The Acceleration of the Expansion of the Universe: A Brief Early History of the Supernova Cosmology Project (SCP)
aa r X i v : . [ a s t r o - ph . C O ] J u l The Acceleration of the Expansion of theUniverse: A Brief Early History of theSupernova Cosmology Project (SCP)
Gerson Goldhaber Lawrence Berkeley National Laboratory and Physics Department University of California,Berkeley, CA 94720
Abstract.
It is now about 10 years since the evidence, based on Type Ia supernovae, for theacceleration of the expansion of the Universe was discovered. I will discuss some aspects of thework and events in the Supernova Cosmology Project (SCP), during the period 1988 to 1998, whichled to this discovery.
Talk presented at Dark Matter Conference DM08, Marina del Rey, Feb. 20, 2008.
PREAMBLE
In the past I have written papers that described new physics results. In this report,however, on this tenth anniversary of the discovery of dark energy, I present the historyof our discovery as I saw it . I will recount parts of this many-year project that I rememberparticularly vividly as important in the work I was doing. Of course, the other membersof the team would recall different mixes of events and efforts in which they wereparticularly engaged, and I hope they will find the opportunity to give their recollections.When the bubble chamber work in Luis Alvarez’s group at LBNL came to an end inthe early 1970s, there started an interest in astrophysics. Rich Muller became interestedin the Cosmic Microwave Background (CMB). He was later joined by George Smoot.After convincing Luis Alvarez of the feasibility of a CMB measurement and obtaininghis support, Rich Muller and George Smoot and their collaborators studied the CMBwith detectors placed on U2 airplane flights. This led to their discovery of the CMBdipole asymmetry. Smoot then went on to work with the COBE satellite, and was theleader in the discovery of the CMB anisotropy.Meanwhile, Rich Muller, joined by Carl Pennypacker (a staff member at the SpaceScience Laboratory), set up an automated search for nearby supernovae (SNe). Thiswork was primarily under Pennypacker’s responsibility and was helped by RichardTreffers of the Berkeley Astronomy department and others. They were later joined byMuller’s student Saul Perlmutter who carried out his thesis, a search for "Nemesis", thesuspected companion star to the sun, as well as developing the analyses and software [email protected] 1 or the SNe search. During the period 1980-1988, they demonstrated that the method,originally suggested by Stirling Colgate et al. [1], worked, and they were able to discoverabout 25 nearby SNe. This work became the prototype of later automated SNe searches.In 1988, a National Science Foundation (NSF) Center for Particle Astrophysics(CfPA) was being formed on the UC Berkeley campus. Pennypacker and Perlmutter,working in Muller’s group, had developed a new experiment to discover the “Fate ofthe Universe” through a study of distant Type Ia SNe, claimed to be “standard candles.”This experiment was included as one of the elements of the new CfPA [2]. THE SCP GROUP AND THE DISCOVERY OF DISTANT SNE
In 1989, as I was thinking about my next experiment, I was invited by Carl Pennypackerto join in the search for the discovery of the “Fate of the Universe” in what was atthat time called the "Deep Supernova Search." The subject appealed to me, as did theproposed technique, since it involved evaluating images, something I have been doingthroughout my career. In 1990-91 the entire group at Berkeley (now known as the Supernova CosmologyProject (SCP)), consisted of Carl Pennypacker, Saul Perlmutter, Heidi Marvin, myself,and Rich Muller (shown in Fig. 1, l. to r.) .Soon after my joining the group, there were major changes in its makeup. In 1991Rich Muller decided to spend his efforts in research on the ice ages and global weatherpatterns, and Carl Pennypacker founded the "Hands on Universe" Program for highschool students and began to devote most of his time to educational activities. This leftSaul and me to carry on with graduate student Heidi Marvin Newberg and later AlexKim. Shortly thereafter, Bob Cahn, at the time the physics division director at LBNL,discussed with me the question of a group leader. With my strong support, he decided toappoint Saul to that position.Perhaps because Rich, Saul, Carl, and I were not established as astronomers, it was atfirst difficult for us to get time on the large premier telescopes. Before I joined the group,on Carl’s initiative , Carl, Rich and Saul, together with two colleagues in Australia, BrianBoyle (starting out in the UK) and Warrick Couch, did manage to obtain scheduled timeon the 3.9m Anglo Australian Telescope and built a focus reducing lens and CCD camerafor it. During the three years we observed at this telescope, while the system worked well(and became one of the most-used instruments at the telescope for years after), there wasno identified SN candidate. Unfortunately, 80% of our scheduled nights were lost due tobad weather. The lens and camera construction and installation and the techniques wedeveloped, as well as an unidentified candidate found later at the 2.5m Isaac Newton In 1989 my co-group leader George Trilling and I had come to an end of a nearly twenty-year collab-oration with Burton Richter and Martin Perl and coworkers of SLAC and LBL where we had discoveredthe psion family, charmed mesons, and the tau lepton among many others. Evaluating images is something I have been doing throughout my physics career, beginning withphotographic emulsions in the 1950s, to bubble chambers in the 1960s, to computer-reconstructed particleevents in the 1970s, and 1980s. Thus, the supernova experiment with its computer-reconstructed opticalimages, seemed to be a nice fit with my experience and inclinations. 2
IGURE 1.
