upgrade of results from monte carlo study of the o(4) invariant λΦ4 model in the broken phase
TRANSCRIPT
Nuclear Physics B (Proc. Suppl.) 9 (1989) 21-25 21 North-Holland, Amsterdam
U P G R A D E OF RESULTS F R O M M O N T E CARLO S T U D Y OF T H E O(4) I N V A R I A N T A# 4 MODEL IN T H E B R O K E N P H A S E
T H O M A S N E U H A U S
Fakult~t fiir Physik, Universit~t Bielefeld, D-4800 Bielefeld, F. 1%. Germany
A B S T R A C T : I summarize recent results from large scale Monte Carlo simulations of scalar 0(4) theory in the broken phase. The wave function renormalization constant Za, the mass of the scalar particle m~ and the scalar field expectation value ~ are determined. The effect of the finite lattice size on the scalar mass is estimated. The upper bound on the ratio rna/ fa is 2.6(1) at a value of the cut-off Acut ~ 2.5, corresponding to an triviality bound of 8.2(5) on the ratio of the Higgs mass over W mass. First results of an extensive calculation on various lattice sizes at finite small external source j and at a value of fG "" .4 are presented. Taking into account the effect of the Goldstone states via analytic formulae, the scalar field expectation value ~ is determined. The agreement of this value calculated at finite j and the calculation using rotated fields at j=0 is impressive.
1. I N T R O D U C T I O N
It is now understood that the four component 0(4) A# 4
model serves as the heart of the spontaneous symmetry break-
ing and vector boson mass generation mechanism of the Weinberg
Salam model defined at the Gaussian fixed point (GFP) [1]. While
it is almost rigorously proven that the renormalized quartic cou-
pling of the scalar model vanishes in D=4 in the infinite cut-off
limit, one can define an effective and interacting theory if one al-
lows for a finite and physical value of the cut-off Acu~. In such
an effective theory the ratio of the scalar mass mq over fa , the
Goldstone equivalent of the pion decay constant, is basically a free
parameter. It is expected however that R = ma/ fG is bounded
from above as one varies the bare quartic coupling at any fixed
value of the cut-off and this bound is saturated at a value oo for
the bare quartic coupling. Lowering the cut-off will increase the
bound. When A~,t is of the order of m~ quantities defined in the
lattice regularization will show significant deviations from a scal-
ing behavior characteristic of the Gaussian fixed point. Here the
effective theory loses its meaning and as a consequence a physi-
cally sensible value of R is bounded from above by R,na=, leading
to an upper bound of (rnH/rnw)rnaz = 2 X Rrnaz/gw [2], where gw denotes the renormalized SU(2) gauge field coupling constant
g~v - - . 4 .
Those considerations motivate the recent high precision
Monte Carlo (MC) calculations [3-5] of the scalar field expectation
value ~, the scalar mass ma, the wave function renormalization
constant of the scalar particle Z~, and of the Goldstone mode
z a = ~ l f~ . In analogy to QCD and its ~r meson is the low energy be-
havior of the 0(4) model governed by the Goldstone modes and
can be described by an 0(4) invariant non-linear o'-modeh Thus replacing the original theory by a low energy one with universal
properties [6], depending on its coupling constant fc , which is
the analog to f~ in the chiral model. Perturbation theory in the
finite box in ( f a x L) -1 then predicts formula for certain 0(4)
invariant correlation functions which allow e.g. the determination
of za [7,8]. In an alternative approach one can study the response of the
system to an external finite but small source j introducing an
explicit breaking of the 0(4) symmetry but still in a region of
couplings where a universal low energy model provides a sensible
description. The projection of the magnetisation to the direction
of the external source then becomes a well defined and positive
quantity even on the finite lattice. It can be used for the determi-
nation of the infinite volume ~ by comparing the measured data
with finite volume theorems for the magnetisation derived in the
framework of the low energy model at very small values of the
external source [9].
The work which I describe in these conference proceedings
was done in colloboration with A. Hasenfratz, K. Jansen, J. Jers~.k
, C. B. Lang and also in part with H. Yoneyama. Most of the
results at zero value of the external source have been published
already [3]. So I concentrate on the infinite volume extrapolation
of certain quantities and the comparison of our work with other
groups. The second part of this paper contains interesting new
results with a nonzero source, which will be published soon.
