Weekly interdisciplinary seminars from speakers in academia, government labs, and industry involved in the computing sciences.
During the spring 2018 semester seminar will be held Thursdays from 3:05-4:05 in room CCP 259 in the City Center Plaza Building in downtown Boise, unless otherwise noted below.
Directions and transportation options for getting to the City Center Plaza Building can be found at this link: http://coen.boisestate.edu/cs/aboutccp/
You can subscribe to the Computing Colloquium Google Calendar at this link: https://goo.gl/wr2DTh
Past Colloquium presentations can be viewed on our Youtube Channel: https://www.youtube.com/channel/UCbIwk9__d0L21BVRgRP7EQQ
Dr. Joshua A. Anderson – Research Area Specialist, Biointerfaces Institute, University of Michigan
Attaining maximum GPU execution performance in the HOOMD-blue particle simulation toolkit
I wrote HOOMD-blue from scratch ten years ago specifically for GPUs when CUDA version 0.8 beta was first released to the public. In the beginning, there was only CUDA, explicitly managed memory, very slow PCIe links, no GPU-accelerated libraries, and the only cache was for texture images. GPUs have changed a lot since the beginning. Modern GPUs make programming easy, with unified memory, high speed NVLINK interconnects, many available libraries, large on-GPU cache, and directives-based programming models like OpenACC. We have rewritten the core computation kernels in HOOMD-blue several times to attain the best possible performance on each GPU generation. In this talk, I will introduce GPU computing and discuss how researchers can take advantage of all the conveniences that modern (P100+) GPUs provide to get fast codes up and running with minimal development effort. Then I will detail specific micro-optimizations we use to reach maximum performance in HOOMD-blue, including warp-synchronous programming, auto-tuning, multiple thread per particle parallelism, bounding volume hierarchy data structures, and in-block work queues. These are all general optimization techniques that could be applied to a variety of research codes.
Dr. Foteini Baldimtsi – Assistant Professor, Computer Science, George Mason University
Indistinguishable Proofs of Work or Knowledge
We introduce a new class of protocols called Proofs of Work or Knowledge (PoWorKs). In a PoWorK, a prover can convince a verifier that she has either performed work or that she possesses knowledge of a witness to a public statement without the verifier being able to distinguish which of the two has taken place.
We formalize PoWorK in terms of three properties, completeness, f-soundness and indistinguishability (where f is a function that determines the tightness of the proof of work aspect) and present a construction that transforms 3-move HVZK protocols into 3-move public-coin PoWorKs. To formalize the work aspect in a PoWorK~protocol we define cryptographic puzzles that adhere to certain uniformity conditions, which may also be of independent interest. We instantiate our puzzles in the random oracle (RO) model as well as via constructing “dense” versions of suitably hard one-way functions.
We then showcase PoWorK~protocols by presenting a number of applications. We first show how non-interactive PoWorKs can be used to reduce spam email by forcing users sending an e-mail to either prove to the mail server they are approved contacts of the recipient or to perform computational work. As opposed to previous approaches that applied proofs of work to this problem, our proposal of using PoWorKs is privacy-preserving as it hides the list of the receiver’s approved contacts from the mail server. Our second application, shows how PoWorKs can be used to compose cryptocurrencies that are based on proofs of work (“Bitcoin-like”) with cryptocurrencies that are based on knowledge relations (these include cryptocurrencies that are based on “proof of stake”, and others). The resulting PoWorK-based cryptocurrency inherits the robustness properties of the underlying two systems while PoWorK-indistinguishability ensures a uniform population of miners. Finally, we show that PoWorK~protocols imply straight-line quasi-polynomial simulatable arguments of knowledge and based on our construction we obtain an efficient straight-line concurrent 3-move statistically quasi-polynomial simulatable argument of knowledge.
Dr. Paul Simmonds – Assistant Professor, Depts. of Physics and Material Science, Boise State University
Unbreakable Cryptography Using Quantum Dots
Quantum cryptography will allow us to encode information in such a way that it is completely secure. One proposed strategy for achieving this goal relies on entangled photons. In quantum physics, to entangle a pair of photons is to cause them to interact in such a way that even if we separate them by an arbitrarily large distance, their polarizations remain linked. After entanglement, the two photons exhibit a superposition of both horizontal and vertical polarizations. Any subsequent measurement of one of the two photons to discover its polarization instantaneously causes the other photon to assume the opposite polarization. Einstein famously referred to this phenomenon as “spooky action at a distance”!