The SCP group at Berkeley in 1990-91. Carl Pennypacker, Saul Perlmutter, Heidi Marvin,Gerson Goldhaber, and Rich Muller l. to r.
Telescope, are discussed in great detail in Heidi Marvin Newberg’s thesis [3].While for the first three years of our efforts we did not find any identified supernovae, this time was not wasted, as it allowed us to develop the techniques which led to theeventual success of the project.In particular, Saul Perlmutter developed the technique for finding “SNe on demand” or“batch mode". This involved taking a CCD “reference" image just after a new moon, andcoming back to take CCD “discovery” images before the next new moon (See Fig. 2).The data were sent back to Berkeley the same night they were taken at the telescope.On the next morning we ran a program that subtracted the reference image from thediscovery image and flagged possible candidates. In a few hours of hand scanning wewere able to select the promising SN candidates. Thus as soon as SN candidates werediscovered one could send observer(s) to the Keck 10m telescope to measure spectraand, after that, start to take data points for the light curve at ground based telescopes.(See Fig. 2 for list of follow-up telescopes.) For the first time, new supernovae couldbe guaranteed to be discovered on a certain date, and while they were still brightening–making it possible to propose for scheduled nights on the follow-up telescopes. This However, our general approach received important validation with the discovery by Norgaard-Nielsenet al. [4] of a single high z supernova (red shift of 0.31) after a two year effort. This SN was howeverdiscovered well after maximum light. 3 echnique, or some variation of it, was adopted by all subsequent SN searches.It is instructive to compare Saul’s new technique to what was being invented as thestate of the art in the early 1990s at much lower redshifts, Hamuy et al.’s Calan/TololoSupernova Search (CTSS). This search observed fields (with photographic plates orfilm) twice a month, and then organized follow-up observation campaigns – but unlikeSaul’s technique is was not aimed at producing guaranteed batches of (on-the-rise) SNdiscoveries on a given date. Thus, towards the end of the CTSS search, Hamuy et al. [5]wrote: “Unfortunately, the appearance of a SN is not predictable. As a consequence ofthis we cannot schedule the followup observations a priori , and we generally have to relyon someone else’s telescope time. This makes the execution of this project somewhatdifficult.”As pointed out in a review article by Perlmutter and Schmidt, Saul’s technique ad-dressed a different problem: “The SCP targeted a much higher redshift range, z > . IGURE 2.
The strategy developed by Saul Perlmutter for finding SNe on demand from repeated CCDimages.
In 1993 we collaborated with Alexei Filippenko, of the UC Berkeley AstronomyDepartment, for measurements of spectra at the Keck Telescope in Hawaii.In 1994 Don Groom, from the Particle Data Group at LBL, joined our group, as wellas Susana Deustua. Don Groom adapted the CERN program MINUIT to fit our SNe tothe light curve and to determine the stretch parameter.Graduate students Alex Kim, Matthew Kim and programmer Ivan Small were alsoessential in this early work.By 1995, now that we had became successful in finding distant supernovae, we wereable to hire two postdocs, Rob Knop and Peter Nugent, and later, Greg Aldering. Thesenew people, as well as Susana, all had a background in astronomy.By 1998 our group had grown to an international collaboration with 32 members Gregory Aldering, Brian Boyle, Patricia Castro, Warrick Couch, Susana Deustua, Richard Ellis, Se-bastien Fabbro, Alexei Filippenko, Andrew Fruchter, Gerson Goldhaber, Ariel Goobar, Donald Groom,Isobel Hook, Mike Irwin, Alex Kim, Matthew Kim, Robert Knop, Julia Lee, Chris Lidman, Thomas5 hat signed the Dark Energy discovery paper [7]. In 2007 Saul and the other 31 were co-recipients of the Gruber prize in cosmology for the discovery of Dark Energy, togetherwith Brian Schmidt and the Hi-Z Team [8].
