parts. If tic aberration, Eq. (5), is approximately proportional to the camerais movingtowardsthe object,atasignificant 1−(v/c)cosθ , for v/c ≪ 1, and occurs before the dis- fractionofthe speed of light, the near parts reflectwhen c tance hastime to change. Thereforethe initialview isof they are significantly closer and hence look bigger than objectsshrinking,andthisisinterpretedasthemmoving thefarpartswhichreflectedwhentheywerefurtheraway away, and hence as backwards movement of the viewer. and hence looked smaller. If we are moving directly to- Astheycontinuetoaccelerate,nearbyobjectseventually wards the middle of an object the net result is that the passby,andtheperceptionofforwardmotionisrestored. middlelooksfatterthantheends,seeFig.4. Iftheobject is off to one side it is curved into a circular arc. Colors change due the Doppler effect Eq. (4), but the Currently, Real Time Relativity is limited to all ob- headlight effect quickly saturates the scene with bright jects being at rest in the world frame. This means that light, dominating all other effects, due to its dependence relativistic dynamics is not within its capabilities. on the third power of the Doppler factor, Eq. (6). Con- sequently, it is useful to be able to turn it off. Although this goes against the principle of making the simulation VI. THE RELATIVITY OF SIMULTANEITY as realistic as possible, it is difficult to see some other effects if it is left on. The relativity of simultaneity has been identified as a The Doppler effect depends on the viewing angle, particularly difficult concept for students to learn from Eq. (4). There is a particular angle to the direction of passive instruction.2,3,4 In order that students might ac- motion θ for which there is no effect, since the Doppler 0 tively discover the relativity of simultaneity for them- factor D =1 when selves, the Real Time Relativity simulation includes cosθ =(c/v)(1−γ−1). (7) clocks in the world frame. Even when the camera is 0 at rest in the world frame, clocks at different distances ◦ For v = 0.5c this angle is θ = 1.3 radians (74 ). If a from the camera are seen to read different times due to 0 student looks at at a pure colored object at this angle the light propagation delay, see the top frame of Fig. 5. 6 FIG. 4: Screenshot from Real Time Relativity showing rela- tivistic aberration.The camera is moving towards the center of the column with v =0.9682c, so that γ =4. The Doppler and headlight effects are off. Studentsgenerallyhavenodifficultyrecognizinganduti- lizing this fact.2,3,4 Note that clocks the same distance FIG. 5: Screenshots from Real Time Relativity explaining from the camera read the same time. the relativity of simultaneity. Top frame. The effect of light propagation delay on observed clocks. The camera is at rest The middle frame shows the same view of the clocks, relative to the clocks, which are lined up perpendicular to but with the camera moving with v = 0.5c parallel to the line of sight to the central clock. The clocks are 5 light- the clocks from left to right. The camera is looking per- secondsapartandreadseconds. Middleframe. Thecamerais pendicular to its direction of motion. Note that the eye movingfrom left torightparallel totheclockswithv=0.5c. gets confusingcues fromthis image,asthe clocksarero- Theperpendiculardistancetotheclocksisthesameasinthe tatedasifwe werelookingatthemfrominfront,butwe topframe(about31light-seconds). Themajorcontributorto are not. This effect is a result of relativistic aberration the different clock readings is the relativity of simultaneity. knownas“Terrellrotation”.21 Lengthcontractionbythe However, light delay causes clocks to the left to differ more factor γ−1 =0.87 is also apparent. fromthecentralclockthanthosetotheright. Bottomframe. Thecamerahasbeenbroughttorestimmediatelyaftertaking The relativity of simultaneity is apparent from the themiddleframe,althoughsometimethenelapsedbeforethe readingsontheclocksinthemiddle frameofFig.5. The image wastaken. Intheclocks’ rest framethedifferent clock right-most clock is ahead of the left-most by 10 seconds. readings are entirely a consequence of the light propagation This cannot be explained by light delay in the camera delay. The field of view is thesame in each frame. frame, as