Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallel optimization”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 451 records · Page 25

Evaluating In-Clique and Topological Parallelism Strategies for Junction Tree-Based Bayesian Inference Algorithm on the Cray XMT

Long viewed as a strong statistical inference technique, Bayesian networks have emerged to be an important class of applications for high-performance computing. We have applied an architecture-conscious approach to parallelizing the Lauritzen-Spiegelhalter Junction Tree algorithm for exact inferencing in Bayesian networks. In optimizing the Junction Tree algorithm, we have implemented both in-clique and topological parallelism strategies to best leverage the fine-grained synchronization and massive-scale multithreading of the Cray XMT architecture. Two topological techniques were developed to parallelize the evidence propagation process through the Bayesian network. One technique involves performing intelligent scheduling of junction tree nodes based on its topology and relative size. The second technique involves decomposing the junction tree into a much finer tree-like representation to offer much more opportunities for parallelism. We evaluate these optimizations on five different Bayesian networks and report our findings and observations. Another important contribution of this paper is to demonstrate the application of massive-scale multithreading for load balancing and use of implicit parallelism-based compiler optimizations in designing scalable inferencing algorithms.

99 GENERAL AND MISCELLANEOUS↗

HPC formulations of optimization algorithms for tensor completion

Tensor completion is a powerful tool used to estimate or recover missing values in multi-way data. It has seen great success in domains such as product recommendation and healthcare. Tensor completion is most often accomplished via low-rank sparse tensor factorization, a computationally expensive non-convex optimization problem which has only recently been studied in the context of parallel computing. In this work, we study three optimization algorithms that have been successfully applied to tensor completion: alternating least squares (ALS), stochastic gradient descent (SGD), and coordinate descent (CCD++). We explore opportunities for parallelism on shared- and distributed-memory systems and address challenges such as memory- and operation-efficiency, load balance, cache locality, and communication. Among our advancements are a communication-efficient CCD++ algorithm, an ALS algorithm rich in level-3 BLAS routines, and an SGD algorithm which combines stratification with asynchronous communication. Furthermore, we show that introducing randomization during ALS and CCD++ can accelerate convergence. We evaluate our parallel formulations on a variety of real datasets on a modern supercomputer and demonstrate speedups through 16384 cores. Further, these improvements reduce time-to-solution from hours to seconds on real-world datasets. We show that after our optimizations, ALS is advantageous on parallel systems of small-to-moderate scale, while both ALS and CCD++ provide the lowest time-to-solution on large-scale distributed systems.

97 MATHEMATICS AND COMPUTING↗

Simulation, Characterization, and Optimization of Metabolic Models with the High Performance Systems Biology Toolkit

The High Performance Systems Biology Toolkit (HiPer SBTK) is a collection of simulation and optimization components for metabolic modeling and the means to assemble these components into large parallel processing hierarchies suiting a particular simulation and optimization need. The components come in a variety of different categories: model translation, model simulation, parameter sampling, sensitivity analysis, parameter estimation, and optimization. They can be configured at runtime into hierarchically parallel arrangements to perform nested combinations of simulation characterization tasks with excellent parallel scaling to thousands of processors. We describe the observations that led to the system, the components, and how one can arrange them. We show nearly 90% efficient scaling to over 13,000 processors, and we demonstrate three complex yet typical examples that have run on {approx}1000 processors and accomplished billions of stiff ordinary differential equation simulations. This work opens the door for the systems biology metabolic modeling community to take effective advantage of large scale high performance computing resources for the first time.

59 BASIC BIOLOGICAL SCIENCES↗

Materials Design using an Active Subspace-based Batch Bayesian Optimization Approach

Integrated computational materials engineering (ICME) calls for integrating simulation tools and/or experiments to develop new materials and materials systems. However, implementation of ICME approaches is challenging mainly due to the considerable computational expense of such frameworks and large dimensionality of the design space. Addressing these challenges is thus critical to the success of ICME initiatives. We present here a specific Bayesian optimization framework designed to address these two challenges. In particular, we propose an active subspace batch Bayesian optimization framework. The framework makes use of dimension reduction via the active subspace method and makes use of the ability to query in parallel via the batch Bayesian optimization approach. Here, the integration of these techniques leads to significant efficiency improvements while maintaining accuracy.

