Engineering topics
Laanait, Nouamane
Publications and source records attributed to Laanait, Nouamane.
Evaluating Robust Entanglement on a Trapped Ion Platform
We present proof of principle results for evaluating an entanglement volumetric benchmark [1] on trapped ion platforms. The benchmark quantifies the robustness of multipartite and bipartite entanglement using stabilizer measurements and witness functions. Each n-qubit graph state is prepared and used to evaluate n state-specific stabilizer strings. These stabilizer measurements are used to evaluate entanglement witness functions. The entanglement benchmark defines families of graph states associated with an initial sub-graph of the hardware qubit connectivity - with all-to-all connectivity, trapped ion systems provide a flexibility in the choice of this initial graph, and this will affect the associated family. In this work, we present results targeting several classes of entangled states: 1) 1-D cluster states, 2) n-qubit GHZ states, and 3) cycle graph states. These states have been found in the literature as standard hardware benchmarks, and have connections to many near-term applications.
Accelerating Collective Communication in Data Parallel Training across Deep Learning Frameworks
This work develops new techniques within Horovod, a generic communication library supporting data parallel training across deep learning frameworks. In particular, we improve the Horovod control plane by implementing a new coordination scheme that takes advantage of the characteristics of the typical data parallel training paradigm, namely the repeated execution of collectives on the gradients of a fixed set of tensors. Using a caching strategy, we execute Horovod’s existing coordinator-worker logic only once during a typical training run, replacing it with a more efficient decentralized orchestration strategy using the cached data and a global intersection of a bitvector for the remaining training duration. Next, we introduce a feature for end users to explicitly group collective operations, enabling finer grained control over the communication buffer sizes. To evaluate our proposed strategies, we conduct experiments on a world-class supercomputer — Summit. We compare our proposals to Horovod’s original design and observe 2x performance improvement at a scale of 6000 GPUs; we also compare them against tf.distribute and torch.DDP and achieve 12% better and comparable performance, respectively, using up to 1536 GPUs; we compare our solution against BytePS in typical HPC settings and achieve about 20% better performance on a scale of 768 GPUs. Finally, we test our strategies on a scientific application (STEMDL) using up to 27,600 GPUs (the entire Summit) and show that we achieve a near-linear scaling of 0.93 with a sustained performance of 1.54 exaflops (with standard error +- 0.02) in FP16 precision.
Scalable balanced training of conditional generative adversarial neural networks on image data
Here, we propose a distributed approach to train deep convolutional generative adversarial neural network (DC-CGANs) models. Our method reduces the imbalance between generator and discriminator by partitioning the training data according to data labels, and enhances scalability by performing a parallel training where multiple generators are concurrently trained, each one of them focusing on a single data label. Performance is assessed in terms of inception score, Fréchet inception distance, and image quality on MNIST, CIFAR10, CIFAR100, and ImageNet1k datasets, showing a significant improvement in comparison to state-of-the-art techniques to training DC-CGANs. Weak scaling is attained on all the four datasets using up to 1000 processes and 2000 NVIDIA V100 GPUs on the OLCF supercomputer Summit.
Smoky Mountain Data Challenge 2020: An Open Call to Solve Data Problems in the Areas of Neutron Science, Material Science, Urban Modeling and Dynamics, Geophysics, and Biomedical Informatics
The 2020 Smoky Mountains Computational Sciences and Engineering Conference enlists research scientists from across Oak Ridge National Laboratory (ORNL) to be data sponsors and help create data analytics challenges for eminent data sets at the laboratory. This work describes the significance of each of the seven data sets and their as- sociated challenge questions. The challenge questions for each data set were required to cover multiple difficulty levels. An international call for participation was sent to students, and researchers asking them to form teams of up to four people to apply novel data analytics techniques to these data sets.
Deep Learning and Structural Imaging of Materials
Deep learning has had a transformative effect on numerous domains and is actively utilized by many scientists in data-intensive fields such as high-energy physics and cosmology. Materials science, and in particular, the structural imaging of materials with electrons and X-rays are projected to enter the age of scientific data torrents, positioning them as new application spaces for modern artificial intelligence. In this contribution, we provide a synopsis on the foundations and latest progress in deep learning and present an in-depth application of utilizing modern deep artificial neural networks in scanning transmission electron microscopy to extract structural material properties. We use this case study to expose the strengths of deep learning-based models and discuss their current limitations, in the process highlighting their potential use in other data-intensive structural imaging modalities.
