Playing with computers
Once (1984-1986) I had the fascinating experience to
be involved with the building of a special purpose computer
called "PERCOLA". The computer was running until recently in Saclay and has
been giving data for over ten years. The first were on the
conductivity and superconductivity exponents of percolation
published
in J. Stat. Phys., Vol.52, p.441-446 (1988) and
published
in Int.J.Mod.Phys. C, Vol. 1, 207-214 (1990) respectively.
The latest results have been
published in
Int.J.Mod.Phys. C, Vol. 6, 813-818 (1995).
In the past I also did a lot of bit-manipulation for multi-spin coding
on vector computers. With John Zabolitzky we established in a
paper (J.Comp.Phys. 76, 426-447 (1988))
some world records in size and speed of Ising simulations
(on the Cray-2).
In 1992 I was involved in the European Teraflop Initiative and did
benchmarking and tests of parallel machines of various type.
One of these comparisons on Connexion Machines is given in
the following report .
A paper
which much impact
on computer algorithms is with P.M.C. de Oliveira and T. Penna
(Braz. J. of Phys. 26, 677-683 (1997)) about a new canonical
Monte Carlo histogram technique which has
good precision over
a much larger temperature range than usual histogram methods.
With my former PhD student
Jose Daniel Muñoz who is now professor at the Universidad
Nacional de Bogota we extended this method to systems with
continuous degrees of freedom as published in a
paper
in Int.J.Mod.Phys. C, Vol. 10, 95-114 (1999) for the XY model and in
a paper
published in Comp. Physcs Comm., Vol. 121, p. 13-15 (1999) and in the
proceedings
of the XII meeting on Computer Simulation Studies in Condensed Matter
Physics, p.174-178 (1999) for the more general case.