36 MATERIALS SCIENCE↗

Optimizing Management of Persistent Data Structures in High-Performance Analytics

Large-scale data analytics workflows ingest massive input data into various data structures, including graphs and key-value datastores. These data structures undergo multiple transformations and computations and are typically reused in incremental and iterative analytics workflows. Persisting in-memory views of these data structures enables reusing them beyond the scope of a single program run while avoiding repetitive raw data ingestion overheads. Memory-mapped I/O enables persisting in-memory data structures without data serialization and deserialization overheads. However, memory-mapped I/O lacks the key feature of persisting consistent snapshots of these data structures for incremental ingestion and processing. The obstacles to efficient virtual memory snapshots using memory-mapped I/O include background writebacks outside the application’s control, and the significantly high storage footprint of such snapshots. To address these limitations, we present Privateer, a memory and storage management tool that enables storage-efficient virtual memory snapshotting while also optimizing snapshot I/O performance. Here, we integrated Privateer into Metall, a state-of-the-art persistent memory allocator for C++, and the Lightning Memory-Mapped Database (LMDB), a widely-used key-value datastore in data analytics and machine learning. Privateer optimized application performance by 1.22× when storing data structure snapshots to node-local storage, and up to 16.7× when storing snapshots to a parallel file system. Privateer also optimizes storage efficiency of incremental data structure snapshots by up to 11× using data deduplication and compression.

Computer science↗

Comparing current cluster, massively parallel, and accelerated systems

Currently there is large architectural diversity in high perfonnance computing systems. They include 'commodity' cluster systems that optimize per-node performance for small jobs, massively parallel processors (MPPs) that optimize aggregate perfonnance for large jobs, and accelerated systems that optimize both per-node and aggregate performance but only for applications custom-designed to take advantage of such systems. Because of these dissimilarities, meaningful comparisons of achievable performance are not straightforward. In this work we utilize a methodology that combines both empirical analysis and performance modeling to compare clusters (represented by a 4,352-core IB cluster), MPPs (represented by a 147,456-core BG/P), and accelerated systems (represented by the 129,600-core Roadrunner) across a workload of four applications. Strengths of our approach include the ability to compare architectures - as opposed to specific implementations of an architecture - attribute each application's performance bottlenecks to characteristics unique to each system, and to explore performance scenarios in advance of their availability for measurement. Our analysis illustrates that application performance is essentially unrelated to relative peak performance but that application performance can be both predicted and explained using modeling.

97 MATHEMATICS AND COMPUTING↗

HyperSpace

Machine learning (ML) models often contain numerous hyperparameters, free parameters that must be set before the models can be trained. As the number of model hyperparameters increases, their optimization becomes significantly more challenging as we face a combinatorial increase in potential model configurations. Similarly, there is an increased chance that our models’ hyperparameters interact in complex ways. HyperSpace allows a user to optimize complex machine learning algorithms based on their hyperparameters. HyperSpace works by parallelizing parameter search spaces, running Bayesian model based optimization (SMBO) over each of these spaces in parallel. It was designed to be as minimally invasive as possible such that very little change to existing code will be needed to get a user started.

Ramanathan, Arvind↗

Advances in global optimization of high brightness beams

High brightness electron beams play an important role in accelerator-based applications such as driving X-ray free electron laser (FEL) radiation.Here in this paper, we report on advances in global beam dynamics optimization of an accelerator design using start-to-end simulations and a new parallel multi-objective differential evolution optimization method. The global optimization results in significant improvement of the final electron beam brightness.

43 PARTICLE ACCELERATORS↗

Performance Optimization of a Beam Dynamics PIC Code On Hybrid Computer Architectures

The self-consistent multi-particle tracking based on particle-in-cell method (PIC) has been widely used in particle accelerator beam dynamics study. However, the PIC simulation is time-consuming and needs to use modern parallel computers for high resolution applications. In this paper, we implemented and optimized a parallel beam dynamics PIC code on two types of hybrid parallel computer architectures: one is the GPU and GPU cluster, while the other is the "Knight Landing" CPU cluster.

