Ination, Quantum Cosmology

and the Anthropic Principle1

Andrei Linde

arXiv:hep-th/0211048v2 8 Nov 2002

Department of Physics, Stanford University, Stanford, CA 94305, USA

Contents
1 Introduction

2 Chaotic ination

3 Inationary quantum uctuations

4 Eternal chaotic ination

5 Baby universes

12

6 From the universe to the multiverse

15

7 Double universe model and the cosmological constant problem

18

8 Cosmological constant, dark energy, and the anthropic principle

21

9 Problem of calculating the probabilities

23

10 Does consciousness matter?

25

11 Why is mathematics so ecient?

27

12 Why quantum?

29

To appear in Science and Ultimate Reality: From Quantum to Cosmos, honoring John
Wheelers 90th birthday. J. D. Barrow, P.C.W. Davies, & C.L. Harper eds. Cambridge University
Press (2003)

Introduction

One of the main desires of physicists is to construct a theory that unambiguously

predicts the observed values for all parameters of all elementary particles. It is very
tempting to believe that the correct theory describing our world should be both
beautiful and unique.
However, most of the parameters of elementary particles look more like a collection of random numbers than a unique manifestation of some hidden harmony
of Nature. For example, the mass of the electron is 3 orders of magnitude smaller
than the mass of the proton, which is 2 orders of magnitude smaller than the mass
of the W-boson, which is 17 orders of magnitude smaller than the Planck mass Mp .
Meanwhile, it was pointed out long ago that a minor change (by a factor of two
or three) in the mass of the electron, the ne-structure constant e , the strong2
interaction constant s , or the gravitational constant G = Mp would lead to a
universe in which life as we know it could never have arisen. Adding or subtracting even a single spatial dimension of the same type as the usual three dimensions
would make planetary systems impossible. Indeed, in space-time with dimensionality d > 4, gravitational forces between distant bodies fall o faster than r 2 , and
in space-time with d < 4, the general theory of relativity tells us that such forces
are absent altogether. This rules out the existence of stable planetary systems for
d = 4. Furthermore, in order for life as we know it to exist, it is necessary that the
universe be suciently large, at, homogeneous, and isotropic. These facts, as well
as a number of other observations, lie at the foundation of the so-called anthropic
principle (Barrow and Tipler, 1986; Rozental, 1988; Rees, 2000) According to this
principle, we observe the universe to be as it is because only in such a universe
could observers like ourselves exist.
Until very recently, many scientists were ashamed of using the anthropic principle in their research. A typical attitude was expressed in the book The Early
Universe by Kolb and Turner: It is unclear to one of the authors how a concept as
lame as the anthropic idea was ever elevated to the status of a principle (Kolb,
1990).
This critical attitude is quite healthy. It is much better to nd a simple physical
resolution of the problem rather that speculate that we can live only in the universes
where the problem does not exist. There is always a risk that the anthropic principle
does not cure the problem, but acts like a painkiller.
On the other hand, this principle can help us to understand that some of the
most complicated and fundamental problems may become nearly trivial if one looks
at them from a dierent perspective. Instead of denying the anthropic principle
or uncritically embracing it, one should take a more patient approach and check
whether it is really helpful or not in each particular case.
2

There are two main versions of this principle: the weak anthropic principle
and the strong one. The weak anthropic principle simply says that if the universe
consists of dierent parts with dierent properties, we will live only in those parts
where our life is possible. This could seem rather trivial, but one may wonder
whether these dierent parts of the universe are really available. If it is not so, any
discussion of altering the mass of the electron, the ne structure constant, and so
forth is perfectly meaningless.
The strong anthropic principle says that the universe must be created in such
a way as to make our existence possible. At rst glance, this principle must be
faulty, because mankind, having appeared 1010 years after the basic features of
our universe were laid down, could in no way inuence either the structure of the
universe or the properties of the elementary particles within it.
Scientists often associated the anthropic principle with the idea that the universe
was created many times until the nal success. It was not clear who did it and why
was it necessary to make the universe suitable for our existence. Moreover, it would
be much simpler to create proper conditions for our existence in a small vicinity of
a solar system rather than in the whole universe. Why would one need to work so
hard?
Fortunately, most of the problems associated with the anthropic principle were
resolved (Linde, 1983a,1984b,1986a) soon after the invention of inationary cosmology. Therefore we will remember here the basic principles of inationary theory.

Chaotic ination

Inationary theory was formulated in many dierent ways, starting with the models
based on quantum gravity (Starobinsky, 1980) and on the theory of high temperature phase transitions with supercooling and exponential expansion in the false
vacuum state (Guth, 1981; Linde, 1982a; Albrecht and Steinhardt, 1982). However,
with the introduction of the chaotic ination scenario (Linde, 1983b) it was realized that the basic principles of ination actually are very simple, and no thermal
equilibrium, supercooling, and expansion in the false vacuum is required.
To explain the main idea of chaotic ination, let us consider the simplest model
2
of a scalar eld with a mass m and with the potential energy density V () = m 2 ,
2
see Fig. 1. Since this function has a minimum at = 0, one may expect that the
scalar eld should oscillate near this minimum. This is indeed the case if the
universe does not expand. However, one can show that in a rapidly expanding
universe the scalar eld moves down very slowly, as a ball in a viscous liquid,
viscosity being proportional to the speed of expansion.
There are two equations which describe evolution of a homogeneous scalar eld
3

in our model, the eld equation

+ 3H = m2 ,

(1)

and the Einstein equation

H2 +

k
8
=
2
a2
3Mp

1 2
+ V ()
2

(2)

Here H = a/a is the Hubble parameter in the universe with a scale factor a(t) (the

size of the universe), k = 1, 0, 1 for an open, at or closed universe respectively,

2
Mp is the Planck mass, Mp = G, where G is the gravitational constant. The
rst equation becomes similar to the equation of motion for a harmonic oscillator,

where instead of x(t) we have (t). The term 3H is similar to the term describing
friction in the equation for a harmonic oscillator.
If the scalar eld initially was large, the Hubble parameter H was large too,
according to the second equation. This means that the friction term was very large,
and therefore the scalar eld was moving very slowly, as a ball in a viscous liquid.
Therefore at this stage the energy density of the scalar eld, unlike the density of
ordinary matter, remained almost constant, and expansion of the universe continued
with a much greater speed than in the old cosmological theory. Due to the rapid
growth of the scale of the universe and a slow motion of the eld , soon after the
k

beginning of this regime one has 3H , H 2 a2 , 2 m2 2 , so the system of

equations can be simplied:
a

3 = m2 ,
(3)
a
H=

2m
=
a
Mp

.
3

(4)

The last equation shows that the size of the universe a(t) in this regime grows
approximately as eHt , where H = 2m .
Mp
3
This stage of exponentially rapid expansion of the universe is called ination. In
realistic versions of inationary theory its duration could be as short as 1035 seconds. When the eld becomes suciently small, viscosity becomes small, ination
ends, and the scalar eld begins to oscillate near the minimum of V (). As any
rapidly oscillating classical eld, it looses its energy by creating pairs of elementary
particles. These particles interact with each other and come to a state of thermal
equilibrium with some temperature T . From this time on, the corresponding part
of the universe can be described by the standard hot universe theory.
The main dierence between inationary theory and the old cosmology becomes
clear when one calculates the size of a typical inationary domain at the end of
ination. Investigation of this issue shows that even if the initial size of inationary
4

Planck density
A

Figure 1: Motion of the scalar eld in the theory with V () = m 2 . Several

2
dierent regimes are possible, depending on the value of the eld . If the potential
4
energy density of the eld is greater than the Planck density Mp 1094 g/cm3 ,
quantum uctuations of space-time are so strong that one cannot describe it in
usual terms. Such a state is called space-time foam. At a somewhat smaller energy
3
4
density (region A: mMp < V () < Mp ) quantum uctuations of space-time are
small, but quantum uctuations of the scalar eld may be large. Jumps of the
scalar eld due to quantum uctuations lead to a process of eternal self-reproduction
of inationary universe which we are going to discuss later. At even smaller values
2
3
of V () (region B: m2 Mp < V () < mMp ) uctuations of the eld are small; it
slowly moves down as a ball in a viscous liquid. Ination occurs both in the region
A and region B. Finally, near the minimum of V () (region C) the scalar eld
rapidly oscillates, creates pairs of elementary particles, and the universe becomes
hot.