2. LATTICE A C T I O N AND MC C A L C U L A T I O N For the lattice action we choose the form
4
z6AD=I
+~ ~(+~¢~ - 1)= + ~ +~¢~ + ~ ~ ~:, (1) zEA zEA zEA
0920-5632/89/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
22 T. Neuhaus / R e s u l t s from Monte Carlo s t u d y
where to, ,~ > 0 are the hare coupling parameters. The fields
¢~ are real and we use the convention tha t summat ion over the
index a = 1 . . . 4 is implied, the • fields should be resealed to
the fields ~ = x / ~ . Here the emphasis of the presentation is on
results for bare quart ic coupling ,~ = oo and various values of x in
the broken phase, Results for non-zero j are presented at ~ = o¢
and ~ = .355. We have studied ensembles of configurations for
hypercnbic lat t ices A of size 84 up to 164 with direct MC simulat ion
and periodic boundary conditions. The updat ing method was a
2-hit Metropolis with more than .3 x 106 MC-sweeps per point
considered. The calculations at finite j required typically up to
106 sweeps. The calculations were performed on vector computers
CYBER 205, ETA10 and CRAY X-MP/48.
3. R E S U L T S A T J = 0
3.1 ~ E X P E C T A T I O N V A L U E
We measured the magnet isa t ion
1 M a = 7~'.4 Z ~ " (2)
x
and i ts length averaged over the Monte Carlo run ((MaMa')½1 is
close to the scalar field expectat ion value. Following Neuberger
[8] we have at hand the finite size formula
42 ( ( M t ' M a ) ~ ) 2 = ~2(1 + 2~-~) + O(L-41og'rL) (3)
~, I I ' ' ' ' I '
.4
.2
.0
I I I t I I
. 3 . 3 5
Figure 1. ~ as a function of ~ at )~ = co, triangles up are
124, crosses 144 and triangles down 164 lattices. We only plot the
data which were fitted in .305 < x < .34. The curve corresponds
to the fit .
valid for large and periodic lattices. We extract values for Y],
by using eq. (3) to order O(1/L ~) and then f i t our da ta on 124,
144 and 164 lat t ices from Table 1 with MINUIT to the sealing
form
= At, x t x/2 x (lnt) 1/4
t = ~ - ~---~ (4)
assuming exponents and logari thmic corrections of the GFP. The
results from the fit are A= = .822(1) and ge = .304(1), agreeing
very well with [4,5].
3 . 2 S C A L A R M A S S ma
We determine the scalar mass from the exponential decay of
the propagator of the "rotated" scalar field
~ M ~
~o,~ = 1[ M II (5)
a t zero momentum. We find tha t our correlation functions are well
measurable and consistent with the propagat ion of one massive
particle. The effects of resonant behavior due to the coupling of
Goldstone modes can be controlled. For the details of the analysis
I refer to [3]. The obtained values of ma are given in Table 1
showing a pronounced finite size dependence as can be seen in
Figure 2.
ma
0
I i I I I J I
O
O ~ / ~ l l i , i i o / 3~ ~(L~ I ' ' ' _ _ I''1
0 0 ~ " 2 L ~ ~
4 8 z I , , , , I
. 3 . 3 5 f
Figure 2. m,, as function of ~ at ,~ = ~ . Circles are 84,
triangles up 124 , crosses 144 and triangles down 164 lattices. The
curve corresponds to a infinite volume estimate. The inlet shows a
plot of the ratio mo(L) /mo(oo) against the variable z = L x m o ( L )
The data form a common curve supporting the validity of the
infinite volume estimate eq. (7).
The observed large finite size effects make a infinite volume
extrapolat ion for the scalar mass neccessary and we present here
an es t imate of the infinite volume mass. Guided by per turbat ive
arguments [13] we assume tha t the leading contribution to the
finite size dependence of the scalar mass from the Goldstone states
is given by A
m , ( L ) ~-- m,,(oo) + - ~ (6)
T. Neuhaus / Resul ts from Monte Carlo s tudy 23
at fixed values of the hopping parameter r . Here we have however
only few ~¢ values where such an extrapolat ion is leasable and we
are left wi th 3 points where we estimate the infinite volume scalar
mass. F i t t ing our da ta with eq. (6) we obtain m ¢ = .29(1) a t ~ =
.3075, ma = .40(1) at a = .31 and ma -- .62(2) a t ~ = .3175. We
note however tha t the determinat ion of the finite size dependence
of a resonance mass is far from tr ivial and should be investigated
more precisely.
