322
Nuclear Physics B (Proc. Suppl.) 4 (1988) 322-326 North-Holland, Amsterdam
QCD AT FINITE BARYON DENSITY
A. VLADIKAS *Department
of Physics and Astronomy,
The University,
Glasgow G12 8QQ, U.K.
Direct attempts to measure the chlral condensate in QCD wlth dynamical fermlons and non-zero chemical potential fall to predict the chlral symmetry restoration transition point. We propose an alternative method and present some early results.
1.
INTRODUCTION
2. NC SIHULATIONS WITH DYNAMICAL FERNIONS
The work reported here i s being done in
In o r d e r to r e s o l v e
c o l l a b o r a t i o n with lan Barbour of Glasgow U n i v e r s i t y and Clive B a i l l l e of Caltech.
the above d i s c r e p a n c y ,
we i n c l u d e Kogut S u s s k l n d f e r m i o n s i n t h e The
theory.
One t y p i c a l
MC ( m u l t i ) s w e e p c o n s i s t s
computers used are an IBM In Glasgow and the
of visiting
Crays XMP in Rutherford and San Diego.
l a p p i n g a r o u n d e a c h o f them 4 t i m e s ,
We
half
the lattice
have studied the c h l r a l symmetry r e s t o r a t i o n
each of i t s
links
t r a n s i t i o n a t (almost) zero temperature and
we u s e t h e
(exact)
non-zero chemical p o t e n t i a l in SU(3) with
calculate
dynamlcal fermlons.
and t h e d i f f e r e n c e
We have been constrained
hypercubes,
10 t i m e s .
hitting
While we u p d a t e ,
Lanczos a l g o r i t h m
the ratio
of f e r m i o n i c of the traces
[sJ to
determinants of the i n v e r s e
to very small l a t t i c e s and very low s t a t i s t i c s
o f t h e f e r m t o n m a t r i x M.
which, we hope, w i l l reveal the b a s i c mechanism
M-1 and M-2 f o r t h e h y p e r c u b e i n q u e s t i o n .
c h a r a c t e r l s l n g the t r a n s i t i o n , although they
Details
w i l l not give p h y s i c a l l y r e l i a b l e answers.
action
The chemical p o t e n t i a l u (in l a t t i c e u n i t s ) i s
fashion.
included by m u l t i p l y i n g l a t t i c e l i n k s in the
15 m i n u t e s on t h e Cray.
positive e~(e-U)
(negative) [1,2].
The objective is to evaluate
the chiral condensate
as an order parameter
can be f o u n d in [ 6 ] .
For SU(3),
for
Metropolis
such a multisweep takes
problem is that
f o r SU(3),
detM i s c o m p l e x when ~ 0 .
Since the
Metropolis
on c o n f i g u r a t i o n s
algorithm
relies
the restoration
of chlral symmetry with
being generated
increasing u.
This transition
we use W=ldetM{ exp(-Sg).
is expected to
end, we need
The W i l s o n
Sg i s u p d a t e d i n s t a n d a r d
The m o s t s e r i o u s
direction by a factor
To t h i s
via a real
positive
w e i g h t W,
W i t ht h i s weight,
occur at a critical value Uc=(mp/Nc)
(mp:= the
the condensate becomes a r a t i o of two
proton mass, Nc:= number of colours)
[2] but
observables
quenched approximation
( m . : = the pion mass)[3].
values
coincide
for
SU(3)
this
for Is
has been attributed approximation
w
results suggest a value
uc=m./2
SU(2) a
The two
(Nc=2, mp=mw) but
discrepancy,
which
to the quenched
[4].
=
(1)
w
w h e r e ~ i s t h e p h a s e o f detM. imaginary parts
denominator of the above r a t i o identically
[7].
Note t h a t
vanish
We p e r f o r m e d MC m e a s u r e m e n t s
o f t h e n u m e r a t o r and d e n o m i n a t o r o f eqn. b a r e q u a r k mass mq-0.1 and i n v e r s e * I.N.F.N. F r a s c a t l , C.P. 13, 00044 F r a s c a t i , Italy
0920-5632/88/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
the
o f b o t h t h e n u m e r a t o r and
6-1.5 on a 44 l a t t i c e .
