INNER WORKINGS

How fast is the universe expanding? Clashingmeasurements may point to new physicsAnil Ananthaswamy, Science Writer

The siren sounded on August 17, 2017. Astronomerspicked up a burst of gravitational waves from thecollision of two neutron stars (1), an event that DanielHolz had been dreaming about for more than a de-cade. “You write these papers, and it sounds likefantasy. The equations say this might happen, butit’s completely different when nature cooperatesand lets you make the measurement,” says Holz, anastrophysicist at the University of Chicago’s Kavli In-stitute for Cosmological Physics (KICP). “It’s justtoo good. It’s ridiculous. It’s embarrassing. But thereit is.”

The neutron star merger, named GW170817, gaveHolz and his colleagues an entirely new way to mea-sure how fast the universe is expanding. This methodcould settle a simmering dispute between the twoestablished ways of measuring expansion, and it couldmean rethinking the makeup of our universe—per-haps requiring new types of a subatomic particle orunexpected forms of dark matter or dark energy.

Climb the Cosmic LadderAstronomer Edwin Hubble discovered in the 1920sthat our universe is expanding. The expansion iscausing all galaxies to speed away from us, and therate at which a galaxy is receding is equal to its dis-tance times a number called the Hubble constant. Thedistance to the galaxy is calculated in megaparsecs,where 1 megaparsec is about 3.26 million light years,and the constant has units of kilometers per secondper megaparsec. Multiply the two and you get thespeed of the galaxy in kilometers per second.

So the Hubble constant in today’s universe, H0,which is also called the local expansion rate, can bedetermined by measuring the distances to and thespeeds of galaxies. The most direct measurement ofH0 uses a technique called the cosmic distance lad-der, pioneered by Hubble (2). In a multistep process,astronomers use what they call standard candles—particular types of stars and supernovae whose in-trinsic brightness, or luminosity, is revealed by the way

One way of measuring the Hubble constant (H0) uses the CMB, shown here as observed by the European SpaceAgency’s Planck satellite, which launched in 2009 andmeasured the CMB for about 4.5 years. But this technique yields adifferent value for the constant than does an approach that calculates H0 by measuring the distances to and the speedsof galaxies. Image © ESA and the Planck Collaboration.

Published under the PNAS license.

9810–9812 | PNAS | October 2, 2018 | vol. 115 | no. 40 www.pnas.org/cgi/doi/10.1073/pnas.1811473115

INNER

WO

RKIN

GS

Dow

nloa

ded

by g

uest

on

May

23,

202

0

their observed brightness changes with time. Thesestandard candles allow astronomers to bootstrap theirway out of the Milky Way.

For example, the first step on the ladder involvesfinding the distance to a standard candle in our galaxyby using its parallax, which is the difference in theposition of the star when viewed from two differentlocations on Earth. The distance and the observedbrightness let astronomers calculate the intrinsicbrightness or luminosity of the standard candle andrelate it to some other observable property that doesnot depend on its distance from us. The next step thenis to find the same standard candle in a galaxy outsidethe Milky Way and use it to get the distance tothe galaxy.

Once astronomers step on the last rung of theladder, they can determine the luminosity of a su-pernova located in a distant galaxy. From that, as-tronomers can calculate its distance and that of itshost galaxy.

“Once we know the luminosity, then we could,from any supernova, measure the expansion rate ofthe universe,” explains Daniel Scolnic, an astrophysi-cist at KICP and member of SH0ES, a project thatclimbs the distance ladder using the Hubble SpaceTelescope and ground-based telescopes in Chileand Hawaii.

To measure the expansion rate, or H0, astronomersneed one more piece of data: the amount by whichthe light from the supernova is shifted toward longerwavelengths. This redshift is a measure of the speed atwhich the supernova and its host galaxy are recedingfrom us (analogous to how an ambulance’s siren falls inpitch as the ambulance speeds away). The SH0ESteam put all this together and found H0 to be about73.24 kilometers per second per megaparsec (3).

