A5036031916- Distributed and Parallel Computing Systems
- Scientific Computing and Data Management
- Nonlinear Partial Differential Equations
- Cloud Computing and Resource Management
- Advanced Data Storage Technologies
- Advanced Electron Microscopy Techniques and Applications
- Electron and X-Ray Spectroscopy Techniques
- Advanced Fluorescence Microscopy Techniques
- Advanced Mathematical Modeling in Engineering
- Mathematical Inequalities and Applications
- Optimization and Variational Analysis
- Spectral Theory in Mathematical Physics
- Navier-Stokes equation solutions
- Advanced Harmonic Analysis Research
- Numerical methods in inverse problems
- Integrated Circuits and Semiconductor Failure Analysis
- Mathematical Approximation and Integration
- Advanced Mathematical Physics Problems
- Computability, Logic, AI Algorithms
- Macrophage Migration Inhibitory Factor
San Diego Supercomputer Center
2017-2024
University of California, San Diego
2013-2024
TU Dortmund University
2023
Scripps Research Institute
2009-2010
Torrey Pines Institute For Molecular Studies
2009
We describe the design motivation, architecture, deployment, and early operations of Expanse, a 5 Petaflop, heterogenous HPC system that entered production as an NSF-funded resource in December 2020 will be operated on behalf national community for five years. Expanse serve broad range computational science engineering through combination standard batch-oriented services, by extending to broader CI ecosystem gateways, public cloud integration, support high throughput computing, composable...
Voyager is an innovative computational resource designed by the San Diego Supercomputer Center in collaboration with technology partners to accelerate development and performance of artificial intelligence machine learning applications science engineering. Based on Intel's Habana Labs first-generation deep (Gaudi) training (Goya) inference processors, funded National Science Foundation's Advanced Computing Systems & Services Program as a Category II system will be operated for 5 years,...
Abstract We prove improved differentiability results for relaxed minimisers of vectorial convex functionals with <m:math xmlns:m="http://www.w3.org/1998/Math/MathML"> <m:mrow> <m:mo>(</m:mo> <m:mi>p</m:mi> <m:mo>,</m:mo> <m:mi>q</m:mi> </m:mrow> <m:mo>)</m:mo> </m:math> \left(p,q) -growth, satisfying a Hölder-growth condition in <m:mi>x</m:mi> x . consider both Dirichlet and Neumann boundary data. In addition, we obtain characterisation regular points such minimisers. particular, case...
The Comet petascale supercomputer was put into production as an XSEDE resource in early 2015 with the goal of serving a much larger user community than HPC systems similar size. project set audacious reaching over 10,000 users its four years planned operation. That achieved less two years, due large part to adoption policies that favor smaller allocations and science gateways. Here we describe our experiences operating supporting Comet, highlight some important it has enabled, provide...
The Hadoop framework is extensively used for scalable distributed processing of large datasets. This extended abstract provides information on the optimization deployment Gordon data intensive supercomputer, at San Diego Supercomputer Center (SDSC) University California Diego, using myHadoop software. details system configuration, storage and network options (1 Gig-E, IPOIB, UDA), tuning considered, results TestDFSIO, TeraSort benchmarks, bulk copy tests with distcp are presented in this abstract.
Influenced by the advances in data and computing, scientific practice increasingly involves machine learning artificial intelligence driven methods which requires specialized capabilities at system-, science- service-level addition to conventional large-capacity supercomputing approaches. The latest distributed architectures built around composability of data-centric applications led emergence a new ecosystem for container coordination integration. However, there is still divide between...
We establish universality of the renormalised energy for mappings from a planar domain to compact manifold, by approximating subquadratic polar convex functionals form $\int_\Omega f(|\mathrm{D} u|)\,\mathrm{d} x$. The analysis relies on condition that vortex map ${x}/{\lvert x\rvert}$ has finite and $t\mapsto f (\sqrt{t})$ is concave. derive leading order asymptotics provide detailed description convergence $\mathrm{W}^{1,1}$-almost minimisers, characterization second-order asymptotics. At...
We prove improved differentiability results for relaxed minimisers of vectorial convex functionals with $(p, q)$-growth, satisfying a H\"older-growth condition in $x$. consider both Dirichlet and Neumann boundary data. In addition, we obtain characterisation regular points such minimisers. particular, case homogeneous conditions, this allows us to deduce partial regularity on smooth domains radial integrands. also some non-homogeneous conditions.
The Comet petascale system is an XSEDE resource with the goal of serving a large user community. project has served number users while using traditional supercomputing as well science gateways. In addition to these offerings, also includes non virtual machine framework that allows access entire Virtual Clusters instead just focusing on individual machines. integrates custom administration interface, novel image management back-end, industry standard hardware virtualization technology and...
Influenced by the advances in data and computing, scientific practice increasingly involves machine learning artificial intelligence driven methods which requires specialized capabilities at system-, science- service-level addition to conventional large-capacity supercomputing approaches. The latest distributed architectures built around composability of data-centric applications led emergence a new ecosystem for container coordination integration. However, there is still divide between...
We give a direct harmonic approximation lemma for local minima of quasiconvex multiple integrals that entails their $\mathrm{C}^{1,\alpha}$ or $\mathrm{C}^{\infty}$-partial regularity. Different from previous contributions, the method is fully and elementary, only hinging on $\mathrm{L}^{p}$-theory strongly elliptic linear systems Sobolev's embedding theorem. Especially, no heavier tools such as Lipschitz truncations are required.
A User Portal is being developed for NSF-funded Expanse supercomputer. The portal based on the Open OnDemand HPC platform which has gained widespread adoption at centers. will provide a gateway launching interactive applications such as MATLAB, RStudio, and an integrated web-based environment file management job submission. This paper discusses early experience in deploying customizations that were made to accommodate requirements of user community.
We will establish an $\varepsilon$-regularity result for weak solutions to Legendre-Hadamard elliptic systems, under the a-priori assumption that gradient $\nabla u$ is small in $\mathrm{BMO}.$ Focusing on case of Euler-Lagrange systems simplify exposition, regularity results be obtained up boundary, and global consequences explored. Extensions general quasilinear higher-order integrands also discussed.
A partial regularity theorem is presented for minimisers of $k$th-order functionals subject to a quasiconvexity and general growth condition. We will assume natural condition governed by an $N$-function satisfying the $\Delta_2$ $\nabla_2$ conditions, assuming no quantitative estimates on second derivative integrand; this new even in $k = 1$ case. These results also be extended case strong local minimisers.