universe was as small as the Plank size lP 1033 cm, after 1035 seconds of
12
ination the universe acquires a huge size of l 1010 cm. This makes our universe
almost exactly at and homogeneous on large scale because all inhomogeneities were
12
stretched by a factor of 1010 .
This number is model-dependent, but in all realistic models the size of the
universe after ination appears to be many orders of magnitude greater than the
size of the part of the universe which we can see now, l 1028 cm. This immediately
solves most of the problems of the old cosmological theory (Linde, 1990a).
Consider a universe which initially consisted of many domains with chaotically
distributed scalar eld (or if one considers dierent universes with dierent values
of the eld). Those domains where the scalar eld was too small never inated,
so they do not contribute much to the total volume of the universe. The main
contribution to the total volume of the universe will be given by those domains
which originally contained large scalar eld . Ination of such domains creates
huge homogeneous islands out of the initial chaos, each homogeneous domain being
much greater than the size of the observable part of the universe. That is why I
called this scenario chaotic ination.
There is a big dierence between this scenario and the old idea that the whole
universe was created at the same moment of time (Big Bang), in a nearly uniform
state with indenitely large temperature. In the new theory, the condition of uniformity and thermal equilibrium is no longer required. Each part of the universe
could have a singular beginning (see (Borde et al, 2001) for a recent discussion of
this issue). However, in the context of chaotic ination, this does not mean that
the universe as a whole had a single beginning. Dierent parts of the universe could
come to existence at dierent moments of time, and then grow up to the size much
greater than the total size of the universe. The existence of initial singularity (or
singularities) does not imply that the whole universe was created simultaneously in
a single Big Bang explosion. In other words, we cannot tell anymore that the whole
universe was born at some time t = 0 before which it did not exist. This conclusion
is valid for all versions of chaotic ination, even if one does not take into account
the process of self-reproduction of the universe discussed in Section 4.
The possibility that our homogeneous part of the universe emerged from the
chaotic state initial state has important implications for the anthropic principle.
Until now we have considered the simplest inationary model with only one scalar
eld. Realistic models of elementary particles involve many other scalar elds.
For example, according to the standard theory of electroweak interactions, masses
of all elementary particles depend on the value of the Higgs scalar eld in our
universe. This value is determined by the position of the minimum of the eective
potential V () for the eld . In the simplest models, the potential V () has only
one minimum. However, in general, the potential V () may have many dierent
minima. For example, in the simplest supersymmetric theory unifying weak, strong
6

and electromagnetic interactions, the eective potential has several dierent minima
of equal depth with respect to the two scalar elds, and . If the scalar elds
and fall to dierent minima in dierent parts of the universe (the process called
spontaneous symmetry breaking), the masses of elementary particles and the laws
describing their interactions will be dierent in these parts. Each of these parts
may become exponentially large because of ination. In some of these parts, there
will be no dierence between weak, strong and electromagnetic interactions, and
life of our type will be impossible there. Some other parts will be similar to the one
where we live now (Linde, 1983c).
This means that even if we will be able to nd the nal theory of everything,
we will be unable to uniquely determine properties of elementary particles in our
universe; the universe may consist of dierent exponentially large domains where
the properties of elementary particles may be dierent. This is an important step
towards the justication of the anthropic principle. A further step can be made if
one takes into account quantum uctuations produced during ination.

Inationary quantum uctuations

According to quantum eld theory, empty space is not entirely empty. It is lled
with quantum uctuations of all types of physical elds. The wavelengths of all
quantum uctuations of the scalar eld grow exponentially during ination. When
the wavelength of any particular uctuation becomes greater than H 1 , this uctuation stops oscillating, and its amplitude freezes at some nonzero value (x)

because of the large friction term 3H in the equation of motion of the eld . The
amplitude of this uctuation then remains almost unchanged for a very long time,
whereas its wavelength grows exponentially. Therefore, the appearance of such a
frozen uctuation is equivalent to the appearance of a classical eld (x) produced
from quantum uctuations.
Because the vacuum contains uctuations of all wavelengths, ination leads to
the continuous creation of new perturbations of the classical eld with wavelengths
greater than H 1 . An average amplitude of perturbations generated during a time
H
interval H 1 (in which the universe expands by a factor of e) is given by |(x)| 2
(Vilenkin and Ford, 1982; Linde, 1982c).
These quantum uctuations are responsible for galaxy formation (Mukhanov
and Chibisov, 1981; Hawking, 1982; Starobinsky, 1982; Guth and Pi, 1982; Bardeen
et al, 1983). But if the Hubble constant during ination is suciently large, quantum uctuations of the scalar elds may lead not only to formation of galaxies, but
also to the division of the universe into exponentially large domains with dierent
properties.

As an example, consider again the simplest supersymmetric theory unifying

weak, strong and electromagnetic interactions. Dierent minima of the eective
potential in this model are separated from each other by the distance 103 Mp .
The amplitude of quantum uctuations of the elds , and in the beginning of
chaotic ination can be as large as 101 Mp . This means that at the early stages of
ination the elds and could easily jump from one minimum of the potential
to another. Therefore even if initially these elds occupied the same minimum all
over the universe, after the stage of chaotic ination the universe becomes divided
into many exponentially large domains corresponding to all possible minima of the
eective potential (Linde, 1983c, 1984b).

Eternal chaotic ination

The process of the division of the universe into dierent parts becomes even easier
if one takes into account the process of self-reproduction of inationary domains.
The basic mechanism can be understood as follows. If quantum uctuations are
suciently large, they may locally increase the value of the potential energy of
the scalar eld in some parts of the universe. The probability of quantum jumps
leading to a local increase of the energy density can be very small, but the regions
where it happens start expanding much faster than their parent domains, and
quantum uctuations inside them lead to production of new inationary domains
which expand even faster. This surprising behavior leads to the process of selfreproduction of the universe.
This process is possible in the new ination scenario (Steinhardt, 1982; Linde,
1982a; Vilenkin, 1983). However, even though the possibility to use this result for
the justication of the anthropic principle was mentioned in (Linde, 1982a), this
observation did not attract much attention because the amplitude of the uctuations in new ination typically is smaller than 106 Mp . This is too small to probe
most of the vacuum states available in the theory. As a result, the existence of the
self-reproduction regime in the new ination scenario was basically forgotten; for
many years this eect was not studied or used in any way even by those who have
found it.
The situation changed dramatically when it was found that the self-reproduction
of the universe occurs not only in new ination but also in the chaotic ination
scenario (Linde, 1986a). In order to understand this eect, let us consider an
inationary domain of initial radius H 1 containing a suciently homogeneous
eld with initial value Mp . Equations (3), (4) tell us that during a typical
2
Mp
time interval t = H 1 the eld inside this domain will be reduced by = 4 .
H
Comparing this expression with the amplitude of quantum uctuations 2 =
8

one can easily see that for Mp Mp , one has || ||, i.e. the
2
m
motion of the eld due to its quantum uctuations is much more rapid than its
classical motion.
m ,
3Mp

During the typical time H 1 the size of the domain of initial size H 1 containing
the eld grows e times, its volume increases e3 20 times, and almost in a
half of this new volume the eld jumps up instead of falling down. Thus the total
volume of inationary domains containing the eld grows approximately 10
times. During the next time interval H 1 this process continues; the universe enters
an eternal process of self-reproduction. I called this process eternal ination.
In this scenario the scalar eld may wander for an indenitely long time at
the density approaching the Planck density. This induces quantum uctuations of
all other scalar eld, which may jump from one minimum of the potential energy
to another for an unlimited time. The amplitude of these quantum uctuations
can be extremely large, 101 Mp . As a result, quantum uctuations
generated during eternal chaotic ination can penetrate through any barriers, even
if they have Planckian height, and the universe after ination becomes divided into
indenitely large number of exponentially large domains containing matter in all
possible states corresponding to all possible mechanisms of spontaneous symmetry
breaking, i.e. to the dierent laws of the law-energy physics (Linde, 1986a; Linde
et al, 1994).
A rich spectrum of possibilities may appear during ination in Kaluza-Klein and
superstring theories, where an exponentially large variety of vacuum states and ways
of compactication is available for the original 10- or 11-dimensional space. The
type of compactication determines coupling constants, vacuum energy, symmetry
breaking, and nally, the eective dimensionality of the space we live in. As it was
shown in (Linde and Zelnikov, 1988), chaotic ination at a nearly Planckian density
may lead to a local change of the number of compactied dimensions; the universe
becomes divided into exponentially large parts with dierent dimensionality.
Sometimes one may have a continuous spectrum of various possibilities. For example, in the context of the Brans-Dicke theory, the eective gravitational constant
is a function of the Brans-Dicke eld, which also experienced uctuations during
ination. As a result, the universe after ination becomes divided into exponentially large parts with all possible values of the gravitational constant G and the
amplitude of density perturbations (Linde, 1990b; Garcia-Bellido et al 1994).

Ination may divide our universe into exponentially large domains with continuously varying baryon to photon ratio nB (Linde, 1985) and with galaxies having
n
vastly dierent properties (Linde, 1987b). Ination may also continuously change
the eective value of the vacuum energy (the cosmological constant ), which is a
pre-requisite for many attempts to nd an anthropic solution of the cosmological
constant problem (Linde, 1984b,1986b; Weinberg, 1987; Efstathiou, 1995; Vilenkin,
9

