Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Network partitioning”

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 91 records · Page 5

Scaling neural simulations in STACS

Abstract As modern neuroscience tools acquire more details about the brain, the need to move towards biological-scale neural simulations continues to grow. However, effective simulations at scale remain a challenge. Beyond just the tooling required to enable parallel execution, there is also the unique structure of the synaptic interconnectivity, which is globally sparse but has relatively high connection density and non-local interactions per neuron. There are also various practicalities to consider in high performance computing applications, such as the need for serializing neural networks to support potentially long-running simulations that require checkpoint-restart. Although acceleration on neuromorphic hardware is also a possibility, development in this space can be difficult as hardware support tends to vary between platforms and software support for larger scale models also tends to be limited. In this paper, we focus our attention on Simulation Tool for Asynchronous Cortical Streams (STACS), a spiking neural network simulator that leverages the Charm++ parallel programming framework, with the goal of supporting biological-scale simulations as well as interoperability between platforms. Central to these goals is the implementation of scalable data structures suitable for efficiently distributing a network across parallel partitions. Here, we discuss a straightforward extension of a parallel data format with a history of use in graph partitioners, which also serves as a portable intermediate representation for different neuromorphic backends. We perform scaling studies on the Summit supercomputer, examining the capabilities of STACS in terms of network build and storage, partitioning, and execution. We highlight how a suitably partitioned, spatially dependent synaptic structure introduces a communication workload well-suited to the multicast communication supported by Charm++. We evaluate the strong and weak scaling behavior for networks on the order of millions of neurons and billions of synapses, and show that STACS achieves competitive levels of parallel efficiency.

59 BASIC BIOLOGICAL SCIENCES↗

Scaling Irregular Applications through Data Aggregation and Software Multithreading

Bioinformatics, data analytics, semantic databases, knowledge discovery are emerging high performance application areas that exploit dynamic, linked data structures such as graphs, unbalanced trees or unstructured grids. These data structures usually are very large, requiring significantly more memory than available on single shared memory systems. Additionally, these data structures are difficult to partition on distributed memory systems. They also present poor spatial and temporal locality, thus generating unpredictable memory and network accesses. The Partitioned Global Address Space (PGAS) programming model seems suitable for these applications, because it allows using a shared memory abstraction across distributed-memory clusters. However, current PGAS languages and libraries are built to target regular remote data accesses and block transfers. Furthermore, they usually rely on the Single Program Multiple Data (SPMD) parallel control model, which is not well suited to the fine grained, dynamic and unbalanced parallelism of irregular applications. In this paper we present {\bf GMT} (Global Memory and Threading library), a custom runtime library that enables efficient execution of irregular applications on commodity clusters. GMT integrates a PGAS data substrate with simple fork/join parallelism and provides automatic load balancing on a per node basis. It implements multi-level aggregation and lightweight multithreading to maximize memory and network bandwidth with fine-grained data accesses and tolerate long data access latencies. A key innovation in the GMT runtime is its thread specialization (workers, helpers and communication threads) that realize the overall functionality. We compare our approach with other PGAS models, such as UPC running using GASNet, and hand-optimized MPI code on a set of typical large-scale irregular applications, demonstrating speedups of an order of magnitude.

Aggregation↗

Deep Reinforcement Learning Enabled Physical-Model-Free Two-Timescale Voltage Control Method for Active Distribution Systems

Active distribution networks are being challenged by frequent and rapid voltage violations due to renewable energy integration. Conventional model-based voltage control methods rely on accurate parameters of the distribution networks, which are difficult to achieve in practice. This paper proposes a novel physical-model-free two-timescale voltage control framework for active distribution systems. To achieve fast control of PV inverters, the whole network is first partitioned into several subnetworks using voltage-reactive power sensitivity. Then, the scheduling of PV inverters in the multiple sub-networks is formulated as Markov games and solved by a multi-agent soft actor-critic (MASAC) algorithm, where each subnetwork is modeled as an intelligent agent. All agents are trained in a centralized manner to learn a coordinated strategy while being executed based on only local information for fast response. For the slower time-scale control, OLTCs and switched capacitors are coordinated by a single agent-based SAC algorithm using the global information with considering control behaviors of the inverters. Particularly, the two-level agents are trained concurrently with information exchange according to the reward signal calculated from the data-driven surrogate model. Comparative tests with different benchmark methods on IEEE 33-and 123-bus systems and 342-node low voltage distribution system demonstrate that the proposed method can effectively mitigate the fast voltage violations and achieve systematical coordination of different voltage regulation assets without the knowledge of accurate system model.

