You are on page 1of 4

VOLUME 61, NUMBER 2

Particlelike

Centre for Mathematical

PHYSICAL

REVIEW LETTERS

Solutions

of the Einstein- Yang-Mills

Robert Bartnik

Analysis,

Australian

(Received

and John

McKinnon

National

5 February

University,

1988)

Canberra,

We

study

the

static

spherically

symmetric

Einstein-Yang-Mills

equations

Equations

A. C. T. 260l, Australia

with

SU(2) gauge

group

11 JULY 1988

and

find

numerical

solutions

which

are

nonsingular

and

asymptotically

flat.

These

solutions

have

a

high-density

interior

region

with

sharp

boundary,

a near-field

region

where

the metric

is approximately

Reissner-N4rdstrom

with

Dirac

monopole

curvature

source,

and

a far-field

region

where

the

metric

is

approximately

Schwarzschild.

 

PACS numbers:

04.20.Jb, 11.15.Kc

 

Introduction

ither

N. e—

the vacuum

Einstein

equations

nor the

globally regular (i.e. , nonsingular, asymptotically flat)

solutions.

pure

Yang-Mills

equations

Deser

have

nontrivial

this

also

static

by

that

For pure

Yang-Mills

'2

and

fields,

has

was shown

shown

Deser and Coleman,

there are no static Einstein- Yang-Mills (EYM) solutions

in D=2+1. 3 The corresponding result for vacuum

gravity

theorem,

due

and for Einstein-Maxwell

with

modifications

Robinson, '

Israel, 4

fields is Lichnerowicz's

permitting interior horizons Bunting and Masood-ul-

to

behavior

pole

and

modeled

closely

the

metric

is

on

the

very

Reissner-Ndrdstrom

solution.

Dirac

close

In

the

ing"

transition

zone,

Ro & r & R ~,

magnetic

the

mono-

to

extremal

"charge-shield-

the

Reissner-

Nprdstrom charge gradually decays to zero, and in the far-field region, r & R ~, the solution approximates the Schwarzschild solution, with zero YM charge integrals.

from

these results:

There

are

two

important

lessons

to

be

drawn

(1) The

gravitational

interaction

cannot

be dismissed

Alam. b

So it

is

natural

to conjecture

that

the

coupled

as too weak to be of consequence the existence

of these

EYM equations

also

have

no nontrivial

globally

regular

solutions

depends

essentially

on

the

interaction

between

solutions.

the YM and Einstein

equations.

 

In this

paper

we present

strong

(numerical)

evidence

(2) Both

nontrivial

Yang-Mills

vacua

and

"symmetry

for the existence of a discrete family of globally regular

solutions

attraction

of the static EYM equations

can

balance

the

Yang-Mills

the gravitational

(YM)

repulsive

force.

The

solutions

we

find

have

gauge

group

SU(2)

and

three

terior

are

spherically

regions,

symmetric.

Their

behavior

shows

with

two transition

zones.

The in-

by high

YM curva-

r =1 is

zone about

distinct

region,

r & 1, is characterized

The transition

ture and stress-energy.

marked

metric

damped

by

large

fluctuations

These

in

the

connection

where

and

coefficients.

in

the

fluctuations

are 1 & r & Ro,

and

has

rapidly

the

near-field

decays

region,

YM

curvature

polynomially

dominant

breaking"

can

occur

without

involvement

of the

Higgs

mechanism. The equations contain no Higgs fields and are topologically trivial, yet these solutions have nonzero YM curvature, which progressively degenerates su(2)

u(1)

0 as r

EYM equations and boundary conditions. The

SU(2) connection has been de-

spherically symmetry

scribed

the

in many

places.

