National Energy Research Scientific Computing Center Research Articles

Search for two-neutrino double-beta decay of 136Xe to the excited state of 136Ba with the complete EXO-200 dataset* *EXO-200 is supported by DOE and NSF in the United States, NSERC in Canada, SNF in Switzerland, IBS in Korea, DFG in Germany, and CAS in China. EXO-200 data analysis and simulation uses resources of the National Energy Research Scientific Computing Center

A new search for two-neutrino double-beta (2νββ) decay of 136Xe to the excited state of 136Ba is performed with the full EXO-200 dataset. A deep learning-based convolutional neural network is used to discriminate signal from background events. Signal detection efficiency is increased relative to previous searches by EXO-200 by more than a factor of two. With the addition of the Phase II dataset taken with an upgraded detector, the median 90% confidence level half-life sensitivity of 2νββ decay to the state of 136Ba is yr using a total 136Xe exposure of 234.1 kg yr. No statistically significant evidence for 2νββ decay to the state is observed, leading to a lower limit of yr at 90% confidence level, improved by 70% relative to the current world's best constraint.

Planting a Lyman alpha forest on AbacusSummit

ABSTRACT The full-shape correlations of the Lyman alpha (Ly α) forest contain a wealth of cosmological information through the Alcock–Paczyński effect. However, these measurements are challenging to model without robustly testing and verifying the theoretical framework used for analysing them. Here, we leverage the accuracy and volume of the N-body simulation suite AbacusSummit to generate high-resolution Ly α skewers and quasi-stellar object (QSO) catalogues. One of the main goals of our mocks is to aid in the full-shape Ly α analysis planned by the Dark Energy Spectroscopic Instrument (DESI) team. We provide optical depth skewers for six of the fiducial cosmology base-resolution simulations ($L_{\rm box} = 2\, h^{-1}\, {\rm Gpc}$, N = 69123) at z = 2.5. We adopt a simple recipe based on the Fluctuating Gunn–Peterson Approximation (FGPA) for constructing these skewers from the matter density in an N-body simulation and calibrate it against the 1D and 3D Ly α power spectra extracted from the hydrodynamical simulation IllustrisTNG (TNG; $L_{\rm box} = 205\, h^{-1}\, {\rm Mpc}$, N = 25003). As an important application, we study the non-linear broadening of the baryon acoustic oscillation (BAO) peak and show the cross-correlation between DESI-like QSOs and our Ly α forest skewers. We find differences on small scales between the Kaiser approximation prediction and our mock measurements of the Ly α × QSO cross-correlation, which would be important to account for in upcoming analyses. The AbacusSummit Ly α forest mocks open up the possibility for improved modelling of cross-correlations between Ly α and cosmic microwave background (CMB) lensing and Ly α and QSOs, and for forecasts of the 3-point Ly α correlation function. Our catalogues and skewers are publicly available on Globus via the National Energy Research Scientific Computing Center (NERSC) (full link under the section ‘Data Availability’).

Leveraging History to Predict Infrequent Abnormal Transfers in Distributed Workflows.

Scientific computing heavily relies on data shared by the community, especially in distributed data-intensive applications. This research focuses on predicting slow connections that create bottlenecks in distributed workflows. In this study, we analyze network traffic logs collected between January 2021 and August 2022 at the National Energy Research Scientific Computing Center (NERSC). Based on the observed patterns, we define a set of features primarily based on history for identifying low-performing data transfers. Typically, there are far fewer slow connections on well-maintained networks, which creates difficulty in learning to identify these abnormally slow connections from the normal ones. We devise several stratified sampling techniques to address the class-imbalance challenge and study how they affect the machine learning approaches. Our tests show that a relatively simple technique that undersamples the normal cases to balance the number of samples in two classes (normal and slow) is very effective for model training. This model predicts slow connections with an F1 score of 0.926.