24 POWER TRANSMISSION AND DISTRIBUTION↗

PhILMs: Collaboratory on Mathematics and Physics-Informed Learning Machines for Multiscale and Multiphysics Problems

The landscape of computational science and engineering is continually evolving, with the challenge of high-dimensional regression problems standing as a significant hurdle in numerous scientific endeavors. Addressing this challenge, our research, funded by this award, has led to the development of an innovative computational framework known as Probabilistic Partition of Unity Networks (PPOU-Nets). This initiative represents a collaborative effort to harness the potential of mathematics and physics-informed machine learning in tackling multiscale and multiphysics problems prevalent in high-dimensional spaces. Through this work, we have proposed a novel methodology that seamlessly integrates adaptive dimensionality reduction and a mixture of experts model, thereby facilitating a more efficient and accurate approximation of complex functions. This research effort has not only advanced the state of computational science but also opened new avenues for exploration in quantum computing and beyond. This report outlines the motivation, methodology, key findings, and implications of our work, underscoring our contributions to the broader scientific community and the potential pathways for future research.

97 MATHEMATICS AND COMPUTING↗

PCAfold 2.0—Novel tools and algorithms for low-dimensional manifold assessment and optimization

We describe an update to our open-source Python package, PCAfold, designed to help researchers generate, analyze and improve low-dimensional data manifolds. In the current version, PCAfold 2.0, we introduce novel tools and algorithms for assessing and optimizing low-dimensional manifolds. This includes a method that generates a “map” of local feature sizes that can help pinpoint researchers to problematic regions on a manifold. We introduce a novel cost function that characterizes the quality of a manifold topology with a single number. We develop two algorithms for feature selection based on principal component analysis (PCA) that use the cost function as an objective function to minimize. We introduce a quantity of interest (QoI)-aware dimensionality reduction strategy where data projections are computed using an artificial neural network and are directly optimized towards representing various projection-independent and projection-dependent QoIs. We also introduce an implementation of partition of unity networks (POUnets) for efficient reconstruction of QoIs from low-dimensional manifolds based on combining neural network classification with localized polynomial regression. Our software can be broadly applicable in all domains of science and engineering that aim to reduce data dimensionality, as well as in the fundamental research on representation learning.

97 MATHEMATICS AND COMPUTING↗

Global-Address Space Networking for Exascale

GASNet-EX is a portable, open-source, high-performance communication library designed to efficiently support the networking requirements of Partitioned Global Address Space (PGAS) runtime systems and other alternative models in future exascale machines. The library is an evolution of the popular GASNet communication system, building upon over 15 years of lessons learned. GASNet is a language-independent, networking middleware layer that provides network-independent, high-performance communication primitives including Remote Memory Access (RMA) and Active Messages (AM). It has been used to implement parallel programming models and libraries such as UPC, Co-Array Fortran, Titanium, Legion, Chapel, and many others. The interface is primarily intended as a compilation target and for use by runtime library writers (as opposed to end users), and the primary goals are high performance, interface portability, and expressiveness. GASNet stands for "Global-Address Space Networking".

ECP↗

Adaptive bandwidth management for entanglement distribution in quantum networks

Flexible grid wavelength division multiplexing is a powerful tool in lightwave communications to maximize spectral efficiency. In the emerging field of quantum networking, the need for effective resource provisioning is particularly acute, given the generally lower power levels, higher sensitivity to loss, and inapplicability of optical detection and retransmission. In this letter, we leverage flex grid technology to demonstrate reconfigurable distribution of quantum entanglement in a four-user tabletop network. By adaptively partitioning bandwidth with a single wavelength-selective switch, we successfully equalize two-party coincidence rates that initially differ by over two orders of magnitude. Our scalable approach introduces loss that is fixed with the number of users, offering a practical path for the establishment and management of quality-of-service guarantees in large quantum networks.

