1. Critical properties in the 2-dimensional spin models
A relevant part of my research activity deals with the study of
the critical properties of non-linear sigma models in 2 dimensions
with O(N) symmetry.
By assuming some hypotheses concerning the percolation properties
of the model, several
authors have argued that it would undergo a Kosterlitz-Thouless (KT)-type phase
transition at finite temperature and that the model has no
mass-gap, contradicting exact (but not rigorous) analytical results.
In
Nucl. Phys. B500 513 (1997)
a deep analysis of the model is done by high precision
simulations (O(107) decorrelated statistics) with standard
and improved
actions (Symanzik 0-loop) and with effective schemes
with perturbative corrections to the universal scaling up to 4
loops for the standard action and 3 loops for the Symanzik action.
Moreover two different effective schemes are used for checking
purposes. The computer code for the simulation is written in
c language in order to efficiently construct the clusters
of the Wolff algorithm. A modification of this
algorithm allows to use it for the Symanzik action.
The mass gap obtained from the simulation shows an excellent agreement
(within 4%) with the analytical calculation for the O(3) Heisenberg
model and within 5‰ for the O(8) model. Also
the magnetic susceptibility is in excellent agreement with large
N calculations. These results support the scenario where the spin models
have a single UV fixed point, as predicted by perturbation theory.
Moreover it is shown that the Monte Carlo data do not
follow the scaling laws predicted by a KT phase transition.
It is possible to reach this conclusion thanks to the
extremely good precision
data obtained from the high statistics simulation.
In
Phys. Rev. D59 067703 (1999) further studies for the standard
action O(3) model are shown.
Again the data follow the behaviour predicted by perturbation theory
much better than by a KT transition.
In
Phys. Rev. Lett. 83 3669 (1999)
it is shown which of the hypotheses used to
prove that the model cannot have a mass gap can fail. That hypothesis
assumes that a type of clusters (made of spins which are
almost perpendicular to a fixed unit vector)
do not percolate. Our numerical simulations clearly indicate that instead
such clusters percolate, at least at temperatures as low as T=0.5.
The results of the Phys. Rev. Lett. paper have been questioned by
A. Patrascioiu and E. Seiler in their paper
J. Statist. Phys. 106 811 (2002). Unfortunately these
authors perform their simulations at an incredibly low temperature
which, considering that they use lattice sizes of order L∼ 1000,
makes their results insensible. To do that, they actually do not
make use of the standard formulation of the O(3) model on the lattice.
Instead a Lipschitz-type condition is imposed: neighbour spins cannot be more
separated than a prefixed angle. This angle is directly related to the
real temperature of the system (what usually is written as 1/T
in the Boltzmann weight)
in a complicated manner. However their choice for the angle
drives the system to such a low temperature that the correlation
length turns out to be many orders of magnitude larger than the
lattice size itself. Thus, their results are meaningless.
2. Perturbation theory on the lattice
I have extensively worked on perturbation theory
on the lattice. Such calculations, that are particularly heavy due to the
lengthy expressions for vertices and propagators, have been partially
simplified by creating a software which automates some of the
algebraic operations needed to develop the perturbation theory,
see
Nucl. Phys. B413 553 (1994).
Both the software and the analytical calculations have been developed
independently by the other authors and only the final
results were compared for obvious checking purposes.
In
Phys. Lett. B249 490 (1990) and
Phys. Rev. D44 513 (1991) the multiplicative renormalization
Z for the topological
charge, the perturbative tail and mixing coefficients of the topological
susceptibility with gluon condensates for two different
operators on the lattice have been calculated at lowest order.
These results are essential to obtain the correct (and unique)
physical result for the topological
susceptibility from the two different operators.
In
Phys. Lett. B268 241 (1991)
the origin of the multiplicative renormalization
Z is shown. On the other hand, in reference
Phys. Lett. B350 70 (1995)
the corrections from dynamical quarks to the value of
Z are calculated.
In
Nucl. Phys. B413 553 (1994)
the first 3-loop calculation of the perturbative tail for the
topological susceptibility in QCD is shown.
The first calculation of the third term in the free energy of pure
Yang-Mills is done in
Phys. Lett. B324 433 (1994).
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 (see for instance
Phys. Lett. B100 485 (1981) and references inside).
In
Nucl. Phys. B491 498 (1997)
the third coefficient of the lattice beta function
in pure Yang-Mills is calculated. It is the first correction
to the universal scaling on the lattice. A similar result had already
been obtained by M. Lüscher and P. Weisz a few months before but it was convenient to
check it by an independent calculation. For
instance in our paper the integrals are evaluated with a different technique from
that used by M. Lüscher and P. Weisz.
In
Phys. Lett. B426 361 (1998), Erratum-ibid. B553 337 (2003)
the calculation of reference
Phys. Lett. B324 433 (1994) is corrected by adding the
fermionic contributions up to 3 loops. This allows a better precision
in using effective schemes in Monte Carlo simulations with Wilson
fermions.
In references
Nucl. Phys. B500 513 (1997),
Nucl. Phys. B563 213 (1999), Erratum-ibid. B576 658 (2000),
Nucl. Phys. B562 581 (1999) and also the proceeding
Nucl. Phys. (Proc. Suppl.) 83 709 (2000)
a great effort was done to renormalize up to 4 loops the spin O(N) models in 2
dimensions on the lattice with standard action and with Symanzik
0-loop and 1-loop improved actions. These results are extremely
useful to extract physical information from numerical simulations
of the theory, such as mass gap or magnetic susceptibility.
The main motivation is to check the validity of the Symanzik
programme as well as to understand the critical properties of
spin models (see above).
In the paper
Nucl. Phys. B437 627 (1995) a study of the continuum limit in
field theories defined on random lattices was performed. A positive example
was carried through for the case of the O(3) model in 2 dimensions
showing that in effect the
theory is correctly represented on the random lattice. Several levels of
randomness were taken into account. The necessary loop integrals were
performed by using a novel method expressly introduced for the case at hand.
3. Lattice calculation of the topological susceptibility at the transition
In
Phys. Lett. B412 119 (1997)
a lattice simulation was performed to understand the behaviour of the
topological susceptibility in pure gauge theory with SU(2) gauge group
at the finite temperature transition. The drop of the signal after the
transition is less steep than in the SU(3) case (see
Most cited articles) as shown in the figure.
In the paper
Phys. Lett. B483 139 (2000)
an analogous study was performed for full QCD with gauge group SU(3) at
finite temperature. Again the signal drops at the phase transition.
The study was repeated obtaining similar results for full QCD with
gauge group SU(2) at finite temperature and finite density in
Nucl. Phys. B752 124 (2006). In this case a figure was
prepared to show that the three observables, topology, chirality and
confinement (for topology and confinement, it would be more adequate
to speak about "quantities related to" instead of observables, see
paper) undergo a transition at the same value for the density
(or chemical potential):
In this figure a is the lattice spacing, P is the
Polyakov loop, μ is the chemical potential and
〈ΨΨ〉
is the chiral condensate.