You are on page 1of 9

Using Cosmological Data to

Constrain the Dark Energy Equation of State

Vanessa M. D’Amico REU program, College of William and Mary


Josh Erlich College of William and Mary, Dept. of Physics

1 August 2008

Abstract

anks to improved technology in recent years, astrophysical observations such as


those of the cosmic microwave background (CMB), type Ia supernovae, cluster
surveys, and galaxy rotation curves can be used to constrain cosmological parameters
with great precision. Type Ia supernovae are especially useful for placing limits on
dark energy in the universe because of their role as standard candles. Using data from
the Supernova Legacy Survey (SNLS) of high-redshift supernovae as well as data
gathered by the Wilkinson Microwave Anisotropy Probe (WMAP), we impose
constraints on the current value and investigate the possible evolution of the dark
energy equation of state.

Supported by NSF REU grant PHY-0755262

1 Introduction
Cosmology is the study of the large-scale structure of the universe, its formation, and its
evolution. e study of cosmology tells us where we have been, where we are today, and
where we are going based on the current physical properties of the universe. Over the past
few decades, cosmological observations have led us to believe that we live in an expanding,
flat universe. e expansion of the universe can be measured in terms of the Hubble
constant, H0, which expresses the rate of recession of an object as a function of its distance.
If the universe is to stop expanding at some point, the ratio of its total energy density to a
specific critical energy density must be 1. is ratio is referred to as Ω and is comprised of
two components: matter (both baryonic and dark), Ωm, and dark energy, ΩΛ. According to
the currently accepted concordance model, Ωm~ 0.3, ΩΛ~ 0.7 and H0~ 70 km/s/Mpc. Our
constraints on these values will be discussed in Section 5. e concordance model suggests
a so-called Λ-CDM cosmology. Λ stands for dark energy while CDM specifies the
presence of cold dark matter only.
Type Ia supernovae are excellent probes of cosmology because of their role as
standard candles. All type Ia supernovae originate from a binary star system in which a
white dwarf star accretes material from a larger counterpart. Once the white dwarf reaches
the Chandrasekhar limit of about 1.4 solar masses, electron degeneracy pressure can no
longer support it. High density and pressure in the star's core allow carbon-oxygen fusion
to occur in a runaway process and the star explodes into a supernova. Since this is a
standardized process, peak luminosity is the same for all type Ia supernovae and their
distances can be calculated very accurately. is is why they are often called “standard
candles” in astronomy.
It is important to note that it is the presence of dark energy that causes the universe
to undergo accelerated expansion. e brightness-distance relation is therefore a function
of a specified cosmology. In addition, light from distant supernovae is redshifted because it
travels through expanding space. e Doppler shift of light is also a function of a specified
cosmology and indicates how quickly a supernova is receding. For these reasons, type Ia
supernovae are very useful tools with which to investigate dark energy and other
cosmological factors.
Another source of cosmological data is the cosmic microwave background (CMB).
Immediately after inflation, the universe underwent a reheating process. At this time, it was
nothing but a hot plasma of charged particles. It wasn't until about 300,000 years later that
the temperature cooled enough for atoms to form, a period known as recombination. is
was the first time in the history of the universe that light could travel freely. e CMB we
detect today is microwave radiation left over from this primordial light. Projects like the
Wilkinson Microwave Anisotropy Probe (WMAP) map small temperature fluctuations in
the otherwise homogeneous 2.7 K radiation. ese anisotropies are due to quantum
mechanical density fluctuations magnified during inflation. Slightly more dense regions are
hotter than the background today, while slightly less dense regions are cooler. e power
spectrum of these fluctuations is determined by the cosmology in which they were created.
e CMB can therefore be used to tell us a lot about the properties of the universe.

2 eory
In order to effectively speak about the universe on any kind of large scale, it is necessary to
introduce the Robertson-Walker metric [1]

dr 2
ds 2=−c 2 dt 2  R20 a t 2  1−kr r 2 2 r 2 sin 2  d 2 
2 (1)

e RW metric describes the spacetime interval over which events occur in homogeneous
and isotropic space. Here, k indicates whether the universe is open, closed or flat, with
values of -1, +1 or 0, respectively. e dimensionless scale factor a refers to the expansion of
space and is related to the redshift z of an object (which replaces time in cosmology due to
the finite speed of light) by

