1982ApJ. . .2 63L. . .

IP

The Astrophysical Journal, 263:L1-L5, 1982 December 1

© 1982. The American Astronomical Society. All rights reserved. Printed in U.S.A.

LARGE-SCALE BACKGROUND TEMPERATURE AND MASS FLUCTUATIONS

DUE TO SCALE-INVARIANT PRIMEVAL PERTURBATIONS

P. J. E. Peebles
Joseph Henry Laboratories, Physics Department, Princeton University
Received 1982 July 2; accepted 1982 August 13

ABSTRACT
The large-scale anisotropy of the microwave background and the large-scale fluctuations in the
mass distribution are discussed under the assumptions that the universe is dominated by very
massive, weakly interacting particles and that the primeval density fluctuations were adiabatic with
the scale-invariant spectrum P oc wavenumber. This model yields a characteristic mass comparable
to that of a large galaxy independent of the particle mass, mx, if mx > 1 keV. The expected
background temperature fluctuations are well below present observational limits.
Subject headings: cosmic background radiation — cosmology — galaxies: formation

I. INTRODUCTION ~ 104 to make nonlinear density fluctuations form by

It is useful to consider which scenarios for the nature the present epoch is obtained if Û ~ 1 (Silk and Wilson
and evolution of the mass distribution in the universe 1981) because density fluctuations on scales greater than
can fit the observations without undue contrivance. We the matter-radiation Jeans length, \x, grow before de-
may hope that as the observations improve the list of coupling. However, Xx is large so it makes the mass
candidates will narrow, and that this process may in autocorrelation function unacceptably broad (Silk and
time help guide us to a fundamental theory of the origin Wilson 1981; Peebles 1981a). Press and Vishniac (1980)
of structure in the universe. The picture discussed here is emphasized that there is no hope for the development of
motivated by the argument that, if the initial conditions appreciable density fluctuations on scales smaller than
for conventional classical cosmology were set at some Ac-
exceedingly high redshift (perhaps the Planck time or The problem is relieved if the universe is dominated
the grand unified theory epoch), and if the initial condi- by massive, weakly interacting particles because density
tions did not involve exceedingly large or small num- fluctuations on small scales can grow before decoupling.
bers, then the cosmological density parameter ought to This effect has been widely discussed in the case that the
be ß = 1, and, assuming adiabatic density perturba- weakly interacting particle mass, mx, is some tens of
tions, the power spectrum P ought to be proportional to electron volts (e.g., Doroshkevich et al. 1981 and refer-
the wavenumber k (e.g., Hawking 1982). Density ences therein). If ß ~ 1, the mass coherence length is
fluctuations on scales greater than the horizon are de- broad but perhaps not unacceptable (Peebles 1982 a).
fined as in Peebles (1980, § 91, hereafter LSS). If P cc The case mx — 1 keV is discussed by Bond, Szalay, and
perturbations to the geometry diverge only as log k so Turner (1982) and Blumenthal, Pagels, and Primack
the cutoffs can be at very large and small k and the (1982). I discuss here a particularly simple and perhaps
spectrum can be truly scale invariant (Harrison 1970; important limiting case, mx> \ keV. The main results
Peebles and Yu 1970, § Via; Zerdovich 1972). Density are the spectrum of mass fluctuations, which seems quite
fluctuations appearing on the horizon have a fixed value, reasonable for the production of galaxies and clusters of
SM/M ~ 10-4, and, as this number is not greatly dif- galaxies, the statistical character of the background tem-
ferent from unity, we might imagine it is fixed by perature fluctuations, and the expected size of the mass
fundamental physics. density anticorrelation at large separations.
Observational constraints include the measurements
of background temperature fluctuations, ÔT/T < 1 X
10“4 (Boughn, Cheng, and Wilkinson 1981; Melchiorri
et al. 1981) and the estimate of the coherence length of II. CALCULATION
the galaxy distribution. They tell us P oc A: is unaccept- I assume zero cosmological constant and ß = 1, the
able in the usual cosmology with baryons, electrons, mass being mainly in weakly interacting particles, mass
radiation, and massless neutrinos. The observations of mx. Following Davis et al. (1981), I take the particle
8T/T imply e < 1 X 10“4. The wanted growth factor distribution in phase space at u « c in the absence of
LI