I will describe how we can exploit this strange quantum behavior to encrypt information. I will also talk about my research in which we are looking for convenient ways to create entangled photons. We have developed a new type of quantum dot nanomaterial that has some very promising characteristics for emitting entangled photons. I will discuss how we create these nanomaterials, and show some recent results that confirm their suitability for use as generators of entangled photons.
Dr. Yuanzhe Xi – Postdoctoral Research Associate, Department of Computer Science and Engineering, University of Minnesota
Fast and stable algorithms for large-scale computation
Scientific computing and data analytics have become the third and fourth pillars of scientific discovery. Their success is tightly linked to a rapid increase in the size and complexity of problems and datasets of interest. In this talk, I will discuss our recent efforts in the development of novel numerical algorithms for tackling these challenges. In the first part, I will present a stochastic Lanczos algorithm for estimating the spectrum of Hermitian matrix pencils. The proposed algorithm only accesses the matrices through matrix-vector products and is suitable for large-scale computations. This algorithm is one of the key ingredients in the new breed of “spectrum slicing”-type eigensolvers for electronic structure calculations. In the second part, I will present our newly developed fast structured direct solvers for kernel systems and its applications in accelerating the learning process. By exploiting intrinsic low-rank property associated with the coefficient matrix, these structured solvers could overcome the cubic solution cost and quadratic storage cost of standard dense direct solvers and provide a new framework for performing various matrix operations in linear complexity.
Dr. Jonah Reeger – Assistant Professor of Mathematics, Department of Mathematics and Statistics, Air Force Institute of Technology
Numerical Quadrature Over Bounded Smooth Surface
This talk describes a high order accurate method to calculate integrals over smooth surfaces with boundaries. Given data locations that are arbitrarily distributed over the surface, together with some functional description of the surface and its boundary, the algorithm produces matching quadrature weights. This extends on our earlier methods for integrating over the surface of a sphere and over arbitrarily shaped smooth closed surfaces by now considering domain boundaries. The core approach consists of combining RBF-FD (radial basis function-generated finite difference) approximations for curved surface triangles, which together make up the full surface. A discussion of proposed applications is included.
Dr. Michal Kopera – Assistant Researcher in the Dept. of Earth and Planetary Science at UC Santa Cruz and a Visiting Researcher in the Dept. of Applied Math at the Naval Postgraduate School
Adaptive high-order continuous/discontinuous Galerkin model of the ocean with application to Greenland fjords
Interactions of ocean and glaciers inside Greenland’s fjords is one of the key outstanding challenges in modeling studies of climate change. Several-fold increase in the mass discharge through marine-terminating glaciers has been observed; however, the underlying physical mechanisms are yet to be fully understood1. The leading hypothesis is that those changes are driven mainly by increased incursions of warm seawater into the fjords. However, due to orders of magnitude difference in spatial scales between open ocean (~1000km) and fjord (<1km), as well as complex bathymetry and coastline, present-day ocean models are not able to resolve fine-scale processes in the fjords and at the ice/ocean interface.
In this talk, I will present a new computational approach to this problem using high-order continuous/discontinuous Galerkin methods on flexible, adaptive meshes, developed in the NUMO project (Non- hydrostatic Unified Model of the Ocean). To address the problem of ice-sheet/ocean interactions, circulation within the fjord, and exchanges with the open ocean, I have designed a non-hydrostatic ocean model based on the three-dimensional incompressible Navier-Stokes equations. An unstructured mesh is used to realistically represent the geometry of the fjord, while in the areas of particular importance (i.e., glacier front) the resolution is increased by local non-conforming mesh refinement. NUMO aims to mitigate the cost of non-hydrostatic ocean simulations by leveraging modern high-performance computing systems.
I will present model validation results, and discuss outstanding challenges in realistic simulations of the Sermilik Fjord. The long-term goal is to simulate all Greenland fjords and adjacent coastal ocean and couple this simulation to regional or global Earth System Models.
Dr. Varun Shankar – Assistant Professor Lecturer of Mathematics and Adjunct Assistant Professor in School of Computing, University of Utah
A high-order meshfree framework for solving PDEs on irregular domains and surfaces
We present meshfree methods based on Radial Basis Function (RBF) interpolation for solving partial differential equations (PDEs) on irregular domains and surfaces; such domains are of great importance in mathematical models of biological processes. First, we present a generalized high-order RBF-Finite Difference (RBF-FD) method that exploits certain approximation properties of RBF interpolants to achieve significantly improved computational complexity, both in serial and in parallel. Like all RBF-FD methods, our method requires stabilization when applied to solving PDEs. Consequently, we present a robust and automatic hyperviscosity-based stabilization technique to rectify the spectra of RBF-FD differentiation matrices. The amount of hyperviscosity is determined quasi-analytically in two stages: first, we develop a novel mathematical model of spurious solution growth, and second, we use simple 1D Von Neumann analysis to analytically cancel out these spurious growth terms. The resulting expressions for hyperviscosity are a generalization of formulas from both RBF-FD and classical spectral methods. The resulting stabilized RBF-FD method serves as a high-order meshfree framework for solving PDEs on irregular domains. Finally, we present a powerful new RBF-FD technique that allows for the solution of PDEs on surfaces using scattered nodes and Cartesian coordinate systems. In all cases, our methods achieve O(N) complexity for N nodes.