the observed time difference is too large, and the times increase from left to right. However it is ex- plained by light delay in the clocks’frame. Students can mation; in particular, the times read by the clocks when see this by immediately stopping the camera relative to they were emitted. The different times of the different the clocks. Due to relativistic aberrationthey mustthen clocksis understoodby Bobasaresultofthe lightprop- look back to see the clocks: this view is shown in the agation delay over the different distances to the clocks. bottom frame of Fig. 5. In the clocks’ rest frame the However,the clockswereatapproximatelythe same dis- camera is not opposite the clocks, but is to their right. tance from Alice when they emitted the light, so she re- From this perspective it is clear why the clocks read as quiresanotherexplanation. Thisisanewphysicaleffect: they do: the left-most clock is furthest and reads earli- the relativity of simultaneity. The relativity postulate est, while the right-most is closest and reads latest. The ensures that what is true for these clocks is true for any time difference between them is exactly that seen by the clocks, and hence for time itself. A complete discussion moving camera. is given in Appendix A. Let us restate the argumentin terms of two photogra- phers: Alice is moving relative to the clocks, and Bob is stationaryrelativetotheclocks. BothAliceandBobtake VII. LABORATORY EVALUATION photographs of the clocks at an event “CLICK”, chosen sothatAlice,inherownframe,isapproximatelyequidis- tant from the locations of the clocks when they emitted Real Time Relativity was incorporated in a first the photographed light.45 Both Alice and Bob are sam- year laboratory session at The Australian National pling the set of photons originating from the clocks and University.41 Thecourseincludedninelecturesandthree present at CLICK. These photons carry the same infor- tutorials on special relativity. One lecture was devoted 7 to relativistic optics. Theflawsinthesimulationhadabiggernegativeimpact The content of the laboratory has been indicated in on the students than expected. Students sometimes at- section V. Its effectiveness was assessed in three ways. tributedtheir lackofunderstandingofthe physicsofthe First,studentscompletedquestionnairesbeforeandafter simulationto a “bug”,evenwhen there wasnone,rather the laboratory. Second, one of the authors was present than to their need to develop better understanding.13 asanobserverineachlaboratory,recordinghowstudents OurexperiencesuggeststhattheRealTimeRelativity interactedwiththesimulation. Alaboratorydemonstra- simulation canstimulate discoverylearning,and provide torwasalsopresent. Third,studentsrecordedtheirwork complementary learning opportunities to those provided in laboratory log books which were assessed. by lectures and problem solving tutorials. However, re- The pre-laboratory surveys indicated that students alizing its full value will require further cycles of testing usually had prior knowledge of relativity and were eager and development. Next time we use it, we shall require to learn more. However they tended to perceive it as an students to “play” with the simulation as part of the abstract subject. The post-laboratory surveys indicated pre-laboratory preparation, so that they have some fa- that students felt they had learnt about relativity from miliarity with the controls and with the peculiarities of the simulation, andthat it hadstimulated their interest. navigation in a relativistic world. We shall also provide Some students reported that the “concrete” or “visual” moreopportunityforopenendedexploration,asthisap- nature of the simulation was helpful:46 pears to be its strength. “Real Time Relativity is very useful - many people are visual learners.” VIII. CONCLUSION However, students often reported that the laboratory manual was too prescriptive and did not allow them to Real Time Relativity is an immersive physics simu- adequately pursue their own investigations. This criti- lation of a kind that is becoming increasingly accessi- cism focussed on the quantitative exercises: ble due to the improving cost effectiveness of computer technology. It givesstudents the opportunity to discover “Why arewe forcingequations