The topological susceptibility χ
is calculated on the lattice for
the pure SU(3) gauge theory both at zero temperature and across the deconfining
transition. We obtain χ ≈ (170(7)(4) MeV)4
at zero
temperature (the first error comes from our simulation and the second error
derives from the determination of the physical units) while at the transition
the following figure describes our main results:
The horizontal line and the grey band are the value and error of
χ at
zero temperature. Data are obtained from simulations on a
323x 8 lattice
with two different operators (called 1-smear and 2-smear respectively).
The numerical value obtained for χ is in agreement with the
expectation from the analysis of the UA(1) problem.
In the Erratum Nucl. Phys. B679 397 (2004) a few trivial numerical errors are
corrected. The new results and figures are qualitatively identical (the figure
shown above is the corrected one). The old version of the paper together with the Erratum
(appearing as an appendix to the old paper) can be found in
hep-lat/9605013.
The triviality problem of the λΦ4
theory in 4
dimensions with N fields is studied by calculating the Gaussian Approximation of the
Effective Potential of the theory. Under such an approximation two
apparently interacting phases are
found. The bare coupling constant λB
runs with the cutoff Λ as
λB∝ 1/log(Λ/Ω) where Ω
is a mass parameter
related to the scalar particle and is one of the
variational parameters in the trial wave functional of the vacuum.
These two are technical papers for full QCD lattice practitioners.
It is shown that the updating of topological modes is particularly slow
in Monte Carlo simulations of full QCD both with Wilson
(
paper 3) and staggered
(
paper 4) fermions. A comparison of decorrelation efficiencies is
performed with other more local observables showing that many more hits of HMC are needed
for the updating of topology than for the updating of the other
observables.
The coupling constant of strong interactions at the scale μ,
αs(μ), is
an extremely important quantity of QCD. The rate at which
αs(μ) runs
with μ is determined by the scale Λ of QCD.
This is a fundamental
parameter of the standard model. In the paper this parameter is calculated for
the quenched theory obtaining the value
ΛMS=0.31(5) MeV.
First the 3-gluon proper Green function is extracted from
a Monte Carlo simulation in the Landau gauge. Then a renormalization
scheme analogous to the MOM∼ is imposed. The matching
with the MS
scheme is performed by a perturbative
computation in the continuum without using lattice perturbation
theory. The gauge field on the lattice is defined as
Aμ(x)≡ -½ i [Uμ(x) -
Uμ(x)†]traceless
in terms of the links.
In the Monte Carlo simulation the 3-gluon Green function and the gluonic propagator
are evaluated at a particular set of external momenta in such a way that
the proper 3-gluon function can be obtained by a simple algebra. The
rotational invariance of the result has been checked. The calculation of
the full QCD coupling constant by
P. Boucaud et al. is based on the above method.
The paper contains the first analytical calculation of the third
term in the free energy of pure Yang--Mills. Such a result proves
to be very useful to improve
the precision in the extraction of the quenched gluon condensate
and to define effective schemes on the lattice.
The calculation was done by using the software prepared by us
and introduced in
Nucl. Phys. B413 553 (1994). The result has been extended to
full QCD with Wilson fermions in
Phys. Lett. B426 361 (1998) (Erratum:ibid. B553 337 (2003))
where the coefficients are extracted for several fermion masses up to
three loops.
The topological susceptibility χ of a gauge theory can be calculated on the
lattice by measuring the observable 〈QLQL〉
where QL is the lattice regularized topological charge
operator. However, in general, the above expectation value does not yield
the continuum value of the susceptibility but instead the combination
Z2 a4χ + M where a is the lattice
spacing and Z and M are multiplicative and additive
renormalization constants respectively. Unless these two renormalization
constants are known accurately enough, it will not be possible to extract
the sought topological susceptibility from numerical simulation data.
The above expression can be justified in the context of perturbation
theory by using the Product Expansion Theorem.
Regularizations of the topological charge density operator have been
defined around mid 90's for which M vanishes and Z is unity.
Z and M have been determined by using perturbation theory
(see for instance my paper
Nucl. Phys. B413 553 (1994) where M is calculated at
three loops for the naïve definition of QL).
However the method is intrinsically ambiguous since there is
little control on the higher order terms. This problem, which is common
in every perturbative calculation, is particularly acute in the case of lattice
regularization. In the paper
Phys. Rev. D48 2284 (1993)
an improvement of the method to evaluate Z and M
is utilized. The idea was put forward in a seminal paper by two of the
authors,
Phys. Lett. B275 429 (1992).
This is a non-perturbative
technique which avoids the systematic errors implicit in perturbation
theory calculations at few loops.
In a nutshell the renormalization constants Z and M are
simulated on configurations with a well-known instanton content in such a
way to be able to impose a renormalization scheme such as the
MS where
the topological charge in the presence of one instanton is fixed to be 1.
Needless to say, the validity of this procedure lies on an (yet unproved)
extension of the Operator Product Theorem beyond perturbation theory.
In
Phys. Rev. D54 1044 (1996) a consistency check of the method is
performed. In
Phys. Rev. D74 094503 (2006) another check
is shown where the calibration of the topological content of configurations
is done by counting fermionic zero modes and the result compared with the
usual calibration based on standard cooling. In
Phys. Rev. D77 056008 (2008) a further check is done. In this case the main
goal was to verify the so-called Haldane conjecture which states that the
mass gap of the O(3) non-linear sigma model in 2 dimensions with a θ
term shall vanish when θ=π. Only by using the values of
Z calculated with the present method, we were able to recover the
correct position (θ=π) where the mass gap closes.