.4
.3
0 .01 1/L~ Figure 3. Values of mo as function of 1/L 2 at ~¢ = .3075
on lattices ranging from 84 to 164 . The straight line shows the fit
according to eq. (6) leading to ma = .29(1)
Assuming finally critical exponents and logari thmic scaling
corrections of the GFP we fit the infinite volume masses to the
scaling Ansatz
mo = ASM x (t) 112 x (lnt) -114 (z)
where t has the same meaning as in eq. (4) and ~:c = .304 is fixed.
From the fit we obtain ASM = 3.94(9). We have also considered
mff using the Ganssian model relat ion m f f = 2 x sinh(m~/2).
Here we find ARSM = 3.98(9). The last ampl i tude can be compared
with [5] and is consistent with the number given there.
3.3 W A V E F U N C T I O N R E N O R M A L I Z A T I O N
The wave function renormalizat ion of the scalar field Za is
determined by the residue of the propagator of the "rotated" field
eq. (5) at nonzero momentum. For the wave function renormal-
ization of the Goldstone mode Z c we s tudy the operator
1
arEA,
and its 0 (4) invariant zero momentum correlation function having
contributions from the Goldstone modes and the massive s ta te as
well. Lowest order per turbat ion theory in ( f c L ) -1 in the frame-
work of the low energy model predicts the following parabolic de-
pendence
< 0~(0)0~(~) >
= C + Z a x ~ x ( ~ - )2+... (i0)
of the 0 (4) invariant correlation function on r for large values of
the distance r allowing the determinat ion of Za . The results for
Zc are given in Table 1, for more details see [3].
3.4 RESULT FOR THE RATIO AND COMPARISON
In Figure 4. we present our final result for the the ratio
R = ~- x Z~ at a value of A = o0 as a function of the cut-off in
units of the scalar mass. The triangles correspond to 124 , crosses
to 144 and triangles down to 164 lattices. In a region offthe cut-off
1/m~, ranging from 2-3 the upper bound on R is ]?~na~ - 2.6(1). The values for R may however be little overestimated as can be
seen if we include our estimate of possible finite size effects. Here
the main effect comes from the finite size dependence of the scalar
mass. Using the fits to E and mo according to equations (4) and
(7) and alternatively replacing ma by rnff everywhere we arrive at
curves labeled 1 and 2 in the plot. We assumed Za to be .97 for the
graph. We did not include errors in the curves, but the statistical
error is basically given by the error of the scalar mass while the
systematic errors of the FS extrapolation is unknown. We learn
that the finite size corrections of R at values of the cut-off 2-3 may
be as large as 5 percent even on a 164 lattice. Here one should
keep in mind that for the moment we are only in the position to
give a reasonable estimate of the finite size effect. Work in this
direction is in progress.
We can now compare our data [3] with the high statistics
numerical simulations of the group around J. Kuti [4,5]. Our
infinite volume estimate for R and their infinite volume extrap-
olation which corresponds to curve 3 are very close indeed. In
addition one can state that the numerical simulations coincide in
their raw data obtained on finite lattices [10]. Finally we also
compare the analytic calculation of Liischer and Weisz [11] with
the various curves. We find impressive agreement showing that
numerical and analytical methods are consistent within a few per-
cent in their determination of R at the considered values of the
cutoff.
~ I I I I
3 . 5
3
2 I 1 2 3 4
A
Figure 4. The ratio R as function of the cut-off Curves I and 2 are ezplained in the tezt. Symbols have the same meaning as
24 T. Neuhaus / Results from Monte Carlo study
in Fig. ~. The 3 diamond shaped data points are from the analytic
calculation by Lf=scher and Weiss using g1¢ = 3 x R 2 [11]. Curve 3
is taken from the fits to mo and ~ in [5] where I took the values of
the amplitudes given there and used a common xe = .304 for the
scaling curves. Z was chooses as above. We add to our graph the
result from a calculation in the SU(~) coupled gauge field Higgs
model at a value of fl = 8. [1~] (fat circles ), assuming m~t = mo undrape= 2 2 2 9 z ~g f~ with = .5.
