Supernovae, Dark Energy, and the Accelerating UniverseSaul Perlmutter Citation: Phys. Today 56(4), 53 (2003); doi: 10.1063/1.1580050 View online: http://dx.doi.org/10.1063/1.1580050 View Table of Contents: http://www.physicstoday.org/resource/1/PHTOAD/v56/i4 Published by the American Institute of Physics. Additional resources for Physics TodayHomepage: http://www.physicstoday.org/ Information: http://www.physicstoday.org/about_us Daily Edition: http://www.physicstoday.org/daily_edition

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

For millennia, cosmology has been a theorist’s domain,where elegant theory was only occasionally endangered

by inconvenient facts. Early in the 20th century, AlbertEinstein gave us new conceptual tools to rigorously ad-dress the questions of the origins, evolution, and fate of theuniverse. In recent years, technology has developed to thepoint where these concepts from general relativity can besubstantiated and elaborated by measurements. For ex-ample, measurement of the remnant glow from the hot,dense beginnings of the expanding universe—the cosmicmicrowave background—is yielding increasingly detaileddata about the first half-million years and the overallgeometry of the cosmos (see the news story on page 21 ofthis issue).

The standard model of particle physics has also begunto play a prominent role in cosmology. The widely acceptedidea of exponential inflation in the immediate aftermathof the Big Bang was built on the predicted effect of certainputative particle fields and potentials on the cosmic ex-pansion. Measuring the history of cosmic expansion is noeasy task, but in recent years, a specific variety of super-novae, type Ia, has given us a first glimpse at that his-tory—and surprised us with an unexpected plot twist.

Searching for a standard candleIn principle, the expansion history of the cosmos can be de-termined quite easily, using as a “standard candle” any dis-tinguishable class of astronomical objects of known in-trinsic brightness that can be identified over a widedistance range. As the light from such beacons travels toEarth through an expanding universe, the cosmic expan-sion stretches not only the distances between galaxy clus-ters, but also the very wavelengths of the photons en route.By the time the light reaches us, the spectral wavelengthl has thus been redshifted by precisely the same incre-mental factor z � Dl/l by which the cosmos has beenstretched in the time interval since the light left its source.That time interval is the speed of light times the object’sdistance from Earth, which can be determined by com-paring its apparent brightness to a nearby standard of thesame class of astrophysical objects.

The recorded redshift and brightness of each such ob-ject thus provide a measurement of the total integrated ex-

pansion of the universe since the timethe light was emitted. A collection ofsuch measurements, over a sufficientrange of distances, would yield an en-tire historical record of the universe’sexpansion.

Conceptually, this scheme is a re-markably straightforward means to a

profound prize: an empirical account of the growth of ouruniverse. A spectroscopically distinguishable class of ob-jects with determinable intrinsic brightness would do thetrick. In Edwin Hubble’s discovery of the cosmic expansionin the 1920s, he used entire galaxies as standard candles.But galaxies, coming in many shapes and sizes, are diffi-cult to match against a standard brightness. They cangrow fainter with time, or brighter—by merging with othergalaxies. In the 1970s, it was suggested that the brightestmember of a galaxy cluster might serve as a reliable stan-dard candle. But in the end, all proposed distant galacticcandidates were too susceptible to evolutionary change.

As early as 1938, Walter Baade, working closely withFritz Zwicky, pointed out that supernovae were extremelypromising candidates for measuring the cosmic expansion.Their peak brightness seemed to be quite uniform, andthey were bright enough to be seen at extremely large dis-tances.1 In fact, a supernova can, for a few weeks, be asbright as an entire galaxy. Over the years, however, asmore and more supernovae were measured, it becameclear that they were a rather heterogeneous group with awide range of intrinsic peak brightnesses.

In the early 1980s, a new subclassification of super-novae emerged. Supernovae with no hydrogen features intheir spectra had previously all been classified simply astype I. Now this class was subdivided into types Ia and Ib,depending on the presence or absence of a silicon absorp-tion feature at 6150 Å in the supernova’s spectrum.2 Withthat minor improvement in typology, an amazing consis-tency among the type Ia supernovae became evident. Theirspectra matched feature-by-feature, as did their “lightcurves”—the plots of waxing and waning brightness in theweeks following a supernova explosion.3,4

The uniformity of the type Ia supernovae became evenmore striking when their spectra were studied in detail asthey brightened and then faded. First, the outermost partsof the exploding star emit a spectrum that’s the same forall typical type Ia supernovae, indicating the same ele-mental densities, excitation states, velocities, and so forth.Then, as the exploding ball of gas expands, the outermostlayers thin out and become transparent, letting us see thespectral signatures of conditions further inside. Eventu-ally, if we watch the entire time series of spectra, we getto see indicators that probe almost the entire explosiveevent. It is impressive that the type Ia supernovae exhibitso much uniformity down to this level of detail. Such a “su-pernova CAT-scan” can be difficult to interpret. But it’s

© 2003 American Institute of Physics, S-0031-9228-0304-030-4 April 2003 Physics Today 53

Saul Perlmutter is a senior scientist at the Lawrence BerkeleyNational Laboratory and leader of the Supernova CosmologyProject.

