Krzysztof Boryczko

• AGH University of Krakow

In this work we propose rank densities, a novel density measure for non-metric data. Unlike typical measures employed in density clustering algorithms, rank densities are defined via ranks of patterns in k-nearest neighbor lists. Therefore they depend on the structure of the k-nearest neighbor graph, rather than exact values of the similarity measu...

Biclustering is a technique of discovering local similarities within data. For many years the complexity of the methods and parallelization issues limited its application to big data problems. With the development of novel scalable methods, biclustering has finally started to close this gap. In this paper we discuss the caveats of biclustering and...

Recent years have seen a growing trend towards the introduction of more advanced manycore processors. On the other hand, there is also a growing popularity for cheap, credit-card-sized, devices offering more and more advanced features and computational power. In this paper we evaluate Parallella -- a small board with the Epiphany manycore coprocess...

Text data mining is the process of extracting valuable information from a dataset consisting of text documents. Popular clustering algorithms do not allow detection of the same words appearing in multiple documents. Instead, they discover general similarity of such documents. This article presents the application of a hybrid biclustering algorithm...

Hybrid methods are a branch of biclustering algorithms that emerge from combining selected aspects of pre-existing approaches. The syncretic nature of their construction enriches the existing methods providing them with new properties. In this paper the concept of hybrid biclustering algorithms is explained. A representative hybrid biclustering alg...

Bielustering, which Is simultaneous clustering of columns and rows in data matrix, became an issue when classical clustering algorithms proved not to be good enough to detect similar expressions of genes under subset of conditions. Bielustering algorithms may be also applied to different datasets, such as medical, economical, social networks etc. I...

Visualization of a large-scale protein databases may help biologists in discovering similarity between sequences of different organisms. In this article we present a complex approach for visually representing relations between proteins in large scale databases. Our approach includes sequence alignment, mutual distance measurement, clustering and cl...

Parallel computing architectures are proven to significantly shorten computation time for different clustering algorithms. Nonetheless, some characteristics of the architecture limit the application of graphics processing units (GPUs) for biclustering task, whose function is to find focal similarities within the data. This might be one of the reaso...

Extreme Environmental Events, Springer Verlag, 126-149, 2011

The new methods of diagnosis are often more sensitive and speed up its process allowing for a more effective treatment. Yet they need to be carefully tested before they can be used in practice. The paper concerns the problems of developing nanorobots, which would circulate in the bloodstream of the human body gathering information about its conditi...

This work presents an efficient implementation of affinity propagation (AP) on clusters of graphical processing units (GPUs). AP is a state-of-the-art method for finding exemplars in data sets described by similarity matrices. It is typically employed in crisp clustering applications. However, when finding exemplars in an n-pattern data set with de...

Multiple different filesystems — including disk-based, network, distributed, abstract — arean integral part of every operating system. They are usually written as kernel modules and abstracted to the user via a virtual filesystem switch. In this paper we analyse the feasibility of reimplementing the virtual filesystem switch as a userspace daemon a...

In the paper an algorithm of triangle mesh generation for a full three-dimensional volumetric data is presented. Calculations are performed in real time using graphics processors. The method is very well suited for the visu- alization of dynamic data, as the calculations use only the current frame data (not including data from previous frames). Due...

This article focuses on implementation of Affinity Propagation, a state of the art method for finding exemplars in sets of patterns, on clusters of Graphical Processing Units. When finding exemplars in dense, non-metric data Affinity Propagation has O(n 2) memory complexity. This limits the size of problems that can fit in the Graphical Processing...

In this paper we present an interactive GPU-based, GUI client, working with rendering server employing ray tracing based global illumination. The client is designed to guarantee interactivity (namely 1/60sec response time) no matter how slow the rendering server is. The client dynamically adjusts image resolution to match the server performance and...

The SPH method for simulating incompressible fluids is presented in this article. The background and principles of the SPH method are explained and its application to incompressible fluids simulations is discussed. The parallel implementation of the SPH simulation with OpenMP and MPI environments are demonstrated. Both models of parallel implementa...

In this paper we present an improved approach to full spectral rendering. The technique is optimized for quasi-Monte Carlo ray tracing, however the underlying physical theory can be applied to any global illumination scheme. We start with explanation of the necessity of full spectral rendering in any correct global illumination system. Then we pres...

A novel technique based on cluster analysis of the multi-resolutional structure of earthquake patterns is developed and applied to observed and synthetic seismic catalogs. The observed data represent seismic activities situated around the Japanese islands in the 1997-2003 time interval. The synthetic data were generated by numerical simulations for...

The adaptations of the SPH method for simulating incompressible fluids which focuse on two features: the surface tension and artificial viscosity, are presented in this article. The background and principles of the SPH method are explained and its application to incompressible fluids simulations is discussed. The methodology and implementation of a...

Classical clustering algorithms often perform poorly on data harboring background noise, i.e. large number of observations distributed uniformly in the feature space. Here, we present a new density-based algorithm for approximate clustering of such noisy data. The algorithm employs Shared Nearest Neighbor Graphs for estimating local data density an...