Liu, Zhicong↗

Designing Critical Experiments Using Gaussian Process Optimization

New nuclear applications may lack similarity between the application and existing ICSBEP benchmarks. Without accurate benchmark data, predictive simulations will have high uncertainty in the estimate of a parameter of interest (such as k eff ). The design of critical experiments tuned for specific applications of interest is one solution to reducing uncertainty in the necessarily nuclear data. We propose an optimization technique to design critical experiments that match the cross-section sensitivities of an application model while minimizing the number of simulation runs in the optimization. The optimization of a critical experiment is slow because the potential design space covers many qualitative variables (types of material and geometries) and corresponding quantitative variables (e.g., material dimensions) and the evaluation of any specific design requires running an expensive code like MCNP1 and Whisper. The optimization of expensive black-box computer codes with quantitative variables has been extensively explored in the literature of design of computer experiments. We propose a modified version of standard Gaussian process (GP) optimization that optimizes over multiple discrete configurations of qualitative variables in parallel which reduces the computational cost of the entire optimization. The goal of this work is to minimize the number of MCNP evaluations while still exploring many different design configurations in parallel.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Ultrafast stimulated Raman parallel adiabatic passage by shaped pulses

We present a general and versatile technique of population transfer based on parallel adiabatic passage by femtosecond shaped pulses. Their amplitude and phase are specifically designed to optimize the adiabatic passage corresponding to parallel eigenvalues at all times. We show that this technique allows the robust adiabatic population transfer in a Raman system with the total pulse area as low as 3{pi}, corresponding to a fluence of one order of magnitude below the conventional stimulated Raman adiabatic passage process. This process of short duration, typically picosecond and subpicosecond, is easily implementable with the modern pulse shaper technology and opens the possibility of ultrafast robust population transfer with interesting applications in quantum information processing.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

...And Eat it Too: High Read Performance in Write-Optimized HPC I/O Middleware File Formats

As HPC applications run on increasingly high process counts on larger and larger machines, both the frequency of checkpoints needed for fault tolerance and the resolution and size of Data Analysis Dumps are expected to increase proportionally. In order to maintain an acceptable ratio of time spent performing useful computation work to time spent performing I/O, write bandwidth to the underlying storage system must increase proportionally to this increase in the checkpoint and computation size. Unfortunately, popular scientific self-describing file formats such as netCDF and HDF5 are designed with a focus on portability and flexibility. Extra care and careful crafting of the output structure and API calls is required to optimize for write performance using these APIs. To provide sufficient write bandwidth to continue to support the demands of scientific applications, the HPC community has developed a number of I/O middleware layers, that structure output into write-optimized file formats. However, the obvious concern with any write optimized file format would be a corresponding penalty on reads. In the log-structured filesystem, for example, a file generated by random writes could be written efficiently, but reading the file back sequentially later would result in very poor performance. Simulation results require efficient read-back for visualization and analytics, and though most checkpoint files are never used, the efficiency of a restart is very important in the face of inevitable failures. The utility of write speed improving middleware would be greatly diminished if it sacrificed acceptable read performance. In this paper we examine the read performance of two write-optimized middleware layers on large parallel machines and compare it to reading data natively in popular file formats. The two I/O middleware layer examined in this paper are the Adaptable IO System (ADIOS), a library-based approach developed at Oak Ridge National Laboratory to provide a high-level IO API that can be used in place of netCDF or HDF5 to do much more aggressive write-behind and efficient reordering of data locations within the file; and the Parallel Log-structured Filesystem (PLFS), a stackable FUSE filesystem approach developed at Los Alamos National Laboratory that decouples concurrent writes to improve the speed of checkpoints. Since ADIOS is an I/O componentization that affords selection of different I/O methods at or during runtime, through a single API, users can have access to MPI-IO, Posix-IO, HDF5, netCDF, and staging methods. The ADIOS BP file format is a new log-file format that has a superset of the features of both HDF5 and netCDF, but is designed to be portable and flexible while being optimized for writing. PLFS, on the other hand, is mounted as a stackable filesystem on top of an existing parallel filesystem. Reads or writes to the PLFS filesystem are transparently translated into operations on per-process log files stored in the underlying parallel filesystem. Since PLFS performs this translation without application modification, users can write in HDF5, netCDF, or app-specific file formats and PLFS will store the writes in a set of efficiently written log-formatted files, while presenting the user with a logical 'flat' file on reads. Despite their different approaches, the commonality behind both of these middleware systems is that they both write to a log file format. As shown in previous publications, writes are fully optimized in both systems, sometimes resulting in 100x improvements over writing data in popular file formats. But as mentioned above, writing is only one part of the story. In this paper we examine the read performance of our middleware layers on large parallel machines and compare these to reading data either natively or from other popular file formats. We compare the reading performance in two different scenarios: (1) Reading back restarts from the same number of processors as wrote the data and (2) Reading back restart data from a different number of processors. We observe that not only can write-optimized I/O middleware be built to not penalize read speeds, but for important workloads, techniques that improve write performance can, perhaps counterintuitively, improve read speeds over reading from a contiguously organized file format. In the remainder of this paper, we investigate this further through case studies of PLFS and ADIOS on simulation checkpoint restarts.