1995b; Martel et al, 1998; Garriga and Vilenkin, 2000,2001b,2002; Bludman and
Roos, 2002; Kallosh and Linde, 2002). Under these circumstances, the most diverse
sets of parameters of particle physics (masses, coupling constants, vacuum energy,
etc.) can appear after ination.
To illustrate the possible consequences of such theories in the context of inationary cosmology, we present here the results of computer simulations of evolution
of a system of two scalar elds during chaotic ination (Linde et al, 1994). The eld
is the inaton eld driving ination; it is shown by the height of the distribution
of the eld (x, y) in a two-dimensional slice of the universe. The eld determines
the type of spontaneous symmetry breaking which may occur in the theory. We
paint the surface black if this eld is in a state corresponding to one of the two
minima of its eective potential; we paint it white if it is in the second minimum
corresponding to a dierent type of symmetry breaking, and therefore to a dierent
set of laws of low-energy physics.
In the beginning of the process the whole inationary domain was black, and
the distribution of both elds was very homogeneous. Then the domain became
exponentially large and it became divided into exponentially large domains with
dierent properties, see Fig. 2. Each peak of the mountains corresponds to a
nearly Planckian density and can de interpreted as a beginning of a new Big Bang.
The laws of physics are rapidly changing there, but they become xed in the parts
of the universe where the eld becomes small. These parts correspond to valleys
in Fig. 2. Thus quantum uctuations of the scalar elds divide the universe into
exponentially large domains with dierent laws of low-energy physics, and with
dierent values of energy density.
As a result of quantum jumps of the scalar elds during eternal ination, the
universe becomes divided into innitely many exponentially large domains with
dierent laws of low-energy physics. Each of these domains is so large that for all
practical purposes it can be considered a separate universe: Its inhabitants will live
exponentially far away from its boundaries, so they will never know anything about
the existence of other universes with dierent properties.
If this scenario is correct, then physics alone cannot provide a complete explanation for all properties of our part of the universe. The same physical theory
may yield large parts of the universe that have diverse properties. According to
this scenario, we nd ourselves inside a four-dimensional domain with our kind of
physical laws not because domains with dierent dimensionality and with alternate
properties are impossible or improbable, but simply because our kind of life cannot
exist in other domains.
This provides a simple justication of the weak anthropic principle and removes
the standard objections against it. One does not need anymore to assume that some
supernatural cause created our universe with the properties specically ne-tuned
10

Figure 2: A typical distribution of scalar elds and during the process of selfreproduction of the universe. The height of the distribution shows the value of the
eld which drives ination. The surface is painted black in those parts of the
universe where the scalar eld is in the rst minimum of its eective potential,
and white where it is in the second minimum. Laws of low-energy physics are
dierent in the regions of dierent color. The peaks of the mountains correspond
to places where quantum uctuations bring the scalar elds back to the Planck
density. Each of such places in a certain sense can be considered as a beginning of
a new Big Bang.

11

to make our existence possible. Inationary universe itself, without any external
intervention, may produce exponentially large domains with all possible laws of
low-energy physics. And we should not be surprised that the conditions necessary
for our existence appear on a very large scale rather than only in a small vicinity
of the solar system. If the proper conditions are established near the solar system,
ination ensures that similar conditions appear everywhere within the observable
part of the universe.
The new possibilities that appear due to the self-reproduction of the universe
may provide a basis for what I called the Darwinian approach to cosmology (Linde,
1987a; Vilenkin, 1995; Garcia-Bellido and Linde, 1995). Mutations of the laws of
physics may lead to formation of the domains with the laws of physics that allow
a greater speed of expansion of the universe; these domains will acquire greater
volume and may host a greater number of observers.
On the other hand, the total volume of domains of each type grows indenitely
large. This process looks like a peaceful coexistence and competition, and sometimes even like a fruitful collaboration, when the fastest growing domains produce
many slower growing brothers. In this case a stationary regime is reached, and the
speed of growth of the total volume of domains of each type becomes equally large
for all of the domains (Linde et al, 1994).

Baby universes

As we have seen, ination allows one to justify the weak anthropic principle by
ensuring that all vacuum states and, consequently, all possible laws of elementary
particle physics that are allowed by the basic theory are realized in some exponentially large and locally uniform parts of our universe.
Note, however, that here we are talking not about the choice among many
dierent theories, but about the choice among many possible vacuum states, or
phases, that are allowed by a given theory. This is similar to the possibility to nd
water in a gaseous, liquid or solid state. These states look very dierent (sh cannot
live in ice), but their basic chemical composition is the same. Similarly, despite the
fact that some of the theories may have extremely large number of vacuum states,
our freedom of choice is still limited by the unique fundamental law that is supposed
to remain the same in every corner of our universe.
Now its time to make the next step and ask whether the basic theory was
in fact xed from the very beginning and could not change? A very interesting
set of ideas related to this question was developed in the end of the 80s. It was
called the baby universe theory (Coleman, 1988a,1988b; Banks, 1988; Giddings and
Strominger, 1988,1989). For a short time, this theory was immensely popular, but
12

then it was almost completely forgotten. In our opinion, both extremes were due
to the over-reaction with respect to the uncritical use of the Euclidean approach to
quantum cosmology. But if one distinguishes between this method and the rest of
the theory, one can nd something very interesting and instructive.
The main idea of the baby universe theory is that our universe can split into
disconnected pieces due to quantum gravity eects. Baby universes created from
the parent universe can carry from it an electron-positron pair, or some other
combinations of particles and elds, unless it is forbidden by conservation laws.
Such a process can occur in any place in our universe. Many ways were suggested
to describe such a situation. The simplest one is to say that the existence of baby
universes leads to a modication of the eective Hamiltonian density.
H(x) = H0 ((x)) +

Hi [(x)]Ai .

(5)

The Hamiltonian (5) describes the elds (x) on the parent universe at distances
much greater than the Planck scale. H0 is the part of the Hamiltonian which does
not involve topological uctuations. Hi () are some local functions of the elds
, and Ai are combinations of creation and annihilation operators for the baby
universes. These operators do not depend on x since the baby universes cannot
carry away momentum. Coleman argued (Coleman, 1988a,1988b) that the demand
of locality, on the parent universe,
[H(x), H(y)] = 0

(6)

for spacelike separated x and y, implies that the operators Ai must all commute.
Therefore, they can be simultaneously diagonalized by the states:
Ai |i = i |i .

(7)

If the state of the baby universe is an eigenstate of the Ai , then the net eect of the
baby universes is to introduce innite number of undetermined parameters (the i )
into the eective Hamiltonian (5): one can just replace the operators Ai by their
eigenvalues. If the universe initially is not in the Ai eigenstate, then, nevertheless,
after a series of measurements the wave junction soon collapses to one of the Ai
eigenstates (Coleman, 1988a,1988b; Giddings and Strominger, 1988,1989).
This gives rise to an extremely interesting possibility related to the basic principles of physics. We were accustomed to believe that the main purpose of physics is
to discover the Lagrangian (or Hamiltonian) of the theory that correctly describes
our world. However, the question arises: if our universe did not exist sometimes
in a distant past, in which sense could one speak about the existence of the laws
of Nature which govern the universe? We know, for example, that the laws of our
biological evolution are written in our genetic code. But where were the laws of
physics written at the time when there was no universe (if there was such time)?
13

The possible answer now is that the nal structure of the (eective) Hamiltonian
becomes xed only after measurements are performed, which determine the values
of coupling constants in the state in which we live. Dierent eective Hamiltonians
describe dierent laws of physics in dierent (quantum) states of the universe, and
by making measurements we reduce the variety of all possible laws of physics to
those laws that are valid in the (classical) universe where we live.
We will not discuss this issue here any further, since it would require a thorough
discussion of the dierence between the orthodox (Copenhagen) and the manyworld interpretation of quantum mechanics. We would like to mention only that
this theory opens a new interesting possibility to strengthen the anthropic principle
by allowing all fundamental constants to take dierent values in dierent quantum
states of the universe.
But if it is so interesting, why dont we hear about this theory anymore? In order to answer this question we must remember why it became so popular in the end
of the 80s. The most interesting application of this theory was the possible explanation of the vanishing of the cosmological constant (Coleman, 1988a,1988b). The
main idea is closely related to the previous suggestion by Hawking. According to
(Hawking, 1984), the cosmological constant, like other constants, can take dierent
values, and the probability to nd ourselves in the universe with the cosmological
constant = V () is given by
P () exp(2SE ()) = exp

4
3MP
,

(8)

where SE is the action in the Euclidean version of de Sitter space. However, Coleman pointed out that one should not only take into account one-universe Euclidean
congurations. Rather one should sum over all congurations of babies and parents
connected by Euclidean wormholes. This nally gives (Coleman, 1988a,1988b)
P () exp exp

4
3MP

(9)

Eqs. (8) and (9) suggest that it is most probable to live in a quantum state of
the universe with = 0. This would be a wonderful solution of the cosmological
constant problem.
Unfortunately, the use of the Euclidean approach in this context was not well
justied. The whole trick was based on the fact that Euclidean action SE has a
wrong (negative) sign (Hartle and Hawking, 1983). Usually Euclidean methods
work well for SE > 0 and become very problematic for SE < 0 (Linde, 1984a,1998;
Vilenkin, 1984). After playing with this method for a while, most of the people
became dissatised and abandoned it. Sometimes one can obtain sensible results
by replacing SE by |SE | (Linde, 1984a; Vilenkin, 1984), but this would not yield
any interesting results with respect to in the context of the baby universe theory.
14

Moreover, current observations suggest that the cosmological constant may be

non-vanishing. As a result, the baby universe theory was nearly forgotten.
From our point of view, however, the basic idea that the universe may exist
in dierent quantum states corresponding to dierent laws of physics may be very
productive. But this idea is still somewhat complicated because it pre-assumes that
one can deal with the issues like that only at the level of the so-called third quantization (Coleman, 1988a,1988b; Banks, 1988; Giddings and Strominger, 1988,1989),
with quantum eld theory applied not only to particles but also to the universes.
This is a rather radical assumption. A somewhat dierent approach to quantum
cosmology and variation of fundamental constants was suggested later in (Linde,
1990a; Vilenkin, 1995; Garcia-Bellido and Linde, 1995). Still it was usually emphasized that these approaches are based on quantum cosmology, which is a rather
complicated and controversial science. Thus, it would be helpful to simplify these
ideas a bit, and to present them in an alternative form that may allow further
generalizations.