' If we let z;, i =1, 2, 3, denote

the usual polar coordi-

usual

basis of su(2), 8 and

nates on S2, and parametrize the space of 5 orbits of

p

the symmetry

group

by (r, t ), the connection

can be writ-

ten

3 =az3dr+ bz3 dr+

(cz~+ dz2)d8+ (cot8z3+ cz2 d z~ )sin8dp,

where

a,

b,

c, and

d are functions

of (r, t). This connec-

tion arises

from the global

symmetry

group

SU(2) rather

than SO(3). '

 

By changing

coordinates,

we

can

write

the

spherical

metric

as

2 T 2dt 2+R 2 dr 2+r 2(d82+sin28d~2)

(2)

(r

(R

R

'T

'Ta')'

2RT(c +d )a= 0,

'c')'+RTa2c+r

2RT

cd' dc' =0.

'(1 c

d)c =0,

We consider only time-independent

per, and

solutions

in this

so the functions

a, b, c, d, R, and

T depend

pa-

now

only

on r.

The connection

A has a residual

U(1) gauge

freedom,

h

'Ah+h

'dh

use to impose

the radial

where

gauge

h(r) =exp(tirz3),

b:0.

which

we

The YM curvatu—re

Using

the

=g'bg'

remaining

with

from

gauge

at

freedom

r =0.

The

we

can

EYM

set

equations

d =0,

c=w E R

derived

w~0

the

f ( R+ I F I )v gdx,

Lagrangean

F„Fba, are

Ricab =2F .Ff

2

I

F I

gab

1988 The American

Physical

Society

141

VOLUME 61, NUMBER 2

PHYSICAL

REVIEW LETTERS

11 JULY 1988

We now

assume

a =0, so that the YM curvature

is purely

magnetic

and there

are not

dyons. '

In fact, one can show

that this assumption follows from suitable asymptotics and finite energy. If we introduce m(r) by R = (1 the static spherically symmetric EYM equations reduce to

2m/r)

m ' = (1 2m/r )w '

+

'

—,

(1 w ) /r,

r

(1

2m/r)w

'+[2m (1 w ) /r]w'+(1 w )w =0,

with

the supplementary

equation

2r(1 2m/r)T'/T=(1

w

and YM curvature

tensor

)

/r

2(1 2m/r)w'

2m/r,

F =w'r~ dr Ad8+ w'r2dr Asin8dg

(1 w

) r3d8Asin8dg.

We

tures,

introduce

the

radial

(1 w')

r2

'

and

angular

w'

rR

magnetic

curva-

and

satisfies

so

R/T

is

increasing

and

T&R) 1.

Since

(6)

T

and note that

the local energy

4&Too= I~r I

+ 2

I&L I

density

is

We impose

the

asymptotic

conditions

as

two explicit solutions satisfying w= 1, then

r

~, so

that

R(r)

1, and

T(r)

these

m(r) =M, constant,

and R =T=(1

(7)

m(r) ~ M & ~

1.

There

conditions.

2m/r)—

are

If

which is just the Schwarzschild

YM curvature.

constant,

If w=0, then

and

2m(r) =2M (1+e )/r,

metric,

with

a(r)

a=an

vanishing

e/r,

ao,

e

R =T=(1 2m/r)

This describes the Reissner-Ngrdstron (RN) metric with

mass

M, electric

YM curvature

charge

e, magnetic

charge

g =1, and

F = (e/r

) r3dr Adt r3d8Asin8dp

Since F is u(1) valued,

monopole

by Harnad,

with

e =0, this represents

a solution

a Dirac

source for the RN metric,

Shnider,

and Tafel. '5

also noted

Of course, the RN and Schwarzschild solutions both

condi-

have

tions

quirement

singularities.

We now consider

for nonsingular

the boundary

solutions.

at r =0 needed

that

The

re-

the density

Too be finite implies

2m(r) =O(r ), w(r) =1+0(r ), w'(r) =O(r) as r~ 0.

Together

regular

with

at

T'(0) =0, this

a result

smoothly

ensures

that

the

metric

now

is

r =0, and

extends

of Uhlenbeck across r =0.

shows

that the bundle

There

are

some

elementary

observations

about

these

equations.

We have

(lnR/T)'

=2w' /r,

=2(1

2m/r—)w' /RT+(R/r

T)(w

1)w

we

for

see that

T' & 0

w

'

—,

in

the

form

[(w')'/RT]'

if

w

~1. If we

write

the

equation

=w'2/RT+

(R/r

T)(w

1)w

w

either

ww"

oscillating.