The Other Shoe DropsBut another way of measuring H0 is at odds withSH0ES. This technique uses the cosmic microwavebackground (CMB). About 380,000 years after the BigBang, the universe cooled down enough for the freeelectrons and protons to combine and form the firstatoms of neutral hydrogen. Photons, which until thenhad been constantly interacting with free electronsand thus unable to travel far, were unshackled. Nowstretched to microwave wavelengths, they form theCMB. “This piece of physics is very well understood,”says Silvia Galli of the Institute of Astrophysics in Paris.“It works extremely well to describe the CMB.”

The European Space Agency’s Planck satellite,launched in 2009, measured the CMB for about4.5 years. The satellite looked for tiny deviations in thetemperature of the CMB photons from place to placein the sky. The deviations have their roots in thephysics of the early universe, and their angular scale inthe sky can be used to infer the local Hubble constant(4). Galli and her Planck colleagues calculate H0 to beabout 67.8. It has proved impossible to reconcile thiswith the SH0ES value of 73, says Galli.

Of course, both teams may have overlooked somesystematic errors in their measurements, and this

discrepancy or tension in the data may go away withfurther refinements. Nonetheless, the fact that thetension hasn’t evaporated despite two generations ofexperimentation is troubling, say researchers, becauseit could undermine some of our assumptions aboutthe universe. “We have seen a lot of tensions comeand go,” says cosmologist Miguel Zumalacarregui ofthe University of California, Berkeley. “This is the mostserious that I have seen.”

The Planck result depends on the standard model ofcosmology, which says that the universe is 68.3% darkenergy (the energy of the vacuum of space–time), 26.8%dark matter (the unseen matter whose gravitational in-fluence can be detected in the motions of galaxies andgalaxy clusters), and 4.9% normal matter. Althoughthese parameters are tightly constrained by the CMBdata, the standard model also assumes that the amountof dark matter hasn’t changed. What if it has?

Torsten Bringmann of the University of Oslo inNorway and his colleagues showed earlier this yearthat if a tiny percentage of the universe’s dark matterhas decayed into undetectable radiation, that would

bump up the estimates of H0 from the Planck data,bringing it in line with that of SH0ES (5). Favored darkmatter candidates do not decay, so this would requiresome new piece of fundamental physics. For example,to accommodate decaying dark matter, physicistswould have to introduce a new fundamental force thatallows dark matter particles to decay.

Another solution could be to add a particle.Among that 4.9% normal matter are three knowntypes of subatomic particles called neutrinos. If an as-yet-undetected type of neutrino exists, this would in-crease the Planck value of H0. However, all existingempirical data from cosmology and particle physicsare consistent with three types (6).

The most intriguing change to the standard modelwould involve dark energy, a concept astronomersintroduced in the late 1990s to explain the accelerat-ing expansion of the universe.

In the standard model, the amount of dark energyper unit volume of the cosmos doesn’t change withtime, but in other models it is dynamic. One class oftheories attributes acceleration to the changing natureof gravity as the universe evolves, thanks to a hy-pothesized field called the Galileon (7). These theoriesboost the value of the CMB-estimated H0 to bring itcloser to distance-ladder measurements.

The Third WayWhat’s needed is an entirely new way to measureH0—and that is where GW170817 comes in. “Thistension is staying and both sides are digging in. Thereare no signs at all of anyone moving,” says Scolnic.

“This tension is staying and both sides are digging in.There are no signs at all of anyone moving.”

—Daniel Scolnic

Ananthaswamy PNAS | October 2, 2018 | vol. 115 | no. 40 | 9811

Dow

nloa

ded

by g

uest

on

May

23,

202

0

“Everyone is extremely excited about what this gravi-tational wave measurement will be able to say.”Importantly, it will be an alternate, independentmeasurement of the local value of the Hubbleconstant, Scolnic notes.

GW170817 was the outcome of two neutron stars,each a dense remnant of a past supernova explosion,that had been orbiting each other closely. They spiraledinward, radiating ripples in the fabric of space–timeknown as gravitational waves that were picked up onEarth by the Laser Interferometer Gravitational-WaveObservatory (LIGO). The neutron stars eventuallycollided in a massive explosion, generating a finalburst of gravitational waves and gamma rays.