Scalable training of graph convolutional neural networks for fast and accurate predictions of HOMO-LUMO gap in molecules

Graph Convolutional Neural Network (GCNN) is a popular class of deep learning (DL) models in material science to predict material properties from the graph representation of molecular structures. Training an accurate and comprehensive GCNN surrogate for molecular design requires large-scale graph datasets and is usually a time-consuming process. Recent advances in GPUs and distributed computing open a path to reduce the computational cost for GCNN training effectively. However, efficient utilization of high performance computing (HPC) resources for training requires simultaneously optimizing large-scale data management and scalable stochastic batched optimization techniques. In this work, we focus on building GCNN models on HPC systems to predict material properties of millions of molecules. We use HydraGNN, our in-house library for large-scale GCNN training, leveraging distributed data parallelism in PyTorch. We use ADIOS, a high-performance data management framework for efficient storage and reading of large molecular graph data. We perform parallel training on two open-source large-scale graph datasets to build a GCNN predictor for an important quantum property known as the HOMO-LUMO gap. We measure the scalability, accuracy, and convergence of our approach on two DOE supercomputers: the Summit supercomputer at the Oak Ridge Leadership Computing Facility (OLCF) and the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC). We present our experimental results with HydraGNN showing (i) reduction of data loading time up to 4.2 times compared with a conventional method and (ii) linear scaling performance for training up to 1024 GPUs on both Summit and Perlmutter.

Hidden physicists

Like me, my friend Vincent Kargatis earned a PhD in high-energy astrophysics and worked at NASA’s Goddard Space Flight Center in Greenbelt, Maryland. Also like me, Kargatis left Goddard in the late 1990s to pursue a career outside of astronomy—in his case, computer programming. Now, 23 years later, he manages a team that ensures the quality of software produced by Oracle Utilities.Since I joined Physics Today in 1997, I’ve been obliged to keep up with developments in physics and to nurture contacts with physicists and their close professional relatives. Still, neither Kargatis nor I think of ourselves as physicists any longer.Debbie Bard leads the data science engagement group at the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory. Before then, she worked at Imperial College London and SLAC. She earned a PhD in experimental particle physics at Edinburgh University in Scotland.Bard’s team helps researchers who need supercomputers for their data analysis and simulations. The job entails performing two translations: from what scientists need into what NERSC’s systems engineers can provide and from what NERSC’s hardware and software can do into what the scientists can implement.“My work is very far from actual physics,” Bard told me by email. “I’m mainly a manager of people and projects these days. But if someone asks me what I do, I still answer ‘physicist.’ It’s pretty deeply ingrained in me.”I learned about Bard from Spotlight on Hidden Physicists, an ongoing series of articles in Radiations, the magazine of Sigma Pi Sigma, the physics honor society. Each article is about a person with a physics degree who, like Bard, Kargatis, and me, went on to do other things. The range of other things is wide. Among the 37 hidden physicists are a jazz singer, a Disney animator, a Wall Street analyst, and a theologian.Kargatis qualifies as a hidden physicist. Indeed, people who work in computer software of one kind or another form the largest group in the Radiations series. Computer software is also the private-sector profession that employs the most PhD physicists, according to surveys conducted by the American Institute of Physics’s Statistical Research Center. (AIP publishes Physics Today and hosts Sigma Pi Sigma.)Sigma Pi Sigma’s Spotlight on Hidden Physicists succeeds in showing the variety of rewarding careers that people with a bachelor’s, master’s, or PhD in physics choose for themselves. High school seniors contemplating what subject to major in can be justifiably confident that a physics degree opens doors to careers beyond physics itself.But when the professional physics community contemplates hidden physicists, it should be careful. Although Bard continues to think of herself as a physicist, Kargatis does not. When he and I get together, we talk mostly about music, food, and wine and occasionally about physics. He is not a prime candidate for becoming a Physics Today subscriber or for rejoining the American Astronomical Society.Of the physicists in the private sector, how many still consider themselves to be physicists? I don’t know, but my hunch is that the answer has to do with whether they continue to do physics. This past May Google announced its intention to spend billions of dollars to build a quantum computer that can perform large-scale, error-free calculations by 2029. Out of curiosity, I looked up Google’s jobs site. To accomplish that ambitious goal, the company sought to recruit just six additional physicists.Two other companies in physics-flavored industries, IBM and Tesla, had no postings for physicists. That’s not to say either company had no jobs for which physicists were qualified. Tesla is currently looking for someone to do computer-aided engineering and computational fluid dynamics. Some physicists meet the position’s requirements, but so, too, do engineers and applied mathematicians.If “hidden physicist” is intended to mean someone outside academia who identifies as a physicist and who does physics, some adjustment might be needed. There could be fewer of them than expected.© 2021 American Institute of Physics.