From the universe to the multiverse

Usually one describes a physical theory by presenting its action. One may write,
for example,
d4 x g(x)

S=N

R(x)
+ L((x)) ,
16G

(10)

where N is a normalization constant, R(x) is the general relativity Lagrangian with

2G
2
G = Mp , and L() is a Lagrangian for the usual matter elds. One obtains the
Einstein equations by variation of the action S with respect to the metric g , and
one nds the equations of motion for the matter elds by variation of the action
S with respect to .
Let us now do something very unusual and add to our original action many
other actions describing dierent elds i with dierent Lagrangians Li living in k

dierent universes of dierent dimensions ni with dierent metrics gi and dierent

gravitational constants Gi :
S = N

d4 x g(x)

R(x)
+ L((x))
16G

Ni

dni xi gi (xi )

i=1

(11)

R(xi )
+ Li (i (xi )) .
16Gi

One may wonder whether this modication will aect our life in the universe described by the original action (10)? The answer is that it will have no impact
15

whatsoever on the physical processes in our universe. Indeed, equations of motion

for and g will not change because the added parts do not depend on and g ,
so their variation with respect to and g vanishes.
This implies that the extended action (11) describes all events in our universe
in the same way as the original action (10). This is very encouraging. So let us
continue our exercise and add to this action an innite sum of all possible actions
describing all possible versions of quantum eld theory and M/string theory. If our
original theory successfully described our universe, it will continue doing so even
after all of these modications.
But why would anybody want to add all of these extra terms if they do not
aect our universe?
There are two related answers. First of all, one may simply reply: Why not? In
some countries, everything that is not explicitly allowed, is forbidden. In some other
countries (and in science), everything that is not explicitly forbidden, is allowed.
We live in one of such countries, so why dont we use the freedom if it does not
make us any harm?
But the second answer is more interesting. Now we know that the theory (11)
and all of its possible extensions are exactly equivalent to the theory (10) with
respect to the processes in our universe (assuming that it is described by (10)). So
we can take a step back, look at all the dierent universes described by Eq. (11),
just as we would look for our car among many dierent cars in a parking lot, and
ask:
As a matter of fact, which one of these universes is ours? Are we sure that it
is the rst one?
From a purely theoretical point of view, the rst universe described by the theory
(10) is not any better than any other universe. However, we can live only in those
universes that are compatible with the existence of life as we know it. When we will
search for our universe, rst of all we will look for those Lagrangians Li (i (xi )) that
can describe elementary particles similar to the ones that we see around. Then we
will specify our search even further by nding the Lagrangians describing particles
with masses and coupling constants that are consistent with our existence. Since
we have all universes with all possible laws of physics described by our extended
action, we will certainly nd the universe where we can live. But that is exactly
what we need to justify the validity of the strong anthropic principle.
Let us summarize our progress so far. Inationary theory allows our universe
to be divided into dierent parts with dierent laws of low-energy physics that
are allowed by the unique fundamental theory. Most importantly, it makes each
of such domains exponentially large, which is a necessary part of justication of
the anthropic principle. The diversity of possible laws of physics can be very high,

16

especially in the models of eternal chaotic ination where quantum uctuations can
have an extremely large amplitude, which makes the transition between all possible
states particularly easy.
In addition to that, one can consider dierent universes with dierent laws of
physics in each of them. This does not necessarily require introduction of quantum
cosmology, many-world interpretation of quantum mechanics, and baby universe
theory. It is sucient to consider an extended action represented by a sum of
all possible actions of all possible theories in all possible universes. One may call
this structure a multiverse. This could sound like a very complicated and radical
proposal, but in fact it is pretty trivial since each part of the innite sum does not
aect other parts. However, it establishes a rm formal background for the further
development of the anthropic principle.
But the main reason why we are introducing this structure is not the anthropic
principle. As we already mentioned, we need to know what emerged rst at the
moment of the universe formation: the universe or the law describing the universe.
It is equally hard to understand how any law could exist prior to the universe
formation, or how the universe could exist without a law. One may assume that
there is only one possible law, and it exists in some unspecied way even prior to the
emergence of the universe. However, this would be similar to having elections with
only one name on the ballot. Perhaps a better possibility would be to consider all
logically possible combinations of the universes, the laws describing them, and the
observers populating these universes. Given the choice among dierent universes
in this multiverse structure, we can proceed by eliminating the universes where our
life would be impossible. This simple step is sucient for understanding of many
features of our universe that otherwise would seem miraculous.
There are some additional steps that one may want to make. In our analysis we
still assumed that any evolution must be described by some kind of action. Meanwhile there are some theories where equations of motion are known even though the
action is unavailable. One may consider other models of evolution, based, e.g., on
cellular automata. One can go even further, and consider all possible mathematical
structures (Tegmark, 1998), or, following Wheeler, consider all logical possibilities
and the concept of it from bit; see (Wheeler, 1990) and references therein.
But before doing so we would like to show that the concept of a multiverse
may have interesting consequences going beyond the justication of the anthropic
principle. In order to do it we must learn whether the dierent universes may
interact with each other.

17

Double universe model and the cosmological

constant problem

Let us consider the double-universe model (Linde, 1988). This model describes two
universes, X and Y, with coordinates x and y , respectively (, = 0, 1, . . . , 3)

and with metrics g (x) and g (y), containing elds (x) and (y) with the action

of the following unusual type:

S=N
[

d4 xd4 y g(x) g (y)

2
Mp
M2

R(x) + L((x)) p R(y) L((y))] .

16
16

(12)

Here N is some normalization constant. This action is invariant under general

coordinate transformations in each of the universes separately. A novel symmetry of

the action is the symmetry under the transformation (x) (x), g (x) g (x)

and under the subsequent change of the overall sign, S S. We call this the
antipodal symmetry, since it relates to each other the states with positive and
negative energies.
An immediate consequence of this symmetry is the invariance under the change

of the values of the eective potentials V () V () + c, V () = V () + c, where

c is some constant. Consequently, nothing in this theory depends on the value of

the eective potentials V () and V () in their absolute minima 0 and 0 . (Note,

that 0 = 0 and V (0 ) = V (0 ) due to the antipodal symmetry.) This is the

basic reason why it proves possible to solve the cosmological constant problem in
our model.
However, our main reason to invoke this new symmetry was not just to solve
the cosmological constant problem. Just as the theory of mirror particles originally
was proposed in order to make the theory CP-symmetric while maintaining CPasymmetry in its observable sector, the theory (10) is proposed in order to make
the theory symmetric with respect to the choice of the sign of energy. This removes
the old prejudice that, even though the overall change of sign of the Lagrangian
(i.e. both of its kinetic and potential terms) does not change the solutions of the
theory, one must say that the energy of all particles is positive. This prejudice
was so strong, that many years ago physicists preferred to quantize particles with
negative energy as antiparticles with positive energy, which caused the appearance
of such meaningless concepts as negative probability. We wish to emphasize that
there is no problem to perform a consistent quantization of theories which describe
particles with negative energy. All diculties appear only when there exist interacting species with both signs of energy. In our case no such problem exists, just as
there is no problem of antipodes falling down from the opposite side of the earth.

The reason is that the elds (y) do not interact with the elds (x), and the equa18


tions of motion for the elds (y) are the same as for the elds (x) (the overall

minus sign in front of L((y)) does not change the Lagrange equations). Similarly,
gravitons from dierent universes do not interact with each other. However, some
interaction between the two universes does exist. Indeed, the Einstein equations in
our case are:
1
1

R (x) g R(x) = 8GT (x) g R(y) + 8GL((y)) ,

(13)
2
2
1
1

R (y) g R(y) = 8GT (y) g R(x) + 8GL((x))

2
2

(14)

Here T is the energy-momentum tensor of the elds (x), Ta is the energy

momentum tensor of the elds (y), the sign of averaging means
R(x) =

R(y) =

d4x g(x)R(x)
d4 x g(x)
d4 y g (y)R(y)

d4 y g (y)

(15)

(16)

and similarly for L(x) and L(y) . Thus, the novel feature of the theory (10) is
the existence of a global interaction between the universes X and Y: The integral
over the whole history of the Y-universe changes the vacuum energy density of the
X-universe.
In general, the computation of the averages of the type (15), (16) may be a
rather sophisticated problem. Fortunately, however, in the inationary theory (at
least, if the universe is not self-reproducing, see below), this task can be rather
trivial. Namely, the universe after ination becomes almost at and its lifetime
becomes exponentially large. In such a case, the dominant contribution to the
average values R and L comes from the late stages of the universe evolution
at which the elds (x) and () relax near the absolute minima of their eective
a
potentials. As a result, the average value of L((x)) almost exactly coincides
with the value of the eective potential V () in its absolute minimum at = 0 ,
and the averaged value of the curvature scalar R(x) coincides with its value at the
late stages of the universe evolution, when the universe transforms to the state
corresponding to the absolute minimum of V (). Similar results are valid for the

average values of L((y)) and of R(y) as well. In such a case one can easily show
(Linde, 1988) that at the late stages of the universe evolution, when the elds (x)

and (y) relax near the absolute minima of their eective potentials, the eective
cosmological constant automatically vanishes,
R(x) = R(y) =

32

G[V (0 ) V (0 )] = 0
3
19

(17)