4. RESULTS AT F I N I T E J
4.1 M A G N E T I S A T I O N S PARALLEL C O M P O N E N T
The coupling of an external source to the 0(4) model provides
the laboratory of an alternative, more rigorous and even more aes-
thetical description of the models properties in the broken phase.
One immediate consequence is now that the expectation value of
the magnetisations parallel component
1 < ~ - - ~ > (11) < Mp > = ~-4 m
is a positive number even on the finite lattice. Only if the external
source is 0 its average value is 0 showing that there is no sponta-
neous symmetry breaking on the finite lattice. In the case of the
j=0 simulations this difficulty was circumvented by assuming that
the length of the magnetisation vector serves as an estimate of Z,
however strictly spoken assuming that the state prepared on the
periodic lattice is a pure state in the infinite volume limit. Intro-
ducing the external source, this assumption is no longer needed
and one can try to determine ~ in a different way. It is possible in the framework of the low energy effective
model and analytic tbrmulae have been worked out by Leutwyler
and Gasser [6,9]. For the very small values of the external source the lowest order large volume theorem for the parallel component
of the magnetisation gives
1 ~Xz(s) (12) < Mp > - - ~ x 2 x X2(s)
where X2 (s) = I1 (2s)/s is a Bessel function of the scaling variable
s = ½ x E x f. We introduced f = j /(2n) "s. The formula describes
how the magnetization < Mp > vanishes in the "thirst limit"
j --* 0 on the finite lattice. It assumes that the lattice is large
compared to the massive a-field
1 - - << L (13) IT~ a
and that mo '(~ me. Here we perform the calculation at a value
of ~ = .355 and A = ~ where the scalar mass takes a value of
ma ~- 1.2 and fG -~ .4. The condition of eq. (13) is therefore
easily fulfilled if we choose lattices ranging in size from L=8 to
L=12. We also decided to choose such small values of j, that the
lattices were finite compared to the Goldstone modes
1 - - >> L (14) m G
with
m~ = ~ x f (15) J a
allowing for large finite size effects and a genuine study of those
effects in the chiral limit. Typical values of j were smaller than
.005. Such small values of the external source require a high statis-
tics calculation of < NIp > on the already mentioned lattices
and we have typically used 106 sweeps on the 84 and 104 lattices,
.3 x 106 sweeps on the 124 lattices. We performed a careful error
analysis of the magnetisation by introducing blocks of data with a
typical block length of a few thousand sweeps. We observed sat-
uration of the error determined from the blocked data at a block
length of the order of 104 sweeps. The results for < Mp > are
given in Figure 4., nicely exhibiting the huge finite size effect in
vicinity of the j=0 point.
We fitted all the data using eq. (12) and determined :E by a
global one parameter MINUIT fit. Using the lowest order formula
eq. (12) we obtain the value E = .409(4) from the fit, which
one can compare with the value E = .403(1) coming from the
infinite volume extrapolation at j=0, eq. (3). In addition one also
knows the next higher order corrections to eq. (12) [6]. Using
these one obtains ~ = .404(4), for a detailed analysis I refer to a
forthcoming paper. The agreement of the fitted numbers coming
from the calculation at finite source and the calculation at j=0 is
impressive.
I ' I ' I
.4 •
(Mp)
. 2
0
I , I , 1 , 0 . 0 0 2 . 0 0 4 j
Figure 5. < Mp > as function of the external source j at
A = go and t¢ = .355. Circles are 84, triangles 124 and crosses
124 lattices. The 3 curves correspond to the 1 parameter fit using
eq. (1~}. The fat circle plotted at j=O corresponds to the infinite
volume ~ value.
5. C O N C L U S I O N S
We determined the triviality bound on the ratio 1% of the
scalar boson mass ma over the analog of the "pion decay constant"
fa at a value 1~ms = 2 - 3. We obtain an upper bound of 2.6(1),
which for a renormalized gauge coupling of g~v = 0.4 corresponds
T. Neuhaus / Resul ts from Monte Cado s tudy 25
to a maximal ratio m H / m w = 8.2(5). We estimated possible finite size effects and our calculation is in very good agreement with other groups [4,5,11].