Strategies to Deploy and Scale Deep Learning on the Summit Supercomputer
The rapid growth and wide applicability of Deep Learning (DL) frameworks poses challenges to computing centers which need to deploy and support the software, and also to domain scientists who have to keep up with the system environment and scale up scientific exploration through DL. We offer recommendations for deploying and scaling DL frameworks on the Summit supercomputer, currently atop the Top500 list, at the Oak Ridge National Laboratory Leadership Computing Facility (OLCF). We discuss DL software deployment in the form of containers, and compare performance of native-built frameworks and containerized deployment. Software containers show no noticeable negative performance impact and exhibit faster Python loading times and promise easier maintenance. To explore strategies for scaling up DL model training campaigns, we assess DL compute kernel performance, discuss and recommend I/O data formats and staging, and identify communication needs for scalable message exchange for DL runs at scale. We recommend that users take a step-wise tuning approach beginning with algorithmic kernel choice, node I/O configuration, and communications tuning as best-practice. We present baseline examples of scaling efficiency 87% for a DL run of ResNet50 running on 1024 nodes (6144 V100 GPUs).
STEMDL
The code "STEMDL" provides highly-optimized multi-gpu implementations of DL models of applicability in electron microscopy via the use of mixed-precision arithmetic, native graph construction and evaluation in Tensorflow, and adaptive scaling of learning parameters to avoid numerical instabilities due to reduced precision. STEMDL also uses low-level methods from the Horovod library for gradient communication and reduction and adds an additional optimization by allowing the user to scheduling the gradient reduction operations based on their position within the layers of a neural net. STEMDL provides optimized I/O routines for data staging and data reading/writing in both Tensorflow and Pytorch using multi-processing and asynchronous transfer between host and device. .
NAMSA
In NAMSA, we introduce two main solutions to address the limitations of MSA-type algorithms for simulation of electron scattering/microscopy data: just-in-time compilation and work partitioning via MPI and GPU-to-GPU communication. First, the just-in-time (JIT) compilation strategy produces optimally tuned CUDA C++ kernels due to the data layout being perfectly known at runtime and compiled via calls from the Python interpreter. NAMSA automatically optimizes the 2-D FFT planning, and CUDA multi-streams scheduling/launching of probe simulations. With NAMSA, we can simulate electron scattering/microscopy on hundreds to thousands of GPUs, while removing the major limitations of memory and bandwidth from MSA.
Competing phases in epitaxial vanadium dioxide at nanoscale
Phase competition in correlated oxides offers tantalizing opportunities as many intriguing physical phenomena occur near the phase transitions. Owing to a sharp metal-insulator transition (MIT) near room temperature, the correlated vanadium dioxide (VO 2 ) exhibits a strong competition between insulating and metallic phases, which is important for practical applications. However, the phase boundary undergoes a strong modification when strain is involved, yielding complex phase transitions. Here, we report the emergence of nanoscale M2 phase domains in VO 2 epitaxial films under anisotropic strain relaxation. The competing phases of the films are imaged by multilength-scale probes, detecting the structural and electrical properties in individual local domains. Competing evolution of the M1 and M2 phases indicates the critical role of lattice-strain on both the stability of the M2 Mott phase and the energetics of the MIT in VO 2 films. This study demonstrates how strain engineering can be utilized to design phase states, which allow deliberate control of MIT behavior at the nanoscale in epitaxial VO 2 films.
USID and Pycroscopy – Open Source Frameworks for Storing and Analyzing Imaging and Spectroscopy Data
Not provided.
Towards Native Execution of Deep Learning on a Leadership-Class HPC System
Large parallel machines generally offer the best parallel performance with "native execution" that is achieved using codes developed with the optimized compilers, communication libraries, and runtimes offered on the machines. In this paper, we report and analyze performance results from native execution of deep learning on a leadership-class high-performance computing (HPC) system. Using our new code called DeepEx, we present a study of the parallel speed up and convergence rates of learning achieved with native parallel execution. In the trade-off between computational parallelism and synchronized convergence, we first focus on maximizing parallelism while still obtaining convergence. Scaling results are reported from execution on up to 15,000 GPUs using two scientific data sets from atom microscopy and protein folding applications, and also using the popular ImageNet data set. In terms of the traditional measure of parallel speed up, excellent scaling is observed up to 12,000 GPUs. Additionally, accounting for convergence rates of deep learning accuracy or error, a deep learning-specific metric called "learning speed up" is also tracked. The performance results indicate the need to evaluate parallel deep learning execution in terms of learning speed up, and point to additional directions for improved exploitation of high-end HPC systems.
Giant thermally-enhanced electrostriction and polar surface phase in L a 2 M o 2 O 9 oxygen ion conductors
Ferroelectrics possess spontaneous electric polarization at macroscopic scales which nonetheless imposes strict limitations on the material classes. Recent discoveries of untraditional symmetry-breaking phenomena in reduced material dimensions have indicated feasibilities to extend polar properties to broader types of materials, potentially opening up the freedom for designing materials with hybrid functionalities. Here in this paper, we report the unusual electromechanical properties of La 2 Mo 2 O 9 (LAMOX) oxygen ion conductors, systematically investigated at both bulk and surface length levels. We first observed giant electrostriction effects in La 2 Mo 2 O 9 bulk ceramics that are thermally enhanced in concert with their low-energy oxygen-vacancy hopping dynamics. Moreover, while no clear bulk polarization was detected, the surface phases of LAMOX were found to be manifestly polar, likely originating from the coupling between the intrinsic structural flexibilities with strain gradients (i.e., flexoelectricity) and/or chemical heterogeneities present in the materials. These findings identify La 2 Mo 2 O 9 as a promising electromechanical material system and suggest that the flexible structural and chemical configurations in ionically active materials could enable fundamentally different venues to accommodate electric polarization.