Using very distant supernovae as standard candles, onecan trace the history of cosmic expansion and try to findout what’s currently speeding it up.

Saul Perlmutter

Supernovae, Dark Energy, and theAccelerating Universe

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

clear that essentially the same physical processes are oc-curring in all of these explosions.

The detailed uniformity of the type Ia supernovae im-plies that they must have some common triggering mech-anism (see the box on page 56). Equally important, thisuniformity provides standard spectral and light-curvetemplates that offer the possibility of singling out those su-pernovae that deviate slightly from the norm. The complexnatural histories of galaxies had made them difficult tostandardize. With type Ia supernovae, however, we sawthe chance to avoid such problems. We could examine therich stream of observational data from each individual ex-plosion and match spectral and light-curve fingerprints torecognize those that had the same peak brightness.

Within a few years of their classification, type Ia su-pernovae began to bear out that expectation. First, DavidBranch and coworkers at the University of Oklahomashowed that the few type Ia outliers—those with peakbrightness significantly different from the norm—couldgenerally be identified and screened out.4 Either theirspectra or their “colors” (the ratios of intensity seenthrough two broadband filters) deviated from the tem-plates. The anomalously fainter supernovae were typicallyredder or found in highly inclined spiral galaxies (or both).Many of these were presumably dimmed by dust, whichabsorbs more blue light than red.

Soon after Branch’s work, Mark Phillips at the CerroTololo Interamerican Observatory in Chile showed thatthe type Ia brightness outliers also deviated from the tem-plate light curve—and in a very predictable way.5 The su-pernovae that faded faster than the norm were fainter attheir peak, and the slower ones were brighter (see figure1). In fact, one could use the light curve’s time scale to pre-dict peak brightness and thus slightly recalibrate each su-pernova. But the great majority of type Ia supernovae, asBranch’s group showed, passed the screening tests andwere, in fact, excellent standard candles that needed nosuch recalibration.6

Cosmological distancesWhen the veteran Swiss researcher Gustav Tammann andhis student Bruno Leibundgut first reported the amazinguniformity of type Ia supernovae, there was immediate in-terest in trying to use them to determine the Hubble con-stant, H0, which measures the present expansion rate ofthe cosmos. That could be done by finding and measuringa few type Ia supernovae just beyond the nearest clustersof galaxies, that is, explosions that occurred some 100 mil-lion years ago. An even more challenging goal lay in the

tantalizing prospect that we could find such standard-candle supernovae more than ten times farther away andthus sample the expansion of the universe several billionyears ago. Measurements using such remote supernovaemight actually show the expected slowing of the expansionrate by gravity. Because that deceleration rate would de-pend on the cosmic mean mass density rm, we would, in ef-fect, be weighing the universe.

If mass density is, as was generally supposed a decadeago, the primary energy constituent of the universe, thenthe measurement of the changing expansion rate wouldalso determine the curvature of space and tell us aboutwhether the cosmos is finite or infinite. Furthermore, thefate of the universe might be said to hang in the balance:If, for example, we measured a cosmic deceleration bigenough to imply a rm exceeding the “critical density” rc(roughly 10–29 gm/cm3), that would indicate that the uni-verse will someday stop expanding and collapse toward anapocalyptic “Big Crunch.”

All this sounded enticing: fundamental measure-ments made with a new distance standard bright enoughto be seen at cosmological distances. The problem was thattype Ia supernovae are a pain in the neck, to be avoided ifanything else would do. At the time, a brief catalog of rea-sons not to pursue cosmological measurement with type Iasupernovae might have begun like this: � They are rare. A typical galaxy hosts only a couple oftype Ia explosions per millennium.� They are random, giving no advance warning of whereto look. But the scarce observing time at the world’s largesttelescopes, the only tools powerful enough to measurethese most distant supernovae adequately, is allocated onthe basis of research proposals written more than sixmonths in advance. Even the few successful proposals aregranted only a few nights per semester. The possible oc-