99 GENERAL AND MISCELLANEOUS↗

High-Level Synthesis of Irregular Applications: A Case Study on Influence Maximization

The Influence Maximization problem is the problem of identifying a small cohort of actors from a broader population that, when initially activated in a diffusion process, are expected to result in a large number of activations in the population. While the problem is known to be NP-hard, several approximation algorithms have been devised by leveraging its submodular structure. While these algorithms are theoretically efficient, they are computationally very expensive in practice. This work advances the current state-of-the-art parallelization scheme for the IMM algorithm by devising the adoption of custom hardware accelerators implemented on FPGAs by leveraging High Level Synthesis from OpenCL. We study the performance of our proposed approach by exploring optimizations tailored at improving the parallel efficiency of the accelerators and highlight their effects and limitations in accelerating complex graph analytic applications. Our experimental evaluation shows that FPGA acceleration can improve the performance of the LT diffusion model up to 1.72x for the entire application and up to 2.90x for its most important kernel with respect to a CPU only parallel execution. The FPGA acceleration of the LT model shows also a 1.54x reduction in energy consumption when compared to a parallel CPU only run.

Neff, Reece W.↗

Wakefield Computations for the CLIC PETS using the Parallel Finite Element Time-Domain Code T3P

In recent years, SLAC's Advanced Computations Department (ACD) has developed the high-performance parallel 3D electromagnetic time-domain code, T3P, for simulations of wakefields and transients in complex accelerator structures. T3P is based on advanced higher-order Finite Element methods on unstructured grids with quadratic surface approximation. Optimized for large-scale parallel processing on leadership supercomputing facilities, T3P allows simulations of realistic 3D structures with unprecedented accuracy, aiding the design of the next generation of accelerator facilities. Applications to the Compact Linear Collider (CLIC) Power Extraction and Transfer Structure (PETS) are presented.

43 PARTICLE ACCELERATORS↗

Noise-aware optimization in nominally identical manufacturing and measuring systems for high-throughput parallel workflows

Device-to-device variability in experimental noise critically impacts reproducibility, especially in automated, high-throughput systems like additive manufacturing farms. While manageable in small labs, such variability can escalate into serious risks at larger scales, such as architectural 3D printing, where noise may cause structural or economic failures. This contribution presents a noise-aware decision-making algorithm that quantifies and models device-specific noise profiles to manage variability adaptively. It uses distributional analysis and pairwise divergence metrics with clustering to choose between single-device and robust multi-device Bayesian optimization strategies. Unlike conventional methods that assume homogeneous devices or enforce generic robustness, the proposed framework explicitly determines whether shared optimization across devices is appropriate based on the degree of inter-device noise heterogeneity. This enables improved performance, reproducibility, and efficiency. An experimental case study involving three nominally identical 3D printers (same brand, model, and close serial numbers) demonstrates reduced redundancy, lower resource usage, and improved reliability, along with improved convergence stability and solution quality through the selection of the appropriate optimization strategy based on the degree of inter-device noise heterogeneity. Overall, this framework establishes a general approach for precision- and resource-aware optimization in scalable, automated experimental platforms, demonstrated here on a representative multi-device 3D printing case study.

Schenk, Christina↗

Optimal quantum learning of a unitary transformation

We address the problem of learning an unknown unitary transformation from a finite number of examples. The problem consists in finding the learning machine that optimally emulates the examples, thus reproducing the unknown unitary with maximum fidelity. Learning a unitary is equivalent to storing it in the state of a quantum memory (the memory of the learning machine) and subsequently retrieving it. We prove that, whenever the unknown unitary is drawn from a group, the optimal strategy consists in a parallel call of the available uses followed by a 'measure-and-rotate' retrieving. Differing from the case of quantum cloning, where the incoherent 'measure-and-prepare' strategies are typically suboptimal, in the case of learning the 'measure-and-rotate' strategy is optimal even when the learning machine is asked to reproduce a single copy of the unknown unitary. We finally address the problem of the optimal inversion of an unknown unitary evolution, showing also in this case the optimality of the 'measure-and-rotate' strategies and applying our result to the optimal approximate realignment of reference frames for quantum communication.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