Numerical

trivial solutions ing the formal

& 1

& 0 if w'

or

w

~1 if the total

=0 and

w

& 1, which

mass

is finite.

is consistent

Also,

with

w

solutions

Our .

method

of

of (3)-(9) is quite simple

power-series

expansion

about

finding

minded.

r =0,

non-

Us-

2m=4b

r

+ "

,

b

w

'b

=i+br +(

r

+O(r

+ —, ', b

)r

),

+O(r

)

b g R,

we

construct

tolerance

initial

data

b,

the

at

we

r =0.01 for

a

"shoot"

solver.

global

were

found

standard-

By adjust-

solutions.

package

ing

the

With

ordinary-differential-equation

free

parameter

10

',

to vary

continuously and regularly with b, indicating that this numerical procedure is well behaved.

solutions

For b & 0.0706, the solver

broke down

at

r & 1

with

w' large

and

negative.

For

solution

oscillated

in

Iw

I

0.706 & b & 0, the generic

& 1 before

crossing

Iw

I

=1

and

rapidly

going

to infinity.

However,

at a discrete

set

of values

range, seen to be asymptotic

in this

the solutions

to w = +

after

oscillating

are

1

(Fig. 1). We can index

these solutions by k =number

of zeroes

of w.

There

are

three

r & 1; the

field region

distinguished

near-field

regions:

region

the

inner

core

II, r & 1, w-0; and

III, r»1, w ~ 1.

region

I,

the far-

The stress-energy

but

decays

rapidly

density

(Fig. 2).

(7) is large In region

cating

that

the

solution

approximates

in the inner

core

II, BL»BT, indi-

the

U(1) Dirac

monopole.

142

VOLUME 61, NUMBER 2

PHYSICAL

REVIEW LETTERS

11 JULY 1988

6 L~x

RADIUS

2'

Wj

FIG. 1. Connection

parameter

 

3

 

RADIUS

wj„k =1, 3, 5.

FIG.

2.

Energy

density

in the interior,

k

3.

The

mass

function

m(r)

increases

to

a limit

Mk & 0,

where

the total

mass Mq was found empirically to satisfy

'

.

a large

The

peak

metric

coefficient

at

r- 1,

the

size

R =(1

of the

Mi, =1 1.055e

2m/r)

'i2 has

peak depending on k. As suggested above, in the near-

field

e

region

II, R approximates

magnetic

charge

g

the

RN

coefficient

with

=0 and

=1. This

can be clearly charge,

seen by definition

of the RN (magnetic)

g'(r) =2r(Mk m).

Notice

(Fig. 3) that

the charge

is shielded

in the transi-

re-

gion III the metric is approximately Schwarzchild with mass Mi, . EYM-Higgs perturbations of the extremal RN metric about the internal infinity have been studied

near

tion

between

regions

and

may

II and

III, and

in

to model

the

far-field

by Hajicek'

r =1.

apply

the behavior

In the

far-field

region

sion,

for some

c & 0

we have

the

asymptotic

expan-

w

1

c/r,

IBTI -2c/»',

m -M c /r, IBL I -c/»'

Thus the YM curvature decays polynomially and the to- tal YM charge integrals vanish.

Discussion

length

Altho.

ugh

the

model

has

an

cm)

unspecified

radius

scale (we have

is fixed

c =G = I), the ratio mass

=1 (by radius

we

mean

i

of

ra-

these solutions

dius of the inner-core region). Assignment cle radius 1 Planck length (1.6X10

the

of the parti-

leads

to

a

There

are

strong

similarities

with

magnetic

monopole

models, ' but there are also some significant differences. In the near-field region, the dominant contribution BL to

the

YM curvature

closely

approximates

the U(1) Dirac

monopole, but the transverse component

small,

tially.

appears

In the

to decay

far-field

polynomially

region

and

the curvature

BT, although

not

exponen-

decay

is still

polynomial but O(r

) rather

than

the

U(l)

monopole

O(r ), and so the YM charges vanish trivially.