Tim Long – Director of Enterprise Data Science at Micron Technology
Manufacturing with AI – Industry 4.0
Micron is a world leader in memory solutions that transform how the world uses information. However, to remain a leader, innovation must reach deeper than new product design. Manufacturers must innovate internal operations, creating smart factory processes driven by data and algorithms.
In this presentation, Tim Long, Micron’s Director of Enterprise Data Science, will discuss Micron’s advancement to smart manufacturing (also known as Industry 4.0) and share projects that have helped Micron to improve customer satisfaction, achieve faster yield maturity for new products by 25 percent, and increase overall factory output by 10 percent.
Dr. Liljana Babinkostova- Professor, Boise State Mathematics
Grostl, Permutations and Transversals
Permutations are important building blocks for any cryptographic schema. In this talk we will focus on the properties of permutations that dictate the security of the Grostl – cryptographic hash function designed in response to the Cryptographic Hash Algorithm Competition initiated by NIST in 2007.
Grostl is an iterated hash function, where the compression function is built from two fixed permutations. These permutations are crucial for the resistance of Grostl against large classes of cryptanalytic attacks.
In this talk we present a nice analogy of the problem of existence of a “good” permutations in Grostl to the problem of counting transversals in latin squares.
This is joint work with several undergraduate and graduate students.
Tammi Vacha-Haase – Dean of the Graduate College, Boise State University
Mentoring and Advising: Being an Informed Consumer
Explore what graduate students need to know about getting the most out of mentoring and advising relationships. This seminar will focus on the value of a mentor relationship, what to look for in a mentor, and your role as a mentee. Learn what to do when there are challenges within the mentoring relationship, strategies for good communication, and the value of clarifying expectations.
Dr. Donna Calhoun – Associate Professor, Boise State Mathematics
Volcanic Ash Transport Using ForestClaw
As the eruption of Eyjafjallajokull in Iceland in 2010 demonstrated, “zero ash tolerance” policies imposed on civil aviation can lead to significant disruption for travelers and loss of revenue for airlines. To mitigate the hazards associated with volcanic eruptions, numerical volcanic ash transport and dispersal models are routinely used to forecast ash concentration loads in the atmosphere. However, delivering high fidelity predictions within operational time frames, especially for long-lasting eruptions that spread ash over wide areas with dense air traffic, places severe demands on computational resources.
Ash3d , developed at the Cascade Volcanic Observatory (Vancouver, WA) is one of several volcanic ash transport models in operational use. Available through a web-based portal, Ash3d solves a set of advection-diffusion-deposition equations to transport one or more classes of ash particles on a regional latitude/longitude grid. Current meteorological data is used to define wind fields for the transport equations and second order finite volume schemes are used to update the evolving ash plume.
To improve the efficiency of the single grid, serial Ash3d code, we have ported Ash3d to our parallel, adaptive software library ForestClaw . In this approach, fine level grids in a quadtree hierarchy are placed in areas with the highest ash concentration. The solution on each patch is updated using the single grid routines from Ash3d, and the grid hierarchy is updated every few time steps to track the evolving ash cloud.
We present results showing the scalability of the Ash3d extension of ForestClaw and discuss issues related to porting legacy codes to modern parallel, adaptive frameworks. Results will include tests to simulate the eruption of Mt. St. Helen’s and the Icelandic volcano Eyja. We will also discuss our approach to ensuring conservation on the adaptive grids and our on-going efforts to extend the Ash3d code to ForestClaw’s cubed sphere grid.
 Hans F. Schwaiger and Roger P. Denlinger and Larry G. Mastin, “Ash3d : A finite-volume, conservative, numerical model for ash transport and tephra deposition”, Journal of Geophys.
Res. Vol. 117 (BO4204). (2012).
 D. Calhoun, C. Burstedde “ForestClaw : A parallel adaptive library for logically Cartesian, mapped multi-block domains”, (www.forestclaw.org).