(1) a t
coupling
We work in strong
coupling so as to be able to compare with
A. Vladikas / QCD at finite baryon density similar
results
[8] and a l s o
decrease
correlations.
updating
each hypercube
measurements/sweep) values
of u=0.3,
uncorrelated) present
distribution
(i.e.
for
0.9,
language of refs.[ 3,4]
1.0.
the massless
5
Because
function
fluctuating,
p of the numerator
In t e r m s o f p ( x ) ,
any observable
x c a n be w r i t t e n
the results
is
0/0 answer.
and
the average
of
for
X
xi
i=1
= I
the case where of eiffenvalues In Figure
with
the
a nd
In thls
as
the
indeterminate
The same b e h a v i o u r
u = 0 . 4 a nd 0 . 9 .
of
2 we
Since the distributions
with being symmetric,
occurs
region,
at
t h e mass i s
fCOS(~) 16
max
N 1
is
In the
the denominator
compatible
X
:=
this
the strip
at ,=0.5.
condensate
expectations.
fermion matrix.
are compatible
we
the normalised
denominator.
plot
numerator
( w h i c h w e r e f o u n d t o be
through
with naive
t h e ma s s i s o u t s i d e
32
5 s w e e p s and f o r
0.5,
were wildly
results
which agrees
to
We m e a s u r e a f t e r
0.4,
our measurements
in order
323
(2)
x p(x) dx
Xmin
1./.
for a large number of measurements
N.
In Figure I we show p(cos¢) for .=0.3.
Its
1"2
asymmetry resulted in our being able to obtain a condensate of about 200 (in lattice units),
1.0
5'0
0-8
e(cosm) 0"6 ~0
tto[, 3.0
cos(C) -1-0 Figure
20
-0"6 2a:
-0'2
0-6
Same a s i n F I g .
~(Re[ei# Tz M" ])
{
0-2
10
1 for .=0 5
,0"01 ,0008
10
1ooo6t
}
t
t
I i I
-.-oa ~-06' t-oL, , -oz-O~
I
I
01 0-2
I I
cos(m) 0/,,
06
08
1"0
f
0.001 Re[el# Tz Mm ]
Figure I: The normallsed distribution of measurements for the denominator of the condensate at u=0.3
-114 F i g u r e 2b: Same a s i n F i g . 2 a of the condensate
112 for
the numerator
A. Vladikas / QCD at jqnite baryon density
324
inside the strip of eigenvalues [3,4],
of M(mq=O)
and MC appears to be failing.
behaviour,
j~(Re[ei~Tz M-i]}
This
1.0
a result of a rapidly fluctuating
phase of detff, is in disagreement analysis of [7].
wlth the
But at u=l.0, Figure 3
suggests that the numerator has a symmetric distribution denominator
(i.e. it vanishes) does not.
0.6
but the
Thus the condensate
either very small or compatible with O.
Is
O.t,
In
this case the mass sits inside the cavity formed by the elgenvalues
[3,4].
The above results suggest that in the region , ½,
u¢(0.3,1.0),
the condensate
cannot be measured
-1.o
-o.~
~4
o
25 Figure 3b: by MC;
.p(cos ~ )
~. Re,ei#,z . ~ 1- H") 1.45
Same as in Fig.2b for u-l.0
this behaviour may signal that we are
near the critical
region.
At any rate, our
results differ from the quenched result
2.0
~c-0.3.
This is mildly encouraging.
3. THE DETERMINANT
EXPANSION
In order to deal wlth the above problems,
we
use the fact that detM can be expanded in terms of Wilson loops and Polyakov lines.
,1.5
We group
these expansion terms according to the number of time-like Polyakov lines they contain. Each such Polyakov
llne contributes
exp(Nt# ) or exp(-Ntu)
a factor
and thus we have
M
1.0
detM=Co+ ~ [c m=l
exp(mNt~)
+ c: exp(-mNt~)]
where M:= Ns3Nc in obvious notation. coefficients space-llke tlme-llke
O.S
contain contributions
loops and lines, as well as from
+t as in the -t direction.
coefficients eqn.(3).
-1.0
-~6
-0-2
0.2
0.6
because
This explains
of the +~ and -~ exponentials
The last terms'
Same as in Flg.2a for ~-1.0
coefficient
it consists of the determinant
product of all the time-like Polyakov
interpreted
as representing
the
of the
Each of the terms in the expansion Figure 3a:
The
from
loops which have as many links in the
complex conjugate relationship
cos @
(3)
m
in
is I, of the lines.
can be
a contribution
from
A. Vladikas / QCD at finite baryon density
quarks
325
(anti-quarks) winding around in the
TABLE 1
forward (backward) time direction m times. Now recall that under Z 3 transformations of the t-like gauge links, we obtain a valid SU(3) configuration which leaves the path integral measure and the pure gauge action invariant but changes d e t N .