Performance Portability for Advanced Architectures

The papers in this special section focus on performance portability for advanced architectures. This is a very timely topic as we continue our journey to exascale computers. The leader on the Top 500 list today is the Fujitsu Fugaku computer in Japan, using a modified ARM architecture and 512-bit scalable vector extension (SVE) instructions. In the USA, the National Energy Research Scientific Computing Center (NERSC) has just dedicated Perlmutter, named for the Berkeley Nobel Laureate Saul Perlmutter. Perlmutter, built by Cray/HPE, contains both CPU-only and CPU/GPU nodes. The CPUs are produced by AMD and the GPUs come from NVIDIA. Frontier at Oak Ridge National Laboratory and El Capitan at Lawrence Livermore National Laboratory will contain new AMD GPUs, and Aurora at Argonne National Laboratory will feature new GPUs from Intel. Los Alamos National Laboratory’s next-generation machine Crossroads will use Intel CPUs with high-bandwidth memory and AVX-512 vector instructions. Optimizing code for three different GPUs is a daunting challenge for application programmers. Further, there are still many computers that do not feature GPUs, and most computational scientists would not be eager to totally abandon such machines.

Hydrodynamic description of D meson production in high-energy heavy-ion collisions * *Supported by the National Key Research and Development Program of China (2020YFE0202002); the National Natural Science Foundation of China (11935007, 11221504, 11861131009, 12075098, 11890714); the Director, Office of Energy Research, Office of High Energy and Nuclear Physics, Division of Nuclear Physics, of the U.S. Department of Energy (DOE) (DE-

The large values and constituent-quark-number scaling of the elliptic flow of low- D mesons imply that charm quarks, initially produced through hard processes, might be partially thermalized through strong interactions with quark-gluon plasma (QGP) in high-energy heavy-ion collisions. To quantify the degree of thermalization of low- charm quarks, we compare the meson spectra and elliptic flow from a hydrodynamic model to experimental data as well as transport model simulations. We use an effective charm chemical potential at the freeze-out temperature to account for the initial charm quark production from hard processes and assume that they are thermalized in the local comoving frame of the medium before freeze-out. mesons are sampled statistically from the freeze-out hyper-surface of the expanding QGP as described by the event-by-event (3+1)D viscous hydrodynamic model CLVisc. Both the hydrodynamic and transport models can describe the elliptic flow of mesons at GeV/c as measured in Au+Au collisions at GeV. Though the experimental data on spectra are consistent with the hydrodynamic result at small GeV/c, they deviate from the hydrodynamic model at high transverse momentum, GeV/c. The diffusion and parton energy loss mechanisms in the transport model can describe the measured spectra reasonably well within the theoretical uncertainty. Our comparative study indicates that charm quarks only approach local thermal equilibrium at small , even though they acquire sizable elliptic flow comparable to light-quark hadrons at both small and intermediate .