This model provided the rst example of a theory with a non-local interaction
of universes. It inspired the baby-universe scenario, and it was forgotten when
the baby-universe scenario failed. However, this model is based on a completely
dierent principle, so it should be considered quite independently.
There are several problems with this model that should be addressed before
taking it too seriously. First of all, in order to solve the cosmological constant
problem in our universe we added a new universe with negative energy density. At
the rst glance, this may not seem very economical. However, during the last several
years the idea that we may have several dierent interacting universes became very
popular in the context of the brane world scenario (Arkani-Hamed et al, 1998,2000;
Antoniadis et al, 1999; Randall and Sundrum, 1999). The cancellation of the
eective cosmological constant on our brane (our universe) is often achieved by
the introduction of the negative tension brane (the universe with a negative energy
density), see e.g. (Randall and Sundrum, 1999). It is not quite clear whether any
symmetry can protect this cancellation against radiative corrections in the brane
world scenario. Meanwhile in our case the theory is fully symmetric with respect
to the choice of the sign of energy, which may protect the cosmological constant
against radiative corrections.
The second problem is more complicated. If the universe is self-reproducing,
one may encounter diculties when computing the averages (15), (16), since they
may become dominated by eternally inating parts of the universe with large V ().
One can avoid this complication in inationary theories where V () grows rapidly
enough at large , since there will be no universe self-reproduction in such theories.
Finally, the cosmological observations indicate that the universe is accelerating
4
as if it has a miniscule positive vacuum energy V () 10123 Mp . Thus we need
to make the vacuum energy cancellation non-exact. This is quite possible: as we
said, the average value of L((x)) almost exactly coincides with the value of
the eective potential V () in its absolute minimum at = 0 . Also, if V () is
very at near its minimum, like in the usual dark energy models, we may move
to the minimum very slowly and at any given moment we will still have a small
non-compensated positive vacuum energy.
We do not know whether this simple model is going to survive in the future.
But this example shows that the multiverse scenario may provide us with new
unexpected possibilities that should be considered very seriously.
Now we will make a step back and discuss the anthropic approach to the cosmological constant problem.

20

Cosmological constant, dark energy, and the

anthropic principle

The rst attempt to solve the cosmological constant problem using the anthropic
principle in the context of inationary cosmology was made in (Linde, 1984b,1986b).
The simplest way to do it is to consider ination driven by the scalar eld (the
inaton eld) and mimic the cosmological constant by the very at potential of the
second scalar eld, . The simplest potential of this type is the linear potential
(Linde, 1986b)
3
V () = Mp .
(18)
If is suciently small, < 10122 , the potential V () is so at that the eld
practically does not change during the last 1010 years, its kinetic energy is very
small, so at the present stage of the evolution of the universe its total potential
energy V () acts exactly as a cosmological constant. This model was one of the
rst examples of what later became known as quintessence, or dark energy.
Even though the energy density of the eld practically does not change at
the present time, it changed substantially during ination. Since is a massless
eld, it experienced quantum jumps with the amplitude H/2 during each time
H 1 . These jumps move the eld in all possible directions. In the context of
the eternal ination scenario this implies that the eld becomes randomized by
quantum uctuations: The universe becomes divided into innitely large number
of exponentially large parts containing all possible values of the eld . In other
words, the universe becomes divided into innitely large number of universes with
all possible values of the eective cosmological constant = V () + V (0 ), where
V (0 ) is the energy density of the inaton eld in the minimum of its eective
4
4
potential. This quantity may change from Mp to +Mp in dierent parts of the
<
universe, but we can live only in the universes with || O(10)0 1028 g/cm3 ,
where 0 is the present energy density in our part of the universe.
Indeed, if < 1028 g/cm3 , the universe collapses within the time much

smaller than the present age of the universe 1010 years (Linde, 1984b,1986b;
Barrow and Tipler, 1986). On the other hand, if 1028 g/cm3 , the universe at
present would expand exponentially fast, energy density of matter would be exponentially small, and life as we know it would be impossible (Linde, 1984b,1986b).
This means that we can live only in those parts of the universe where the cosmological constant does not dier too much from its presently observed value || 0 .
This approach constituted the basis for many subsequent attempts to solve
the cosmological constant problem using the anthropic principle in inationary
cosmology (Weinberg, 1987; Linde, 1990a; Vilenkin, 1995b; Martel et al, 1998;
Garriga and Vilenkin, 2000,2001b,2002).
At rst glance, an introduction of the miniscule parameter < 10122 does not
21

provide a real explanation of the equally miniscule cosmological constant || 0

4
10123 Mp . However, exponentially small parameters like that may easily appear
due to nonperturbative eects. One could even think that a similar exponential
suppression may be the true reason why || is so small. But there are many large
contributions to , due to quantum gravity, due to spontaneous symmetry breaking
in GUTs and in the electroweak theory, due to supersymmetry breaking, QCD
eects, etc. One could appeal to the nonperturbative exponential smallness of
only if all large contributions to the vacuum energy miraculously cancel, like in the
model considered in the previous section. And even if this cancellation is achieved,
we still need to explain why || is suppressed exactly to the level when it becomes
of the same order as the present energy density of the universe. This coincidence
problem becomes resolved in the theory (18) for all suciently small ; instead of
the ne-tuning of we simply need it to be suciently strongly suppressed. A very
clear discussion of the issue of ne-tuning versus exponential suppression can be
found in (Garriga and Vilenkin, 2000) in application to a similar model with the
6
potential m2 2 /2 with m2 10240 Mp | |1 .
Alternative approaches based on the anthropic principle are described in (Bousso
and Polchinski, 2001; Feng et al, 2001; Banks et al, 2001). One can also use a
more general approach outlined in Section 6 and consider a baby-universe scenario,
or a multiverse consisting of dierent inationary universes with dierent values
of the cosmological constant in each of them (Linde, 1989,1990a,1991). In this
case one does not need to consider extremely at potentials, but the procedure
of comparing probabilities to live in dierent universes with dierent becomes
more ambiguous (Vilenkin, 1995; Garcia-Bellido and Linde, 1995). However, if one
makes the simplest assumption that the universes with dierent values of are
equally probable, one obtains an anthropic solution of the cosmological constant
problem without any need of introducing extremely small parameters < 10122
6
or m2 10240 Mp | |1 .
>
The constraint 1028 g/cm3 still remains the strongest constraint on
the negative cosmological constant; for the recent developments related to this
constraint see (Kallosh and Linde, 2002; Garriga and Vilenkin, 2002). Meanwhile,
the constraint on the positive cosmological constant, < 1028 g/cm3 , was made

much more precise and accurate in the subsequent works.

In particular, Weinberg pointed out that the process of galaxy formation occurs
only up to the moment when the cosmological constant begins to dominate the
energy density of the universe and the universe enters the stage of late-time ination
>
(Weinberg, 1987). For example, one may consider galaxies formed at z 4, when
the energy density of the universe was 2 orders of magnitude greater than it is now.
>
Such galaxies would not form if 102 0 1027 g/cm3 .
The next important step was made in a series of works (Efstathiou, 1995;
Vilenkin, 1995b; Martel et al, 1998; Garriga and Vilenkin, 2000,2001b,2002; Blud22

man and Roos, 2002). The authors considered not only our own galaxy, but all
other galaxies that could harbor life of our type. This would include not only the
existing galaxies but also the galaxies that are being formed at the present epoch.
Since the energy density at later stages of the evolution of the universe becomes
smaller, even a very small cosmological constant may disrupt the late-time galaxy
formation, or may prevent the growth of existing galaxies. This allows to strengthen
the constraint on the cosmological constant. According to (Martel et al, 1998), the
probability that an astronomer in any of the universes would nd the presently
observed ratio /0 as small as 0.7 ranges from 5% to 12%, depending on various assumptions. For some models based on extended supergravity, the anthropic
constraints can be strengthened even further (Kallosh and Linde, 2002).

Problem of calculating the probabilities

As we see, the anthropic principle can be extremely useful in resolving some of

the most profound problems of modern physics. However, to make this principle
more quantitative, one should nd a proper way to calculate the probability to live
in a universe of a given type. This step is not quite trivial. One may consider
the probability of quantum creation of the universe from nothing (Hartle and
Hawking, 1983; Linde, 1984a; Vilenkin, 1984), or the results of the baby universe
theory (Coleman, 1988a,1988b), or the results based on the theory of the selfreproduction of the universe and quantum cosmology (Linde et al, 1994; GarciaBellido et al, 1994; Vilenkin, 1995; Vanchurin et al, 2000; Garcia-Bellido and Linde,
1995; Linde and Mezhlumian, 1996; Garriga and Vilenkin, 2001). Unfortunately,
these methods are based on dierent assumptions, and the results of some of these
works signicantly dier from each other. This may be just a temporary setback.
For example, in our opinion, an interpretation of Euclidean quantum gravity used
in (Hartle and Hawking, 1983, Coleman, 1988a,1988b) is not quite convincing. The
method proposed in (Turok, 2002) is basically equivalent to the investigation of
the probability distribution in comoving coordinates Pc (, t) (Linde, 1990a). This
approach ignores information about most of the observers living in our universe, so
it can hardly have any relation to the standard anthropic considerations and misses
the eect of the self-reproduction of the universe. An investigation of creation of
the universe from nothing (Linde, 1984a; Vilenkin, 1984) can be very useful, but
I believe that it should be considered only as a part of the more general approach
based on the stochastic approach to ination.
It is more dicult to make a denite choice between the dierent answers provided by the dierent methods of interpretation of the results obtained by the
stochastic approach to ination (Starobinsky, 1986; Linde et al, 1994; GarciaBellido et al, 1994; Vilenkin, 1995; Garcia-Bellido and Linde, 1995; Linde and
23