54 April 2003 Physics Today http://www.physicstoday.org

–20

–20

–19

–19

–18

–18

–17

–17

–16

–16

–15

–15

AB

SO

LU

TE

MA

GN

ITU

DE

SC

AL

ED

MA

GN

ITU

DE

–20 0 20 40 60DAYS

a

b

Figure 1. Light curves of nearby, low-redshift type Ia super-novae measured by Mario Hamuy and coworkers.7 (a) Ab-

solute magnitude, an inverse logarithmic measure of intrinsicbrightness, is plotted against time (in the star’s rest frame) be-

fore and after peak brightness. The great majority (not all ofthem shown) fall neatly onto the yellow band. The figure

emphasizes the relatively rare outliers whose peak brightnessor duration differs noticeably from the norm. The nesting of

the light curves suggests that one can deduce the intrinsicbrightness of an outlier from its time scale. The brightest

supernovae wax and wane more slowly than the faintest. (b)Simply by stretching the time scales of individual light

curves to fit the norm, and then scaling the brightness by anamount determined by the required time stretch, one gets all

the type Ia light curves to match.5,8

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

currence of a chance su-pernova doesn’t makefor a compelling pro-posal.� They are fleeting.After exploding, theymust be discoveredpromptly and measuredmultiple times within afew weeks, or they willalready have passedthe peak brightnessthat is essential for cal-ibration. It’s too late tosubmit the observingproposal after you’vediscovered the super-nova.

This was a classiccatch-22. You couldn’tpreschedule telescopetime to identify a su-pernova’s type or followit up if you couldn’tguarantee one. But youcouldn’t prove a tech-nique for guaranteeingtype Ia supernova dis-coveries without pre-scheduling telescopetime to identify themspectroscopically.

The list of prob-lems didn’t stop there.The increasing red-shifting of supernovaspectra with distancemeans that the bright-ness of a very distantsupernova measuredthrough a given filter ishard to compare withthe brightness of amuch closer supernova measured through the same filter.(Astronomers call this the K-correction problem.) Dust ina supernova’s host galaxy can dim the explosion’s light.And there were doubts that the spectra of faint distant su-pernovae could be reliably identified as type Ia.

In fact, the results from the first search for very dis-tant type Ia supernovae were not encouraging. In the late1980s, a Danish team led by Hans Nørgaard-Nielsenfound only one type Ia supernova in two years of inten-sive observing, and that one was already several weekspast its peak.

A systematic solutionDaunting as these problems appeared, it seemed crazy tolet the logistics stand in the way, when the tools were athand for measuring such fundamental properties of theuniverse: its mass density, its large-scale curvature, andits fate. After all, we didn’t have to build anything nearlyas formidable as the gargantuan accelerators and detec-tors needed for particle physics. In a project that Carl Pen-nypacker and I began in Richard Muller’s group at theUniversity of California, Berkeley, just before the Danishteam’s 1988 supernova discovery, we started by building awide-field imager for the Anglo–Australian Observatory’s4-meter telescope. The imager would let us study thou-sands of distant galaxies in a night, upping the odds of a

supernova discovery. Contemporary computing and net-working advances just barely made possible the next-dayanalysis that would let us catch supernovae as they firstbrightened.

Finding our first supernova in 1992, we also found asolution to the K-correction problem by measuring the su-pernova in a correspondingly redshifted filter. By playingthis trick with two redshifted filter bands, one could alsoexpect to recognize dust absorption by its wavelength de-pendence. But we still hadn’t solved the catch-22 telescopescheduling problem. We couldn’t preschedule follow-up ob-servations of our first supernova, so we couldn’t obtain itsidentifying spectrum.

In retrospect, the solution we found seems obvious—though much effort was needed to implement it and proveit practical. By specific timing of the requested telescopeschedules (see figure 2), we could guarantee that our wide-field imager would harvest a batch of about a dozen freshlyexploded supernovae, all discovered on a pre-specified ob-serving date during the dark phase of the moon. (A brightmoon is an impediment to the follow-up observation.) Wefirst demonstrated this supernovae-on-demand methodol-ogy in 1994. From then on, proposals for time at majorground-based telescopes could specify discovery dates androughly how many supernovae would be found and fol-

http://www.physicstoday.org April 2003 Physics Today 55

50–100fields

Almost 1000galaxies perfield

Scheduled follow-up spectroscopyat Keck and ESO telescopes

Scheduled follow-upimaging byHubble, Cerro

Tololo, WIYN, IsaacNewton and ESO telescopes

Hubble

Bri

ghtn

ess

TIME

A dozen type 1a supernovaediscovered while stillbrightening

Figure 2. Observing strategy that guarantees batches of about a dozen fresh supernovae on de-mand. A first set of images of adjacent patches of sky containing tens of thousands of galaxies ismade just after a new moon, and then these patches are reimaged just before the next new moon.New bright spots pinpoint supernovae explosions since the first exposures. The timing ensuresthat the supernovae are discovered before or near peak brightness. Scientists can preschedule, formoonless nights just after the second field imaging, the spectral observations at the large tele-scopes in Hawaii and Chile needed to confirm supernova type. By searching through many galax-ies, we can guarantee a dozen or so new supernovae discovered on the second visit. That allowsadvance scheduling of time on the Hubble Space Telescope and other telescopes around theworld, to monitor the light curves as they fade over several months.

