最近の論文から:
Abstract
In this paper we propose
the recursive stochastic state selection method,
an extension of the recently developed
stochastic state selection method in Monte Carlo
calculations for quantum spin systems.
In this recursive method we use intermediate states to define
probability functions for stochastic state selections.
Then we can diminish variances of samplings when we calculate
expectation values of the powers of the Hamiltonian.
In order to show the improvement
we perform numerical calculations of the spin-1/2 anti-ferromagnetic
Heisenberg model on the triangular lattice.
Examining results on the ground state of the 21-site system we confide
this method in its effectiveness.
We also calculate the lowest and the excited energy eigenvalues as well
as the static structure factor for the 36-site system.
The maximum number of basis states kept in a computer memory for this
system is about
J.Phys.Soc.Jpn.73(2004)2245-2251.
T.Munehisa, Y.Munehisa
3.6 x 10^7.
Employing a translationally invariant initial trial state, we evaluate
the lowest energy eigenvalue within 0.5 % of the statistical errors.
Abstract
A new method to construct event-generators based on next-to-leading order QCD matrix-elements and leading-logarithmic parton showers is proposed. Matrix elements of loop diagram as well as those of a tree level can be generated using an automatic system. A soft/collinear singularity is treated using a leading-log subtraction method. Higher order resummation of the soft/collinear correction by the parton shower method is combined with the NLO matrix-element without any double-counting in this method. An example of the event generator for Drell-Yan process is given for demonstrating a validity of this method.
Abstract
We propose a new method for updating units in the Hopfield model. With this method two or more units change at the same time, so as to become the lowest energy state among all possible states. Since this updating algorithm is based on the detailed balance equation, convergence to the Boltzmann distribution is guaranteed.
If our algorithm is applied to finding the minimum energy in constraint satisfaction and combinatorial optimization problems, then there is faster convergence than with the usual algorithm in the neural network. This is shown by experiments with the travelling salesman problem, the four color problem, the N-queen problem and the graph bipartitioning problem.
In constraint satisfaction problems, for which earlier neural networks are effective in some cases, our updating scheme works fine. Even though we still encounter the problem of ending up in local minima, our updating scheme has a great advantage compared with the usual updating scheme used in combinatorial optimization problems.
Also we discuss parallel computing using our updating algorithm.