MILESTONES LEADING TO THE DISCOVERY OF DARKENERGY
Here is a list of some of the important comments and papers which allowed us to makeour discovery. • Suggestions to use Type Ia SNe as standard candles: From 1930s on Zwicky, Baade,Sandage, Kowal, Tammann etc. • • • • • • z < . • • • W m and W L shown to be feasible,fitting apparent magnitudes of SNe Ia at a variety of redshifts extending to z > Matheson, Richard McMahon, Heidi Newberg, Peter Nugent, Nelson Nunes, Reynald Pain, Nino Pana-gia, Carl Pennypacker, Saul Perlmutter, Robert Quimby, Pilar Ruiz-Lapuente, Brad Schaefer, NicholasWalton. (Alexei Filippenko moved to the Hi-Z Team when it was formed in 1995.) 6 nd shows that at each redshift the best-fit confidence region will make a diagonalon the W m vs. W L plane (with a slope that varies with redshift).Here "stretch", introduced by Saul, is the deviation of the measured timescale of alightcurve from a standard average lightcurve. Typical values are s = . .
2. Thelarger s the brighter the SN. The effective B-magnitudes are corrected to a standardlightcurve. SEPT. 24, 1997: A PEAK IN THE W m HISTOGRAM
What has become the conventional way to determine W m and W L is the Goobar-Perlmutter [25] best fit confidence level distribution on the W m , W L plane. This approach,favored by astronomers, requires accurate determinations of the errors and correlated er-rors. When the SNe light curves are first fitted to a template this information is not yetavailable.The alternative approach, similar to the technique used to find resonances in particlephysics, is to look for peaks in the distributions (on histograms) of the variable beingstudied, namely mass in particle physics and W m in our case here. In this approach themeasurement errors are reflected in the width of the resulting peak.In September 1997 we had completed a first pass of the light curve data point analysison 38 SNe, but the detailed error analyses were not yet available. It took several monthsfrom the time a SN was discovered until the light curve points could be evaluated. Wealso took some final reference points a year after discovery. Rob Knop, who was workingon this, was able typically to complete the measurements on one to two SNe per month.On the basis of these 38 Type Ia SNe we obtained an indication that the universeis accelerating, rather than decelerating as we had originally expected. This was theculmination of nine years of work learning how to find high redshift SNe, how tomeasure them, how to fit them, how to K-correct them, how to stretch correct them,how to account for dust reddening, and then finally fit them to a lightcurve template.This was all achieved by a collaborative effort of the entire group. I studied the SNeafter the data points on the lightcurve had been measured, fitted them with the group’slightcurve fitter which gave us the stretch value, and made a table with some 20 attributesfor each SN.The question was next how to obtain the W m distribution from this plot. What I triedwas to plot Hubble curves, for a flat universe, with W m in fine intervals at 0.0, 0.2, 0.4,0.6, 0.8, 1.0. See Fig. 3. A distribution in W m is then obtained by counting (by hand!)all the SNe that fell into each DW m interval giving the histogram shown in Fig. 4. A fewdays later Saul wrote a program, SNLOOK, to have the computer count the number ofSNe in each DW m interval. He also made plots both for a flat universe, and for a universewith W L = IGURE 3.
Our preliminary data of 38 SNe, as of Sep., 24, 1997. The effective B-magnitude vs. log(cz).Here the effective B-magnitude is the K-corrected and stretch-corrected magnitude and z is the redshift.The curves for a series of W m values, in a flat universe, are shown. The sum of the data points in each DW m interval, as counted by hand, is also shown at the right hand upper edge of the figure. flat universe and a W L = W m − W L plot, like the one shownin Figure 9. We can also analyze this data by projecting onto an W m histogram for eitherof the simple cosmology cases, the flat universe line or the W L = ∼ W m in the flat-universe-case histogram,we will be guaranteed to find a negative value for W m in the W L = W m measurements fromgalaxy clusters, etc., that were being made in the late 1990s, where the measurement byitself did not tell you anything about W L .In Fig. 4 the peak we actually saw at W m = . . W L is 0 . .
7. Thus we obtained a significant positive value for W L or possiblya cosmological constant. As we were looking for deceleration we always calculated thedeceleration parameter q = ( / ) W m − W L . Here q = − . − .
55. A negative valuefor q implies acceleration. We calculated q for each of our samples and it always said:ACCELERATION. IGURE 4.