Field enhancement of electronic conductance at ferroelectric domain walls
Ferroelectric domain walls have continued to attract widespread attention due to both the novelty of the phenomena observed and the ability to reliably pattern them in nanoscale dimensions. But, the conductivity mechanisms remain in debate, particularly around nominally uncharged walls. Here, we posit a conduction mechanism relying on field-modification effect from polarization re-orientation and the structure of the reverse-domain nucleus. Through conductive atomic force microscopy measurements on an ultra-thin (001) BiFeO 3 thin film, in combination with phase-field simulations, we show that the field-induced twisted domain nucleus formed at domain walls results in local-field enhancement around the region of the atomic force microscope tip. In conjunction with slight barrier lowering, these two effects are sufficient to explain the observed emission current distribution. Our results suggest that different electronic properties at domain walls are not necessary to observe localized enhancement in domain wall currents.
A Programming Framework for Neuromorphic Systems with Emerging Technologies
Neuromorphic computing is a promising post-Moore's law era technology. A wide variety of neuromorphic computer (NC) architectures have emerged in recent years, ranging from traditional fully digital CMOS to nanoscale implementations with novel, beyond CMOS components. There are already major questions associated with how we are going to program and use NCs simply because of how radically different their architecture is as compared with the von Neumann architecture. When coupled with the implementations using emerging device technologies, which add additional issues associated with programming devices, it is clear that we must define a new way to program and develop for NC devices. In this work, we discuss a programming framework for NC devices implemented with emerging technologies. We discuss how we have applied this framework to program a NC system implemented with metal oxide memristors. We utilize the framework to develop two applications for the memristive NC device: a simple multiplexer and a simple control task (the cart-pole problem). Finally, we discuss how this framework can be extended to NC systems implemented with a variety of novel device components and materials.
Pycroscopy - An Open Source Approach to Microscopy and Microanalysis in the Age of Big Data and Open Science
Explore the source record for details and available documents.
Heteroepitaxial growth of cadmium carbonate at dolomite and calcite surfaces: Mechanisms and rates
Here, the systematic variation of rates and the mechanism of cadmium uptake on the (104) surface of dolomite (CaMg(CO 3 ) 2 ) were investigated using in situ and ex situ atomic force microscopy (AFM), ex situ specular X-ray reflectivity (XR), and ex situ X-ray fluorescence (XRF). Selected experiments were performed on the calcite (CaCO 3 ) (104) surface for comparison. Aqueous solutions of CdCl 2 , CaCl 2 , and NaHCO 3 , undersaturated with respect to calcite and supersaturated with respect to otavite (CdCO 3 ) and the (Cd x Ca 1-x )CO 3 solid solution, were reacted with dolomite surfaces for minutes to days. Calcite substrates were reacted with solutions containing 1-50 μM CdCl 2 , and with no added Ca or CO 3 . Thin carbonate films following the Stranski-Krastanov growth mode were observed on both substrates. Specular XR and XRF revealed the formation of nm-thick Cd-rich carbonate films that were structurally ordered with respect to the dolomite (104) plane. Epitaxial films adopted the calcite crystal structure with a d 104 - spacing (3.00 Å) larger than those of pure dolomite (2.88 Å) and otavite (2.95 Å) indicating either a solid solution with x approximate to 0.5, or a strained Cd-rich carbonate with a composition near that of otavite. The growth rate r of this phase increases with the initial supersaturation of the solution with respect to the solid solution, beta max, and follows the empirical relationship, as determined from XRF measurements, given by: r = 10 -4.88 ± 0.42 (β 2.29 ± 0.24 max - 1), (in units of atoms of Cd/Å 2 /h).The morphology of the overgrowth also varied with β max , as exemplified by AFM observations. Growth at step edges occurred over the entire β max range considered, and additional growth features including 3 Å high monolayer islands and ~ 25 Å high tall islands were observed when log β max > 1. On calcite, in situ XR indicated that this phase is similar to the Cd-rich overgrowth formed on dolomite and images obtained from X-ray reflection interface microscopy (XRIM) reveal the existence of laterally variable Cd-rich domains.
Big, Deep, and Smart Data in Scanning Probe Microscopy
Scanning probe microscopy techniques open the door to nanoscience and nanotechnology by enabling imaging and manipulation of structure and functionality of matter on nanometer and atomic scales. We analyze the discovery process by SPM in terms of information flow from tip-surface junction to the knowledge adoption by scientific community. Furthermore, we discuss the challenges and opportunities offered by merging of SPM and advanced data mining, visual analytics, and knowledge discovery technologies.