MareNostrum4, quantum algorithm implementations, Sunway supercomputer and more
In this regular feature, HPCwire highlights recently published research in the high performance computing community and related fields. From parallel programming to exascale to quantum computing, the details are here.
A parallel algorithm for unilateral contact problems
A multidisciplinary team of Spanish researchers from the Barcelona Supercomputing Center, the Technical University of Catalonia and the International Center for Numerical Methods in Engineering has developed a parallel contact algorithm designed for high performance computing with particular emphasis on its “implementation computational work in a multiphysical finite environment”. item code. The researchers based the algorithm on “the method of partial boundary conditions of Dirichlet-Neumann”. It can “numerically solve a nonlinear contact problem between rigid and deformable bodies in a complete parallel frame”. Spanish researchers validated the algorithm by performing benchmark tests and comparing the “proposed solution to theoretical and other numerical solutions”. For the benchmark tests, the researchers used the MareNostrum4 supercomputer at the BSC to perform the simulations. They also “evaluated the parallel performance of the proposed algorithm in a real-world impact test to show its capabilities for large-scale applications.”
Authors: G. Guillameta, M. Riveroa, M. Zavala-Ake´, M. Vazquez, G. Houzeauxa and S. Ollerb
Tensor Network Quantum Virtual Machine to Simulate Quantum Circuits at Exascale
In this paper by a team of researchers at Oak Ridge National Laboratory and Nvidia Corp., the authors present “a quantum circuit simulator based on a general tensor network capable of modeling both ideal and noisy quantum circuits as well as compute various experimentally accessible properties depending on the tensor network formalism used.The new Tensor Network Quantum Virtual Machine (TNQVM) “serves as a quantum circuit simulation backend in the eXtreme-scale ACCelerator (XACC) framework” The researchers based the version “on the ExaTN (Exascale Tensor Networks) scalable tensor network processing library. The article details the initial benchmarks of the “framework, which include a demonstration of distributed execution, embedding of quantum decoherence models and the simulation of random quantum circuits used for the certification of quantum supremacy on the architecture s Google’s Sycamore Upraconductor”.
Authors: Thien Nguyen, Dmitry Lyakh, Eugene Dumitrescu, David Clark, Jeff Larkin and Alexander McCaskey
Optimized SWAP Networks with Equivalent Circuit Averaging for QAOA
A multidisciplinary team of researchers from the University of California, Berkeley, Lawrence Berkeley National Lab in California, Super.tech, a division of ColdQuanta in Illinois, and the University of Chicago, presents two techniques for streamlining the execution of SWAP networks for Quantum Approximate Optimization Algorithm (QAOA). A SWAP network is a qubit routing sequence used to perform QAOA efficiently. The researcher’s techniques are experimentally validated at the Advanced Quantum Testbed through running QAOA circuits to find the ground state of two- and four-node Sherrington-Kirkpatrick spin-glass models with various randomly sampled parameters. “. The results showed “an average reduction of about 60% in error (total variation distance) for QAOA of depth p=1 over four transmon qubits on a superconducting quantum processor.”
Authors: Akel Hashim, Rich Rines, Victory Omole, Ravi K. Naik, John M. Kreikebaum, David I. Santiago, Frederic T. Chong, Irfan Siddiqi and Pranav Gokhale
Design and implementation of ShenWei Universal C/C++
Chinese researchers from Tsinghua University present ShenWei Universal C/C++ (SWUC), which “is a language extension for C/C++ [developed] to better support heterogeneous programming on ShenWei multi-core processors. ShenWei multi-core series processors (which now have SW26010 and SW26010pro) provide the necessary computing power that the Sunway supercomputer needs. The language reduces engineer effort and SWUC “enables smooth programming across management processing element (MPE) and computation processing element (CPE) boundaries”. The researchers demonstrate that “through the use of several new compiler attributes and directives, users are able to write code running on MPE and CPE in a single file.” Additionally, SWUC allows the use of “available Athread library interfaces, easing the learning curve for original ShenWei users”.
Authors: Huanqi Cao and Jiajie Chen
Stable Parallel Training of Wasserstein Conditional Generative Adversarial Neural Networks
Researchers at Oak Ridge National Laboratory in Tennessee are developing a “stable parallel approach to training Wasserstein’s Conditional Generative Adversarial Neural Networks (W-CGANs) under the constraint of a fixed computational budget.” Their proposed approach “avoids inter-process communications, reduces the risk of mode collapse, and improves scalability by using multiple generators, each of which is trained simultaneously on a single data tag.” Numerical experiments and scalability tests were performed on the Summit supercomputer at the Oak Ridge Leadership Computing Facility. “The researchers’ use of the Wasserstein metric also reduces the risk of cycling by stabilizing the entrainment of each generator.” By using the CIFAR10, CIFAR100 and ImageNet1k reference image datasets, the researchers were able to retain the “original resolution of the images for each dataset”.
Authors: Massimiliano Lupo Pasini and Junqi Yin
Next-generation computational tools for modeling and designing exascale particle accelerators
In this article, researchers at Lawrence Berkeley National Laboratory (LBNL) detail three computational tools used for “modeling and designing particle accelerators, preparing the codes for next-generation machines in the Exascale era.” The first is the Beam pLasma Accelerator Simulation Toolkit (BLAST) open-source software toolkit developed by LBNL researchers, which “provides modeling tools for modeling hybrid accelerators, containing both plasma elements and line elements conventional lights. Second, ABLASTR “is a modern C++17 library used to share in-cell particle routines between simulation codes”. Finally, ImpactX was “developed to succeed IMPACT-Z as a new, sbased on beam dynamics code with intrinsic GPU, mesh refinement and tight coupling with time-based codes and AI/ML capabilities. Still in its infancy, ImpactX can already “model much larger sets of particles than its predecessors,” the researchers conclude. Further developments are planned.
Authors: Axel Huebl, Remi Lehe, Chad E. Mitchell, Ji Qiang, Robert D. Ryne, Ryan T. Sandberg and Jean-Luc Vay
Quantum Algorithm Implementations for Beginners
The Los Alamos National Laboratory researchers are “revisiting the principles of quantum programming, which are quite different from classical programming, with simple algebra that makes understanding the fascinating underlying principles of quantum mechanics optional.” In this article, the authors provide an “introduction to quantum computing algorithms and their implementation on real quantum hardware”. They summarize 20 quantum algorithms with an overview of how to implement them on IBM’s quantum computer, then examine the “implementation” results with respect to the differences between the simulator and the actual material executions”. The code is publicly available on GitHub at https://github.com/lanl/quantum_algorithms.
Authors: Abhijith J., Adetokunbo Adedoyin, John Ambrosiano, Petr Anisimov, William Casper, Gopinath Chennupati, Carleton Coffrin, Hristo Djidjev, David Gunter, Satish Karra, Nathan Lemons, Shizeng Lin, Alexander Malyzhenkov, David Mascarenas, Susan Mniszewski, Balu Nadiga , Daniel O’malley, Diane Oyen, Scott Pakin, Lakshman Prasad, Randy Roberts, Phillip Romero, Nandakishore Santhi, Nikolai Sinitsyn, Pieter J. Swart, James G. Wendelberger, Boram Yoon, Richard Zamora, Wei Zhu, Stephan Eidenbenz, Andreas Bärtschi, Patrick J. Coles, Marc Vuffray and Andrey Y. Lokhov
Do you know of any research that should be included in next month’s list? If so, send us an email at [email protected]. We look forward to hearing from you.