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

lowed up. This approach also made it possible to use theHubble Space Telescope for follow-up light-curve observa-tions, because we could specify in advance the one-square-degree patch of sky in which our wide-field imager wouldfind its catch of supernovae. Such specificity is a require-ment for advance scheduling of the HST. By now, theBerkeley team had grown to include some dozen collabo-rators around the world, and was called the SupernovaCosmology Project (SCP).

A community effortMeanwhile, the whole supernova community was makingprogress with the understanding of relatively nearby su-pernovae. Mario Hamuy and coworkers at Cerro Tololotook a major step forward by finding and studying manynearby (low-redshift) type Ia supernovae.7 The resultingbeautiful data set of 38 supernova light curves (someshown in figure 1) made it possible to check and improveon the results of Branch and Phillips, showing that typeIa peak brightness could be standardized.6,7

The new supernovae-on-demand techniques that per-mitted systematic study of distant supernovae and the im-proved understanding of brightness variations amongnearby type Ia’s spurred the community to redouble its ef-forts. A second collaboration, called the High-Z SupernovaSearch and led by Brian Schmidt of Australia’s MountStromlo Observatory, was formed at the end of 1994. Theteam included many veteran supernova experts. The tworival teams raced each other over the next few years—oc-casionally covering for each other with observations whenone of us had bad weather—as we all worked feverishly tofind and study the guaranteed on-demand batches of supernovae.

At the beginning of 1997, the SCP team presented theresults for our first seven high-redshift supernovae.8 Thesefirst results demonstrated the cosmological analysis tech-niques from beginning to end. They were suggestive of anexpansion slowing down at about the rate expected for thesimplest inflationary Big Bang models, but with error barstoo large to permit definite conclusions.

By the end of the year, the error bars began to tighten,as both groups now submitted papers with a few more su-pernovae, showing evidence for much less than the ex-pected slowing of the cosmic expansion.9–11 This was be-ginning to be a problem for the simplest inflationarymodels with a universe dominated by its mass content.

Finally, at the beginning of 1998, the two groups pre-sented the results shown in figure 3.12,13

What’s wrong with faint supernovae? The faintness—or distance—of the high-redshift super-novae in figure 3 was a dramatic surprise. In the simplest

56 April 2003 Physics Today http://www.physicstoday.org

26

24

22

20

18

16

140.01 0.02 0.04 0.1

0.2 0.4 0.6 1

OB

SE

RV

ED

MA

GN

ITU

DE

22

21

200.2 0.4 0.6 1.0

Acceleratinguniverse

Deceleratinguniverse

with vacuum

energ

y

without v

acuumenergy

Mas

s de

nsi

ty0

rc

Empt

y

REDSHIFT z

0.8 0.7 0.6 0.5LINEAR SCALE OF THE UNIVERSE RELATIVE TO TODAY

Supernova CosmologyProject

High-Z SupernovaSearch

Hamuy et al.

0.0001

0.001

0.01

0.1

1

RE

LA

TIV

E B

RIG

HT

NE

SS

Exploding White Dwarfs

Aplausible, though unconfirmed, scenario would explainhow all type Ia supernovae come to be so much alike,

given the varied range of stars they start from. A lightweightstar like the Sun uses up its nuclear fuel in 5 or 10 billionyears. It then shrinks to an Earth-sized ember, a white dwarf,with its mass (mostly carbon and oxygen) supported againstfurther collapse by electron degeneracy pressure. Then itbegins to quietly fade away.

But the story can have a more dramatic finale if the whitedwarf is in a close binary orbit with a large star that is stillactively burning its nuclear fuel. If conditions of proximityand relative mass are right, there will be a steady stream ofmaterial from the active star slowly accreting onto the whitedwarf. Over millions of years, the dwarf’s mass builds upuntil it reaches the critical mass (near the Chandrasekharlimit, about 1.4 solar masses) that triggers a runaway ther-monuclear explosion—a type Ia supernova.