STAR Data Production Workflow on HPC: Lessons Learned & Best Practices

The Solenoidal Tracker at RHIC (STAR) is a multi-national supported experiment located at Brookhaven National Lab. The raw physics data captured from the detector is on the order of tens of PBytes per data acquisition campaign, which makes STAR fit well within the definition of a big data science experiment. The production of the data has typically run on standard nodes or on standard Grid computing environments. All embedding simulations (complex workflow mixing real and simulated events) have been run on standard Linux resources at the National Energy Research Scientific Computing Center (NERSC) aka PDSF. However, HPC resources such as Cori have become available for STAR’s data production as well as embedding, and STAR has been the very first experiment to show feasibility of running a sustainable data production campaign on this computing resource. The use of Docker containers with Shifter is required to run on HPC @ NERSC – this approach encapsulates the environment in which a standard STAR workflow runs. From the deployment of a tailored Scientific Linux environment (requiring many of its own libraries and special configurations required to run) to the deployment of third-party software and the STAR specific software stack, it has become impractical to rely on a set of containers containing each specific software release. To this extent, solutions based on the CERN VM File System (CVMFS) for the deployment of software and services have been employed in HENP, but one needs to make careful scalability considerations when using a resource like Cori, such as not allowing all software to be deployed in containers or bare node. Additionally, CVMFS clients are not compatible on Cori nodes and one needs to rely on an indirect NFS/DVS mount scheme. In our contribution, we will discuss our strategies from the past and our current solution based on CVMFS. Furthermore, running on HPC is not a simple task as each aspect of the workflow must be enabled to scale, run efficiently, and the workflow needs to fit within the boundaries of the provided queue system (SLURM in this case). Lastly, we will also discuss what we have learned so far about what is the best method for grouping jobs to maximize a single 48 core HPC node within a specific time frame and maximize our workflow efficiency.

Interactions of Ionophoric Polyphenols with Human Serum Albumin

The distribution of a drug is determined, at least in part, by its interactions with serum proteins. Human serum albumin (HSA), the most abundant blood plasma protein, serves as the most common mode of transport for drugs to their target. In our previous work, we have synthesized and studied important characteristics of a series of ionophoric polyphenols as potential anti‐Alzheimer’s disease agents. As part of our ongoing investigations on the biological activity of these compounds, we are now exploring possible transport mechanisms. In our current work, the interactions of the ionophoric polyphenols with HSA were studied via in vitro and in silico methods. In line with in vitro studies, the interaction of the ionophoric polyphenols with HSA were studied by fluorescence and circular dichroism (CD) spectroscopies to obtain binding affinities, bimolecular quenching constants, theoretical number of binding sites and HSA secondary structural changes. Moreover, in line with in silico studies, a preferred mode of binding was predicted by molecular docking and the predicted systems were subject to a 5 ns MD simulation; consequently, RMSD studies were performed to attest for stability of the systems, MM/GBSA binding free energies were calculated to correlate with experimental binding affinities and per‐residue energy decomposition analyses were carried out to analyze residues critical to binding. Using resveratrol, a polyphenolic compound which has demonstrated strong neuroprotective and anti‐Alzheimer disease properties, as a control, in vitro results showed the formation of adducts (static quenching), one theoretical binding site, binding constants (Ka) ranging 104 to 106 M−1 and only slight secondary structure modifications between HSA and the ionophoric polyphenols. Further, in silico results revealed the ionophoric polyphenols and resveratrol favorably binding subdomain IIA of HSA, forming critical interactions to Trp 214 and Asp451. RMSD studies showed the HSA‐IIA system plateauing after a short time indicating stable binding events; in addition, a Pearson correlation coefficient of R=0.80 between the experimental binding affinity and computational binding free energies showed a strong correlation between in vitro and in silico results. Overall, our results seem to indicate that there is a ionophoric polyphenol‐serum albumin interaction, which could have a potential impact on the use of ionophoric polyphenols as therapeutics for Alzheimer’s disease.Support or Funding InformationThis research used resources of the National Energy Research Scientific Computing Center (NERSC), a U.S. Department of Energy Office of Science User Facility operated under Contract No. DE‐AC02‐05CH11231.