Fluid flow can be realistically simulated by physical models. We present a method for simplifying the Navier-Stokes equations by relaxing the incompressibility constraint. Our method allows for low-cost real-time simulation of two-dimensional fluid flow with accuracy sufficient for computer graphics. The implementation takes advantage of recent pro...

We present an adoption of the bucket sort algorithm capable of running entirely on GPU architecture. Our implementation employs render-to-texture to enable scatter operation. Linked lists of elements in each bucket are build and stored directly in video memory. We show also the use of this sorting method in a particle-based simulation. Dissipative...

This paper presents a new method for detecting clusters of microcalcification s in high- resolution digital mammograms. Using cluster analysis, we have designed a descriptive set of mammogram image features which enables precise recognition of microcalcifi cations. These features are fed into the Support Vector Machine classifier traine d to discri...

Mesoscopic features embedded within macroscopic phenomena in colloids and suspensions, when coupled together with micro-structural dynamics and boundary singularities, produce complex multi-resolution patterns, which are difficult to capture with the continuum model using partial differential equations, i.e., the Navier-Stokes equation and the Cahn...

In this paper we compare timing results for short-range 12/6 Lennard-Jones parallel molecular dynamics 31) simulation using two programming environments PVM and MPI on HP Exemplar SPP computers. The results obtained on Intel Paragon XP/S are also reported.

In the paper we present a simple dynamic load-balancing procedure applied in short-range 12/6 Lennard-Jones molecular dynamics (MD) simulation on clusters of computers. For study fluid and mixtures in microcapillaries or growth of optical fibres, the computational box is arranged as a long cylinder. First, simulation of the cluster load has been us...

Parallel algorithms and programs for lattice gas automata and molecular dynamics for fluid flow simulation were developed and investigated using PVM, Express, Network Linda and p4 on a network of workstations and a CONVEX C3210.

Microstructural dynamics and boundary singularities generate complex multiresolution patterns, which are difficult to model with the continuum approaches using partial differential equations. To provide an effective solver across the diverse scales with different physics the continuum dynamics must be augmented with atomistic models, such as non-eq...

In this paper we compare efficiency results for molecular dynamics simulations obtained on multiprocessor supercomputers: Convex Exemplar SPP1000/XA and SPP1200/XA, Intel Paragon XP/S and Connection Machine 5 using two approaches — data-parallel programming and message-passing paradigm.

In the paper we compare timing results for short-range 12/6 Lennard-Jones molecular dynamics simulation for two kinds of data structures resulting from two different kinds of 3-D computational box. For parallel and distributed computing the domain decomposition method and message-passing paradigm were applied. The programs were run on parallel comp...

We discuss a parallel preconditioner for the algebraic systems of equations arising from Newton-Raphson linearization and finite element (FE) discretization of stabilized formulation for the highly nonlinear Stokes equations. We compare SSOR/Jacobi and FSAI(1)/Jacobi preconditioners for the parallel MINRES accelerator implemented using C++/OpenMP....

Fluid particle with fibrin monomers coarse graining of many molecules Fibrin monomer m= 1.02×10 -20 kg Fluid particle 150 nm coarse graining of many plasma molecules m= 3.13×10 -17 kg Wall and red blood cells particles F=k(r-r 0)+F D +F B Fluid particles F=F C +F D +F B F C -repulsive Fluid with fibrins F=F C +F D +F B F C -repulsive Fluid with fib...

We present a novel technique based on a multi-resolutional clustering and nonlinear multi-dimensional scal-ing of earthquake patterns to investigate observed and syn-thetic seismic catalogs. The observed data represent seis-mic activities around the Japanese islands during 1997–2003. The synthetic data were generated by numerical simulations for va...

In this article we present a new method for recognition of subtle microcalcifications in high-resolution digital mammograms. The identification of suspicious regions in mammogram images is carried out using a method based on the 2D discrete wavelet transform. For classification of regions we use the SVC algorithm. Initially, a number of statistical...

Molecular biology is a source of vast quantities of information. Nucleotide sequences, gene expression patterns, protein abundances, sequences and structures, drug activities, gene and metabolic networks are being harvested at laboratories throughout the world. The collected data can be represented by multidimensional feature vectors or by descript...

Mesoscopic features embedded within macroscopic phenomena in colloids and suspensions, when coupled together with microstructural dynamics and boundary singularities, produce complex multiresolutional patterns, which are difficult to capture with the continuum model using partial differential equations, i.e., the Navier-Stokes equation and the Cahn...

Excessive clotting can cause bleeding over a vast capillary area. We study the mesoscopic dynamics of clotting by using the fluid particle model. We assume that the plasma consists of fluid particles containing fibrin monomers, while the red blood cells and capillary walls are modeled with elastic mesh of "solid" particles. The fluid particles inte...

Excessive clotting can cause bleeding over a vast capillary areas contributing to stroke, heart attack or blindness. We study the mesoscopic dynamics of clotting by using a discrete-particle model. We assume that the plasma consists of fluid particles containing fibrin monomers, while the red blood cells and capillary walls are modeled using elasti...