The W m distribution for the data points in Fig. 3, as presented at our SCP group meeting onSep. 24, 1997, is shown. The values for the first 7 SNe, which gave a considerably larger value for W m ,are shown cross-hatched. Here W L is given by 1 − W m . The low redshift distribution of SN absolute magnitudes available at that time alreadyshowed a strong peak that indicated a large fraction of supernovae were not sufferingextinction. The strong peak in the W m distribution was likely also to be due to unex-tincted SNe, but we understood that, whatever this extinction distribution, as long asit was the same at low and high redshift the cosmology results would be correct. Westudied the color distributions at both redshift ranges, and found that the distributionswere sufficiently consistent. This histogram of Figure 4 (and the others shown below )therefore accounts for dust in this statistical approach, rather than correcting extinctionSN by SN. (Our other analyses at that time examined this point with further studies ofthe measured colors of the distant SNe, performing analyses correcting individual SNbrightnesses and analyses comparing groups of SNe. We concluded that our cosmologyresult was robust – and that the correct way to analyze color and dust for all the data setsat that time was not to use a Baysean prior on the dust distribution, as others had done.)It is also interesting to note that the stretch correction, which becomes very important inlater precision measurements, does not change these preliminary values significantly. Itonly tends to broaden the observed peak. At any rate we did apply the stretch correctionto these data. his startling result was naturally treated with some skepticism within the group,when these plots were discussed in the weekly SCP meetings at the end of September1997. Although we had already submitted a Nature paper[33] indicating evidence fora cosmological constant, we were still surprised that this result was settling in on thelower end of the W m range from our first seven SNe [11, 12, 13] which had centered ona high value for W m . These seven SNe are shown as cross hatched in Fig. 4. and indeedoccur at the upper end of the distribution.Again the fact that the deceleration coefficient was negative, indicating acceleration,was hard to swallow at first. I had been “bump hunting” for the past 30 years, and foundthe observed peak completely convincing. I believe that Saul convinced himself afterhe wrote the program, SNLOOK and by independently studying our data. He later inOctober and December presented such a histogram in colloquia giving W m and W L .Some other colleagues, Rob Knop and Carl Pennypacker, appeared convinced; whileGreg Aldering was not fully convinced, he stated however that this histogram “helpedto galvanize the effort within our group". Indeed, as the result of a major effort, the SCPgroup was able to calculate the best fit confidence levels, for all 40 SNe, in time for theJan. 8, 1998 AAS meeting.For confirmation, I asked my colleagues to check these results, in case there couldhave been a mistake. However, all the 20 entries in the so called “Gerson-table” for eachof the 38 SNe were confirmed as correct. Later Greg Aldering added more columns tothis table, giving the correlated errors, and Richard Ellis added a column with the bestestimate of the nature of each host galaxy.Although as mentioned above, Goobar and Perlmutter (1995)[25] had shown that anySN measurements indicating low W m for a flat universe will also indicate negative W m for a W L = W L = W L = W m histogram. Howevernow it occurs for negative W m values. In Fig. 6. is shown the W m histogram, this timecalculated with Saul’s SNLOOK program, for this case. As can be noted, 25 out of the38 SNe give a negative and hence un-physical value for W m . This clearly demonstratesthat W L cannot be zero but must be greater than zero. Figures 5 and 6 are from mynotebook from October 1997 and were shown informally to group members. This resultwas also presented in my seminar at Santa Barbara in December, 1997.One can also turn this argument around and calculate the location of the peak in an W m histogram for a series of different increasing W L values and find for which W L valuethe peak begins to lie at positive (hence physical) values. This approaches, and goesbeyond, W m , W L values consistent with the above values for a flat universe. This methodis equivalent to traveling along the central line for the confidence level “ellipses”, startingat negative W m values, shown in Fig. 9. In retrospect, it is instructive to revisit the question of what happened with those first seven supernovae.As shown shaded in Fig. 4 they are all at high W m values. Two turned out to be outliers with one probablynot a Type Ia. The other 5 moved slightly towards lower mass densities with remeasurement, but for themost part they simply happened to lie on the tail of the distribution. The lesson is: beware of statistics ofsmall numbers! 10 IGURE 5.
The Hubble curves for both a flat universe (solid curves) and an W L = FIGURE 6.
The W m distribution for a W L = W m and hence an un-physical value. This shows clearly and independently of the flat universe condition that W L must be greater than zero. 11 ABLE 1.