This slow, relentless approach to a sudden cataclysmicconclusion at a characteristic mass erases most of the orig-inal differences among the progenitor stars. Thus the lightcurves (see figure 1) and spectra of all type Ia supernovaeare remarkably similar. The differences we do occasionallysee presumably reflect variations on the common theme—including differences, from one progenitor star to the next,of accretion and rotation rates, or different carbon-to-oxy-gen ratios.

Figure 3. Observed magnitudeversus redshift is plotted for

well-measured distant12,13 and(in the inset) nearby7 type Ia su-pernovae. For clarity, measure-ments at the same redshift are

combined. At redshifts beyondz = 0.1 (distances greater thanabout 109 light-years), the cos-

mological predictions (indi-cated by the curves) begin to

diverge, depending on the as-sumed cosmic densities of

mass and vacuum energy. Thered curves represent models

with zero vacuum energy andmass densities ranging from thecritical density rc down to zero(an empty cosmos). The best fit

(blue line) assumes a mass density of about rc /3 plus a

vacuum energy density twicethat large—implying an accel-

erating cosmic expansion.

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

cosmological models, the expansion history of the cosmosis determined entirely by its mass density. The greater thedensity, the more the expansion is slowed by gravity. Thus,in the past, a high-mass-density universe would have beenexpanding much faster than it does today. So one should-n’t have to look far back in time to especially distant (faint)supernovae to find a given integrated expansion (redshift).

Conversely, in a low-mass-density universe one wouldhave to look farther back. But there is a limit to how lowthe mean mass density could be. After all, we are here, andthe stars and galaxies are here. All that mass surely putsa lower limit on how far—that is, to what level of faint-ness—we must look to find a given redshift. The high-redshift supernovae in figure 3 are, however, fainter thanwould be expected even for an empty cosmos.

If these data are correct, the obvious implication isthat the simplest cosmological model must be too simple.The next simplest model might be one that Einstein en-tertained for a time. Believing the universe to be static, hetentatively introduced into the equations of general rela-tivity an expansionary term he called the “cosmologicalconstant” (L) that would compete against gravitational col-lapse. After Hubble’s discovery of the cosmic expansion,Einstein famously rejected L as his “greatest blunder.” Inlater years, L came to be identified with the zero-pointvacuum energy of all quantum fields.

It turns out that invoking a cosmological constant al-lows us to fit the supernova data quite well. (Perhaps therewas more insight in Einstein’s blunder than in the best ef-forts of ordinary mortals.) In 1995, my SCP colleague ArielGoobar and I had found that, with a sample of type Ia su-pernovae spread over a sufficiently wide range of dis-tances, it would be possible to separate out the competingeffects of the mean mass density and the vacuum-energydensity.14

The best fit to the 1998 supernova data (see figures 3and 4) implies that, in the present epoch, the vacuum en-ergy density rL is larger than the energy density attribut-able to mass (rmc2). Therefore, the cosmic expansion is nowaccelerating. If the universe has no large-scale curvature,

as the recent measurements of the cosmic microwave back-ground strongly indicate, we can say quantitatively thatabout 70% of the total energy density is vacuum energyand 30% is mass. In units of the critical density rc, oneusually writes this result as

WL � rL/rc � 0.7 and Wm � rm/rc � 0.3.

Why not a cosmological constant?The story might stop right here with a happy ending—acomplete physics model of the cosmic expansion—were itnot for a chorus of complaints from the particle theorists.The standard model of particle physics has no naturalplace for a vacuum energy density of the modest magni-tude required by the astrophysical data. The simplest es-timates would predict a vacuum energy 10120 times greater.(In supersymmetric models, it’s “only” 1055 times greater.)So enormous a L would have engendered an accelerationso rapid that stars and galaxies could never have formed.Therefore it has long been assumed that there must besome underlying symmetry that precisely cancels the vac-uum energy. Now, however, the supernova data appear torequire that such a cancellation would have to leave a re-mainder of about one part in 10120. That degree of fine tun-ing is most unappealing.

The cosmological constant model requires yet anotherfine tuning. In the cosmic expansion, mass density be-comes ever more dilute. Since the end of inflation, it hasfallen by very many orders of magnitude. But the vacuumenergy density rL, a property of empty space itself, staysconstant. It seems a remarkable and implausible coinci-dence that the mass density, just in the present epoch, iswithin a factor of 2 of the vacuum energy density.