Mezhlumian, 1996; Vanchurin et al, 2000; Garriga and Vilenkin, 2001a). We believe that all of these dierent answers in a certain sense are correct; it is the choice
of the questions that remains problematic.
To explain our point of view, let us study an example related to demographics.
One may want to know what is the average age of a person living now on the
Earth. In order to nd it, one should take the sum of the ages of all people and
divide it by their total number. Naively, one could expect that the result of the
calculation should be equal to 1/2 of the life expectancy. However, the actual
result will be much smaller. Because of the exponential growth of the population,
the main contribution to the average age will be given by very young people. Both
answers (the average age of a person, and a half of the life expectancy) are correct
despite the fact that they are dierent. None of these answers is any better; they
are dierent because they address dierent questions. Economists may want to
know the average age in order to make their projections. Meanwhile each of us, as
well as the people from the insurance industry, may be more interested in the life
expectancy.
Similarly, the calculations performed in (Linde et al, 1994; Garcia-Bellido et al,
1994; Vilenkin, 1995; Garcia-Bellido and Linde, 1995; Linde and Mezhlumian, 1996;
Vanchurin et al, 2000; Garriga and Vilenkin, 2001a) dissect all possible outcomes of
the evolution of the universe (or the multiverse) in many dierent ways. (Unlike the
method suggested in (Turok, 2002), these methods cover the whole universe rather
that its innitesimally small part.) Each of these ways is quite legitimate and leads
to correct results, but some additional input is required in order to understand
which of these results, if any, is most closely related to the anthropic principle.
In the meantime one may take a pragmatic point of view and consider this investigation as a kind of theoretical experiment. We may try to use probabilistic
considerations in a trial-and-error approach. If we get unreasonable results, this
may serve as an indication that we are using quantum cosmology incorrectly. However, if some particular proposal for the probability measure will allow us to solve
certain problems which could not be solved in any other way, then we will have a
reason to believe that we are moving in the right direction. But we are not sure
that any real progress in this direction can be reached and we will be able to learn
how to calculate the probability to live in one of the many universes without having
a good idea of what is life and what is consciousness (Linde, 1990a; Garcia-Bellido
and Linde, 1995; Linde and Mezhlumian, 1996; Linde et al, 1996).
A healthy scientic conservatism usually forces us to disregard all metaphysical
subjects that seem unrelated to our research. However, in order to make sure that
this conservatism is really healthy, from time to time one should take a risk to
abandon some of the standard assumptions. This may allow us either to rearm
our previous position, or to nd some possible limitations of our earlier point of
view.
24

10

Does consciousness matter?

A good starting point for our brief discussion of consciousness is quantum cosmology, the theory that tries to unify cosmology and quantum mechanics.
If quantum mechanics is universally correct, then one may try to apply it to
the universe in order to nd its wave function. This would allow us nd out which
events are probable and which are not. However, it often leads to paradoxes. For
example, the essence of the Wheeler-DeWitt equation (DeWitt, 1967), which is the
Schrdinger equation for the wave function of the universe, is that this wave function
o
does not depend on time, since the total Hamiltonian of the universe, including the
Hamiltonian of the gravitational eld, vanishes identically. This result was obtained
in 1967 by Bryce DeWitt. Therefore if one would wish to describe the evolution
of the universe with the help of its wave function, one would be in trouble: The
universe as a whole does not change in time.
The resolution of this paradox suggested by Bryce DeWitt is rather instructive
(DeWitt, 1967). The notion of evolution is not applicable to the universe as a
whole since there is no external observer with respect to the universe, and there is
no external clock that does not belong to the universe. However, we do not actually
ask why the universe as a whole is evolving. We are just trying to understand our
own experimental data. Thus, a more precisely formulated question is why do we
see the universe evolving in time in a given way. In order to answer this question
one should rst divide the universe into two main pieces: i) an observer with his
clock and other measuring devices and ii) the rest of the universe. Then it can be
shown that the wave function of the rest of the universe does depend on the state
of the clock of the observer, i.e. on his time. This time dependence in some sense
is objective: the results obtained by dierent (macroscopic) observers living in
the same quantum state of the universe and using suciently good (macroscopic)
measuring apparatus agree with each other.
Thus we see that without introducing an observer, we have a dead universe,
which does not evolve in time. This example demonstrates an unusually important
role played by the concept of an observer in quantum cosmology. John Wheeler
underscored the complexity of the situation, replacing the word observer by the
word participant, and introducing such terms as a self-observing universe.
Most of the time, when discussing quantum cosmology, one can remain entirely
within the bounds set by purely physical categories, regarding an observer simply
as an automaton, and not dealing with questions of whether he/she/it has consciousness or feels anything during the process of observation. This limitation is
harmless for many practical purposes. But we cannot rule out the possibility that
carefully avoiding the concept of consciousness in quantum cosmology may lead to
an articial narrowing of our outlook.
25

Let us remember an example from the history of science that may be rather
instructive in this respect. Prior to the invention of the general theory of relativity,
space, time, and matter seemed to be three fundamentally dierent entities. Space
was thought to be a kind of three-dimensional coordinate grid which, when supplemented by clocks, could be used to describe the motion of matter. Space-time
possessed no intrinsic degrees of freedom, it played secondary role as a tool for the
description of the truly substantial material world.
The general theory of relativity brought with it a decisive change in this point
of view. Space-time and matter were found to be interdependent, and there was no
longer any question which one of the two is more fundamental. Space-time was also
found to have its own inherent degrees of freedom, associated with perturbations
of the metric gravitational waves. Thus, space can exist and change with time in
the absence of electrons, protons, photons, etc.; in other words, in the absence of
anything that had previously (i.e., prior to general relativity) been called matter.
Of course, one can simply extend the notion of matter, because, after all, gravitons
(the quanta of the gravitational eld) are real particles living in our universe. On
the other hand, the introduction of the gravitons provides us, at best, with a tool
for an approximate (perturbative) description of the uctuating geometry of spacetime. This is completely opposite to the previous idea that space-time is only a
tool for the description of matter.
A more recent trend, nally, has been toward a unied geometric theory of all
fundamental interactions, including gravitation. Prior to the end of the 1970s,
such a program seemed unrealizable; rigorous theorems were proven on the impossibility of unifying spatial symmetries with the internal symmetries of elementary
particle theory. Fortunately, these theorems were sidestepped after the discovery
of supersymmetry and supergravity. In these theories, matter elds and space-time
became unied within the general concept of superspace.
Now let us turn to consciousness. The standard assumption is that consciousness, just like space-time before the invention of general relativity, plays a secondary,
subservient role, being just a function of matter and a tool for the description of the
truly existing material world. But let us remember that our knowledge of the world
begins not with matter but with perceptions. I know for sure that my pain exists,
my green exists, and my sweet exists. I do not need any proof of their existence,
because these events are a part of me; everything else is a theory. Later we nd out
that our perceptions obey some laws, which can be most conveniently formulated
if we assume that there is some underlying reality beyond our perceptions. This
model of material world obeying laws of physics is so successful that soon we forget
about our starting point and say that matter is the only reality, and perceptions
are nothing but a useful tool for the description of matter. This assumption is
almost as natural (and maybe as false) as our previous assumption that space is
only a mathematical tool for the description of matter. We are substituting reality
26

of our feelings by the successfully working theory of an independently existing material world. And the theory is so successful that we almost never think about its
possible limitations.
Guided by the analogy with the gradual change of the concept of space-time,
we would like to take a certain risk and formulate several questions to which we do
not yet have the answers (Linde, 1990a; Page, 2002):
Is it possible that consciousness, like space-time, has its own intrinsic degrees
of freedom, and that neglecting these will lead to a description of the universe that
is fundamentally incomplete? What if our perceptions are as real (or maybe, in a
certain sense, are even more real) than material objects? What if my red, my blue,
my pain, are really existing objects, not merely reections of the really existing
material world? Is it possible to introduce a space of elements of consciousness,
and investigate a possibility that consciousness may exist by itself, even in the
absence of matter, just like gravitational waves, excitations of space, may exist in
the absence of protons and electrons?
Note, that the gravitational waves usually are so small and interact with matter
so weakly that we did not nd any of them as yet. However, their existence is
absolutely crucial for the consistency of our theory, as well as for our understanding
of certain astronomical data. Could it be that consciousness is an equally important
part of the consistent picture of our world, despite the fact that so far one could
safely ignore it in the description of the well studied physical processes? Will it not
turn out, with the further development of science, that the study of the universe
and the study of consciousness are inseparably linked, and that ultimate progress
in the one will be impossible without progress in the other?
Instead of discussing these issues here any further, we will return back to a
more solid ground and concentrate on the consequences of eternal ination and the
multiverse theory that do not depend on the details of their interpretation. As an
example, we will discuss here two questions that for a long time were considered
too complicated and metaphysical. We will see that the concept of the multiverse
will allow us to propose possible answers to these questions.

11

Why is mathematics so ecient?