IMA Workshop – Atomic Motion to Macroscopic Models, Aril 11-15, 2005 University of Minnesota

We discuss the use of current shared-memory systems for discrete-particle modeling of heterogeneous mesoscopic complex fluids in irregular geometries. This has been demonstrated by way of mesoscopic blood flow in bifurcating capillary vessels. The plasma is represented by fluid particles, while the other blood constituents are made of "solid" parti...

We use modern and novel techniques to study the problems associated with detection and analysis of multitudinous seismic events, which form the background for isolated great earthquakes. This new approach involves multivariate analysis of low and large magnitude events recorded in space over a couple of centuries in time. We propose here the deploy...

We have modeled the dynamics of a 3-D system consisting of red blood cells (RBCs), plasma and capillary walls using a discrete-particle approach. The blood cells and capillary walls are composed of a mesh of particles interacting with harmonic forces between nearest neighbors. We employ classical mechanics to mimic the elastic properties of RBCs wi...

We investigate the mechanism of aggregation of red blood cells (RBC) in capillary vessels. We use a discrete-particle model in 3D to model the flow of plasma and RBCs within a capillary tube. This model can accurately capture the scales from 0.001 to 100 microm, far below the scales that can be modeled numerically with classical computational fluid...

Simulating natural phenomena at greater accuracy results in an explosive growth of data. Large- scale simulations with particles currently involve ensembles consisting of between 106-109 particles, which cover 105-106 timesteps. Thus, the data files produced in a single run can reach from tens of gigabytes to hundreds of terabytes. This data bank a...

We report results of numerical simulations of complex fluids, using a combination of discrete-particle methods. Our molecular modeling repertoire comprises three simulation techniques: molecular dynamics (MD), dissipative particle dynamics (DPD), and the fluid particle model (FPM). This type of model can depict multi-resolution molecular structures...

Dissipative particle dynamics (DPD) and its generalization - the fluid particle model (FPM) - represent the "fluid particle" approach for simulating fluid-like behavior in t he mesoscale. Unlike particles from molecular dynamics (MD) method, the "fluid particle" can be viewed as a "droplet" consisting of liquid molecules. In FPM, "fluid particles"...

In this paper we present some methods and algorithms for large scale computing which cover different areas of computational and computer sciences. They concern particle models, CFD computing, animation, monitoring and predicting of application performance as well as scientific visualization and scientific data storing and retrieving.

Advantages and disadvantages of discrete (particle) models applied in numerical simulations of porous material penetrated by liquid metal in foundry process are discussed. Emphasis is put on the usage of Dissipative Particle Dynamics (DPD) method and the basics of this technique is presented. 2-D flow model is analyzed for different porous material...

Flow through porous media belongs to phenomena tough for classical approach. This happens especially when one wants the model to be more realistic by assuming that the porous medium can change dynamically due to the fluid flow. The model of the porous medium can be very much sophisticated, as granules forming it can have different shapes and sizes,...

Scientific visualization is an important aspect of numerically intensive computing. While a vast amount of simulation algorithms exists, visualization systems are still a field of scientific interest (e.g. [1,2]). Molecular dynamics algorithms (and others, like dissipative particle dynamics) while used for, e.g., thermal instabilities (e.g. [3,4])...

. In this paper we compare timing results for short-range 12/6 Lennard-Jones parallel molecular dynamics 3D simulation using two programming environments PVM and MPI on HP Exemplar SPP computers. The results obtained on Intel Paragon XP/S are also reported. 1 Introduction One of the fast developing approaches to material science is application of p...

In this paper we present a parallel algorithm for solving the Euler equations on unstructured grids. Time discretization is organized with a linear, implicit version of Taylor-Galerkin time scheme, while finite elements are employed for space discretization of one step problem. In the algorithm we propose two levels of parallelism. The first is the...

We compare execution time for typical numerical problems, obtained on different parallel architectures, including typical vector supercomputers, multiprocessors and network of computers with PVM environment.

Transmission rates for two communication schemes: ping/pong and ring transfer have been measured on multiprocessors and clusters of workstations. Performance of a network of computers as a tool for numerically intensive computing has been evaluated on the basis of SLALOM and EP-NAS parallel benchmarks.

In this paper we present parallel algorithms for molecular dynamics and lattice gas simulations for hydrodynamics phenomena at the microscopic level. Characteristics of the algorithms (execution time, efficiency and speedup) are shown for the transputer networks and for the networks of SUN SPARC and IBM RS/6000 workstations. Some of the results are...

In this paper we present the simulator of crawler excavator which enables in- teractive simulation with real-time reactions to operator's steering decisions. We introduce a new simple kinematic model of cutting ofi soil from ground, which gives good results in real-time even on modest ofi-the-shelf computers. The computa- tional model takes into ac...

Simulation of interacting particle systems has been a well established method for many years now. Such systems can span different scales, including microscopic (where particles represent atoms, as in Molecular Dynamics simulations) as well as macroscopic. In the latter case, growing interest is put into Smoothed Particle Hydrodynamics approach. Tra...