Computational Studies of Glycosylated Photosensitizers with Plasma Proteins

Photodynamic therapy (PDT) is a promising treatment against cancer, it utilizes photosensitizers that when excited by light undergo photochemical reactions that can kill abnormal cancer cells. However, in order for photosensitizers to reach their target, they must effectively bind serum albumin, the most abundant plasma protein circulated in an organism’s blood stream. In our previous study, we synthesized three glycosylated photosensitizers (PGlu, CGlu and IGlu) and performed in vitro spectroscopy studies that revealed each glycosylated photosensitizer effectively binding bovine serum albumin and human serum albumin (BSA and HSA). The aim of this work was to implement an in silico approach to study how PGlu, CGlu, and IGlu are binding onto BSA and HSA. We performed molecular docking simulations to predict a favorable mode of binding between each of our glycosylated photosensitizers to BSA and HSA. Further, we performed a 5 ns molecular dynamics (MD) simulation for each of our predicted systems; consequently, we calculated the RMSD of the systems to attest for stability, performed binding free energy calculations to correlate with experimental binding constant results and analyzed residue interactions critical to each binding event. We observed PGlu, CGlu and IGlu favorably binding the hemin site located in subdomain IB of BSA, forming strong interactions with the Trp134 residue, as well as favorably binding Sudlow Site 1 located in subdomain IIA of HSA, forming strong interactions to Trp214. RMSD studies revealed the BSA‐IB and HSA‐IIA systems plateauing after a short time indicating stable binding events; in addition, a Pearson correlation coefficient of R=0.87 between the experimental binding constants and computational binding free energies showed a strong correlation between in vitro and in silico results. Overall, our computational results corroborated our experimental results, revealing that our glycosylated photosensitizers can bind serum albumin effectively through critical tryptophan residue interactions and can potentially aid in the progression of drug development for PDT.Support or Funding InformationThis research used resources of the National Energy Research Scientific Computing Center (NERSC), a U.S. Department of Energy Office of Science User Facility operated under Contract No. DE‐AC02‐05CH11231.

Managing a Cray supercomputer as a git branch

SummaryModern software engineering/DevOps techniques applied to Cray XC Series systems management enable higher fidelity translation from test to production environments, reduce administration costs by avoiding duplicate efforts, and increase reliability. This can be done by using a git repository to track and manage all system configurations on the SMW(s) (System Management Workstation) and then adapt a gitflow‐like development methodology. Using git improves change management with peer review of changes, identification of who made a change, knowing why a change was made, facilitating easier reversion of bad changes, and enables a workflow using the ideas from gitflow with individual branches for preparing, maintaining, and recording releases. Using branches enables multiple contributors to work on different types of changes without getting unexpected changes from another feature area. We will describe a process similar to that used at NERSC (National Energy Research Scientific Computing Center) to extract/abstract configuration data, manage it using git, and then re‐apply changes to the SMW(s).

Multi-threaded ATLAS simulation on Intel Knights Landing processors

The Knights Landing (KNL) release of the Intel Many Integrated Core (MIC) Xeon Phi line of processors is a potential game changer for HEP computing. With 72 cores and deep vector registers, the KNL cards promise significant performance benefits for highly-parallel, compute-heavy applications. Cori, the newest supercomputer at the National Energy Research Scientific Computing Center (NERSC), was delivered to its users in two phases with the first phase online at the end of 2015 and the second phase now online at the end of 2016. Cori Phase 2 is based on the KNL architecture and contains over 9000 compute nodes with 96GB DDR4 memory. ATLAS simulation with the multithreaded Athena Framework (AthenaMT) is a good potential use-case for the KNL architecture and supercomputers like Cori. ATLAS simulation jobs have a high ratio of CPU computation to disk I/O and have been shown to scale well in multi-threading and across many nodes. In this paper we will give an overview of the ATLAS simulation application with details on its multi-threaded design. Then, we will present a performance analysis of the application on KNL devices and compare it to a traditional x86 platform to demonstrate the capabilities of the architecture and evaluate the benefits of utilizing KNL platforms like Cori for ATLAS production.