fromthe sim- and confront their misconceptions about relativity, and ulation?” to construct resolutions. Our experience with Real Time Relativity suggests There were also many complaints about the difficulty of that it provides new perspectives on special relativity. using the program,and the inadequate time available to Thismaybeparticularlyvaluabletostudentswhoprefer develop proficiency with navigation through the virtual the concrete over the abstract. Important physics, such world: as the relativity of simultaneity, can be introduced with minimal mathematics. This may broaden the group of “Thecontrolswerereally,reallyhardtouse.” students who can learn relativity. However, the educa- The laboratory observer enabled a testing and refine- tional value of first person simulations, like Real Time mentcycle. We identified problems,andcorrectedthem, Relativity, is an interesting area for further physics edu- before the next student group took the laboratory. In cation research. particular, students often tried to push simulations to the limits to see what happened, behavior noted by the APPENDIX A: THE RELATIVITY OF University of Colorado Physics Education Technology group.13,14 Ifasimulationdoesnotrespondsensibly,stu- SIMULTANEITY dentscanloseconfidenceinitsreliability. Observerswere able to monitor what engaged students, and what frus- In this appendix we expand on the explanation of the trated them. The most engaging aspect was the explo- relativity of simultaneity in terms of light delays that ration of a novel and open ended world. Amongst the was introduced in section VI. It uses the aberrationand morefrustratingthingswerethesimulation’scontrolsnot length contraction formulae. In the context of discovery behaving in ways students considered natural. learning with the Real Time Relativity simulation each The log books completed during the laboratory did of these formulae may, in principle, be deduced from ob- not capture the excitement that was observed in work- servations. Along the way we also deduce time dilation. ing laboratory groups. However, successful quantitative WewillrefertoFig.6,whichshowsschematicdiagrams measurements were generally made: for example, of the of the scenario shown in Figs. 5. At event CLICK both Doppler effect and of length contraction as a function of AliceandBobtakephotographsoftheclocks. Wechoose speed. CLICKto be the co-incidentoriginsofAlice’s andBob’s Our experience confirmed the importance of develop- rest frames, which we assume to be in standard configu- ing educational software through a testing and refine- ration with relative velocity v. Therefore CLICK occurs mentcycle.13,14 Students usedthe simulationinwayswe at times t = t = 0. In the notation of section III Al- A B had not anticipated, and had different ideas to the au- ice’s frame is the camera frame, and Bob’s frame is the thors about what constituted a natural user interface. world frame. 8 gationtimefromC toBob,thetimeonC ’sphotograph 1 1 will be that it read at time t = −d /c = −γd/c. This B B differs from the time deduced by Alice by the time dila- tion factor γ. Thus we obtain time dilation fromaberra- tion. However the focus here is on the relativity of simul- taneity. The path length difference ∆d between the B paths from clocks C and C may be approximated by 1 2 a method familiar from diffraction theory. We drop a perpendicular to C from the line between clock C and FIG.6: Schematicdiagramsfortherelativityofsimultaneity. 2 1 BothpanelsrefertothetimeofeventCLICK,indicatedby*s, Bob. Thedistancealongthislinefromtheperpendicular whenthephotographsaretaken. ThelinesfromclocksC1and to C1 is the approximate path length difference. Using C2 toAliceandBobarethepathstakenbythelightforming the corresponding right-angle triangle with hypotenuse thephotographsintheirrespectiveframes. (a)Alice’sframe. L and angle π−θ we have B B (b) Bob’s frame in which theclocks are at rest. ∆d =L cos(π−θ )=−L cosθ =L (v/c), (A3) B B B B B B Fig. 6(a) shows the light paths taken from the clocks where we again used the aberration formulae, Eqs. (3), C and C to Alice, for whom they are moving from with θ = π/2, to find cosθ = −v/c. The correspond- 1 2 A B right to left with speed v. She looks perpendicular to ing light propagation time