42 ENGINEERING↗

State Space Truncation with Quantified Errors for Accurate Solutions to Discrete Chemical Master Equation

The discrete chemical master equation (dCME) provides a general framework for studying stochasticity in mesoscopic reaction networks. Since its direct solution rapidly becomes intractable due to the increasing size of the state space, truncation of the state space is necessary for solving most dCMEs. It is therefore important to assess the consequences of state space truncations so errors can be quantified and minimized. Here we describe a novel method for state space truncation. By partitioning a reaction network into multiple molecular equivalence groups (MEGs), we truncate the state space by limiting the total molecular copy numbers in each MEG. We further describe a theoretical framework for analysis of the truncation error in the steady-state probability landscape using reflecting boundaries. By aggregating the state space based on the usage of a MEG and constructing an aggregated Markov process, we show that the truncation error of a MEG can be asymptotically bounded by the probability of states on the reflecting boundary of the MEG. Furthermore, truncating states of an arbitrary MEG will not undermine the estimated error of truncating any other MEGs. We then provide an overall error estimate for networks with multiple MEGs. To rapidly determine the appropriate size of an arbitrary MEG, we also introduce an a priori method to estimate the upper bound of its truncation error. This a priori estimate can be rapidly computed from reaction rates of the network, without the need of costly trial solutions of the dCME. As examples, we show results of applying our methods to the four stochastic networks of (1) the birth and death model, (2) the single gene expression model, (3) the genetic toggle switch model, and (4) the phage lambda bistable epigenetic switch model. We demonstrate how truncation errors and steady-state probability landscapes can be computed using different sizes of the MEG(s) and how the results validate our theories. Overall, the novel state space truncation and error analysis methods developed here can be used to ensure accurate direct solutions to the dCME for a large number of stochastic networks.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Configuring compute nodes of a parallel computer in an operational group into a plurality of independent non-overlapping collective networks

Methods, apparatus, and products are disclosed for configuring compute nodes of a parallel computer in an operational group into a plurality of independent non-overlapping collective networks, the compute nodes in the operational group connected together for data communications through a global combining network, that include: partitioning the compute nodes in the operational group into a plurality of non-overlapping subgroups; designating one compute node from each of the non-overlapping subgroups as a master node; and assigning, to the compute nodes in each of the non-overlapping subgroups, class routing instructions that organize the compute nodes in that non-overlapping subgroup as a collective network such that the master node is a physical root.

Archer, Charles J↗

Detecting multiple communities using quantum annealing on the D-Wave system

A very important problem in combinatorial optimization is the partitioning of a network into communities of densely connected nodes; where the connectivity between nodes inside a particular community is large compared to the connectivity between nodes belonging to different ones. This problem is known as community detection, and has become very important in various fields of science including chemistry, biology and social sciences. The problem of community detection is a twofold problem that consists of determining the number of communities and, at the same time, finding those communities. This drastically increases the solution space for heuristics to work on, compared to traditional graph partitioning problems. In many of the scientific domains in which graphs are used, there is the need to have the ability to partition a graph into communities with the “highest quality” possible since the presence of even small isolated communities can become crucial to explain a particular phenomenon. We have explored community detection using the power of quantum annealers, and in particular the D-Wave 2X and 2000Q machines. It turns out that the problem of detecting at most two communities naturally fits into the architecture of a quantum annealer with almost no need of reformulation. This paper addresses a systematic study of detecting two or more communities in a network using a quantum annealer.

97 MATHEMATICS AND COMPUTING↗

Partitioning of ocean and land uptake of CO2 as inferred by delta C-13 measurements from the NOAA Climate Monitoring and Diagnostics Laboratory Global Air Sampling Network

Using delta C-13 measurements in atmospheric CO2 from a cooperative global air sampling network, we determined the partitioning of the net uptake of CO2 between ocean and land as a function of latitude and time. The majority of delta C-13 measurements were made at the Institute of Arctic and Alpine Research (INSTAAR) of the University of Colorado. We perform an inverse deconvolution of both CO2 and delta C-13 observations, using a two-dimensional model of atmospheric transport. Also, the discrimination against C-13 by plant photosynthesis, as a function of latitude and time, is calculated from global runs of the simple biosphere (SiB) model. Uncertainty due to the longitudinal structure of the data, which is not represented by the model, is studied through a bootstrap analysis by adding and omitting measurement sites. The resulting error estimates for our inferred sources and sinks are of the order of 1 GTC (1 GTC = 10(exp 15) gC). Such error bars do not reflect potential systematic errors arising from our estimates of the isotopic disequilibria between the atmosphere and the oceans and biosphere, which are estimated in a separate sensitivity analysis. With respect to global totals for 1992 we found that 3.2 GTC of carbon dissolved into the ocean and that 1.5 GTC were sequestered by land ecosystems. Northern hemisphere ocean gyres north of 15 deg N absorbed 2.7 GTC. The equatorial oceans between 10 deg S and 10 deg N were a net source to the atmosphere of 0.9 GTC. We obtained a sink of 1.6 GTC in southern ocean gyres south of 20 deg S, although the deconvolution is poorly constrained by sparse data coverage at high southern latitudes. The seasonal uptake of CO2 in the northern gyres appears to be correlated with a bloom of phytoplankton in surface waters. On land, northern temperate and boreal ecosystems between 35 deg N and 65 deg N were found to be a major sink of CO2 in 1992, as large as 3.5 GTC. Northern tropical ecosystems (equator-30 deg N) appear to be a net source to the source to the atmosphere of 2 GTC which could reflect biomass burning. A small sink, 0.3 GTC, was inferred for southern tropical ecosystems (30 deg S-equator).