SCP SNe Ia Discoveries with z > . We thus have two ways to visualize with histograms the evidence for a positive valuefor W L . The first histogram essentially looks at where the diagonal confidence region (in,e.g., Fig. 9) crosses the flat universe line, and the second looks at where it crosses the W L = W L must be positive (since the W L = q the supernova data clearlyindicated acceleration.In Table 1 are shown the SN discoveries over the years 1989 to 1997. Saul’s SNe ondemand method really came into its own when we obtained time on the Cerro/Tololo 4mtelescope combined with the “big throughput” camera (BTC) of Bernstein and Tyson.Table 1 shows that over a nine-year period we discovered 42 SNe with z > . Oct. 1997. Two papers by the SCP and Hi-Z Team as well as a jointpaper submitted with hints, from very small samples of SNe, for anon-zero W L The very first analysis of multiple high-redshift SNe indicated a high W m value,albeit with large statistical error bars: using the 5 of these first 7 SNe with lightcurvetimescales that were calibrated at low redshift yielded W m = . + . − . for a flat universe[11, 12, 13]. The following year, we added one very-well-observed type Ia supernovaethat we discovered and followed with the Hubble Space Telescope (HST), SN 1997ap,that was at that time again the most distant that had ever been observed, z = .
83, withthe most leverage for this measurement. The addition of this single SN allowed us torecalculate W m for 5+1 supernovae, giving W m = . ± . Nature on Oct. 6, 1997 [33] andit appeared in the January 1, 1998 issue.Garnavich et al. [34] submitted a paper by the Hi-Z Team on Oct. 13, 1997, basedon three confirmed high redshift SNe measured with the HST, and one measured fromthe ground. They concluded that matter alone is insufficient to produce a flat universe.Specifically for W m + W L = W m is less than 1 with more than 95% confidence, andtheir best estimate of W m is − . ± . W L = .5 4.0 4.5 5.0 5.5 log(cz) e ff ec ti v e m B redshift z Hamuy et al(A.J. 1996)
SupernovaCosmologyProject
Perlmutter, et al.
Nature (1998) -0.5 Ω Λ = Ω M = F l a t Λ = FIGURE 7.
Hubble diagram from the Nature paper[33] showing the effect of the very well-measuredSN at z = .
83 on the determination of cosmological parameters.
21, 1997 and accepted Apr. 17, 1998, showed that the “snapshot” method works. Thatinvolves taking a single photometry point and a single spectrum on the same night anddeducing the magnitude at maximum light from this limited information. On the basisof four SNe handled in this fashion, as well as the SNe from the above two papers, theydeduced that for a flat universe W m = . + . − . and W m = − . + . − . for an W L = W L they relied on a very small number of SNe. The hints from the snapshot paper similarlyrelied on four SNe using this less precise method.All this work was going on more or less contemporaneously with my study of the38 SNe. My colleague Peter Nugent, who did some of the fits of these confidencelevel distributions, felt that these 3 “hints” were a more convincing and possibly earlierevidence for a non-zero W L than the peak in the W m histogram. My feeling is that weneeded the larger statistics since we had amply demonstrated that with a very smallnumber of SNe (7 SNe) one can be way off! OCT. 23 TO DEC. 14, 1997: THREE COLLOQUIA AND ASEMINAR
The first four public presentations of the SCP evidence for W L > and acceleration,based on the peak in the W m histogram for 40 SNe as well as the unphysical resultsobtained when W L is assumed to be zero. rom October to November, we re-measured and re-fitted all of the 38 SNe and addedtwo more, giving 40 (while three were identified as outliers). We also changed fromthe Burstein and Heiles [36] version of dust extinction in our galaxy to the more recentSchlegel, Finkbeiner and Davis [37] version. I then revised the tables to reflect all thesechanges. Both Saul and I plotted the Hubble plots, based on the revised tables, and usedSNLOOK to obtain the latest version of the histogram.The first public presentation of our results was a colloquium by Saul on October 23,1997 at the Physics Department, UC San Diego. This was followed by Saul’s colloquiaat the Physics Department UC Berkeley on Dec. 1, 1997 and at the Physics DepartmentUC Santa Cruz on Dec. 11, 1997 (See Fig. 8). FIGURE 8. ( left ) The figure of the W m distribution shown at Colloquia at UC San Diego in October1997 and at UC Berkeley and Santa Cruz in early December 1997 by Saul Perlmutter. The x-axis gives W m for a flat universe. All the SNe given in Fig. 4. have been re-fitted. The Gaussian fit to this preliminarydata gave W m = . ± .
06 with a s = .