There is an old problem that bothered many people thinking about the foundations
of mathematics: Why is mathematics so ecient in helping us to describe our world
and predict its evolution?
This question arises at the moment when one introduces numbers and uses them
to count. Then a similar question appears when one introduces calculus and uses
it to describe the motion of the planets. Somehow there are some rules that help
27

us to operate with mathematical symbols and relate the results of these operations
to the results of our observations. Why does it work so well?
Of course, one could always respond that it is just so. But let us consider several
other questions of a similar type. Why is our universe so large? Why parallel lines
do not intersect? Why dierent parts of the universe look so similar? Thirty years
ago such questions would look too metaphysical to be considered seriously. Now
we know that inationary cosmology provides a possible answer to all of these
questions. Let us try it again.
Before we do it, we should give at least one example of a universe where mathematics would be inecient. Here it is. Suppose the universe can be in a stable
4
or metastable vacuum state with a Planckian density Mp 1094 g/cm3 . According to quantum gravity, quantum uctuations of space-time curvature in this
regime are of the same order as the curvature itself. In simple terms, this means
that the rulers are bending, shrinking and extending in a chaotic and unpredictable
way due to quantum uctuations, and this happens faster than one can measure
the distance. The clocks are destroyed faster than one can measure the time. All
records about the previous events become erased, so one cannot remember anything, record it, make a prediction and compare the prediction with experimental
results.
A similar situations occurs in a typical non-inationary closed universe. There is
1
only one natural parameter of dimension of length in quantum gravity, lp = Mp ,
4
and only one natural parameter of dimension of energy density, p = Mp . If
one considers a typical closed universe of a typical initial size lp with a typical
initial density p , one can show that its total lifetime until it collapses is t tp =
1
Mp 1043 seconds, and throughout all of its short history the energy density
4
remains of the order of Mp or greater. Such a universe can incorporate just a few
elementary particles (Linde, 1990a), so one cannot live there, cannot build any
measuring devices, record any events and use mathematics to describe events in
such a universe.
In the cases described above, mathematics would be rather inecient because it
would not help anybody to relate dierent things and processes to each other. More
generally, if the laws of physics inside some parts of the universe disallow formation
of stable long-living structures, then mathematics will not be very useful there, and
there will be no observers (long-living conscious beings capable of remembering and
thinking) who would be able to tell us about it.
Fortunately, among all possible domains of the universe (or among all possible universes) there are some domains where ination is possible. Energy density
4
inside such domain gradually drops down many orders of magnitude below Mp .
These domains become exponentially large and can live for an exponentially long
time. Our life is possible only in those exponentially large domains (or universes)
28

where the laws of physics allow formation of stable long-living structures. The very
concept of stability implies existence of mathematical relations that can be used
for the long-term predictions. The rapid development of the human race became
possible only because we live in the universe where the long-term predictions are
so useful and ecient that they allow us to survive in the hostile environment and
win in the competition with other species.
To summarize, in the context of the multiverse theory, one can consider all
possible universes with all possible laws of physics and mathematics. Among all
possible universes, we can live only in those where mathematics is ecient.

12

Why quantum?

Now we will discuss the famous Wheelers question: Why quantum?

Before doing so, I would like to remember the question often asked by Zeldovich: Do we have any experimental evidence of proton instability and baryon
non-conservation?
In accordance to the unied theories of weak, strong and electromagnetic interactions, protons and other baryons can be unstable. They can decay to leptons.
But the decay rate is so small that we still did not nd any direct evidence of the
proton instability. People were watching protons in thousands of tons of water, and
did not nd any of them decaying. Thus the simple-minded answer to Zeldovichs
question would be No.
However, the true answer is dierent. To make it sound a little bit more challenging, I will formulate it in a way slightly dierent from the formulation used by
Zeldovich, but conveying the same basic idea.
The main experimental evidence of the baryon number non-conservation is provided by the fact that parallel lines do not intersect.
What? Is it a joke? What is the relation?
Well, the fact that the parallel lines do not intersect and remain parallel to each
other is a consequence of the spatial atness of the universe. In a closed universe the
parallel line would intersect, in an open universe they would diverge at innity. The
only known explanation of the atness of the universe is provided by inationary
cosmology. This theory implies that at the end of the exponential expansion of the
universe, the number density of all elementary particles becomes vanishingly small.
All matter surrounding us was produced due to the decay of the scalar eld after
ination (Dolgov and Linde, 1982; Abbott et al, 1982; Kofman et al, 1994,1997;
Felder et al, 2001). The density of protons in our part of the universe is much
greater than the density of antiprotons. This means that at the present time the
29

total baryon number density is not zero. It would be impossible to produce these
baryons from the post-inationary state with the vanishing baryon density if the
baryon number were conserved.
Thus, the only available explanation of the observed atness and homogeneity
of the universe requires baryon number non-conservation. In this sense, the fact
that the parallel lines do not intersect is an observational evidence of the proton
instability.
This is a strange and paradoxical logic, but we must get used to it if we want
to understand the properties of our universe.
Now let us return to Wheelers question. At the rst glance, this question is so
deep and metaphysical that we are not going to know the answer any time soon.
However, in my opinion, the answer is pretty simple.
The only known way to explain why our universe is so large, at, homogeneous
and isotropic requires ination. As we just said, after ination the universe becomes
empty. All matter in the universe was produced due to quantum processes after
the end of ination. All galaxies were produced by quantum uctuations generated
at the last stages of ination. There would be no galaxies and no matter in our
universe if not for the quantum eects. One can formulate this result in the following
way:
Without ination, our universe would be ugly. Without quantum, our universe
would be empty.
But there is something else here. As we already discussed in Section 4, quantum uctuations lead to the eternal process of self-reproduction of the inationary
universe.
Quantum eects combined with ination make the universe innitely large and
immortal.
This provides a possible answer to Wheelers question.

Isnt it amazing that dierent, apparently unrelated things can match together
to form a beautiful and self-consistent pattern? Are we uncovering the universal
truth or simply allow this beauty to deceive us? This is one of the questions that
will remain with us for some time. We need to move carefully and slowly, constantly
keeping in touch with solid and well established facts, but from time to time allowing
ourselves to satisfy our urge to speculate, following the steps of John Wheeler.

30

References
Abbott, L. F., Farhi, E. and Wise, M. B. (1982) Particle Production In The
New Inationary Cosmology, Phys. Lett. B 117, 29.
Albrecht, A. and Steinhardt, P. J. (1982) Cosmology For Grand Unied Theories With Radiatively Induced Symmetry Breaking, Phys. Rev. Lett. 48, 1220.
Antoniadis, I., Arkani-Hamed, N., Dimopoulos, S. and Dvali, G. R. (1998) New
dimensions at a millimeter to a Fermi and superstrings at a TeV, Phys. Lett. B
436, 257 [arXiv:hep-ph/9804398].
Arkani-Hamed, N., Dimopoulos, S. and Dvali, G. R. (1998) The hierarchy
problem and new dimensions at a millimeter, Phys. Lett. B 429, 263 [arXiv:hepph/9803315].
Arkani-Hamed, N., Dimopoulos, S., Dvali, G. R. and Kaloper, N. (2000) Manyfold universe, JHEP 0012, 010 [arXiv:hep-ph/9911386].
Banks, T. (1988) Prolegomena To A Theory Of Bifurcating Universes, Nucl.
Phys. B 309, 493.
Banks, T., Dine, M., and Motl, L. (2001) On anthropic solutions of the cosmological constant problem, JHEP 0101, 031 [arXiv:hep-th/0007206].
Bardeen, J. M., Steinhardt, P. J. and Turner, M. S. (1983) Spontaneous Creation Of Almost Scale - Free Density Perturbations In An Inationary Universe,
Phys. Rev. D 28, 679.
Barrow,J. D. and Tipler, F. J. (1986) The Anthropic Cosmological Principle,
Oxford University Press, New York.
Bludman, S. A. and Roos, M. (2002) Quintessence cosmology and the cosmic
coincidence, Phys. Rev. D 65, 043503 [arXiv:astro-ph/0109551].
Borde, A., Guth, A. H. and Vilenkin, A. (2001) Ination is not past-eternal,
arXiv:gr-qc/0110012.
Bousso, R. and Polchinski, J. (2000) Quantization of four-form uxes and dynamical neutralization of the cosmological constant, JHEP 0006, 006 [arXiv:hepth/0004134].
Coleman, S. R. (1988a) Black Holes As Red Herrings: Topological Fluctuations
And The Loss Of Quantum Coherence, Nucl. Phys. B 307, 867.
Coleman, S. R. (1988b) Why There Is Nothing Rather Than Something: A
Theory Of The Cosmological Constant, Nucl. Phys. B 310, 643.
DeWitt, B. S. (1967) Quantum Theory Of Gravity. 1. The Canonical Theory,
Phys. Rev. 160, 1113 (1967).