Ciais, Philippe↗

Beyond Music Sharing: An Evaluation of Peer-to-Peer Data Dissemination Techniques in Large Scientific Collaborations

The avalanche of data from scientific instruments and the ensuing interest from geographically distributed users to analyze and interpret it accentuates the need for efficient data dissemination. A suitable data distribution scheme will find the delicate balance between conflicting requirements of minimizing transfer times, minimizing the impact on the network, and uniformly distributing load among participants. We identify several data distribution techniques, some successfully employed by today's peer-to-peer networks: staging, data partitioning, orthogonal bandwidth exploitation, and combinations of the above. We use simulations to explore the performance of these techniques in contexts similar to those used by today's data-centric scientific collaborations and derive several recommendations for efficient data dissemination. Our experimental results show that the peer-to-peer solutions that offer load balancing and good fault tolerance properties and have embedded participation incentives lead to unjustified costs in today's scientific data collaborations deployed on over-provisioned network cores. However, as user communities grow and these deployments scale, peer-to-peer data delivery mechanisms will likely outperform other techniques.

97 MATHEMATICS AND COMPUTING↗

System theoretical investigation of human epidermal growth factor receptor-mediated signalling

The partitioning of biological networks into coupled functional modules is gaining increasing attention in the biological sciences. This approach has the advantage that predicting a system level response does not require a mechanistic description of the internal dynamics of each module. Identification of the input-output characteristics of the network modules and the connectivity between the modules provide the necessary quantitative representation of system dynamics. However, determination of the input-output relationships of the modules is not trivial; it requires the controlled perturbation of module inputs and systematic analysis of experimental data. In this report, we apply a system theoretical analysis approach to derive the causal input-output relationships of the functional module for the human epidermal growth factor receptor (HER) mediated Erk and Akt signaling pathways. Using a library of cell lines expressing varying levels of EGFR and HER2, we show that a transfer function-based representation can be successfully applied to quantitatively characterize information transfer in this system.

59 BASIC BIOLOGICAL SCIENCES↗

Optimizing Energy Efficiency of Plants: Quantitative Analysis of Carbon Flux

The efficiency in harvesting light energy is essential for both crop and feedstock productivity. Key targets for optimization of energy efficiency include energy capture, conversion, translocation and storage. Surprisingly little is known about the mechanisms of carbon allocation and the communication between source and sink. We surmise that regulatory networks that coordinate carbon partitioning and allocation might be control points and thus suitable for optimization by engineering. At a first step towards identifying such regulatory networks, our lab identified key players in missing steps of allocation, specifically cellular sugar uptake (SUT family, founding members identified by my team) and cellular efflux carriers (SWEET superfamily, founding members identified by my team). In plants, both classes of transporters are encoded by multigene families. We show that specific subsets of SWEETs serve critical roles in cellular sucrose efflux in the leaf for phloem loading, in seeds for seed filling, as well as in nectar section and vacuolar storage; similarly SUTs play critical roles in cellular uptake during phloem loading and seed filling. Access to the plant SWEETs provided the clues needed for identifying novel bacterial and human sugar transporters. Moreover, we found that pathogens hijack SWEETs to access the plants energy resources. In contrast to SUTs, which are predominantly proton symporters, SWEETs appear to function as facilitators, and thus must be tightly regulated to prevent spillover of carbon into cell walls. One of the key goals of this project was to identify such carriers, characterize the role of key members in carbon allocation. To decrypt the biological circuits we need novel diagnostic tools that, analogous to oscilloscopes in electrical engineering of circuit boards, allow us to quantify fluxes; here by monitoring sugars. Such probes can provide crucial information on systems properties such as metabolic impedance. To measure steady state metabolite levels with minimal invasive methods, and to develop impedance measurements in vivo, we constructed and optimized a set of probes, namely Förster Resonance Energy Transfer (FRET) nanosensors for glucose and sucrose. The probes monitor the concentration of a specific metabolite with high temporal resolution, and, since encoded genetically, can be targeted to specific cells or even specific cellular compartments and thus obtain measurements with cellular and subcellular resolution. We also developed a new platform for monitoring such sensors in vivo, the RootChip.