If we take stability ergy functional,

to mean

that

the Hessian

of the en-

1(w)=(IBLI +2IBTI')y dy,

is positive definite, then simple heuristics

observed numerical behavior

(1Q)

the

at

If

based

on

(Fig.

1) indicate

that

can be stable.

best only

the

k =1 and

k =2 solutions

we assume w-0 on the interval [1;8], the Hessian can be approximated by

a'l(w)(y, y) =& Jp

y' —, y'

2

dr

8 I has negative

eigenvalues by testing linear function &=1

and linear elsewhere. This indicates that the solutions

for p E Co ([1;8]). We readily see that

for

with the continuous, piecewise

2 &r &4,

=0 for

r & l, r &8,

with

w-0 then

k ~ 3

are

unstable,

includes

[1;8].

as the

near-field

region

where

mass of 1 Planck mass (2.2X10

g).

Changing

of the

coupling

constant

does

not

affect

this

ratio:

The

NASS

Lagrangean

f( R+X IF I

)Jg

has

the

well-known

scaled solutions

mq(r) =km(r/A. ),

wq(r) =w(r/X),

Tq(r) =T(r/A, ),

where

m,

w, and

T are static

spherical

solutions

for the

original

Lagrangean.

This

rescaling

does,

of

course,

affect

satisfies

the

YM

B,(r) =~

curvature the

'B(r/~).

rescaled

magnetic

field

10-1

1

io

X

io'

FIG. 3. Mass, RN charge,

io'

and

10

w for

RADIUS

10

k =3.

143

VOLUME 61, NUMBER 2

PHYSICAL

REVIEW LETTERS

11 JULY 1988

By

analogy

the

with

review

the

superposition

expect

solutions

superposition

which

EYM

ex-

for

solu-

Finally, we

tremal RN solutions,

tions for the static EYM equations as well.

mention

some "no go" results

These results

show that it is a strictly nonperturbative effect.

we may

article

for

of Malec, '

the

static

presents

equations.

do not, of course,

cover our solution,

but do

We

would

like

to thank

our

colleagues

in

the

Centre

for Mathematical Analysis for their advice and com- ments, especially Nalini Joshi and Lars Wahlbin.

'S. Deser, Phys. Lett. 64B, 463-465 (1976). 2S. Coleman, in Ne~ Phenomenon in Subnuclear Physics,

edited

by A. Zichichi

(Plenum,

New York, 1975).

3S. Deser, Class. Quantum. Grav. 1, Ll-L2 (1984).

4W. Israel, Commun. Math. Phys. 8, 245-260 (1968). 5D. Robinson, Gen. Relativity Gravitation 8, 695-698

(1977).

G. Bunting

and

A. Masood-ul-Alam,

tation

19, 147-154 (1987).

Gen. Relativity

Gravi-

7S. Hawking

and

D. Ellis,

The

Large

Sca-le

Structure

of

Spacetime

(Cambridge

Univ. Press, Cambridge,

8P. Chrusciel

1881 (1987).

and

W. Kondracki,

Phys.

Rev.

1973).

D 36, 1874-

9E. Witten,

'

P. Forgacs

Phys. Rev. Lett. 38, 121 124 (1977).

and

N.

Manton,

Commun.

Math.

Phys.

72,

15-35 (1980).

"J. Harnad,

S. Shnider,

and

L. Vinet,

J. Math. Phys.

(N.Y.)

21, 2719-2724 (1980).

' J. McKinnon,

Honors

thesis,

Australian

National

Universi-

ty, 1987 (unpublished).

'

R. Bartnik,

to be published.

'4B. Julia and A. Zee, Phys. Rev. D 11, 2227-2232 (1975). '5J. Harnad, S. Shnider, and J. Tafel, J. Math. Phys. (N.Y.)

21, 2236-2240 (1980).

'6P. Hajicek, J. Phys. A 16, 1191-1205 (1983).

A.

Jaffe

(Birkhauser,

and

Boston,

C.

Taubes,

1980).

Vortices

and

Monopoles

'SE. Malec, Acta Phys. Pol. B 18, 1017-1064 (1987).

144