31

Dolgov, A. D. and Linde, A. D. (1982) Baryon Asymmetry In Inationary

Universe, Phys. Lett. B 116, 329.
Donoghue, J. F. (2000) Random values of the cosmological constant, JHEP
0008, 022 [arXiv:hep-ph/0006088].
Efstathiou, G. (1995) M.N.R.A.S. 274, L73.
Felder, G., Garcia-Bellido, J., P. B. Greene, Kofman, L., Linde, A. D. and
Tkachev, I. (2001) Dynamics of symmetry breaking and tachyonic preheating,
Phys. Rev. Lett. 87, 011601, hep-ph/0012142.
Feng, J. L., March-Russell, J., Sethi, S. and Wilczek, F. (2001) Saltatory relaxation of the cosmological constant, Nucl. Phys. B 602, 307 [arXiv:hep-th/0005276].
Garcia-Bellido, J., Linde, A. D. and Linde, D. A. (1994) Fluctuations of the
gravitational constant in the inationary Brans-Dicke cosmology, Phys. Rev. D
50, 730 [arXiv:astro-ph/9312039].
Garcia-Bellido, J. and Linde, A. D. (1995) Stationarity of ination and predictions of quantum cosmology, Phys. Rev. D 51, 429 [arXiv:hep-th/9408023].
Garriga, J. and Vilenkin, A. (2000), On likely values of the cosmological constant, Phys. Rev. D 61, 083502 [arXiv:astro-ph/9908115].
Garriga, J. and Vilenkin, A. (2001a), A prescription for probabilities in eternal
ination, Phys. Rev. D 64, 023507 [arXiv:gr-qc/0102090].
Garriga, J. and Vilenkin, A. (2001b), Solutions to the cosmological constant
problems, Phys. Rev. D 64, 023517 [arXiv:hep-th/0011262].
Garriga, J. and Vilenkin, A. (2002) Testable anthropic predictions for dark
energy, arXiv:astro-ph/0210358.
Giddings, S. B. and Strominger, A. (1988) Loss Of Incoherence And Determination Of Coupling Constants In Quantum Gravity, Nucl. Phys. B 307, 854.
Giddings, S. B. and Strominger, A. (1989) Baby Universes, Third Quantization
And The Cosmological Constant, Nucl. Phys. B 321, 481.
Guth, A. H. (1981) The Inationary Universe: A Possible Solution To The
Horizon And Flatness Problems, Phys. Rev. D 23, 347.
Guth, A. H. and S. Y. Pi (1982) Fluctuations In The New Inationary Universe, Phys. Rev. Lett. 49, 1110.
Hawking, S. W. (1982) The Development Of Irregularities In A Single Bubble
Inationary Universe, Phys. Lett. B 115, 295.
Hawking, S. W. (1984) The Cosmological Constant Is Probably Zero, Phys.
Lett. B 134, 403.
Hartle, J. B. and Hawking, S. W. (1983). Wave Function Of The Universe,
32

Phys. Rev. D 28, 2960

Kallosh, R. and Linde, A. D. (2002) M-theory, cosmological constant and anthropic principle, arXiv:hep-th/0208157.
Kofman, L., Linde, A. D. and Starobinsky, A. A. (1994) Reheating after ination, Phys. Rev. Lett. 73, 3195 [arXiv:hep-th/9405187].
Kofman, L., Linde, A. D. and Starobinsky, A. A. (1997) Towards the theory
of reheating after ination, Phys. Rev. D 56, 3258 [arXiv:hep-ph/9704452].
Kolb, E. W. and Turner, M. S. (1990) The Early Universe, Addison-Wesley,
New York.
Linde, A. D. (1982a) A New Inationary Universe Scenario: A Possible Solution Of The Horizon, Flatness, Homogeneity, Isotropy And Primordial Monopole
Problems, Phys. Lett. B 108, 389.
Linde, A. D. (1982b) Nonsingular Regenerating Inationary Universe, Print82-0554 (Cambridge University).
Linde, A. D. (1982c) Scalar Field Fluctuations In Expanding Universe And
The New Inationary Universe Scenario, Phys. Lett. B 116, 335 (1982).
Linde, A. D. (1983a) The New Inationary Universe Scenario, In: The Very
Early Universe, ed. G.W. Gibbons, S.W. Hawking and S.Siklos, pp. 205-249 Cambridge University Press.
Linde, A. D. (1983b) Chaotic Ination, Phys. Lett. 129B, 177.
Linde, A. D. (1983c) Ination Can Break Symmetry In Susy, Phys. Lett. B
131, 330.
Linde, A. D. (1984a) Quantum Creation Of The Inationary Universe, Lett.
Nuovo Cim. 39, 401.
Linde, A. D. (1984b) The Inationary Universe, Rep. Prog. Phys. 47, 925.
Linde, A. D. (1985) The New Mechanism Of Baryogenesis And The Inationary
Universe, Phys. Lett. B 160, 243.
Linde, A. D. (1986a) Eternally Existing Self-reproducing Chaotic Inationary
Universe, Phys. Lett. B 175, 395.
Linde, A. D. (1986b) Ination And Quantum Cosmology, Print-86-0888 (June
1986), in: Three hundred years of gravitation, (Eds.: Hawking, S.W. and Israel, W.,
Cambridge Univ. Press, 1987), 604-630.
Linde, A. D. (1987a) Particle Physics and Inationary Cosmology, Physics
Today 40, 61.
Linde, A. D. (1987b) Ination And Axion Cosmology, Phys. Lett. B 201, 437
(1988).
33

Linde, A. D. (1988) The Universe Multiplication And The Cosmological Constant Problem, Phys. Lett. B 200, 272.
Linde, A. D. (1989) Life After Ination And The Cosmological Constant Problem, Phys. Lett. B 227, 352.
Linde, A. D. (1990a) Particle Physics and Inationary Cosmology, Harwood
Academic Publishers, Chur, Switzerland.
Linde, A. D. (1990b) Extended Chaotic Ination And Spatial Variations Of
The Gravitational Constant, Phys. Lett. B 238, 160.
Linde, A. D. (1991) Ination And Quantum Cosmology, Phys. Scripta T36,
30.
Linde, A. D. (1998) Quantum creation of an open inationary universe, Phys.
Rev. D 58, 083514 [arXiv:gr-qc/9802038].
Linde, A. D., Linde, D. A. and Mezhlumian, A. (1994) From the Big Bang
theory to the theory of a stationary universe, Phys. Rev. D 49, 1783 [arXiv:grqc/9306035].
Linde, A. D., Linde, D. A. and Mezhlumian, A. (1996) Nonperturbative Amplications of Inhomogeneities in a Self-Reproducing Universe, Phys. Rev. D 54,
2504 [arXiv:gr-qc/9601005].
Linde, A. D. and Mezhlumian, A. (1996) On Regularization Scheme Dependence of Predictions in Inationary Cosmology, Phys. Rev. D 53, 4267 [arXiv:grqc/9511058].
Linde, A. D. and Zelnikov, M. I. (1988) Inationary Universe With Fluctuating
Dimension, Phys. Lett. B 215, 59.
Martel, H., Shapiro P. R. and Weinberg, S. (1998) Likely Values of the Cosmological Constant, Astrophys. J. 492, 29. arXiv:astro-ph/9701099.
Mukhanov, V. F. (1985) Gravitational Instability Of The Universe Filled With
A Scalar Field, JETP Lett. 41, 493.
Mukhanov, V. F. and G. V. Chibisov, Quantum Fluctuation And Nonsingular
Universe, JETP Lett. 33, 532 (1981) [Pisma Zh. Eksp. Teor. Fiz. 33, 549 (1981)].
Page, D. N. (2002) Mindless Sensationalism: A Quantum Framework for
Consciousness, in: Consciousness: New Philosophical Essays, Eds. Q. Smith and
A. Jokic. Oxford, Oxford University Press. arXiv:quant-ph/0108039.
Randall, L, and Sundrum, R. (1999) A large mass hierarchy from a small extra
dimension, Phys. Rev. Lett. 83, 3370 [arXiv:hep-ph/9905221].
Rees, M. (2000) Just Six Numbers: The Deep Forces that Shape the Universe,
Basic Books, Perseus Group, New York.

34

Rozental, I.L. (1988) Big Bang, Big Bounce, Springer Verlag.

Starobinsky, A.A. (1980) A New Type Of Isotropic Cosmological Models Without Singularity, Phys. Lett. B 91, 99.
Starobinsky, A. A. (1982) Dynamics Of Phase Transition In The New Inationary Universe Scenario And Generation Of Perturbations, Phys. Lett. B 117,
175.
Starobinsky, A. A. (1986) Stochastic De Sitter (Inationary) Stage In The
Early Universe, in: Current Topics in Field Theory, Quantum Gravity and Strings,
Lecture Notes in Physics, eds. H.J. de Vega and N. Sanchez (Springer, Heidelberg)
206, p. 107.
Steinhardt, P. (1983) Natural Ination, In: The Very Early Universe, ed.
G.W. Gibbons, S.W. Hawking and S.Siklos, Cambridge University Press.
Tegmark, M. (1998) Is *the theory of everything* merely the ultimate ensemble
theory? Annals Phys. 270, 1 [arXiv:gr-qc/9704009].
Turok, N. (2002) A Critical Review Of Ination, Class. Quant. Grav. 19, 3449
(2002).
Vanchurin, V., Vilenkin, A. and Winitzki, S. (2000) Predictability crisis in
inationary cosmology and its resolution, Phys. Rev. D 61, 083507 [arXiv:grqc/9905097].
Vilenkin, A. (1983) The Birth Of Inationary Universes, Phys. Rev. D 27,
2848.
Vilenkin, A. (1984) Quantum Creation Of Universes, Phys. Rev. D 30, 509.
Vilenkin, A. (1995) Predictions From Quantum Cosmology, Phys. Rev. Lett.
74, 846 [arXiv:gr-qc/9406010].
Vilenkin, A. and Ford, L. H. (1982) Gravitational Eects Upon Cosmological
Phase Transitions, Phys. Rev. D 26, 1231.
Weinberg, S. (1987) Anthropic Bound On The Cosmological Constant, Phys.
Rev. Lett. 59, 2607.
Wheeler, J A. (1990) Information, Physics, Quantum: The Search for Links,
in: Complexity, Entropy and the Physics of Information, ed. W.H Zurek, AddisonWesley, pp. 3-28.

35