We present MacKenzie, a HPC-driven multi-cluster workflow system that was used repeatedly to configure and execute fine-grained US national-scale epidemic simulation models during the COVID-19 pandemic. Mackenzie supported federal and Virginia policymakers, in real-time, for a large number of "what-if" scenarios during the COVID-19 pandemic, and continues to be used to answer related questions as COVID-19 transitions to the endemic stage of the disease. MacKenzie is a novel HPC meta-scheduler that can execute US-scale simulation models and associated workflows that typically present significant big data challenges. The meta-scheduler optimizes the total execution time of simulations in the workflow, and helps improve overall human productivity. As an exemplar of the kind of studies that can be conducted using Mackenzie, we present a modeling study to understand the impact of vaccine-acceptance in controlling the spread of COVID-19 in the US. We use a 288 million node synthetic social contact network (digital twin) spanning all 50 US states plus Washington DC, comprised of 3300 counties, with 12 billion daily interactions. The highly-resolved agent-based model used for the epidemic simulations uses realistic information about disease progression, vaccine uptake, production schedules, acceptance trends, prevalence, and social distancing guidelines. Computational experiments show that, for the simulation workload discussed above, MacKenzie is able to scale up well to 10K CPU cores. Our modeling results show that, when compared to faster and accelerating vaccinations, slower vaccination rates due to vaccine hesitancy cause averted infections to drop from 6.7M to 4.5M, and averted total deaths to drop from 39.4K to 28.2K across the US. This occurs despite the fact that the final vaccine coverage is the same in both scenarios. We also find that if vaccine acceptance could be increased by 10% in all states, averted infections could be increased from 4.5M to 4.7M (a 4.4% improvement) and total averted deaths could be increased from 28.2K to 29.9K (a 6% improvement) nationwide.

Mass spectrometry (MS) based omics data analysis require significant time and resources. To date, few parallel algorithms have been proposed for deducing peptides from mass spectrometry-based data. However, these parallel algorithms were designed, and developed when the amount of data that needed to be processed was smaller in scale. In this paper, we prove that the communication bound that is reached by the parallel algorithms is , where and are the dimensions of the theoretical database matrix, and are dimensions of spectra, and is the number of processors. We further prove that communication-optimal strategy with fast-memory can achieve but is not achieved by any existing parallel proteomics algorithms till date. To validate our claim, we performed a meta-analysis of published parallel algorithms, and their performance results. We show that sub-optimal speedups with increasing number of processors is a direct consequence of not achieving the communication lower-bounds. We further validate our claim by performing experiments which demonstrate the communication bounds that are proved in this paper. Consequently, we assert that next-generation of , and demonstrated superior parallel algorithms are urgently needed for MS based large systems-biology studies especially for meta-proteomics, proteogenomic, microbiome, and proteomics for non-model organisms. Our hope is that this paper will excite the parallel computing community to further investigate parallel algorithms for highly influential MS based omics problems.

3D image registration is one of the most fundamental and computationally expensive operations in medical image analysis. Here, we present a mixed-precision, Gauss-Newton-Krylov solver for diffeomorphic registration of two images. Our work extends the publicly available CLAIRE library to GPU architectures. Despite the importance of image registration, only a few implementations of large deformation diffeomorphic registration packages support GPUs. Our contributions are new algorithms to significantly reduce the run time of the two main computational kernels in CLAIRE: calculation of derivatives and scattered-data interpolation. We deploy (i) highly-optimized, mixed-precision GPU-kernels for the evaluation of scattered-data interpolation, (ii) replace Fast-Fourier-Transform (FFT)-based first-order derivatives with optimized 8th-order finite differences, and (iii) compare with state-of-the-art CPU and GPU implementations. As a highlight, we demonstrate that we can register 256 clinical images in less than 6 seconds on a single NVIDIA Tesla V100. This amounts to over 20× speed-up over the current version of CLAIRE and over 30× speed-up over existing GPU implementations.

In prior works, stochastic dual coordinate ascent (SDCA) has been parallelized in a multi-core environment where the cores communicate through shared memory, or in a multi-processor distributed memory environment where the processors communicate through message passing. In this paper, we propose a hybrid SDCA framework for multi-core clusters, the most common high performance computing environment that consists of multiple nodes each having multiple cores and its own shared memory. We distribute data across nodes where each node solves a local problem in an asynchronous parallel fashion on its cores, and then the local updates are aggregated via an asynchronous across-node update scheme. The proposed double asynchronous method converges to a global solution for -Lipschitz continuous loss functions, and at a linear convergence rate if a smooth convex loss function is used. Extensive empirical comparison has shown that our algorithm scales better than the best known shared-memory methods and runs faster than previous distributed-memory methods. Big datasets, such as one of 280 GB from the LIBSVM repository, cannot be accommodated on a single node and hence cannot be solved by a parallel algorithm. For such a dataset, our hybrid algorithm takes less than 30 seconds to achieve a duality gap of 10 on 16 nodes each using 12 cores, which is significantly faster than the best known distributed algorithms, such as CoCoA+, that take more than 160 seconds on 16 nodes.

