Learning short synfire chains by self-organization

A model of cortical neurons capable of sustaining a low level of spontaneous activity is investigated. Without learning the activity of the network is chaotic. We report on attempts to learn synfire chains in this type of network by introducing a Hebbian learning mechanism and exciting a small set of neurons at random intervals. We discuss the types of instabilities that can arise and prevent the formation of long synfire chains and also discuss various biologically plausible mechanisms which to some extent cure these instabilities.     

Three‐dimensional localization for the MagneBike inspection robot

The MagneBike inspection robot is a climbing robot equipped with magnetic wheels. The robot is designed to drive on three‐dimensional (3D) complexly shaped pipe structures; therefore it is necessary to provide 3D visualization tools for the user, who remotely controls the robot out of sight. The localization system is required to provide a 3D map of the unknown environment and the 3D location of the robot in the environment's map. The localization strategy proposed in this paper consists of combining 3D odometry with 3D scan registration. The odometry model is based on wheel encoders and a three‐axis accelerometer. Odometry enables the tracking of the robot trajectory between consecutive 3D scans and is used as a prior for the scan matching algorithm. The 3D scan registration facilitates the construction of a 3D map of the environment and refines the robot position computed with odometry. This paper describes in detail the implementation of the localization concept. It presents the lightweight, small‐sized 3D range finder that has been developed for the MagneBike. It also proposes an innovative 3D odometry model that estimates the local surface curvature to compensate for the absence of angular velocity inputs. The different tools are characterized in detail based on laboratory and field experiments. They show that the localization concepts reliably track the robot moving in the specific application environment. We also describe various techniques to optimize the 3D scanning process, which is time consuming, and to compensate for the identified limitations. These techniques are useful inputs for the future automatization of the robot's control and optimization of its localization process. © 2010 Wiley Periodicals, Inc.     

Heuristics for the multi-period orienteering problem with multiple time windows

We present the multi-period orienteering problem with multiple time windows (MuPOPTW), a new routing problem combining objective and constraints of the orienteering problem (OP) and team orienteering problem (TOP), constraints from standard vehicle routing problems, and original constraints from a real-world application. The problem itself comes from a real industrial case. Specific route duration constraints result in a route feasibility subproblem. We propose an exact algorithm for this subproblem, and we embed it in a variable neighborhood search method to solve the whole routing problem. We then provide experimental results for this method. We compare them to a commercial solver. We also adapt our method to standard benchmark OP and TOP instances, and provide comparative tables with state-of-the-art algorithms.     

Computing efficiently the lattice width in any dimension

We provide an algorithm for the exact computation of the lattice width of a set of points K in Z² in linear-time with respect to the size of K. This method consists in computing a particular surrounding polygon. From this polygon, we deduce a set of candidate vectors allowing the computation of the lattice width. Moreover, we describe how this new algorithm can be extended to an arbitrary dimension thanks to a greedy and practical approach to compute a surrounding polytope. Indeed, this last computation is very efficient in practice as it processes only a few linear time iterations whatever the size of the set of points. Hence, it avoids complex geometric processings.     

A prismoid framework for languages with resources

Inspired by the Multiplicative Exponential fragment of Linear Logic, we define a framework called the prismoid of resources where each vertex is a language which refines the λ-calculus by using a different choice to make explicit or implicit (meta-level) the definition of the contraction, weakening, and substitution operations. For all the calculi in the prismoid we show simulation of β-reduction, confluence, preservation of β-strong normalisation and strong normalisation for typed terms. Full composition also holds for all the calculi of the prismoid handling explicit substitutions. The whole development of the prismoid is done by making the set of resources a parameter of the formalism, so that all the properties for each vertex are obtained as a particular case of the general abstract proofs.     

Generation of Efficient Nested Loops from Polyhedra

Automatic parallelization in the polyhedral model is based on affine transformations from an original computation domain (iteration space) to a target space-time domain, often with a different transformation for each variable. Code generation is an often ignored step in this process that has a significant impact on the quality of the final code. It involves making a trade-off between code size and control code simplification/optimization. Previous methods of doing code generation are based on loop splitting, however they have nonoptimal behavior when working on parameterized programs. We present a general parameterized method for code generation based on dual representation of polyhedra. Our algorithm uses a simple recursion on the dimensions of the domains, and enables fine control over the tradeoff between code size and control overhead.     

NMscatt: a program for calculating inelastic scattering from large biomolecular systems using classical force-field simulations

Computational tools for normal mode analysis, which are widely used in physics and materials science problems, are designed here in a single package called NMscatt (Normal Modes & scattering) that allows arbitrarily large systems to be handled. The package allows inelastic neutron and X-ray scattering observables to be calculated, allowing comparison with experimental data produced at large scale facilities. Various simplification schemes are presented for analyzing displacement vectors, which are otherwise too complicated to understand in very large systems.

Program summary

Title of program:NMscattCatalogue identifier:ADZA_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/ADZA_v1_0.htmlProgram obtainable from:CPC Program Library, Queen's University of Belfast, N. IrelandLicensing provisions:noNo. of lines in distributed program, including test data, etc.:573 535No. of bytes in distributed program, including test data, etc.:4 516 496Distribution format:tar.gzProgramming language:FORTRAN 77Computer:x86 PCOperating system:GNU/Linux, UNIXRAM:Depends on the system size to be simulatedWord size:32 or 64 bitsClassification:16.3External routines:LAPACKNature of problem: Normal mode analysis, phonons calculation, derivation of incoherent and coherent inelastic scattering spectra.Solution method: Full diagonalization (producing eigen-vectors and eigen-values) of dynamical matrix which is obtained from potential energy function derivation using finite difference method.Running time: About 7 hours per one k-point evaluation in sampling all modes dispersion curves for a system containing 3550 atoms in the unit cell on AMD Athlon 64 X2 Dual Core Processor 4200+.     

A new approach of data clustering using a flock of agents

This paper presents a new bio-inspired algorithm (FClust) that dynamically creates and visualizes groups of data. This algorithm uses the concepts of a flock of agents that move together in a complex manner with simple local rules. Each agent represents one data. The agents move together in a 2D environment with the aim of creating homogeneous groups of data. These groups are visualized in real time, and help the domain expert to understand the underlying structure of the data set, like for example a realistic number of classes, clusters of similar data, isolated data. We also present several extensions of this algorithm, which reduce its computational cost, and make use of a 3D display. This algorithm is then tested on artificial and real-world data, and a heuristic algorithm is used to evaluate the relevance of the obtained partitioning.