Given these two fine-tuning coincidences, it seemslikely that the standard model is missing some funda-mental physics. Perhaps we need some new kind of accel-erating energy—a “dark energy” that, unlike L, is not con-stant. Borrowing from the example of the putative“inflaton” field that is thought to have triggered inflation,theorists are proposing dynamical scalar-field models andother even more exotic alternatives to a cosmological con-

http://www.physicstoday.org April 2003 Physics Today 57

LIN

EA

R S

CA

LE

OF

UN

IVE

RS

E R

EL

AT

IVE

TO

TO

DA

Y

1.5

1.0

0.5

0.0–20 –10 0 +10

BILLIONS OF YEARS FROM TODAY

RELATIVE BRIGHTNESS OF SUPERNOVAE

0.00

01

0.00

1

0.01 0.

11

0

0.5

11.5

2

3

RE

DS

HIF

Tz

Eternal expansion Even

tual collapse

}

}

. .. o

ral

way

sde

cele

rate

s

. .. o

ral

way

sde

cele

rate

s

Exp

an

a

s

en ac

i

decrates, th

ot

n f sir

ce

ele

ler tes

Exp

an

a

s

en ac

i

decrates, th

ot

n f sir

ce

ele

ler tes

Figure 4. The history of cosmic expansion, as measured by thehigh-redshift supernovae (the blackdata points), assuming flat cosmicgeometry. The scale factor R of theuniverse is taken to be 1 at pres-ent, so it equals 1/(1 + z). Thecurves in the blue shaded regionrepresent cosmological models inwhich the accelerating effect ofvacuum energy eventually over-comes the decelerating effect ofthe mass density. These curves as-sume vacuum energy densitiesranging from 0.95 rc (top curve)down to 0.4 rc. In the yellowshaded region, the curves repre-sent models in which the cosmicexpansion is always deceleratingdue to high mass density. They as-sume mass densities ranging (left toright) from 0.8 rc up to 1.4 rc. Infact, for the last two curves, the ex-pansion eventually halts and re-verses into a cosmic collapse.

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

stant, with the goal of solving the coincidence problems.(See the Reference Frame article by Michael Turner onpage 10 of this issue.)

The experimental physicist’s life, however, is domi-nated by more prosaic questions: “Where could my meas-urement be wrong, and how can I tell?” Crucial questionsof replicability were answered by the striking agreementbetween our results and those of the competing team, butthere remain the all-important questions of systematic un-certainties. Most of the two groups’ efforts have been de-voted to hunting down these systematics.15,16 Could thefaintness of the supernovae be due to intervening dust?The color measurements that would show color-dependentdimming for most types of dust indicate that dust is not amajor factor.12,13 Might the type Ia supernovae have beenintrinsically fainter in the distant past? Spectral compar-isons have, thus far, revealed no distinction between theexploding atmospheres of nearby and more distant super-novae.9,12

Another test of systematics is to look for even moredistant supernovae, from the time when the universe wasso much more dense that rm dominated over the dark en-ergy and was thus still slowing the cosmic expansion. Su-pernovae from that decelerating epoch should not get asfaint with increasing distance as they would if dust or in-trinsic evolutionary changes caused the dimming. The firstfew supernovae studied at redshifts beyond z = 1 have al-ready begun to constrain these systematic uncertainties.17

(See PHYSICS TODAY, June 2001, page 17.) By confirming the flat geometry of the cosmos, the re-

cent measurements of the cosmic microwave backgroundhave also contributed to confidence in the accelerating-uni-verse results. Without the extra degree of freedom providedby possible spatial curvature, one would have to invoke im-probably large systematic error to negate the supernova re-sults. And if we include the low rm estimates based on in-ventory studies of galaxy clusters, the Wm–WL parameterplane shows a reassuring overlap for the three independ-ent kinds of cosmological observations (see figure 5).

Pursuing the elusive dark energyThe dark energy evinced by the accelerating cosmic ex-pansion grants us almost no clues to its identity. Its tinydensity and its feeble interactions presumably precludeidentification in the laboratory. By construction, of course,it does affect the expansion rate of the universe, and dif-ferent dark-energy models imply different expansion ratesin different epochs. So we must hunt for the fingerprintsof dark energy in the fine details of the history of cosmicexpansion.

The wide-ranging theories of dark energy are often

characterized by their equation-of-state parameterw � p/r, the ratio of the dark energy’s pressure to its energy density. The deceleration (or acceleration) of an expanding universe, given by the general relativisticequation

R�� /R = –4/3pGr(1 + 3w),

depends on this ratio. Here R, the scale factor of the ex-panding universe, can be thought of as the mean distancebetween galaxy clusters not bound to each other. Thus theexpansion accelerates whenever w is more negative than–1/3, after one includes all matter, radiation, and dark-en-ergy components of the cosmic energy budget.