The real part of the terms (C3m exp(3mNtu)} of the determinant expansion (eqn.(6)), for typical configurations at different values of u. The lattice size is 23x4. m 0
~--0.2 5.25 0.74E-1 0.32E-2 -0.48E-4 -0 11E-6 -0 32E-9 -0.16E-12 0.42E-15 0.14E-19
We may thus write the partition
function as follows I Z = ~ ~ [DU] exp(-Sg)
3 ~ det(Mi) i=1
(4)
where the 3 determinants differ by such a Z 3 transformation.
2 3 4 5 6 7 8
u=0.4 12 8 0.81E-1 0.37E-1 0.12E-1 0.20E-3 0.11E-5 0.55E-7 0.34E-9 0.25E-12
u=O.6S
u=0.5 0.08 0.30 0.81 -0.97 -0,17 -0 27E-2 -0,49E-4 0.75E-6 0 11E~7
0.75E-4 -0.40E-2 -0.14E-1 0.88 -0.86 -0.67 -0.20E-1 0.59E-2 0.53E-3
Thus, the problematic exp(u) takes over and more terms come into
denominator of eqn.(1) can now be obtained as 3
play.
3
These terms correspond to the lattice
being gradually filled up by quarks that wind i[DU]exp(_Sg) idetMll~, det M i - I d e t M11
=
<
~ - det M i > w I d e t M11
(5)
around 3,6,9 etc times in the standard picture of ref.[ 2], until at large enough u, the whole
with the same MC weight as before.
The
lattice is completely filled up (i.e. the last
important fact is that the sum of 3
term with exp(NtM~) dominates).
determinants which appears in eqns.(4) and (5), when evaluated using eqn.(3),
becomes:
values u~0.6 are characterlsed by cancelling
3 M ~detMi=3co+~. [C3m exp(3mNt~)+C;m exp(-3mNt~)](6) i =1 m=l
"noise" which is
If, however, the noise is
similar
u=0.4 v a l u e
lattice.
It
again
difficult.
To assess how relevant these considerations
even
at
when'dealing
with
a 44
may thus
be necessary
strength of each contribution from the positive (6).
Table I we show such contributions
from
clear
directions.
Then, instead of dealing with the
sum of 3 determinants,
our is given by Early trial
runs have shown that a few results obtained in
very different from those obtained in Section 2
expectations of [2].
Work which is currently
in progress along these lines will hopefully Note
clarify the situation.
t h a t f o r small u the z e r o t h o r d e r term c0 dominates.
Thus,
f o r t h i s small l a t t i c e we
and found it to agree with its ~=0 value. the strength of the powers of
ACKNOWLEDGEMENTS I would llke to thank my collaborators,
were a b l e to measure the condensate a t u=0.4
increases,
up
and possibly in accordance with the theoretical
In
equilibrium configurations at different f o r a v e r y small 23x4 l a t t i c e .
to
the
this way at a=0.4 on a 4 ~ lattice are certainly
are in practice, we examine the relative
exponential terms in the expansion
We
difficulties
the sum of 34 determinants.
cancelled expllcitly.
values,
is
trick of Z 3 group rotations in all four
statistics to cancel and may be hiding our
we evaluate using eqn.(6),
results
the signal from further noise by using the same
present in eqn.(1) and will require enormous
signal when evaluating eqn.(2).
reliable
smaller
The cancelling
terms of eqn.(3) are, however,
dominant terms, which means that obtaining
encountered
i.e. only loops that wind around the lattice 3,6,9 etc times contribute.
Unfortunately,
we see from Table 1 that the intermediate
As
Clive Baillie and Ian Barbour for their contributions to the above.
I am also
A. Vladikas / QCD at finite baryon density
326
grateful to Phil Gibbs, Christine Davies and Peter Hasenfratz for useful discussions.
5.
REFERENCES 1. P. Hasenfratz and F. Karsch, Phys. Lett. B125 (1983) 308. 2. J.B. Kogut et. al., Nucl. Phys. B225[FS9] (1983) 93. 3.
I.M. B a r b o u r e t . (1986) 296.
4. P.E. Gibbs, "Understanding Finite Baryonic Density Simulations in Lattice QCD', Glasgow preprlnt, April 1986 (unpublished) and Phys. Lett. B182 (1986) 369.
al.,
Nucl.
P h y s . B275[FS17]
I.M. B a r b o u r e t . (1987) 227.
al.,
J.
Comput. P h y s . 68
6. C. B a i l l i e e t . 8 1 . , "The C h J r a i C o n d e n s a t e i n SU(2) QCD a t F l n l t e D e n s i t y " , G l a s g o w preprlnt, to appear in Physics Letters. 7.
B. Berg e t .
al.,
8.
E. D a g o t t o e t . (1986) 1292.
Z. P h y s . al.,
C3___11 (1986)
P h y s . Rev. L e t t .
167. 5_77