© American Astronomical Society • Provided by the NASA Astrophysics Data System

1982ApJ. . .2 63L. . .IP

L2 PEEBLES Vol. 263

perturbations to be proximated by the formulae

91= (expmxvc/kTx ± 1) \ (1) P = Ak(\ + ak + ßk2) 2

,

meaning the particles thermally decoupled from the a = 6(t//î)2 Mpc, ß = 2.65(t//i)4 Mpc2. (6)
radiation when the particles were relativistic. The
parameter Tx is adjusted to make Q = \. Hubble’s con- Here and below, k is expressed in units of radians per
stant and the present radiation temperature are written megaparsec at the present epoch.
as

//= lOOÄkms-1 Mpc"1, T — 2.7 r K. (2) c) Large-Scale Background Temperature

Fluctuations
If the primeval density fluctuations are adiabatic, the
a) Free Streaming angular distribution of the background temperature is
The smoothing of the mass distribution by the mo- (LSS eq. [93.35])
tions of the free particles has been widely discussed (e.g.,
Wasserman 1981). For either sign in equation (1), one T(0,<t>)/Tb- 1 = -^H22k~2Skcxpik x. (7)
finds that, to good accuracy, the rms peculiar velocity is
This equation assumes that the density fluctuations on
v = (\2)X/\2^hpY2m-^\\ + z), (3) scales of interest appear on the horizon when pr < px
and that the universe subsequently has behaved like the
where v c, pc is the Einstein-de Sitter density, and z is Einstein-de Sitter model. The term cc Y{” due to our
the redshift. The rms coordinate displacement is peculiar motion is not included. The vector x with
length 2 cH~1 points in the direction of observation. In
r= f‘0dtv(t)(\+z), (4) the linear perturbation approximation, the present mass
density is
expressed in units of proper displacement at the present
epoch. The dominant part of this integral comes be- P = Pfcl1 + S^ftCxp ik ■ r). (8)
tween the epochs zx, where v first drops appreciably The expansion of the background temperature in
below c, and zeq, where pr — px (the universe becomes spherical harmonics is
matter dominated). Taking account only of this interval,
one gets (for h — r— 1)
= rÈ(i + K^m),
r ~ (0.8 -h 0.3 log mx)m~A/3> Mpc, (5)
a? = (9)
where mx is expressed in units of kilo-electron volts. k
Thus, ii mx> \ keV, thermal motions are unimportant With the normalization
on the characteristic scale, r~3 h~x Mpc, of large
galaxies. This is seen also in Figure 2 of Peebles (1982 a),
Figure 1 of Bond, Szalay, and Turner (1982), and Figure 2-* (k2dkdQk/4v, (10)
1 of Blumenthal, Pagels, and Primack (1982). k J
we get
b) Spectrum of Mass Distribution /* 0
= (\a^\2)= nH4J dk k~2 \ Sk \2j,(kx) . (11)
When thermal motions are negligible, the free par- *'0
ticles can be described as a zero pressure ideal fluid. At
high redshift, the density fluctuations are supposed to be At large angular scales, | 0^ |2 oc k, and the integral is
adiabatic with spectrum P cc k. The part that appears
within the horizon while pr > px stops growing until 2
Jr^-j,(y) = [21(1+')]-'■ (12)
pr ~ px (because the radiation distribution is oscillating o y
as an acoustic wave). This tilts the mass spectrum to
P oc k~3 at large k (LSS § 92D). At small k, the It is convenient to express the normalization in terms of
spectrum keeps its primeval shape, P cc k. The results of the quadrupole moment a2. Then (eq. [6]),
a numerical integration of the shape of the spectrum
between these limiting cases (Peebles 1982 a) are ap- A = l2(a2)2/(7rH4), (13)

