Fatih Erdogan Sevilgen

Parallel domain decomposition and load balancing using space-filling curves

Partitioning techniques based on space filling curves have received much recent attention due to their low running time and good load balance characteristics. The basic idea underlying these methods is to order the multidimensional data according to a space filling curve and partition the resulting one dimensional order. However, space filling curves are defined for points that lie on a uniform grid of a particular resolution. It is typically assumed that the coordinates of the points are representable using a fixed number of bits, and the run times of the algorithms depend upon the number of bits used. We present a simple and efficient technique for ordering arbitrary and dynamic multidimensional data using space filling curves and its application to parallel domain decomposition and load balancing. Our technique is based on a comparison routine that determines the relative position of two points in the order induced by a space filling curve. The comparison routine could then be used in conjunction with any parallel sorting algorithm to effect parallel domain decomposition.

PHISTO: pathogen–host interaction search tool

SUMMARY Knowledge of pathogen-host protein interactions is required to better understand infection mechanisms. The pathogen-host interaction search tool (PHISTO) is a web-accessible platform that provides relevant information about pathogen-host interactions (PHIs). It enables access to the most up-to-date PHI data for all pathogen types for which experimentally verified protein interactions with human are available. The platform also offers integrated tools for visualization of PHI networks, graph-theoretical analysis of targeted human proteins, BLAST search and text mining for detecting missing experimental methods. PHISTO will facilitate PHI studies that provide potential therapeutic targets for infectious diseases. AVAILABILITY http://www.phisto.org. CONTACT saliha.durmus@boun.edu.tr SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

A provably optimal, distribution-independent parallel fast multipole method

The Fast Multipole Method (FMM) is a robust technique for the rapid evaluation of the combined effect of pairwise interactions of n data sources. Parallel computation of the FMM is considered a challenging problem due to the dependence of the computation on the distribution of the data sources, usually resulting in dynamic data decomposition and load balancing problems. In this paper, we present the first provably efficient and distribution-independent parallel algorithm for the FMM on distributed memory parallel computers. Our algorithm does not require any dynamic data decomposition or load balancing step. We present our algorithm in terms of a few basic and well understood primitive operations such as sorting and parallel prefix.

Dynamic Compressed Hypertoctrees with Application to the N-Body Problem

Hyperoctree is a popular data structure for organizing multidimensional point data. The main drawback ofthi s data structure is that its size and the run-time ofo perations supported by it are dependent upon the distribution of the points. Clarkson rectified the distribution-dependency in the size ofh yperoctrees by introducing compressed hyperoctrees. He presents an O(n log n) expected time randomized algorithm to construct a compressed hyperoctree. In this paper, we give three deterministic algorithms to construct a compressed hyperoctree in O(n log n) time, for any fixed dimension d. We present O(log n) algorithms for point and cubic region searches, point insertions and deletions. We propose a solution to the N-body problem in O(n) time, given the tree. Our algorithms also reduce the run-time dependency on the number of dimensions.

Randomized Hydrodynamic Load Balancing Approach

Load balancing is performed to achieve the optimal use of the existing computational resources as much as possible whereby none of the resources remains idle while some other resources are being utilized. Balanced load distribution can be achieved by the immigration of the load from the source nodes which have surplus workload to the comparatively lightly loaded destination nodes. Applying load balancing during run time is called dynamic load balancing (DLB). DLB can be realized both in a direct or iterative manner according to the execution node selection. In iterative methods, the final destination node is determined through several iteration steps, while in direct methods it is selected in one step. This paper presents the randomized hydrodynamic load balancing (RHLB) method which is a hybrid method that takes advantage of both direct and iterative methods. Using random load migration as a direct method, RHLB approach intends to solve the problems derived from the exceptional instantaneous load rises, and diffuse the surplus workload to relatively free resources. Besides, using hydrodynamic approach as an iterative method, RHLB aims to consume minimum possible system resources to balance the common workload distributions. The results of the experiments designate that, RHLB outruns other iterative based methods in terms of both balance quality and the total time of the load balancing process.