We also present for the first time results at finite external source and a determination of E from those data using low energy and large volume theorems. The gain is twofold. Firstly we find a remarkable agreement of the determined value for Z with the calculation at zero source, implying that the assumptions made there are sound and justified, at least for the large values of fa . Secondly we find that those theorems provide a consistent way to describe some properties of the 0(4) scalar theory in the broken phase. Work in this direction is in progress. A C K N O W L E D G E M E N T
The support of the Computer Centers at Florida State Uni- versity and KFA Jiilich, where the necessary computations were performed, is acknowledged. 6. R E F E R E N C E S
[1] A. Hasenfratz and P. I:Iasenfratz, Phys. Rev. D34 (1986) 3160.
[2] 1%. Dashen and H. Neuberger, Phys. Rev. Lett. DS0 (1983) 1897.
[3] A. Hasenfratz, K. Jansen, C.B. Lang, T. Neuhaus and H. Yoneyama, Phys. Lett. 199B (1987) 531 ; A. tIasenfratz, K. Jansen, J. Jers~,k, C.B. Lang, T. Neuhaus and It. Yoneyama, preprint HLRZ Jiilieh 88-02 and UNIGRAZ-UTP-03-88 to appear in Nucl. Phys. B.
[4] J. Kuti, L. Lin, Y. Shen, Nucl. Phys. (proe. Suppl.) B4 (1988) 397; Phys. Rev. Lett. 61 (1988) 678; U.C. San Diego preprints, UCSD/PTIt 88-05; UCSD/PTH 88-07; UCSD/PTH 88-06 (with S. Meyer).
[5] J. Kuti, San Diego preprint UCSD/PTH 88-12 , October 1988, invited talk at Munich conference, August 4-10,1988.
[6] H. Leutwyler, private communication ; P. Hasenfratz, talk at this conference.
[7] H. Leutwyler, Nucl. Phys. B4 (Proc. Suppl.) (1988) 248. [8] H. Neuberger, Phys. Rev. Lett. 60 (1988) 889; Nuel. Phys.
B300 [FS22] (1988) 180. [9] J. Gasser and H. Leutwyler, Ann. of Phys. 158 (1984) 142; J.
Gasser and It. Leutwyler, Phys. Lett. B184 (1987) 83; Phys. Lett. B188 (1987) 477; Univ. Bern preprint BUTP-87/22.
[10] J. Kuti, private communication. [11] M. Liiseher and P. Weisz, preprint DESY 88-083 to appear
in Phys. Lett. B; preprint DESY 88-146 (October 1988). [12] A. Hasenfratz and T. Neuhaus, Nuel. Phys. B297 (1988)
205. [13] L. Lin, Talk at this conference.
7. TABLES
L x < ~ > mo Zo R
12 .3075 .178(1) .38(1) .96(1) 2.67(7) 12 .3100 .214(1) .47(1) .97(1) 2.75(6) 12 .3175 .292(1) .67(1) .96(1) 2.82(5) 12 .3250 .347(1) .80(1) .96(1) 2.80(4) 12 .3330 .392(1) .92(2) .97(1) 2.83(7) 12 .3550 .480(i) 1.22(4) .97(1) 2.97(9) 14 .3075 .169(2) .35(1) 2.64(8) 14 .3100 .209(1) .43(1) .96(1) 2.56(6) 14 .3175 .289(1) .64(1) 2.78(5) 14 .3200 .310(1) .71(1) .97(1) 2.82(4) 14 .3250 .345(1) .81(2) 2.91(7) 14 .3300 .374(1) .87(2) .97(1) 2.82(7) 14 .3350 .400(1) .92(2) 2.81(6) 16 .3060 .135(1) .27(1) .96(1) 2.50(9) 16 .3075 .165(1) .33(1) .96(1) 2.50(7) 16 .3300 .373(1) .85(5) .97(1) 2.76(15) 16 .3550 .478(1) 1.09(10) .96(1) 2.65(23)
Table 1 : High precision data in the 0(4) model at A =
and in the broken phase. L is the lattice size, x the hopping
paramter, < ~ > denotes < (MaMa)½ > /(2x) 's, R is R = 1 k
mo x - :-r- x Z~ . At few data points on the 144 lattice we <(M~Mo)~ > did not evaluate Za for technical reasons. We then assumed the value .97.