1
a= (2)
1 z

e RW metric allows us to write Einstein's equations of motion in terms of two different


equations involving a(t). e first is the conservation equation [1]
̇i 3 i  pi  H 0=0 (3)

is equation relies on the assumption that the universe and all its components act as
perfect fluids, with energy densities ρ and pressures p. e relationship between fluid
density and pressure is expressed by the equation of state

pi =w i (4)

is equation represents a linearized approximation of the actual behavior of matter and
energy in the universe. e equation of state parameter w is of great importance to
cosmology because specific values of w correspond to different sets of circumstances in the
universe. For instance, w=-2/3 corresponds to the presence of domain walls, while w=-1/3
suggests the existence of cosmic strings. e cosmological constant w=-1 seems to be
favored by current data; however, if w < -1, the universe will accelerate exponentially at
some point, leading to a “Big Rip”. One way out of this problem is to introduce the
possibility that w varies with time, and will tend toward w=-1. A time-varying w can be
constrained by adding a time-dependent expression w(z) to Eq. 4, which would be defined


w z=w 0 1z  (5)

is is only a power law approximation and we assume that ֗ẇ/w is small compared to H0.
If α 1, this is a good estimation and w can be replaced by w(z) in Eqs. 4, 11 and 14. is
adiabatic approximation allows us to ignore the dw/dz and dz/dt components of Einstein's
equations. However, if α is of order 1 or larger, the power law estimate is no longer valid
and the behavior of a time-dependent w must be considered more rigorously.
In addition to the conservation equation, the second solution to Einstein's equations
is the Friedmann equation [2]

2
ȧ 8 G  k
H 02=  =  − 2 (6)
a 3 3 a0

where the subscript 0 represents the value of each variable at a=1 (today). e dimensionless
energy density parameters are defined as follows [2]:

G  −k (7)
m=8   ,  =
2 0 2
, k = 2 2
3H0 3H0 a0 H 0

Eq. 6 can now be reduced to

1=m k (8)

Because data indicates that we live in a flat universe, Ωk=0, and Eq. 8 becomes

1=m  (9)
Taking into account Eqs. 2 and 4, integrating Eq. 3 with respect to time gives

31wi 
i = 0i 1 z (10)

Eq. 6 then becomes

H  z=H 0 1z  1 i i 1z 


2 2 2 13w i
−1 (11)

Since photons are massless particles and therefore follow null geodesics, from Eq. 1:

2 2 2
c dt dr
2=
2 2
(12)
R0 a t  1−kr

Using the Friedmann equation to relate a to H (which relies on z), we can take the square
root of both sides and integrate to yield

rs zs
dr c dz
∫ =∫ (13)
0  1−kr 2
0 R0 H  z

where rs is the cosmological proper distance of an astronomical object at redshift zs. Using
Eq. 11 and solving for rs yields

zs
c dz
r s z s = ∫ (14)
R0 H 0 0

1 z  1 i i 1z 
13wi
−1

for a flat universe. e values of Ω and w will vary if the matter and energy characteristics
of the universe are changed; therefore, rs is a function of a specified cosmology.
is theoretical framework is of great importance to our project because the
astronomical objects we analyze are supernovae. We define the luminosity distance of a
supernova as the effective distance between it and the telescope observing it. is effective
distance is a function of the flux, or the amount of light that reaches the telescope. It takes
the form [1]

d L=1 z s r s  z s  R0 (15)

e observed magnitude, or brightness, of a supernova is a function of dL and is expressed


by
H0dL
m=m0 5 log 10   (16)
c

Eq. 16 expresses a theoretical magnitude that can be taken with Eqs. 14 and 15 to
constrain the values of Ω and w, which is the focus of this project.
3 Technique
Data for this project was taken from the Supernova Legacy Survey (SNLS) first year
results. e survey successfully imaged the light curves of 71 type 1a supernovae at redshift
z < 1 and used spectroscopic analysis to determine their redshifts. e initial observed
magnitude m*B of each was then corrected for brighter-slower and brighter-bluer relations
[3]. We use the corrected magnitude µB in our analysis. e relation between µ and z for
the SNLS first year data is shown in Figure 1.

Figure 1: Magnitude µ and redshift z of 71 high-redshift


supernovae as measured by the SNLS.

e probability that our theoretical model sufficiently explains the observed data is
determined by performing a χ2 statistical analysis, in which both theoretical measurements
and observed measurements are weighed together with errors as follows [4]:

2
ni −v i 
 =i
2 (17)
 i2

for measured values ni, theoretical values vi, and errors σi. e specific χ2 definition used by
the SNLS is [3]:
2
H d
B−m0 5 log 10 0 L 
c (18)
2 =  i 2
i

e value of χ2 is then minimized over all variables and significance fits are generated based
on the function ∆χ2 = χ2 -χ2min at each data point.