3. This analysis accounts for dust with the more correct statisticalapproach discussed in the text above concerning Fig. 4, rather than correcting extinction SN by SN. ( right )A reproduction of the actual figure of the W m distribution shown at the ITP Santa Barbara on Dec. 14, 1997by Gerson Goldhaber. The lower scale gives W m for a flat universe. The upper scale gives W m for an W L = W m . This demonstratesthat W L must be greater than zero by a method which does not require the flat universe condition. Next, I gave a seminar at the Institute for Theoretical Physics (ITP) at UC SantaBarbara on Dec. 14, 1997 (See Fig. 8). In all four of these talks, as stated above, theconventional description of the confidence level distribution on the W m , W L plane wasnot yet available for the 40 SNe. So in all four talks the W m histograms, re-evaluatedwith more careful re-measurements, were shown. We showed that the result was robustagainst the likely sources of systematic error, but we still considered our result aspreliminary as we had not yet completely explored every possible effect. t was not easy to convince the astrophysics community of this result, as it wascontrary to all ingrained beliefs. Acceleration rather than deceleration as expected!The question has been posed: who remembers the details of these talks? I know thatKim Griest remembers Saul’s talk at UC San Diego and was enthusiastic about the result.I also know Joel Primack and Michael Riordan remember Saul’s talk at UC Santa Cruz.As former and current particle physicists they understood the significance of a peak ona histogram. Joel Primack was particularly enthusiastic and stressed the importance ofthis discovery after Saul’s colloquium. Dave Branch remembers my talk and subsequentdiscussions and understood its significance. David Gross asked me after my talk how Icould come to such a momentous conclusion on the basis of just 40 SNe.As is usual, Saul’s Colloquium at UC Berkeley was video taped (DVD available).Saul first talked about the “hint” from our 5+1 SNe fit I mentioned above. Then basedon a gaussian fit to the observed peak on the histogram, Saul quoted W m = . ± . W L = .
71. In my presentation at the ITP I did not try for anaccurate value of W m , but rather showed Fig. 8 and indicated that W m = . W L =
0. As seen from the upper scale in Fig. 8 this assumption gives negative and henceunphysical values for W m . As stated earlier, given the slope of the confidence region forany supernova data, such values for W m in the two cases considered demonstrate that W L must be greater than zero, and does not depend on assuming a flat universe.In his talk Saul went on to point out that while on the one hand 0.7 for W L corre-sponded to a large fraction of the universe but on the other hand it was a very smallvalue, by a factor 10 , compared to the value expected from virtual particle vacuumfluctuations.It is remakable that in Saul’s talk on Dec. 1, 1997 we had already presented, in public,evidence for the whole story of what was later called Dark Energy by Michael Turner. AtDM08 I presented a four-minute video excerpt of Saul’s talk, showing him making thesepoints based on the histogram. (This video is given in the DM08 conference proceedings[41].)Not much has changed qualitatively in the 10 years since then. With many more SNestudies [42] as well as concordance with CMB and evidence from cluster studies [43]and later Baryon Acoustic Oscillation (BAO) [44] studies, we now know W m to moresignificant figures, but qualitatively there is no change from the value we quoted in1997. We have learned that the equation of state parameter w is very close to -1 and havesome limits on the variation of w with z. But after all that, just as 10 years ago, we stilldo not understand the nature of Dark Energy [45]. Examples of exceptions to the W L = W L could be comparable to the currently observed value. Krauss andTurner [39] argued primarily on the basis of the age of the universe and other then available cosmologicalmeasurements that W L should lie between 0.6 and 0.7. Ostriker and Steinhardt [40] derived a concordancemodel based on all the then available cosmological data and concluded W L = . ± .
1. 15
AN. 8, 1998: FIRST PUBLIC PRESENTATION OF THE SCPRESULTS AT AN AMERICAN ASTRONOMICAL SOCIETYMEETING
The fifth public presentation of the SCP evidence for W L > and acceleration, based onthe fit to the confidence level distribution on the W m , W L plane for 40 SNe. From October to late in December 1997, the SCP group at Berkeley re-measured andre-checked and completed all error and correlated error calculations on what had bythen become 40 SNe. This effort was led by Saul, Greg Aldering, Peter Nugent, andRob Knop. Each group member as well as the group members in Baltimore, Stockholm,Paris, England, Australia, Spain and Chile convinced himself/herself by direct studyof the data that indeed W L > W m , W L plane (Fig. 9). Here it should be noted that the confidence level distribution doesnot require the flat universe condition to give W L greater than zero. This was the firstpresentation at an official conference of evidence for W L >
0, and hence from finding q <
0, the acceleration of the expansion of the universe. To take possible systematicuncertainties into account, we allowed for a very generous systematic error and theywere shown for the case where all systematics conspired in the same direction. This isshown by the dotted confidence level curves in Fig. 9. This figure was reproduced byJames Glanz in his report on our presentation in Science [48]. Furthermore, using thesame W m histogram for a flat universe as in Fig. 8, Saul also showed the studies of theeffects of systematics: Malmquist bias, stretch dependence, and color/dust dependence. FEB. 18, 1998: PUBLIC PRESENTATIONS AT THE DARKMATTER 1998 (DM98) CONFERENCE.