Holz and colleagues found the distance toGW170817 by tracking the rate of change of the fre-quency of the gravitational waves and their amplitude.The changing frequency as the neutron stars orbiteach other faster and faster gives astronomers an in-dication of the mass of the binary star system and,therefore, the absolute amplitude of the waves.Comparing the absolute amplitude with the observedamplitude gives the distance.

Because gravitational waves are more like waves ofsound than light, Holz calls such events standard si-rens. Measuring H0 with a standard siren involvesphysics completely different from what is used by ei-ther Planck or SH0ES—it’s simple general relativitywith no assumptions. “It’s very clean in that sense,”says Holz. That means it can potentially be an arbiterbetween the CMB and supernova approaches.

The data from GW170817 give a distance to it ofabout 40 megaparsecs. By combining that with theredshift of its host galaxy, the LIGO team ends up with

a value for H0 of about 70 kilometers per second permegaparsec. That is right in the middle of the Planckand SH0ES values and with error bars large enough toaccommodate either (8).

So this lone siren does not settle the dispute. But ithas narrowed the exotic physics options. In mostGalileon models and in some other theories of darkenergy gravitational waves don’t travel at the speedof light. But telescopes saw the light from the colli-sion arrive at almost exactly the same time as LIGOsaw the gravitational waves. Based on this observa-tion, Zumalacarregui and his colleagues, as well asother teams, ruled out such models of dark energy (9).

Soon the method should be able to do muchbetter. Holz says that once LIGO comes back online inJanuary 2019 after an upgrade, it should detect manymore standard sirens. The previous run of LIGO wassensitive to neutron–star mergers out to a distance ofabout 80 megaparsecs. The upgraded LIGO will seeout to 120 megaparsecs, encompassing more thanthree times the volume. Finding more standard sirenswill help Holz and others calculate H0 with increasingprecision. That could support Planck, in which case theSH0ES team is probably doing something wrong, orsupport SH0ES, in which case Planck may have toreexamine its error bars, or—“by far the most excit-ing” option, according to Scolnic—the discrepancy isreal, and suspicions fall squarely on the assumptions inthe standard model.

The consequences are hard to overstate. Such ascenario would suggest that “there is physics beyondthe standard model,” says Scolnic. “Everything is onthe table.”

1 Abbott BP, et al.; LIGO Scientific Collaboration and Virgo Collaboration (2017) GW170817: Observation of gravitational waves from abinary neutron star inspiral. Phys Rev Lett 119:161101–161118.

2 Hubble E (1929) A relation between distance and radial velocity among extra-galactic nebulae. Proc Natl Acad Sci USA 15:168–173.3 Riess AG, et al. (2016) A 2.4% determination of the local value of the Hubble constant. Astrophys J 826:56.4 Ade PAR, et al. (2016) Planck 2015 results XIII. Cosmological parameters. Astron Astrophys 594:1–63.5 Bringmann T, et al. (2018) Converting non-relativistic dark matter to radiation. ArXiv:1803.03644.6 Verde L (2015) Neutrino properties from cosmology. J Phys Conf Ser 598:12010–12017.7 Nicolis A, Rattazzi R, Trincherini E (2009) Galileon as a local modification of gravity. Phys Rev D 79:64036-1–64036-21.8 Abbott BP, et al.; LIGO Scientific Collaboration and The Virgo Collaboration; 1M2H Collaboration; Dark Energy Camera GW-EMCollaboration and the DES Collaboration; DLT40 Collaboration; Las Cumbres Observatory Collaboration; VINROUGE Collaboration;MASTER Collaboration (2017) A gravitational-wave standard siren measurement of the Hubble constant. Nature 551:85–88.

9 Ezquiaga JM, Zumalacarregui M (2017) Dark energy after GW170817: Dead ends and the road ahead. Phys Rev Lett119:251304–251306.

9812 | www.pnas.org/cgi/doi/10.1073/pnas.1811473115 Ananthaswamy

Dow

nloa

ded

by g

uest

on

May

23,

202

0