4 Results
As it stands, the theory alone seems to fit observations well. Adjusting the parameter m0 in
Eq. 16 to the curve of the SNLS data and assuming approximate values for w and Ωm yields
the fit shown in Figure 2.
To find the best fit values for the theory, we created confidence plots using the χ2
statistical analysis. ese plots are shown in Figures 3-6.
Figure 7 shows an approximate overlay the WMAP data together with the SNLS
data. e WMAP data used included baryon acoustic oscillations (BAO). Although the
contours of the WMAP confidence plot were not perfect ellipses, we approximated them as
such and used the central ellipse to perform an approximate χ2 analysis [5]. e combination
of WMAP and SNLS confidence plots provided a global fit that more rigorously
constrained the values of w and Ωm.

Figure 2: Sample theoretical fit (w=-1, Ωm=0.3) superimposed


on SNLS data. e parameter m0 affects the y-intercept
of the curve and was fit to the data as m0 = 43.146.
w w

Ωm ΩΛ
Figure 3: Confidence plot assuming Figure 4: Confidence plot assuming
time-independent w at the 68% (orange), 90% time-independent w at the 68% (orange), 90%
(purple) and 95% (yellow) confidence levels. (purple) and 95% (yellow) confidence levels.
Best fit is (Ωm, w) = (0.342, -1.329). Best fit is (ΩΛ, w) = (0.658, -1.329).

α α

w0 Ωm
Figure 5: Confidence plot assuming Figure 6: Confidence plot assuming
time-dependent w at the 68% (orange), 90% time-dependent w at the 68% (orange), 90%
(purple) and 95% (yellow) confidence levels. (purple) and 95% (yellow) confidence levels.
Best fit is (w0, α) = (-0.621, 2.215). Best fit is (Ωm, α) = (0.579, 2.215).
w

ΩΛ
Figure 7: Confidence plot assuming time-independent w.
SNLS data is shown in blue and represents 68%, 90% and 95% confidence
levels. WMAP data is approximate and shown in yellow at the 68% and 95%
confidence levels [5]. Global fit is shown in green and represents 68%, 90%, and 95%
confidence levels. Best fit is (ΩΛ, w) = (0.732, -1.106).

5 Conclusions and Future Work


In this project, we relied on a theoretical framework in which observations of supernovae
depend heavily on a specified cosmology. We used this theory as well as observational data
from the SNLS to probe the characteristics of Ωm, ΩΛ and w. rough a χ2 statistical
analysis, we were able to constrain these parameters around specific best fit values. We
constrained these values for both time-independent and time-dependent forms of w, the
dark energy equation of state parameter. A time-independent w yields best fit values of
(Ωm, ΩΛ, w) = (0.342, 0.658, -1.329). When combined with WMAP+BAO data, SNLS
data constrains these parameters more closely, indicating a best fit of (ΩΛ, w) = (0.732,
-1.106). Constraints on cosmological parameters assuming a time-dependent equation of
state parameter provide a best fit of (Ωm, w0, α) = (0.579, -0.621, 2.215).
Although we were successful in constraining a time-dependent w and its redshift
parameter α, our results indicated large values (α >1) for the latter. is suggests that the
approximation w(z) we used in place of w(t) was not as accurate as we had hoped. One
possibility for future study would be to analyze the relation between w(t) and w(z) more
closely and produce a more accurate expression for a time-varying equation of state
parameter.
In addition, the contour plot we borrowed from WMAP was only an approximation
to the actual WMAP+BAO results. It would be worthwhile to either recreate this plot with
greater accuracy or even to investigate alternate sources of data that could be combined in a
similar way with the SNLS observations. One possibility that we considered was to use
observations of galaxy clusters to produce a second global fit with the SNLS data and
constrain Ωm, ΩΛ and w further.

Acknowledgements
I would like to thank my mentor Professor Josh Erlich for sharing his knowledge of
cosmology and for his encouragement and guidance throughout this process. I would also
like to thank Professor Kossler and the REU program at William & Mary for giving me
the chance to participate this summer.

References
[1] J. Erlich and C. Grojean, Supernovae as a Probe of Particle Physics and Cosmology,
Physical Review D, 65 (2002) [hep-ph/0111335].

[2] P.B. Pal, Determination of Cosmological Parameters: An Introduction for Nonspecialists,


Pramana 54, 79 (2000) [hep-ph/9906447].

[3] P. Astier et al., e Supernova Legacy Survey: Measurement of Omega_M, Omega_Lambda


and w from the First Year Data Set, 447, 31 (2006) [astro-ph/0510447v1].

[4] G. Cowan, Statistics, Physical Review D, 66, 232 (2001)

[5] E. Komatsu et al., Five-Year Wilkinson Microwave Anisotropy Probe (WMAP)


Observations: Cosmological Interpretation, arXiv:0803.0547v1 [astro-ph]

You might also like