The escalating integration of inverter-based resources (IBRs) poses new challenges to power systems by introducing fast dynamics with higher frequencies, which may need to be simulated by an electromagnetic transient (EMT) program. As an alternative to conducting EMT simulations for the entire system, which is typically time consuming, hybrid simulation between EMT and phasor-domain transient stability (TS) can greatly reduce the computational burden while preserving the detailed fast dynamics in the EMT zone. This paper establishes an EMT-TS hybrid simulation platform using open-source tools, specifically ParaEMT, GridPACK, and HELICS, which are the EMT simulator, TS simulator, and interface framework, respectively. Case studies on the 240-bus Western Electricity Coordinating Council (WECC) system demonstrate that the developed ParaEMT-HELICS-GridPACK hybrid simulator can accurately capture both slow electromechanical and fast IBR-driven dynamics with a 2.4× speedup.

There is strong demand on building high performance storage service upon emerging fast storage devices (e.g., NVMe SSDs). Unfortunately, current storage software stack cannot satisfy such requirements and the software overhead becomes a major bottleneck for developing high performance storage applications. According to our performance profiling results, most storage software overhead is caused by kernel I/O stacks due to context switch, data copy, interrupt, resource synchronization and etc. To address these issues, we provide SPDK (storage performance development kit), a set of tools and libraries for writing high performance, scalable, user-mode storage applications. It achieves high performance by moving the necessary drivers into user space and operating them in a polled mode instead of interrupt mode, which eliminates kernel context switch and interrupt handling overhead and also provides lockless resource access. Integrated with SPDK, performance competitive storage applications can be build upon those fast storage devices. In our experiments, the per cpu core IOPS of NVMe device driver in SPDK is about 6X to 10X better than the kernel NVMe driver. Some storage applications are also developed upon SPDK's framework and obtain much better per-core performance compared with the existing solutions. For example, SPDK NVMe over fabrics (NVMe-oF) target is 10X more efficient than Linux Kernel NVMe-oF target solution.

Often in a networked AC system, substation witnessing loss of generation fails to supply the loads, which are sensitive and critical. Creation of dedicated infrastructure for support is quite costly. The paper proposes to augment a Multi-Terminal Medium Voltage DC Sub-Transmission System (MT-MVDCS) between the AC grid and AC distribution system, to enhance unlimited expansion, selection of power input, availability, stability and reliability at low cost. The control algorithm of the proposed system is discussed for control power flow, decoupled power control to input only real power from the mains, supply desired reactive power and cater to the critical load connected to the grid side witnessing the loss of generation. The indirect current control is utilized to control the Voltage Source Converters (VSC) under grid tied conditions, whereas, the VSCs are PWM controlled when supplying power to passive loads. The controller in later case regulates the AC side voltage by tight control on modulation index, and evenly compensates voltage drop on DC bus through injection of third harmonic component. Simulation based results demonstrate that proposed method provides effective control of VSC both in current controlled mode and voltage controlled mode for the aforesaid objectives. The presented results show that seamless power is provided to the critical loads connected to the grid with loss/failure of generation, without any transients in voltage.

This paper focuses on the problem of learning binary embeddings for efficient retrieval of high-dimensional non-negative data. Such data typically arises in a large number of vision and text applications where counts or frequencies are used as features. Also, cosine distance is commonly used as a measure of dissimilarity between such vectors. In this work, we introduce a novel spherical quantization scheme to generate binary embedding of such data and analyze its properties. The number of quantization landmarks in this scheme grows exponentially with data dimensionality resulting in low-distortion quantization. We propose a very efficient method for computing the binary embedding using such large number of landmarks. Further, a linear transformation is learned to minimize the quantization error by adapting the method to the input data resulting in improved embedding. Experiments on image and text retrieval applications show superior performance of the proposed method over other existing state-of-the-art methods.

X-ray crystallography produces molecular models (spheres for every atom of the molecule) from electron density maps (distribution of electrons over 3d space). These maps seldom have enough resolution to identify hydrogens, which are added later using simple geometric rules. For example, the structure refinement toolbox PHENIX places hydrogen at a distance of 0.82 AA from the bonded oxygen nucleus, while the structure validation toolbox Molprobity places this hydrogen at a distance of 1 AA from the bonded oxygen. To harmonize these numbers we wanted to use bond lengths that were consistent with the electron density maps based on quantum calculations and hence at a higher resolution than experimentally determined maps. In this video we describe geometric tools based on curvature and contour trees for identifying spherical patches on the iso-surfaces of electron density. These tools allow us to determine lengths of hydrogen covalent bonds from high resolution electron density maps.

Finding shortest paths and distances on the surface of a mesh in R3 is a well studied problem, with most research aiming to minimize computation time. However for large meshes, such as TIN terrain models in GIS, the major bottleneck is often the memory required by an algorithm.