difference, ∆t = L v/c2, is B B thedirectionofrelativemotiontoseethem,atθ =π/2, the time difference between the clocks in Bob’s photo- A andinfersthatwastheirdirectionwhentheyemittedthe graph. However, it is also the time difference between lightshe images. Letthe perpendiculardistance toclock the clocks in Alice’s photograph, since both images are C be d, and the distance between the clocks, in Alice’s made from the same group of photons; those present at 1 frame, be L . Due to the light propagation delay, the event CLICK. A time on the photograph of clock C will be that it read We can express this time difference in terms of Alice’s 1 at time t = −d/c. The path length difference between quantities by using the length contractionformula L = A B the paths from clocks C and C is γL , 2 1 A ∆dA =qd2+L2A−d≈L2A/(2d), (A1) ∆tB =(γLA)(v/c2)=γ(LAv/c2), (A4) where we have assumed L ≪ d and Taylor expanded which is precisely the term responsible for the relativity A A of simultaneity in the inverse Lorentz transformation, the square root to first order. The corresponding light propagationtimedifferencecanbemadearbitrarilysmall ∆t =γ(∆t +∆x v/c2). (A5) by making L a sufficiently small fraction of d45. B A A A Fig. 6(b) shows the light paths taken from the clocks Thus we have shown how the relativity of simultaneity to Bob, who is at rest relative to them. He looks back can be understood in terms of light propagation delays, at the angle θ to photograph them. Let the distance B and be deduced from direct observations of clocks. to clock C be d . Since lengths perpendicular to the 1 B relative motion are invariant this is given in terms of d by ACKNOWLEDGMENTS d sin(π−θ )=d sinθ =d⇒d =γd, (A2) B B B B B TheauthorsacknowledgetheassistanceofDr.K.Wil- where we used the aberration formulae Eqs. (3), with son with the implementation and evaluation of the Real θ = π/2, to find sinθ = γ−1. Due to the light propa- Time Relativity laboratory. A B ∗ Electronic address: [email protected] ofchangingdeeplyheldstudentbeliefsabouttherelativity 1 The Real Time Relativity web site contains informa- of simultaneity”, Am.J. Phys. 70, 1238-1248 (2002). tion about the program and a downloadable installer: 4 R. E. Scherr, “An investigation of student un- hhttp://www.anu.edu.au/Physics/Savage/RTRi. derstanding of basic concepts in special relativ- 2 R. E. Scherr, P. S. Shaffer, and S. Vokos, “Student un- ity”, Ph.D. dissertation, Department of Physics, derstanding of time in special relativity: simultaneity and University of Washington 2001 (unpublished), reference frames”, Phys. Educ. Res., Am. J. Phys. Suppl. hhttp://www.physics.umd.edu/perg/papers/scherr/ 69, S24-S35 (2001). index.htmli. 3 R. E. Scherr, P. S. Shaffer, and S. Vokos, “The challenge 5 A. Einstein, “On the electrodynamics of moving bodies”, 9 Annalen der Physik 17, 891-921 (1905). Reprinted in En- ofthestarfieldfromarelativisticspaceship”,Am.J.Phys. glish translation in J. Stachel, Einstein’s miraculous year 47, 309-316 (1979). (Princeton University Press, Princeton, 1998). 28 Men-Chou Chang, Feipei Lai and Wei-Chao Chen, “Im- 6 A.Komar,“Foundationsofspecialrelativityandtheshape age shading takingintoaccount relativistic effects”, ACM of thebig dipper”, Am.J. Phys.33, 1024-1027 (1965). Trans. Graphics 15, 265 (1996). 7 A. Peres, “Relativistic telemetry”, Am. J. Phys. 55, 516- 29 T. Mu¨ller, “Visual appearance of a Morris-Thorne- 519 (1987). wormhole”, Am.J. Phys.72, 1045-1050 (2004). 8 H. Blatter and T. Greber, “Aberration and Dopplershift: 30 W. Gekelman, J. Maggs and Lingyu Xu, “Real-time rela- anuncommonwaytorelativity”,Am.J.Phys.56,333-338 tivity”, Computers in Physics 5, 372-385 (1991). (1988). 31 Ping-Kang Hsiung , R. H. Thibadeau, and M. Wu, “T- 9 R. R. Hake, “Interactive-engagement versus traditional buffer: fast visualization of relativistic effects in space- methods: a six-thousand-studentsurveyof mechanicstest time”, ACM SIGGRAPH Computer Graphics 24, 83-88 data for introductory physics courses”, Am. J. Phys. 67, (1990). 755-767 (1999). 