© American Astronomical Society • Provided by the NASA Astrophysics Data System

1982ApJ. . .2 63L. . .IP

No. 1, 1982 SCALE-INVARIANT PRIMEVAL PERTURBATIONS L3

and the angular power spectrum of the background is M/Mo
I010 I012 I014 I016 I018
2 2
(a,) = 6(a2) /[/(/+!)]. (14)

The 5-wave part a0 diverges, corresponding to the loga-

rithmic divergence in the primeval spectrum P cc k. The
/7-wave does not include the effect of our peculiar mo-
tion.
The autocorrelation function of the background tem-
perature after the /7-wave has been eliminated is (LSS
§46)

HOn) = (TxT2)/(T)2 - 1

- S(«/)2(2/+^^(cos Í12)/4W. (15)

/> 1
At 9U « 1 radian, this with equation (14) gives Fig. 1.—The rms fluctuation in the mass found within a sphere
of radius R. The top scale is the mean mass within the sphere. The
w(0) = (3/”)allog(®/6), (16) vertical normalization is adjusted to make 8M/M agree with the
observed xfluctuation, 8N/N, in the count of bright galaxies at
where © is on the order of the size of the field within R = 8 h~ Mpc. This curve is based on the assumptions that the
universe is dominated by very massive, weakly interacting particles
which w(9) is measured. and that the initial density fluctuations were adiabatic with power
spectrum P cc k.
d) Mass Fluctuations
A convenient measure is the rms fluctuation in the
mass found within a randomly placed sphere of radius pr > px. Shortward of the scale r fixed by mx (eq. [5]),
R. We have from equation (8) the power spectrum is truncated by thermal motions so
8M/M is independent of R. At large R, 8M/M oc R~2,
SM/M =32 S*(sin kR - kRcos kR)/{kR)\ which is the primeval spectrum. At 1 < Æ < 30 Mpc,
8M/M varies roughly as R~125.
(17) Figure 1 can be compared to the curve tz = 1 in
Figure 2 of Bond, Szalay, and Turner (1982). The short-
With equations (6) and (10), this gives the rms value wavelength cutoff in the latter curve results from the
assumption mx ~ 1 keV. The Bond et al. curve peaks at
8M M ~ 1012 M0 as does the Jeans mass found by
M Blumenthal, Pagels, and Primack (1982), both of which
agree with the position of the break in Figure 1. This is
k3 dk consistent with the fact that mx — 1 keV is roughly
V(R)f=j 0 equivalent to the limit of very large mx.
(1+ 6k + 2.65k2)2 By equations (18) and (19), the expected quadrupole
moment of the microwave background is
(sin kR — kR cos kR )2
X (18)
{kR)6 a2 = 3.5 X 1(T6. (20)

Figure 1 shows 8M/M for h — t = \ and the normali- For the case of baryon-dominated matter, Silk and
zation Wilson (1981) found û2 “ 4 X 10~5. The larger value is
the result of the much broader mass coherence length,
which increases the integral J3 — fr2 dr £ (Peebles
-^-{R = 8 Mpc) = 1, (19) 1981/7).
The Boughn, Cheng, and Wilkinson (1981) measure-
which agrees with the rms fluctuation, 8N/N, in the ments imply 02 ~ 3 X 10-4, but the more recent mea-
counts of bright galaxies at R = S h~l Mpc (LSS § 59; surements of Lubin (1982) and Fixsen (1982) suggest the
Davis and Peebles 1983). At Æ<0.1 Mpc, 8M/M extragalactic anisotropy may be appreciably less than
varies only slowly, as | log R |1/2, fluctuations with fixed that and hence perhaps not inconsistent with equation
variance per octave of R having been stored when (20).