Towards understanding HPC users and systems: A NERSC case study

High performance computing (HPC) scheduling landscape currently faces new challenges due to the changes in the workload. Previously, HPC centers were dominated by tightly coupled MPI jobs. HPC workloads increasingly include high-throughput, data-intensive, and stream-processing applications. As a consequence, workloads are becoming more diverse at both application and job levels, posing new challenges to classical HPC schedulers. There is a need to understand the current HPC workloads and their evolution to facilitate informed future scheduling research and enable efficient scheduling in future HPC systems.In this paper, we present a methodology to characterize workloads and assess their heterogeneity, at a particular time period and its evolution over time. We apply this methodology to the workloads of three systems (Hopper, Edison, and Carver) at the National Energy Research Scientific Computing Center (NERSC). We present the resulting characterization of jobs, queues, heterogeneity, and performance that includes detailed information of a year of workload (2014) and evolution through the systems’ lifetime (2010–2014).

(Invited) Descriptors, Defects, and DFT for Solid Oxide Fuel Cell Materials

Active oxygen materials that can rapidly transport and exchange oxygen with the environment play a critical role in many existing and developing technologies, e.g., solid oxide fuel cells (SOFCs), gas separation membranes, oxygen sensors, chemical looping devices, and memristors. In particular, reducing the working temperature of SOFCs, which is critical to their increased commercialization, is significantly inhibited by the slow oxygen reduction reaction at the cathode. The steps in the ORR and related oxygen transport are complex enough that it still remains an open challenge how to design improved cathode catalytic activities using either intuition or computational materials design tools. In this presentation we summarize our recent results on using the Oxygen 2p-band (or just “p-band”) as a descriptor for many important properties related to SOFC cathode performance for both perovskite and 214 Ruddlesden-Popper oxides. We demonstrate that the p-band, which is an easy to calculate bulk electronic structure descriptor, correlates with properties that include oxygen vacancy and interstitial defect energies [1, 2], oxygen diffusion coefficients [3, 4], surface exchange coefficients [1, 3], and cathode area specific resistance [1] (e.g., see Figure 1). This descriptor is therefore a powerful tool for active oxide materials design, and we describe its use to search for promising new SOFC cathode materials. The p-band descriptor highlights some of the intriguing differences in trends of defect chemistry and surface exchange behavior between the perovskite and 214 Ruddlesden-Popper phases. Specifically, the dominance of vacancy and interstitial defects, respectfully, in the oxygen exchange and transport in these two classes of materials leads to opposite trends in their behavior with p-band and generally in the behavior with oxygen bond strength and state of oxidation. We show the results of detailed calculations of the defect behavior in 214 Ruddlesden-Popper La2-xSrxMO4+d (M = Co, Ni, Cu) compounds and demonstrate how the crossover in dominant defects with doping can be understood in terms of simple trends in oxidation chemistry. We further demonstrate that the previously identified but somewhat elusive peroxide defects are stable in Ruddlesden-Popper phases over some doping domains and can play a role in the oxygen transport in these materials. Acknowledgement We gratefully acknowledge financial support from the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering under award number DE-SC0001284, Department of Energy (DOE), National Energy Technology Laboratory (NETL), Solid State Energy Conversion Alliance (SECA) Core Technology Program (Funding Opportunity Number DEFE0009435), and the National Science Foundation Software Infrastructure for Sustained Innovation (SI2) (award no. 1148011). Computing in this work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by U.S. National Science Foundation with grant number ACI-1053575, and the National Energy Research Scientific Computing Center (NERSC) allocation of the CNMS at ORNL (CNMS2013-292).

Heat Sink Sunk

This article explores various innovative ways that technology companies are working upon to deal with the large amount of heat of their big data centers. Microsoft’s research and development team is working on designing and building of an underwater data center. Researchers believe that underwater data centers might have power, construction, and performance advantages as well. Berkeley’s National Energy Research Scientific Computing Center (NERSC) recycles its chilled water. NERSC uses air to chill the data center and cooler-running components, such as disk drives, routers, and network servers. With the growing efficiency of data centers, better cooling might not be enough to make underwater operations worthwhile. But cooling is only one of the advantages that Microsoft sees in submerging its data centers. Submersibles also simplify deployment. Microsoft used an underwater cable connected to the electrical grid to power Leona Philpot, a submersible data center. In the future, it may reduce costs through renewable energy, combined with on-site energy storage and backup power from the grid.