09 BIOMASS FUELS↗

Model reduction for fractured porous media: a machine learning approach for identifying main flow pathways

Discrete fracture networks (DFN) are often used to model flow and transport in fractured porous media. The accurate resolution of flow and transport behavior on a large DFN involving thousands of fractures is computationally expensive. This makes uncertainty quantification studies of quantities of interest such as travel time through the network computationally intractable, since hundreds to thousands of runs of the DFN model are required to get bounds on the uncertainty of the predictions. Prior works on the subject demonstrated that the complexity of a DFN could be reduced by considering a sub-network of it (often termed a “backbone” sub-network), one whose flow and transport properties were then shown to be similar to that of the full network. The technique is tantamount to partitioning the complete set of fractures of a network into two disjoint sets, one of which is the backbone sub-network while the other its complement. It is in this context that we present a system-reduction technique for DFNs using supervised machine learning via a Random Forest Classifier that selects a backbone sub-network from the full set of fractures. The in-sample errors (in terms of precision and recall scores) of the trained classifier are found to be very accurate indicators of the out-of-sample errors, thus exhibiting that the classifier generalizes well to test data. Moreover, this system-reduction technique yields sub-networks as small as 12% of the full DFN that still recover transport characteristics of the full network such as the peak dosage and tailing behavior for late times. Most importantly, the sub-networks remain connected, and their size can be controlled by a single dimensionless parameter. Furthermore, measures of KL-divergence and KS-statistic for the breakthrough curves of the sub-networks with respect to the full network show physically realistic trends in that the measures decrease monotonically as the size of the sub-networks increase. In conclusion, the computational efficiency gained by this technique depends on the size of the sub-network, but large reductions in computational time can be expected for small sub-networks, yielding as much as 90% computational savings for sub-networks that are as small as 10-12% of the full network.

97 MATHEMATICS AND COMPUTING↗

Distributed-Memory Sparse Deep Neural Network Inference Using Global Arrays

Partitioned Global Address Space (PGAS) models exhibit tremendous promise in developing efficient and productive distributed-memory parallel applications. They have been used extensively in scientific computations due to conveniently offering a ``shared-memory''-like model and convenient interfaces that separate communication with synchronization. Traditionally, PGAS communication models have been applied to dense/contiguously distributed data, but most modern applications depict varied levels of sparsity. Existing PGAS models require certain adaptations to support distributed sparse computations, since associated computations often require matrix arithmetic, in addition to data movement. The Global Arrays toolkit from Pacific Northwest National Laboratory (PNNL) is one of the earliest PGAS models to combine one-sided data communication and distributed matrix operations and is still used in the popular NWChem quantum chemistry suite. Recently, we have expanded the Global Arrays toolkit to support common sparse operations, like sparse matrix-dense matrix multiplies (SpMM), sparse matrix-sparse matrix multiplication (SpGEMM) and Sampled Dense-Dense Matrix Multiplication (SDDMM). As it turns out, these operations are the bedrock of sparse Deep Learning (DL); sparse deep neural networks and Graph Neural Networks (GNNs) have gained increasing attention recently in achieving speedups on training and inference with reduced memory footprints. Unlike scientific applications in High Performance Computing (HPC), modern (distributed-memory capable) DL toolkits often rely on non-standardized and closed-source vendor software optimizations, creating challenges in software-hardware co-design at scale. Our goal is to support a variety of distributed-memory sparse matrix operations and helper functions in the newly created Sparse Global Arrays (SGA), such that it is possible to build portable and productive Machine Learning scenarios for algorithm/software and hardware codesign purposes. Contemporary data-parallel schemes for training/inference are undergoing a major overhaul since model replication limits scalability and causes resource inefficiencies. As such, we have adopted tensor parallelism in decomposing the model and inputs, to mitigate memory issues. Current implementation is built on top of MPI and uses CPUs to maximize the portability across the platforms.

Distributed computing, machine learning↗