© American Astronomical Society • Provided by the NASA Astrophysics Data System

1982ApJ. . .2 63L. . .IP

L4 PEEBLES Vol. 263

The expected temperature anisotropy at intermediate In the spirit of the scenario, we expect £2=1. The
angular scales is given by equation (16). The rms relative velocity data at r < 1 h~1 Mpc in the Center for
fluctuation in T smoothed over Q — 10° in a sample of Astrophysics (CfA) redshift sample yield £2 ~ 0.3, but
size © = 100° is there is an indication of a more broadly distributed mass
component, so £2 = 1 certainly is not improbable. The
87y:r = w1/2~5 x kt6. (21) p-wave part of the background would be almost entirely
due to our peculiar velocity, v. The expected rms veloc-
ity is ~ 1000 km s~1 if £2 ~ 1 (LSS § 74), which is larger
The mass autocorrelation function is than the observed background /?-wave but not to be
/•OO ruled out: as we are not in a particularly strongly
£(r) = (p(r)p(0))/(p)2 — 1 = J/ k2 dk Psmkr/kr. clustered spot, we might expect our velocity is no greater
o than the median, and as the degree of clustering is
highly variable, we might expect the distribution of v is
(22) broad and hence that the median is well below the rms
value.
At large r, this is dominated by the primeval spectrum,
Another interesting possibility is that the mass auto-
P = Ak, so that (LSS § 42)
correlation function, £(r), vanishes at r > 20 Ä-1 Mpc.
If so, the mass fluctuations at smaller /*, where £ > 0,
è—— (24/7r)al(c/Hr)4 = — (9.4 Mpc/r)4. produce long-range gravitational potential fluctuations
that lead to a value of a 2 that agrees with the present
(23) observations (Peebles 1981Z?, 1982&). This is an attrac-
tive coincidence. Of course, if a 2 were substantially
lowered, this picture would be much less attractive. By
III. DISCUSSION going from a flat spectrum to P cc k in the present
Figure 1 is based on the primeval spectrum P cc k scenario, we have introduced anticorrelated mass fluctu-
that has some theoretically attractive and perhaps im- ations, which strongly reduces a2 (Peebles 1981Z?, § V;
portant properties. It is normalized so 8M/M averaged Silk and Wilson 1981). If this model is correct, a mea-
over a sphere of radius 8 Mpc agrees with the rms surement of the background temperature autocorrela-
fluctuation, 8N/N, in the counts of bright galaxies. This tion function, w(0), at large 0 (eqs. [16] and [21]) will
radius is small enough that 8N/N is fairly well known require a considerable advance in the observations. The
yet large enough that the time evolution of 8M/M in computed mass anticorrelation amounts to £(20 Mpc) ~
the absence of nongravitational forces is accurately given — 0.05. As this number is based on the linear approxi-
by the linear perturbation calculation (Peebles and Groth mation, it is not highly accurate, but certainly £(r) must
1976). If galaxy formation were aided by the Ostriker- be negative at large r to reduce a2. There is a tentative
Cowie (1981) process, the primeval amplitude could be indication of anticorrelation at hr ~ 30 Mpc in the CfA
lower. However, 8M/M is almost flat at Ä < 0.1 Mpc, redshift catalog (Davis and Peebles 1983), but a firm
so if we wanted the first generation of objects to form at test awaits deeper samples.
redshift z > 10, we could not decrease a 2 by a factor of If the extragalactic part of a 2 were found to be
more than about 3 unless we were willing to go to appreciable, say a2>3 X 10~5, several interpretations
exceedingly small values of R. could be considered. It is doubtful that the discrepancy
This scenario yields a characteristic length on the with equation (20) could be due to an underestimate of £
order of 1 Mpc, which certainly is observationally inter- (by a factor —100 at /* — 5 h~x Mpc) for that would
esting. It is fixed by the horizon size when pr~px, conflict with the relative velocity data (Davis and
independent of the free particle mass mx if mx > 1 keV. Peebles 1983). The anisotropy could be inserted by
At R between 3 and 30 Mpc, 8M/M scales roughly as sources along the line of sight. If the source density were
M~x/1 so the similarity argument (LSS §§ 26 and 73) proportional to the mass density, the angular spectrum
suggests this could develop into the observed mass clus- in the present scenario would be al oc Z1/2 (compared to
tering hierarchy (Davis and Peebles 1983). alo: rx for the intrinsic part in eq. [14]). Hogan (1982)
All of this discussion has dealt with the second mo- has pointed out that one could assume that the luminos-
ment of the mass distribution. It remains to be seen ity per unit mass averaged over the mass coherence
whether the rather rapidly decreasing large-scale fluctua- length ~ 5 h~x Mpc is a random variable with broad
tions, 8M/M oc R~2, could produce large-scale features dispersion, so that the background temperature fluctua-
such as the cluster-cluster coherence length, rc ~ 25 h~l tions approximate white noise, al — constant. In these
Mpc (where £cc= 1; Hauser and Peebles 1973; Bahcall two cases, the rms fluctuation in the background tem-
and Soneira 1982). perature smoothed over the angle 0 scales as 8T/T~

