Viewpoint: Can we test inflationary expansion of the early universe?

  • Arthur Kosowsky, Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260, USA
Physics 3, 103
A set of proposed relations among observable quantities may allow strong tests of whether a rapid expansion of the very early universe produced the seeds of the large-scale structure we see today.
Credit: (Left) Adapted from Takahashi et al. [10]; (Right) Brian Keating, UCSD
Figure 1: (Left) Design drawing of the BICEP instrument that measures variations in the polarization of the cosmic microwave background at a level of one part in ten million. The 25-cm aperture telescope, with an angular resolution of around a half degree at a microwave frequency of 150GHz, is designed to limit spurious polarization signals from the instrument at this stringent level. BICEP uses 100 polarization-sensitive bolometers cooled to a fraction of a degree Kelvin to detect the polarized radiation. (Right) Photo of the Background Imaging of Cosmic Extragalactic Polarization (BICEP) experiment, which has given the current best limits on the tensor-scalar ratio r from measuring microwave background polarization alone [12]. Such experiments are working towards measurements of r with precision of 0.01 that are necessary for strong tests of the bootstrap relations proposed by Boyle and Steinhardt. The image shows the Dark Sector Lab at the South Pole, atop which sits BICEP behind the metal ground shield.(Left) Design drawing of the BICEP instrument that measures variations in the polarization of the cosmic microwave background at a level of one part in ten million. The 25-cm aperture telescope, with an angular resolution of around a half degree at a... Show more

Inflation in the early universe is a period of exponential expansion that stretched a small causally-connected patch of the universe by a factor of at least e60 into a size large enough to encompass the visible universe today. At the same time this expansion drives the spatial geometry of the expanding region to flatness (for pedagogical reviews, see Refs. [1,2]). Such a period would explain why the universe today appears very nearly spatially flat on average (even though the dynamics of the expanding universe generically drive it away from flatness). It would also explain why the microwave background today has almost the same temperature in all sky directions (even though in the standard cosmology, most points of origin for the microwave radiation we see today were never within each others’ causal horizon, meaning they were never in causal contact). But is it possible to test whether such an inflationary episode actually occurred? In a paper in Physical Review Letters, Latham Boyle of the Canadian Institute for Theoretical Astrophysics in Toronto and Paul Steinhardt of Princeton University propose connections between observable quantities as a check of inflation [3].

The “flatness” and “horizon” problems are intractable enough in the standard cosmology that most cosmologists believe inflation occurred, even though we have no fundamental physical theory that provides a mechanism to drive inflation. If general relativity is correct, accelerating expansion occurs when p<-ρ/(3c2), where p is the mean pressure and ρ the mean energy density of the universe. Energy density is positive by definition, but pressure can be negative in principle (think of a collection of masses connected by springs stretched from their equilibrium). Cosmologists often invoke a classical scalar field φ, dubbed the “inflaton” field, with some potential V(φ) as a model for the contents of the universe during inflation. The condition for inflation would be satisfied if the potential energy of the scalar field dominates over its kinetic energy. But we do not know what scalar field the inflaton might be, or the energy scale of the field’s potential energy, or indeed if fundamental classical scalar fields even exist in nature.

Inflation is compelling for another reason as well: during a period of exponential expansion, quantum fluctuations in both the energy density and in the gravitational field will inevitably be stretched by the expansion into small-amplitude classical fluctuations on scales larger than the horizon (see Ref. [4] for a review). Any causal influences must propagate at a velocity limited by the speed of light. If the wavelength of a virtual quantum fluctuation gets stretched to a length greater than ct, with t the age of the universe, then the fluctuation’s spatially separated parts cannot communicate, which prevents it from disappearing, and it becomes frozen as a classical perturbation whose wavelength continues to stretch with the expansion of the universe. These fluctuations can be precisely those we see today in the temperature fluctuations of the microwave background; the inflation-generated density fluctuations grow via gravitational instability into the observed large-scale distribution of galaxies, while the gravitational field fluctuations become a stochastic background of very-long-wavelength gravitational waves. The perturbations we can probe today with cosmological observations would have been generated during inflation when the universe was around e60 times smaller than it was at the end of inflation. If inflation occurred, it must have started when the universe was less than 10-12 seconds old, corresponding to the electroweak energy scale of 100GeV, and may very well have begun at an age smaller than 10-32 seconds, corresponding to the scale at which grand unification of the strong and electroweak forces is conjectured to occur (see Chapter 3 of Ref. [5] for a classic discussion of the thermal history of the universe).

The density perturbations reflected in the observed temperature fluctuations of the cosmic microwave background and the large-scale distribution of galaxies are consistent with a power law as a function of scale, P(k)=Askns, with observables of the amplitude As and power law index ns, where k is the wave number corresponding to a given Fourier mode of the primordial density perturbations. We also consider the possibility that the power law index itself can vary with scale, a quantity defined as αs=dns/dlnk [6]. The gravitational wave perturbations (or “tensor” perturbations of the metric in general relativity) can likewise be described as a power law with power spectrum P(k)=Atknt+1 (with, unfortunately, a different historical convention for the power law index!). The precise definition of amplitudes depends on the scale at which they are defined. Assuming that both power spectra are given by these forms, the set of observables we can hope to probe is As, ns, αs, At, and nt, with the first two of these being well measured already. (A variation with scale of the tensor power law is almost surely unobservable.) It has long been appreciated that in simple models of inflation driven by a single scalar field, the two power spectra describing the density and gravitational-wave perturbations are not independent. Expanding expressions for measurable quantities in series of so-called “slow-roll” parameters (formed from the expansion rate and its derivatives during inflation) gives a relation between r=At/As, the ratio of density to tensor perturbation amplitudes, and nt [7]. Unfortunately, obtaining precise measurements of nt seems unlikely.

