Volume 118B, number 4, 5, 6 PHYSICS LETTERS 9 December 1982
PRECOCIOUS SCALING IN LATTICE GAUGE THEORIES
F. GLIOZZI, F. RAVANINI 1 and S. SCIUTO
Istituto di Fisica Teorica dell'universitd di Torino, Turin, Italy
and Sezione di Torino dellTstituto Nazionale di Fisica Nucleare, Turin, Italy
Received 26 July 1982
We propose a method to evaluate numerically and in some cases analytically the two-loop contributions to.physical
quantities without computing Feynman graphs. Such contributions are negligible for the SU(2) Wilson action, which
shows a precocious scaling; on the contrary they are important for other actions (including Manton and heat kernel ones)
and account for the observed violations of universality.
I. Introduction. Many different lattice actions can
be used to regularize the same continuum field theory.
Universality [ 1 ] assures that the ratios among physical
quantities have the same values in any formulation as
long as one is close enough to the continuum limit, so
that the correlation length is large compared to the
lattice spacing a.
Near their critical point (vanishing g2) the whole
set of lattice actions corresponding to an asymptotic-
ally free theory are described in terms of only two
parameters: the bare coupling constant g2 and the
mass scale A, which depends on all the remaining pa-
rameters of the lattice action.
In this region a physical quantity M with the di-
mension of a mass can be written in the following
way:
(Ma) = (rn/A)exp(-1/bog2)(bog2) -b¢2b~
X [1 + O(g2)] . (I)
In eq. (1) b 0 and b 1 are the first two coefficients of
the Callan-Symanzik beta function 13(g) defined by:
(J(g) = a(d/da)g = bo g3 + bl g5 + b2 g7 + .... (2)
It is well known that only b 0 and b 1 have a universal
meaning; instead the other coefficients b n (n >~ 2) de-
pend on the renormalization scheme, that is they de-
1 Scuola di Specializzazione in Fisica Nucleare dell'Universith
di Torino, Turin, Italy.
402
pend on the specific form of the lattice action. The
range ofg where eq. (1) can be applied neglecting the
O(g 2) corrections is called scaling or critical region.
Out of this region the lattice theory looses its univers-
al character as the O(g 2) corrections depend on the
non-universal coefficients of the beta function.
Some years ago Creutz [2] showed that in the
SU(2) and SU(3) lattice gauge theories with Wilson
action the scaling region is large enough to be reached
by Monte Carlo simulations in microscopic crystals
stored in a computer.
Similarly it has been recently shown [3] that in
the associated SU(2) X SU(2) spin system the strong
coupling expansion of the mass gap nicely fits to the
scaling law (1) for rather large values ofg.
Universality provides us with an important consis-
tency check of these numerical results: dimensional
quantities obtained in different renormalization
schemes can be compared if the ratio between the cor.
responding A's is known. Such a ratio has been calcu-
lated by different methods [4-9] and for various lat-
tice actions [10,11] ; the outcome of this comparison
is at first sight rather intriguing. Although in some
cases, like in the asymmetric lattices [3,11,12], the
results are in good agreement with universality, in
many others, like for the Manton, heat kernel [13]
and mixed fundamental-adjoint actions [ 14-17 ],
there is a large discrepancy between the ratios of A's
measured by Monte Carlo experiments and the exact
values calculated at g = 0.
0 031-9163/82/0000-0000/$02.75 © 1982 North-HoUand