Bayesian Sequential Partitioning (BSP) is a statistically effective density estimation method to comprehend the characteristics of a high dimensional data space. The intensive computation of the statistical model and the counting of enormous data have caused serious design challenges for BSP to handle the growing volume of the data. This paper proposes a high performance design of BSP by leveraging a heterogeneous CPU/GPGPU system that consists of a host CPU and a K80 GPGPU. A series of techniques, on both data structures and execution management policies, is implemented to extensively exploit the computation capability of the heterogeneous many-core system and alleviate system bottlenecks. When compared with a parallel design on a high-end CPU, the proposed techniques achieve 48x average runtime enhancement while the maximum speedup can reach 78.76x.

A large number of real world networks exhibit community structure, and different communities may often possess heterogeneity. In this paper, considering the heterogeneity among communities, we construct a new community network model in which the communities show significant differences in average degree. Based on this heterogeneous community network, we propose a novel mathematical epidemic model for each community and study the epidemic dynamics in this network model. We find that the location of the initial infection node only affects the spreading velocity and barely influences the epidemic prevalence. And the epidemic threshold of entire network decreases with the increase of heterogeneity among communities. Moreover, the epidemic prevalence increases with the increase of heterogeneity around the epidemic threshold, while the converse situation holds when the infection rate is much greater than the epidemic threshold.

An edge switch is an operation on a graph (or network) where two edges are selected randomly and one of their end vertices are swapped with each other. Edge switch operations have important applications in graph theory and network analysis, such as in generating random networks with a given degree sequence, modeling and analyzing dynamic networks, and in studying various dynamic phenomena over a network. The recent growth of real-world networks motivates the need for efficient parallel algorithms. The dependencies among successive edge switch operations and the requirement to keep the graph simple (i.e., no self-loops or parallel edges) as the edges are switched lead to significant challenges in designing a parallel algorithm. Addressing these challenges requires complex synchronization and communication among the processors leading to difficulties in achieving a good speedup by parallelization. In this paper, we present distributed memory parallel algorithms for switching edges in massive networks. These algorithms provide good speedup and scale well to a large number of processors. A harmonic mean speedup of 73.25 is achieved on eight different networks with 1024 processors. One of the steps in our edge switch algorithms requires the computation of multinomial random variables in parallel. This paper presents the first non-trivial parallel algorithm for the problem, achieving a speedup of 925 using 1024 processors.

A graph is chordal if every cycle of length greater than three contains an edge between non-adjacent vertices. Chordal graphs are of interest both theoretically, since they admit polynomial time solutions to a range of NP-hard graph problems, and practically, since they arise in many applications including sparse linear algebra, computer vision, and computational biology. A maximal chordal subgraph is a chordal subgraph that is not a proper subgraph of any other chordal subgraph. Existing algorithms for computing maximal chordal subgraphs depend on dynamically ordering the vertices, which is an inherently sequential process and therefore limits the algorithms' parallelizability. In this paper we explore techniques to develop a scalable parallel algorithm for extracting a maximal chordal subgraph. We demonstrate that an earlier attempt at developing a parallel algorithm may induce a non-optimal vertex ordering and is therefore not guaranteed to terminate with a chordal subgraph. We then give a new algorithm that first computes and then repeatedly augments a spanning chordal subgraph. After proving that the algorithm terminates with a maximal chordal subgraph, we then demonstrate that this algorithm is more amenable to parallelization and that the parallel version also terminates with a maximal chordal subgraph. That said, the complexity of the new algorithm is higher than that of the previous parallel algorithm, although the earlier algorithm computes a chordal subgraph which is not guaranteed to be maximal. We experimented with our augmentation-based algorithm on both synthetic and real-world graphs. We provide scalability results and also explore the effect of different choices for the initial spanning chordal subgraph on both the running time and on the number of edges in the maximal chordal subgraph.

Large-scale compute clusters of heterogeneous nodes equipped with multi-core CPUs and GPUs are getting increasingly popular in the scientific community. However, such systems require a combination of different programming paradigms making application development very challenging. In this article we introduce libWater, a library-based extension of the OpenCL programming model that simplifies the development of heterogeneous distributed applications. libWater consists of a simple interface, which is a transparent abstraction of the underlying distributed architecture, offering advanced features such as inter-context and inter-node device synchronization. It provides a runtime system which tracks dependency information enforced by event synchronization to dynamically build a DAG of commands, on which we automatically apply two optimizations: collective communication pattern detection and device-host-device copy removal. We assess libWater's performance in three compute clusters available from the Vienna Scientific Cluster, the Barcelona Supercomputing Center and the University of Innsbruck, demonstrating improved performance and scaling with different test applications and configurations.