The sixth and seventh public presentations of the SCP evidence for W L > and accel-eration based on 42 high redshift SNe were given. The first public presentation by theHi-Z Team of evidence for W L > and acceleration based on 14 independent high red-shift SNe was also given. Both teams quoted results based on fits to the confidence leveldistribution on the W m , W L plane. The following month, on Feb. 18, 1998, both I and Saul Perlmutter gave talks, in thatorder [49], showing our results at the meeting organized by Dave Cline just 10 years ago(Dark Matter 1998) at Marina del Rey. By this time our number of fully analyzed SNehad grown to 42. ( See Table 1.)I spent part of my talk on some work I had done on time dilation in SNe explosions, IGURE 9.
The W L vs W m best fit confidence level distribution shown at the Washington meeting ofthe AAS on Jan, 8 1998 by Saul Perlmutter. The dashed curves represent a very generous estimate of whatpossible systematic errors could do. proving that the redshift is due to the expansion of the universe, rather then to the “tiredlight” hypothesis [11, 12, 14]. I then showed the best fit confidence level distribution onthe W m , W L plane, which for a flat universe implied W m = . + . − . (statistical) + . − . (systematic) and hence, from evaluating q , acceleration of the universe at the presentepoch.In his talk Saul Perlmutter gave more details on our results and on the error calcula-tions, and stressed the studies of the systematic uncertainties.Following our talks, Alexei Filippenko presented the results [50] of the Hi-Z Team,who claimed they had established the acceleration of the universe on the basis of aconfidence level analysis of 16 high redshift SNe. These consisted of 10 well measuredSNe, plus 4 from the “snapshot” method [35] plus 2 which came from our set of 42 [7].Later the refereed publications of the two groups capped these momentous results.The Hi-Z Team paper was submitted on Mar. 13, 1998 [8] and the SCP paper wassubmitted on Sept. 8, 1998 [7]. OW DOES AN W m HISTOGRAM LOOK TODAY?
To see how the method, patterned on particle physics and used to indicate accelerationon Sept. 24, 1997, works with larger statistics I have analyzed (this time with a computerprogram) a sample of 257 SNe with z > . W m . The enormous peak is clearly observed,which completely confirms the validity of the peak for the 38 SNe in Fig. 4. FIGURE 10.
The W m distribution for a sample of 257 SNe with z > . W L is given by 1 − W m . FEB. 20, 2008: SO WHAT IS THE “FATE OF THE UNIVERSE” ?
By now, 10 years later, there are many experiments which have confirmed Dark Energy.We can however still ask: what is the "Fate of the Universe" ?The question is now in the form: is W L constant or does it vary with z ? Or alternatively:is w = −
1, namely are we dealing with Einstein’s cosmological constant ? Or can onemeasure deviations from − w is the equation of state parameter?We hope the Joint Dark Energy Mission (JDEM) will answer this question before thenext decade is up at DM2018. UMMARY OF THE STEPS IN THE SCP DISCOVERY OFEVIDENCE FOR DARK ENERGY • • • • W m value. • Sept. 24, 1997 W m = 0.2 to 0.3 for flat universe from peak in histogram, SCP groupmeeting based on 38 SNe. • Oct. 1997 hints for non zero W L from SCP and Hi-Z Team. • Oct. 23 to Dec. 14, 1997 public talks by Saul Perlmutter and Gerson Goldhabershowing evidence that for a flat Universe W m = 0.3 and hence W L = 0.7 based onthe peak in the W m histogram, Figs. 8, and by then on 40 SNe. • Jan. 8, 1998 SCP presentation by Saul Perlmutter at the AAS meeting in Wash-ington DC showing the W m , W L plane with best fit confidence level distributionsyielding flat universe values of W m = . W L = . • Feb. 18, 1998 Both groups talk at Marina del Rey DM98. Gerson and Saul showevidence for a non-zero W L and acceleration based on 42 SNe, followed by Filip-penko who showed evidence for a non-zero W L and acceleration based on 10 + 4 +2 SNe for the Hi-Z Team. ACKNOWLEDGEMENTS
I wish to thank Kyle Barbary, Alex Gude, David Rubin, Tony Spadafora and Nao Suzukifor help with this manuscript.This work has been supported in part by the Director, Office of Science, Office ofHigh Energy Physics, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.