32 J.D. Owens et al., “A Survey of General- 10 L. C. McDermott and E. F. Redish, “Resource Letter: Purpose Computation on Graphics Hard- PER-1: Physics Education Research”, Am. J. Phys. 66, ware”, Computer Graphics Forum 26 (2007); 64-74 (1998). hhttp://graphics.idav.ucdavis.edu/publications/i. 11 L. C. McDermott, “Oersted medal Lecture 2001: Physics 33 D. Weiskopf, “Visualization of four-dimensional education research - the key to student learning”, Am. J. spacetimes”, Ph.D. dissertation, Univer- Phys. 69, 1127-1137 (2001). sity of Tu¨bingen 2001 (unpublished); 12 E. Mazur, Peer Instruction: A User’s Manual (Prentice- hhttp://www.vis.uni-stuttgart.de/relativity/readingi. Hall, Englewood Cliffs, NJ, 1997). 34 M. Bochers, “Interactive and stereoscopic visu- 13 W.K.Adams,S.Reid,R.LeMaster,S.B.McKagan,K.K. alization in special relativity”, Ph.D. disserta- Perkins and C. E. Wieman, “A Study of Interface Design tion, University of Tu¨bingen 2005 (unpublished); for Engagement and Learning with Educational Simula- hhttp://w210.ub.uni-tuebingen.de/dbt/volltexte/2005/1891i. tions”, Journal of Interactive Learning Research, submit- 35 U. Kraus and C. Zahn, “Relativity visualized: space time ted (2006). travel”, hhttp://www.spacetimetravel.org/i. 14 C. Wieman and K.Perkins, “A powerful tool for teaching 36 D. Weiskopf et al., “Explanatory and illustrative visual- science”, Nature Physics 2, 290-292 (2006). ization of special and general relativity”, IEEE transac- 15 S.Yeo,R.Loss,M.Zadnik,A.Harrison,andD.Treagust, tionsonVisualizationandComputerGraphics12,522-534 “What do students really learn from interactive multime- (2006). dia? A physics case study.”, Am. J. Phys. 72, 1351-1358 37 T. Greber and H. Blatter, “Aberration and Dopplershift: (2004). thecosmic background radiation and its rest frame”, Am. 16 R. N. Steinberg, “Computers in teaching science: to sim- J. Phys. 58, 942-945 (1990). ulate or not to simulate”, Phys. Educ. Res., Am.J. Phys. 38 U.Kraus,“Brightnessandcolorofrapidlymovingobjects: Suppl.68, S37-S41 (2000). the visual appearance of a large sphere revisited”, Am. J. 17 A. Van Heuvelen, “Learning to think like a physicist: a Phys. 68, 56-60 (2000). review of research based instructional strategies”, Am. J. 39 J. Andrews and N. Baker, “Xbox 360 System Architec- Phys. 59, 891-897 (1991). ture”, IEEE Micro 26, 25-37 (2006). 18 ThroughEinstein’sEyes,amultimediawebsiteaboutrel- 40 Folding@home project web site: ativisticoptics.AvailableonCDfromtheauthors,andon- hhttp://folding.stanford.edu/i. line at: hhttp://www.anu.edu.au/Physics/Savage/TEE/i. 41 PHYS1201. Lectures and tutorials were 50 minutes each. 19 N.D.Mermin,It’sabout time(PrincetonUniversityPress, There were 74 studentsin the class. Princeton, 2005). 42 G. Gamow, Mr Tompkins in Paperback (Cambridge U.P., 20 R.Penrose,“Theapparentshapeofarelativistically mov- Cambridge, England, 1965). Note that the figures do not ing sphere”, Proc. Camb. Phil. Soc. 55, 137-139 (1959). show what would actually be seen, as this was not known 21 J. Terrell, “Invisibility oftheLorentzcontraction”, Physi- at thetime thestory was written. cal Review116, 1041-1045 (1959). 43 Although the Sun and Earth are the correct size (Sun’s 22 V.F.Weisskopf,“Thevisualappearanceofrapidlymoving diameter is 4.7 light seconds) the distance between them objects”, Physics Today 13, 24-27 (Sept.,1960). is not to scale. 23 W. Rindler, Relativity: special, general, and cosmological 44 The best way to experience this is to run the simulation. (Oxford UniversityPress, USA,2006). However, a movie of the sequence discussed in the text is 24 Standard configuration of frames S and S′ has theS′ ori- available from the Real Time Relativity web site.1 gin traveling with speed v along the positive x axis of S. 45 Choosing the perpendicular distance to the clocks suffi- The frames’ origins are coincident at t = t′ = 0, and the cientlylarge,comparedtothedistancebetweentheclocks, corresponding coordinate axes are parallel. makes this approximation as good as required. It is only 25 TheinverseLorentztransformationsofthe4-frequencyare within this approximation that pure length contraction obtainedfromEqs.(2)byswappingthewandclabelsand and relativity of simultaneity are observed. changing the sign of v. 46 Quotationsaretakenfromthepost-laboratoryevaluations 26 J.M. McKinley and P. Doherty, “Relativistic transforma- completed by students. tions of light power”, Am.J. Phys.47, 602-605 (1979). 27 J.M.McKinley,“Insearchofthestarbow: theappearance