Molecular dynamics simulation of helium–argon gas mixture under various wall conditions

As computational capabilities increase, molecular dynamics (MD) simulations become important tools of simulating reality. These simulations are especially useful for compressible gas mixture problems. In this study, binary diffusion of helium and argon was examined using a hard-sphere MD simulation method. For the sake of computational speed, low spacing ratios were chosen. Binary mass diffusion of gases in two equally sized halves of a box was simulated for identical initial kinetic energies and number densities. It has been noted that a purely mass diffusion mechanism of different gases is not physically possible. The resultant gas mixtures of several diffusion simulations were used as initial conditions for combined heat transfer – Couette flow, and heating and cooling experiments. The results showed the interesting behaviour of the mixture, which was subjected to various wall conditions. Energy of heavier molecules is found to be more sensitive to the wall velocities and less sensitive to the wall temperatures than lighter molecules. Diffusion, heat transfer, viscosity and heat capacity coefficients are deduced as well.

Data-based Reconstruction of Gene Regulatory Networks of Fungal Pathogens.

In the emerging field of systems biology of fungal infection, one of the central roles belongs to the modeling of gene regulatory networks (GRNs). Utilizing omics-data, GRNs can be predicted by mathematical modeling. Here, we review current advances of data-based reconstruction of both small-scale and large-scale GRNs for human pathogenic fungi. The advantage of large-scale genome-wide modeling is the possibility to predict central (hub) genes and thereby indicate potential biomarkers and drug targets. In contrast, small-scale GRN models provide hypotheses on the mode of gene regulatory interactions, which have to be validated experimentally. Due to the lack of sufficient quantity and quality of both experimental data and prior knowledge about regulator–target gene relations, the genome-wide modeling still remains problematic for fungal pathogens. While a first genome-wide GRN model has already been published for Candida albicans, the feasibility of such modeling for Aspergillus fumigatus is evaluated in the present article. Based on this evaluation, opinions are drawn on future directions of GRN modeling of fungal pathogens. The crucial point of genome-wide GRN modeling is the experimental evidence, both used for inferring the networks (omics ‘first-hand’ data as well as literature data used as prior knowledge) and for validation and evaluation of the inferred network models.

Hydrodynamic based hybrid dynamic load balancing

Balanced load distribution is especially important to attain optimal use of existing computational resources in distributed and parallel applications. In dynamic load balancing (DLB), surplus workload in nodes overwhelmed with work is transferred to relatively free nodes during run time. While in iterative DLB methods, the load reaches to its final execution node through several iteration steps, the execution node is selected directly in one step in direct methods. However, direct methods require immense system state information to perform selection. In this paper, we present two new hybrid dynamic load balancing (HLB) methods that take advantage of both direct and iterative methods. HLB aims to consume minimum possible system resources to balance common workload distributions by using an iterative method, the hydrodynamic approach (HA). Besides, HLB intends to solve the problems derived from exceptional instantaneous load rises by using a direct method which requires only little system knowledge. The excess workload is shared directly with some non-neighboring nodes which are selected randomly or from a fixed distribution list. The experimental results designate that the hybrid methods outrun other iterative methods in terms of performance and whole system utilization.

Efficient methods for database storage and retrieval using space-filling curves

Efficient storage and retrieval of records involving multiple keys is a difficult and well-studied problem. A popular solution employed is to visualize the records as points in multidimensional space and use a mapping from this multidimensional space to the one-dimensional space of block addresses in secondary storage. There is significant interest in performing such a mapping using space-filling curves. Unfortunately, space-filling curves are defined for points that lie on a uniform grid of a particular resolution. As a result, both storage and retrieval algorithms based on space-filling curves depend upon the size of the grid. This makes the run time of such algorithms dependent on the distribution of the points and in fact, unbounded for arbitrary distributions. There are two main contributions in this paper: First, we present a distribution-independent algorithm for storing records with multiple keys using space-filling curves. Our algorithm runs in O(knlog n) time for storing n records containing k key fields. We then present an algorithm for answering range queries with a bounded running time independent of the distribution.

CUDA implementation of the pixel based adaptive segmentation algorithm

The operation of foreground background subtraction, which has a crucial role on video processing, needs to be processed in real time. In this paper, we present an implement of the pixel based adaptive segmentation (PBAS) algorithm on CUDA parallel programming platform to process high resolution videos in real time. The performance of the implementation is enhanced by using Nsight profiler metrics such as consumed time for functions, usage of GPU compute and memory, occupancy, divergence. Experimental evaluation on benchmark videos presents the performance of the implementation; On a Tesla K20 graphic processing unit, 646 fps is achieved for a video with 320×240 resolution when memory transfer to and from the GPU is included. As a result, about 17 time speedup is achieved with regards to single thread version.