Several recent methods have been proposed to obtain significant speed-ups in MRI image reconstruction by leveraging the computational power of GPUs. Previously, we implemented a GPU-based image reconstruction technique called the Illinois Massively Parallel Acquisition Toolkit for Image reconstruction with ENhanced Throughput in MRI (IMPATIENT MRI) for reconstructing data collected along arbitrary 3D trajectories. In this paper, we improve IMPATIENT by removing computational bottlenecks by using a gridding approach to accelerate the computation of various data structures needed by the previous routine. Further, we enhance the routine with capabilities for off-resonance correction and multi-sensor parallel imaging reconstruction. Through implementation of optimized gridding into our iterative reconstruction scheme, speed-ups of more than a factor of 200 are provided in the improved GPU implementation compared to the previous accelerated GPU code.

Genome resequencing with short reads generated from pyrosequencing generally relies on mapping the short reads against a single reference genome. However, mapping of reads from multiple reference genomes is not possible using a pairwise mapping algorithm. In order to align the reads w.r.t each other and the reference genomes, existing multiple sequence alignment(MSA) methods cannot be used because they do not take into account the position of these short reads with respect to the genome, and are highly inefficient for large number of sequences. In this paper, we develop a highly scalable parallel algorithm based on domain decomposition, referred to as P-Pyro-Align, to align such large number of reads from single or multiple reference genomes. The proposed alignment algorithm accurately aligns the erroneous reads, and has been implemented on a cluster of workstations using MPI library. Experimental results for different problem sizes are analyzed in terms of execution time, quality of the alignments, and the ability of the algorithm to handle reads from multiple haplotypes. We report high quality multiple alignment of up to 0.5 million reads. The algorithm is shown to be highly scalable and exhibits super-linear speedups with increasing number of processors.

In this paper we present efficient algorithms for sorting on the Parallel Disks Model (PDM). Numerous asymptotically optimal algorithms have been proposed in the literature. However many of these merge based algorithms have large underlying constants in the time bounds, because they suffer from the lack of read parallelism on PDM. The irregular consumption of the runs during the merge affects the read parallelism and contributes to the increased sorting time. In this paper we first introduce a novel idea called the dirty sequence accumulation that improves the read parallelism. Secondly, we show analytically that this idea can reduce the number of parallel I/O's required to sort the input close to the lower bound of [Formula: see text]. We experimentally verify our dirty sequence idea with the standard R-Way merge and show that our idea can reduce the number of parallel I/Os to sort on PDM significantly.

We present a multi-heuristic evolutionary task allocation algorithm to dynamically map tasks to processors in a heterogeneous distributed system. It utilizes a genetic algorithm, combined with eight common heuristics, in an effort to minimize the total execution time. It operates on batches of unmapped tasks and can preemptively remap tasks to processors. The algorithm has been implemented on a Java distributed system and evaluated with a set of six problems from the areas of bioinformatics, biomedical engineering, computer science and cryptography. Experiments using up to 150 heterogeneous processors show that the algorithm achieves better efficiency than other state-of-the-art heuristic algorithms.

Tomographic imaging and computer simulations are increasingly yielding massive datasets. Interactive and exploratory visualizations have rapidly become indispensable tools to study large volumetric imaging and simulation data. Our scalable isosurface visualization framework on commodity off-the-shelf clusters is an end-to-end parallel and progressive platform, from initial data access to the final display. Interactive browsing of extracted isosurfaces is made possible by using parallel isosurface extraction, and rendering in conjunction with a new specialized piece of image compositing hardware called Metabuffer. In this paper, we focus on the back end scalability by introducing a fully parallel and out-of-core isosurface extraction algorithm. It achieves scalability by using both parallel and out-of-core processing and parallel disks. It statically partitions the volume data to parallel disks with a balanced workload spectrum, and builds I/O-optimal external interval trees to minimize the number of I/O operations of loading large data from disk. We also describe an isosurface compression scheme that is efficient for progress extraction, transmission and storage of isosurfaces.

Computational acceleration on graphics processing units (GPUs) can make advanced magnetic resonance imaging (MRI) reconstruction algorithms attractive in clinical settings, thereby improving the quality of MR images across a broad spectrum of applications. This paper describes the acceleration of such an algorithm on NVIDIA's Quadro FX 5600. The reconstruction of a 3D image with 128(3) voxels achieves up to 180 GFLOPS and requires just over one minute on the Quadro, while reconstruction on a quad-core CPU is twenty-one times slower. Furthermore, relative to the true image, the error exhibited by the advanced reconstruction is only 12%, while conventional reconstruction techniques incur error of 42%.

High-dimensional problems arising from robot motion planning, biology, data mining, and geographic information systems often require the computation of k nearest neighbor (knn) graphs. The knn graph of a data set is obtained by connecting each point to its k closest points. As the research in the above-mentioned fields progressively addresses problems of unprecedented complexity, the demand for computing knn graphs based on arbitrary distance metrics and large high-dimensional data sets increases, exceeding resources available to a single machine. In this work we efficiently distribute the computation of knn graphs for clusters of processors with message passing. Extensions to our distributed framework include the computation of graphs based on other proximity queries, such as approximate knn or range queries. Our experiments show nearly linear speedup with over one hundred processors and indicate that similar speedup can be obtained with several hundred processors.