Boyle and Steinhardt [3] employ a simple alternate strategy: expand the Hubble parameter (which quantifies the expansion rate of the universe) during inflation, H(φ), as a Taylor expansion in the value of the inflaton field φ around the field value generating the perturbation scales of interest, φ* [as in Eq. (2) of their paper]. (Note that, while convenient to parameterize the evolution of H in terms of the effects of some effective inflaton field ϕ, this is not necessary and the results do not depend on inflation being driven by an actual fundamental scalar field.) Then, by truncating this expansion after two, three, or four terms and requiring that the resulting functional form for H(φ) be valid until the end of inflation, Boyle and Steinhardt have generated “bootstrap tests”—sets of numerical relations between the observables r,ns, and αs. Current microwave background temperature power spectrum measurements by, for example, the WMAP satellite combined with the higher-resolution ACT experiment, give ns=0.962±0.013 for a model with αs=0, and ns=1.032±0.039 and αs=-0.034±0.018 for models with both ns and αs as free parameters [8].

The tensor-scalar ratio r can be probed by looking for the “curl” mode in the gradient-curl decomposition of the polarization of the microwave background [9]. So far, WMAP plus ACT measurements give an upper limit r<0.25 at 95% confidence level, based mostly on microwave background temperature fluctuations. Further improvement requires polarization measurements; upcoming experiments (Fig. 1) will likely have the sensitivity and systematic error control to detect a tensor amplitude as small as r=0.01 (provided that contaminating polarized emission at microwave wavelengths from our own galaxy can be understood well enough) [10].

The largest observational hurdle for decisive tests of the full set of bootstrap relations may be measuring αs to sufficient precision; note that the first bootstrap test predicts αs=-5.3×10-4. Future measurements of 21- cm radiation from neutral hydrogen prior to the reionization of the universe by the first stars and galaxies can, in principle, probe αs at the level of 10-4 [11], although this exciting prospect depends partly on control of systematic errors that are not well understood at present.

The bootstrap tests presented here involve only quantities that are potentially observable with current or upcoming experiments. In the absence of inflation, there is no reason why r,ns, and αs should have any particular relationships between them. If the relationships presented in this paper are observed, this is strong evidence in favor of an inflationary origin for primordial perturbations, and constitutes a direct probe of physics at an energy scale far beyond the reach of any particle accelerator. If they are not satisfied, inflation may still be the leading contender for explaining the flatness and horizon problems, but the evolution of H during inflation is then not determined by the first few terms in a Taylor expansion around the field value φ*, corresponding to the scales which we can directly probe observationally. In the words of Boyle and Steinhardt, this “sad circumstance” would mean that the evidence in favor of inflation occurring in the early universe would likely remain circumstantial.


  1. A. R. Liddle, in 1998 ICTP Summer School in High Energy Physics and Cosmology, edited by A. Masiero, G. Senjanovic, and A. Smirnov (World Scientific, Singapore, 1999)[Amazon][WorldCat]
  2. W. Kinney, arxiv:0902.1529
  3. L. Boyle and P. J. Steinhardt, Phys. Rev. Lett. 105, 241301 (2010)
  4. J. E. Lidsey et al., Rev. Mod. Phys. 69, 373 (1997)
  5. E. W. Kolb and M. S. Turner, The Early Universe (Westview Press, Boulder, Colorado, 1994)[Amazon][WorldCat]
  6. A. Kosowsky and M. S. Turner, Phys. Rev. D 52, 1739 (1995)
  7. A. A. Starobinsky, Sov. Astron. Lett. 11, 133 (1985); E. J. Copeland, E. W. Kolb, A. R. Liddle, and J. E. Lidsey, Phys. Rev. D 49, 1840 (1994)
  8. J. Dunkley et al., arxiv:1009.0866
  9. M. Kamionkowski, A. Kosowsky, and A. Stebbins, Phys. Rev. Lett. 78,2058 (1997); U. Seljak and M. Zaldarriaga, 78, 2054 (1997)
  10. Y. D. Takahashi et al., Astrophys. J. 711, 1141 (2010)
  11. Y. Mao et al., Phys. Rev. D 78, 023529 (2008); M. Tegmark and M. Zaldarriaga, 82, 103501 (2010)
  12. H. C. Chiang et al., Astrophys. J. 711, 1123 (2010)

About the Author

Image of Arthur Kosowsky

Arthur Kosowsky is Professor of Physics and Astronomy at the University of Pittsburgh, and Associate Director of the Pittsburgh Particle Physics, Astrophysics, and Cosmology Center (PITT PACC). His research focuses on cosmology, gravitation, and high-energy physics. He is known particularly for work on the cosmic microwave background radiation, and he is a member of the Atacama Cosmology Telescope and the Simons Observatory collaborations to map the microwave sky.

Read PDF

Subject Areas

Particles and FieldsAstrophysicsCosmology

Related Articles

Focus: Hard and Soft Bounces Explain Asteroid’s Surface Structure

Focus: Hard and Soft Bounces Explain Asteroid’s Surface Structure

Experiments and computer simulations show that the segregation of small and large rocks on an asteroid’s surface can arise from the way particles hitting the surface collide with the rocks already present. Read More »

Viewpoint: Spinning Gluons in the Proton
Particles and Fields

Viewpoint: Spinning Gluons in the Proton

Computer simulations indicate that about 50% of the proton’s spin comes from the spin of the gluons that bind its quark constituents. Read More »

Synopsis: Neutrino Flashes from Exploding Stars

Synopsis: Neutrino Flashes from Exploding Stars

Calculations indicate that neutrino emission from a supernova could be detected on Earth, possibly revealing how the star explodes. Read More »

More Articles