Toward Exascale Earthquake Ground Motion Simulations for Near-Fault Engineering Analysis

Modernizing SW4 for massively parallel time-domain simulations of earthquake ground motions in 3D earth models increases resolution and provides ground motion estimates for critical infrastructure risk evaluations. Simulations of ground motions from large (M ≥ 7.0) earthquakes require domains on the order of 100 to500 km and spatial granularity on the order of 1 to5 m resulting in hundreds of billions of grid points. Surface-focused structured mesh refinement (SMR) allows for more constant grid point per wavelength scaling in typical Earth models, where wavespeeds increase with depth. In fact, MR allows for simulations to double the frequency content relative to a fixed grid calculation on a given resource. The authors report improvements to the SW4 algorithm developed while porting the code to the Cori Phase 2 (Intel Xeon Phi) systems at the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory. Investigations of the performance of the innermost loop of the calculations found that reorganizing the order of operations can improve performance for massive problems.

Multidisciplinary Science at the Molecular Foundry

Multidisciplinary Science at the Molecular Foundry

Surface Reaction, Solvent Inhomogeneity, and Ion Transport in Electric Double Layers: Predictions from a Classical Density Functional Theory

Electric double layers (EDLs) are formed at the interface of a charged fluid in contact with an electrified surface or an electrode leading to drastic changes in the fluid structure and local thermodynamic properties. Whereas conventional theories of EDLs are often concerned with the local inhomogeneity in terms of ionic distributions near a planar surface, the simplistic model is insufficient to capture many important properties relevant to emerging electrochemical processes for energy storage and conversion. In this talk, we present a few recent applications of the classical density functional theory (CDFT) that account for not only ionic excluded volume effects and electrostatic correlations ignored in conventional theories of EDLs but also electrode geometry and pore size distributions, the local solvent and dielectric inhomogeneity, surface reactions, and hydrodynamic effects. The CDFT calculations will be illustrated with practical examples for unusual ion transport in nanopores, optimization of capacitive energy extraction processes, design of nanostructured electrodes for flow supercapacitors and pseudo-capacitors. The theoretical predictions are validated with experiment results and provide insights for novel electrochemical behavior that are not captured in conventional EDL theories. Acknowledgements: This research is sponsored by the Fluid Interface Reactions, Structures, and Transport (FIRST) Center and the U.S. Department of Energy (DE-FG02-06ER46296). The numerical calculations were performed at the National Energy Research Scientific Computing Center (NERSC).

NERSC's Impact on Advances of Global Gyrokinetic PIC Codes for Fusion Energy Research

The National Energy Research Scientific Computing Center (NERSC) was originally launched as a computing center for the exclusive support of magnetic confinement fusion research in the US. There is, thus, a long history of computational advances and successes that ties fusion scientists with NERSC and its staff. One example of the numerous computational achievements enabled by NERSC in this field is the development of the global gyrokinetic particle-in-cell approach for the simulation of turbulent transport in tokamak fusion devices. This article describes the historical evolution of this method and the codes implementing it, and the impact that NERSC had in their development and the scientific discoveries that they allow.

Big Bang, Big Data, Big Iron: Fifteen Years of Cosmic Microwave Background Data Analysis at NERSC

The cosmic microwave background (CMB) consists of photons created in the Big Bang and cooled by the subsequent expansion of the universe to an almost perfectly uniform 3-Kelvin sky signal today. These photons experience the entire history of the universe, and each epoch has left its imprint as tiny fluctuations in their temperature and polarization. Decoding these provides unique insight into the history of the universe, constraining the fundamental parameters of both cosmology and high-energy physics. The faintness of these fluctuations requires us "researchers" to gather huge datasets and analyze them on supercomputers. For 15 years, the National Energy Research Scientific Computing Center (NERSC) has provided the resources for the CMB community's most computationally challenging analyses. In this article, the authors describe how their analysis algorithms and implementations have evolved over this time, driven by both the growth in CMB data volumes and the changes in high-performance computing architectures.