Each of the components has its own w: negligible fornonrelativistic matter, +1/3 for radiation and relativisticmatter, and –1 for L. That is, L exerts a peculiar negativepressure! General relativity also tells us that each compo-nent’s energy density falls like R–3(1 + w) as the cosmos ex-pands. Therefore, radiation’s contribution falls away first,so that nonrelativistic matter and dark energy now pre-dominate. Given that the dark-energy density is now abouttwice the mass density, the only constraint on dark-energymodels is that w must, at present, be more negative than–1/2 to make the cosmic expansion accelerate. However,most dark-energy alternatives to a cosmological constanthave a w that changes over time. If we can learn moreabout the history of cosmic expansion, we can hope to dis-criminate among theories of dark energy by better deter-mining w and its time dependence.

Unfortunately, the differences between the expansionhistories predicted by the current crop of dark-energy mod-els are extremely small. Distinguishing among them willrequire measurements an order of magnitude more accu-rate than those shown in figure 3, and extending twice as

58 April 2003 Physics Today http://www.physicstoday.org

No Big Bang

Supernovae

CMB

Clusters

0 1 2 3

3

2

1

0

–1

VA

CU

UM

EN

ER

GY

DE

NS

ITY

WL

MASS DENSITY Wm

Flat

Big B

ang

too

rece

nt

Cosmos expands forever

Recollapses eventually

Figure 5. In the cosmological parameter space of the nor-malized mass and vacuum energy densities Wm and WL,

three independent sets of observations—high-redshift super-novae, galaxy cluster inventories, and the cosmic microwave

background—converge nicely near Wm = 0.3 and WL = 0.7.The small yellow contour in this region indicates how well

we expect the proposed SNAP satellite experiment to furthernarrow down the parameters. The inflationary expectation

of a flat cosmos (Wm + WL = 1) is indicated by the black diagonal. The red curve separates an eternally

expanding cosmos from one that ends in a “Big Crunch.”

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

far back in time.There is no shortage of type Ia supernovae; one ex-

plodes somewhere in the sky every few seconds. In princi-ple, then, the job is simply to study a hundred times asmany supernovae as we have so far. That’s a difficult butnot prohibitive task, if we install dedicated wider-field im-agers and improved spectrographs on dedicated large tel-escopes. However, it’s not just a matter of improving thequantity of measurements. The quality must also take adramatic step forward, because the current measurementaccuracy is not limited simply by statistical errors. Evenwith the number of supernovae we already have in hand,our statistical uncertainties are already close to the sys-tematic uncertainties.

A new challenge The next generation of supernova projects has alreadybegun. Telescope scheduling committees have dramati-cally increased the time allotted them on the largest tele-scopes. With biweekly monitoring of patches of sky for sev-eral years on end at two 4-meter telescopes, it will bepossible to collect almost complete light curves for hun-dreds of 5-billion-year-old type Ia supernovae. Smaller tel-escopes will study the time-varying spectra of much closersupernovae. And imagers on the HST and the 8-m SubaruTelescope in Hawaii are now revealing handfuls of 10-bil-lion-year-old supernovae. A number of large new tele-scopes are devoting extensive observing programs to fol-low-up measurements of this plethora of supernovae. Atthe most extreme distances, only the Hubble telescope canjust barely follow the fading supernovae, redshifted intothe infrared. With this array of efforts, we may know, be-fore too long, whether the time-averaged behavior of thedark energy is consistent with a cosmological constant.

The still harder goal of the third generation of super-nova work, which also has already begun, is to look fortime variations in the dark energy. For this higher-preci-sion work, the systematic uncertainties must be reduceddramatically. The physical details of each individual su-pernova explosion must be pinned down with extensiveand exacting spectral and photometric monitoring. Inter-vening dust must be measured with wavelength coverageextending into the near-infrared. Host galaxies must beclassified to control for environmental effects on the typeIa standard candle. And we will have to study enough su-pernovae in each redshift range to take account of possi-ble gravitational lensing by foreground galaxies that canbrighten or dim a supernova.

These very exacting requirements have pushed us towork above the atmosphere and design a new orbiting op-tical and near-infrared telescope called SNAP (Super-Nova/Acceleration Probe). With a 2-meter mirror, a half-billion-pixel imager, and a high-throughput spectrograph,this space mission can accomplish the unprecedentedsuite of measurements required for measuring thousandsof supernovae with adequately constrained systematicuncertainties.18

We live in an unusual time, perhaps the first goldenage of empirical cosmology. With advancing technology, wehave begun to make philosophically significant measure-ments. These measurements have already brought sur-prises. Not only is the universe accelerating, but it appar-ently consists primarily of mysterious substances. We’vealready had to revise our simplest cosmological models.Dark energy has now been added to the already perplex-ing question of dark matter. One is tempted to speculatethat these ingredients are add-ons, like the Ptolemaicepicycles, to preserve an incomplete theory. With the nextdecade’s new experiments, exploiting not only distant

April 2003 Physics Today 59

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

supernovae, but also the cosmic microwave background,gravitational lensing of galaxies, and other cosmologicalobservations, we have the prospect of taking the next steptoward that “Aha!” moment when a new theory makessense of the current puzzles.