© American Astronomical Society • Provided by the NASA Astrophysics Data System

1982ApJ. . .2 63L. . .IP

No. 1, 1982 SCALE-INVARIANT PRIMEVAL PERTURBATIONS L5

3/2 x
0~ and 0~ respectively. Thus, ii a2 can be detected, (mx < 1 keV), or else that the primeval spectrum was
it ought to be possible to check the spectrum. If these not adiabatic with P cc k.
two models can be ruled out, likely interpretations of a
large a2 will be that the growth of clustering at r < 20 This research was supported in part by the National
h-1 Mpc was hindered, perhaps by thermal motions Science Foundation.

REFERENCES
Bahcall, N. A., and Soneira, R. M. 1982, preprint. Lubin, P. M. 1982, paper presented at the 86th course of the
Blumenthal, G. R., Pagels, H., and Primack, J. R. 1982, Nature, in International School of Physics, Varenna.
press. Melchiorri, F., Melchiorri, B. O., Ceccarelli, C, and Pietranera, L.
Bond, J. R., Szalay, A. S., and Turner, M. S. 1982, Phys. Rev. 1981, Ap. J. {Letters), 250, LI.
Letters, 48, 1636. Ostriker, J. P., and Cowie, L. L. 1981, Ap. J. {Letters), 243, LI27.
Boughn, S. P., Cheng, E. S., and Wilkinson, D. T. 1981, Ap. J. Peebles, P. J. E. 1980, The Large-Scale Structure of the Universe
{Letters), 243, LI 13. (Princeton: Princeton University Press) (LSS).
Davis, M., Lecar, M., Pryor, C, and Witten, E. 1981, Ap. J., 250, . 1981a, Ap. J., 248, 885.
423. . 1981/7, Ap. J. {Letters), 243, LI 19.
Davis, M., and Peebles, P. J. E. 1983, Ap. J., submitted. . 1982a, Ap. J., 258, 415.
Dofoshkevich, A. G., Khlopov, M. Yu., Sunyaev, R. A., Szalay, . 19826, Ap. J., 259, 442.
A. S., and Zel’dovich, Ya. B. 1981, in Proceedings of the Tenth Peebles, P. J. E., and Groth, E. J. 1976, Astr. Ap., 53, 131.
Texas Symposium, Ann. NY Acad. Sei., 375, 32. Peebles, P. J. E., and Yu, J. T. 1970, Ap. J., 162, 815.
Fixsen, D. J. 1982, Ph.D. thesis, Princeton University. Press, W. H., and Vishniac, E. T. 1980, Ap. J., 236, 323.
Harrison, E. R. 1970, Phys. Rev., Dl, 2726. Silk, J., and Wilson, M. L. 1981, Ap. J. {Letters), 244, L37.
Hauser, M. G., and Peebles, P. J. E. 1973, Ap. J., 185, 757. Wasserman, I. 1981, Ap. J., 248, 1.
Hawking, S. W. 1982, preprint. Zel’dovich, Ya. B. 1972, M.N.R.A.S., 160, IP.
Hogan, C. J. 1982, Ap. J. {Letters), 256, L33.

P. J. E. Peebles: Joseph Henry Laboratories, Physics Department, Princeton University, Princeton, NJ 08544

© American Astronomical Society • Provided by the NASA Astrophysics Data System