REFERENCES
1. S. A. Colgate, E. P. Moore, and R. Carlson, PASP , 565 (1975).2. C. Pennypacker, and R. A. Muller, Measurement of the gravitational mass density (1988), a Proposalto the National Science Foundation.3. H. J. M. Newberg, Measuring q0 using supernovae at Z approximately 0.2 , Ph.D. thesis, CaliforniaUniv., Berkeley. Lawrence Berkeley Lab. (1992).4. H. U. Norgaard-Nielsen, L. Hansen, H. E. Jorgensen, A. Aragon Salamanca, and R. S. Ellis, Nature , 523 (1989).5. M. Hamuy, et al., AJ , 2392 (1993).6. S. Perlmutter, and B. P. Schmidt, “Measuring Cosmology with Supernovae,” in
Supernovae andGamma-Ray Bursters , edited by K. Weiler, 2003, vol. 598 of
Lecture Notes in Physics, BerlinSpringer Verlag , p. 195.7. S. Perlmutter, et al., ApJ , 565 (1999).8. A. G. Riess, et al., AJ , 1009 (1998). 19. Y. P. Pskovskii,
Soviet Astronomy , 658 (1984).10. M. M. Phillips, ApJL , L105 (1993).11. S. Perlmutter, et al., and G. Goldhaber, et al. Presented at NATO Advanced Study Inst. on Thermonu-clear Supernovae, Aiguablava, Spain, 20 - 30 June 1995.12. P. Ruiz-Lapuente, R. Canal, and J. Isern, “Thermonuclear supernovae. Proceedings.,” in NATO ASICProc. 486: Thermonuclear Supernovae , 1997.13. S. Perlmutter, et al., ApJ , 565 (1997).14. G. Goldhaber, et al., ApJ , 359 (2001).15. A. G. Riess, W. H. Press, and R. P. Kirshner, ApJL , 17 (1995).16. A. G. Riess, W. H. Press, and R. P. Kirshner, ApJ , 88 (1996).17. B. Leibundgut,
Light curves of supernovae type, I. , Ph.D. thesis, Univ. Basel.137 (1988).18. B. Leibundgut, A&A , 1 (1990).19. M. Hamuy, M. M. Phillips, L. A. Wells, and J. Maza, PASP , 787 (1993).20. S. Perlmutter, et al., ApJL , L41 (1995).21. M. Hamuy, et al., AJ , 1 (1995).22. M. Hamuy, et al., AJ , 2391 (1996).23. A. Kim, A. Goobar, and S. Perlmutter, PASP , 190 (1996).24. P. Nugent, A. Kim, and S. Perlmutter, PASP , 803 (2002).25. A. Goobar, and S. Perlmutter, ApJ , 14 (1995).26. A. H. Guth, Physical Review D , 347 (1981).27. A. D. Linde, Physics Letters B , 389 (1982).28. A. Albrecht, and P. J. Steinhardt,
Physical Review Letters , 1220 (1982).29. E. Torbet, et al., ApJL , L79 (1999).30. S. Hanany, et al., ApJL , L5 (2000).31. P. de Bernardis, et al., Nature , 955 (2000).32. C. L. Bennett, et al., ApJS , 97 (2003).33. S. Perlmutter, et al., Nature , 51 (1998).34. P. M. Garnavich, et al., ApJL , L53 (1998).35. A. G. Riess, P. Nugent, A. V. Filippenko, R. P. Kirshner, and S. Perlmutter, ApJ , 935 (1998).36. D. Burstein, and C. Heiles, AJ , 1165 (1982).37. D. J. Schlegel, D. P. Finkbeiner, and M. Davis, ApJ , 525 (1998).38. S. Weinberg, Reviews of Modern Physics , 1 (1989).39. L. M. Krauss, and M. S. Turner, General Relativity and Gravitation , 1137 (1995).40. J. P. Ostriker, and P. J. Steinhardt, Nature , 749 (2008).43. N. A. Bahcall, R. Cen, R. Davé, J. P. Ostriker, and Q. Yu, ApJ , 1 (2000).44. D. J. Eisenstein, et al., ApJ , 560 (2005).45. D. Rubin, et al., ArXiv e-prints (2008).46. S. Perlmutter, et al.,
ArXiv Astrophysics e-prints (1998), poster presented at the American Astronom-ical Society meeting, Washington, D.C., January 8, 1998.47. S. Perlmutter, et al., “Cosmology From Type IA Supernovae: Measurements, Calibration Techniques,and Implications,” in
Bulletin of the American Astronomical Society , 1997, vol. 29, p. 1351.48. J. Glanz,
Science , 651 (1998).49. G. Goldhaber, and S. Perlmutter, Physics Reports , 325 (1998).50. A. V. Filippenko, and A. G. Riess, Physics Reports307