In references 12 and 13, I have listed in full the members onthe High-Z Supernova Search and Supernova CosmologyProject teams, because each of these scientists should be rec-ognized for important contributions to the discoveries de-scribed here. It has been both an honor and a pleasure towork closely with my SCP colleagues, who dedicated them-selves to this work for years on end, providing creativity andleadership.

References1. W. Baade, Astrophys. J. 88, 285 (1938); C. Kowal, Astron. J.

73, 1021 (1968).2. J. C. Wheeler, R. Levreault, Astrophys. J. Lett. 294, 17

(1985); A. Uomoto, R. Kirshner, Astron. Astrophys. 149, L7(1985); N. Panagia, in Supernovae as Distance Indicators, N.Bartel, ed., Springer-Verlag, Berlin (1985); R. Harkness, J. C.Wheeler, in Supernovae, A. Petschek, ed., Springer-Verlag,New York (1990).

3. B. Leibundgut, PhD Thesis, University of Basel (1988); G.Tammann, B. Leibundgut, Astron. Astrophys. 236, 9 (1990).

4. D. Branch, G. Tammann, Annu. Rev. Astron. Astrophys. 30,359 (1992); D. Miller, D. Branch, Astron. J. 100, 530 (1990);D. Branch, D. Miller, Astrophys. J. Lett. 495, L5 (1993).

5. M. Phillips, Astrophys. J. Lett. 413, 105, (1993); A. Riess, W. Press, R. Kirshner, Astrophys. J. 473 88, (1996). See alsoref. 8.

6. D. Branch, A. Fisher, P. Nugent, Astron. J. 106, 2383 (1993);T. Vaughan, D. Branch, D. Miller, S. Perlmutter, Astrophys.J. 439, 558 (1995).

7. M. Hamuy et al., Astron. J. 106, 2392 (1993); and 109, 1(1995).

8. S. Perlmutter et al. (Supernova Cosmology Project), Astro-phys. J. 483, 565 (1997).

9. S. Perlmutter et al. (Supernova Cosmology Project), Nature391, 51 (1998).

10. P. Garnavich et al. (High-Z Supernova Search), Astrophys. J.Lett. 493, 53 (1998).

11. B. Schmidt et al. (High-Z Supernova Search), Astrophys. J.507, 46 (1998).

12. A. Riess, A. Filippenko, P. Challis, A. Clocchiattia, A. Diercks,P. Garnavich, R. Gilliland, C. Hogan, S. Jha, R. Kirshner, B.Leibundgut, M. Phillips, D. Reiss, B. Schmidt, R. Schommer,R. Smith, J. Spyromilio, C. Stubbs, N. Suntzeff, J. Tonry(High-Z Supernova Search), Astron. J. 116, 1009 (1998).

13. S. Perlmutter, G. Aldering, G. Goldhaber, R. Knop, P. Nu-gent, P. Castro, S. Deustua, S. Fabbro, A. Goobar, D. Groom,I. Hook, A. Kim, M. Kim, J. Lee, N. Nunes, R. Pain, C. Pen-nypacker, R. Quimby, C. Lidman, R. Ellis, M. Irwin, R.McMahon, P. Ruiz-Lapuente, N. Walton, B. Schaefer, B.Boyle, A. Filippenko, T. Matheson, A. Fruchter, N. Panagia,H. Newberg, W. Couch (Supernova Cosmology Project), As-trophys. J. 517, 565, (1999).

14. A. Goobar, S. Perlmutter, Astrophys. J. 450, 14 (1995).15. S. Perlmutter, B. Schmidt, in Supernovae and Gamma Ray

Bursters, K. Weiler, ed., Springer-Verlag, New York (2003),and references therein. Available at http://arXiv.org/abs/astro-ph/0303428.

16. See the Web sites of the Supernova Cosmology Project,http://supernova.LBL.gov, and the High-Z Supernova Search,http://cfa-www.harvard.edu/cfa/oir/Research/supernova/HighZ.html.

17. See, for example, A. Riess et al., Astrophys. J., 560, 49 (2001).18. For more information on SNAP, see http://snap.lbl.gov. �

Downloaded 22 May 2013 to 128.135.12.127. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms