BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201324Z
LOCATION:231
DTSTART;TZID=America/Chicago:20251116T090000
DTEND;TZID=America/Chicago:20251116T173000
UID:submissions.supercomputing.org_SC25_sess223@linklings.com
SUMMARY:16th Workshop on Latest Advances in Scalable Algorithms for Large-
 Scale Heterogeneous Systems (ScalAH'25)
DESCRIPTION:Novel hybrid scalable scientific algorithms are needed with th
 e advent of a variety of novel accelerators, including GPUs and FPGAs, as 
 well as with the growth in size of quantum computing devices, neuromorphic
  chips, and various AI-specific processors. This myriad of devices require
 s a unified approach that allows efficient and scalable hybrid approaches 
 combining classical and novel computing paradigms to be implemented at sca
 le. These extreme-scale heterogeneous systems require these novel scientif
 ic algorithms to hide the complexity, hide network and memory latency, hav
 e advanced communication, and have no synchronization points where possibl
 e. With the advent of AI in the past few years, the need for such scalable
  mathematical methods and algorithms for such hybrid architectures that ar
 e able to handle data and compute-intensive applications at scale becomes 
 even more important.\n\nHigh-Performance and Power-Efficient Emulation of 
 Matrix Multiplication using INT8 Matrix Engines\n\nRecent architectures in
 tegrate high-performance and power-efficient matrix engines. \nThese engin
 es demonstrate remarkable performance in low-precision matrix multiplicati
 on, which is crucial in deep learning. \nSeveral techniques have been prop
 osed to emulate single- and double-precision general matr...\n\n\nYuki Uch
 ino (RIKEN Center for Computational Science (R-CCS)), Katsuhisa Ozaki (Shi
 baura Institute of Technology), and Toshiyuki Imamura (RIKEN Center for Co
 mputational Science (R-CCS))\n---------------------\nA High Performance GP
 U CountSketch Implementation and Its Application to Multisketching and Lea
 st Squares Problems\n\nRandom sketching is a dimensionality reduction tech
 nique that approximately preserves norms and singular values up to some O(
 1) distortion factor with high probability. The most popular sketches in l
 iterature are the Gaussian sketch and the subsampled randomized Hadamard t
 ransform, while the CountSk...\n\n\nAndrew Higgins, Erik Boman, and Ichita
 ro Yamazaki (Sandia National Laboratories)\n---------------------\nEfficie
 nt Embedding Initialization via Dominant Eigenvector Projections\n\nThe em
 bedding layer is essential in deep learning, transforming high-dimensional
  data into compact representations. However, growing datasets and model si
 zes pose challenges in training time, memory, and generalization. We propo
 se a scalable method for embedding initialization via spectral dimension..
 .\n\n\nQuentin Petit (Mines Paris - PSL University France); Chong Li (Huaw
 ei Technologies France); Nahid Emad (Maison de la Simulation, University o
 f Paris-Saclay); and Jack Dongarra (University of Tennessee, Knoxville)\n-
 --------------------\nScalable Hydrodynamics on multiple Field-Programmabl
 e Gate Arrays (FPGAs)\n\nHydroDynamics (HD) and MagnetoHydroDynamics (MHD)
  simulations play a central role in modeling physical processes in fields 
 as diverse as astrophysics, nuclear fusion, and plasma physics. These simu
 lations often involve the resolution of hyperbolic systems of partial diff
 erential equations using fini...\n\n\nFrançois-Xavier Mordant (CEA Saclay)
 , Charles Prouveur (French National Center for Scientific Research (CNRS))
 , Pascal Tremblin (CEA Saclay), and Nicolas GAC (University Paris Saclay)\
 n---------------------\nAfternoon Break - Scalable Algorithms for Large-Sc
 ale Heterogeneous Systems (ScalAH'25)\n---------------------\nWelcome\n\nV
 assil Alexandrov\n---------------------\nInvited Talk: Composable Architec
 tures for Future Computing\n\nHPC, AI, and Quantum Computing have emerged 
 as essential capabilities for future computing. The impact of integration 
 of these capabilities will be revolutionary for the whole spectrum of rese
 arch and enterprise computing. In this presentation we will discuss some o
 f the potential impacts of such in...\n\n\nJames Sexton (IBM Research Euro
 pe, Dublin Laboratory)\n---------------------\nInvited Talk: When the qubi
 ts are ready, will the national labs be?\n\nAbstractEarly demonstrations o
 f operations relevant to quantum error correction and fault-tolerant quant
 um computation signal that the noisy intermediate-scale quantum (NISQ) era
  might be coming to an end. Vendor roadmaps indicate that gigaquop machine
 s with hundreds of logical qubits capable of exe...\n\n\nAndrew Baczewski 
 (Sandia National Laboratories)\n---------------------\nInvited Talk: Reinv
 enting Discovery: Accelerating Science in the Age of Artificial Super-Inte
 lligence\n\nFrontier AI models have crossed a threshold. They no longer me
 rely assist scientists, but now co-design not only which questions to purs
 ue, but how to pursue them. This keynote examines how we can accelerate sc
 ientific discovery using these advanced models. Drawing an analogy to Amda
 hl’s Law, ...\n\n\nRick Stevens (Argonne National Laboratory (ANL), Univer
 sity of Chicago)\n---------------------\nClosing\n\nVassil Alexandrov\n---
 ------------------\nNumerical Properties and Scalability of s-Step Precond
 itioned Conjugate Gradient Methods\n\ns-step Preconditioned Conjugate Grad
 ient (PCG) variants for iteratively solving large sparse linear systems re
 duce the number of global synchronization points of standard PCG by a fact
 or of O(s). Despite improving scalability on large-scale parallel computer
 s, they have worse numerical properties th...\n\n\nViktoria Mayer and Wilf
 ried N. Gansterer (University of Vienna)\n---------------------\nLunch bre
 ak (on your own)\n---------------------\nFast Linear Solvers via AI-Tuned 
 Markov Chain Monte Carlo-based Matrix Inversion\n\nLarge, sparse linear sy
 stems are pervasive in modern science and engineering, and Krylov subspace
  solvers are an established means of solving them. Yet convergence can be 
 slow for ill-conditioned matrices, so practical deployments usually requir
 e preconditioners. Markov chain Monte Carlo (MCMC)-base...\n\n\nAnton Lebe
 dev and Won Kyung Lee (STFC Hartree Centre); Soumyadip Ghosh (IBM Thomas J
 . Watson Research Center); Olha I. Yaman (STFC Hartree Centre); Vassilis K
 alantzis, Yingdong Lu, Tomasz Nowicki, Shashanka Ubaru, and Lior Horesh (I
 BM Thomas J. Watson Research Center); and Vassil Alexandrov (STFC Hartree 
 Centre)\n---------------------\nInvited Talk: AI-for-Science by Integratio
 ns of Simulations/Data/Learning on Heterogeneous Supercomputers\n\nSoftwar
 e for heterogeneous systems. Integration of (S+D+L) by h3-Open-BDEC enable
 s significant reduction of computations and power consumption, compared to
  those by conventional simulations. In January 2025, we started to operate
  the Miyabi system together with University of Tsukuba. Miyabi consists...
 \n\n\nKengo Nakajima (University of Tokyo, RIKEN Center for Computational 
 Science)\n---------------------\nMorning Break - Scalable Algorithms for L
 arge-Scale Heterogeneous Systems (ScalAH'25)\n---------------------\nPost-
 Variational Quantum Neural Networks on a Hybrid HPC-QC System\n\nWe implem
 ent a post-variational quantum neural network on a real HPC-QC system and 
 show the feasibility of fully training this class of algorithms on current
  Noisy Intermediate-Scale Quantum (NISQ) devices, which are limited by noi
 se, low number of qubits, and scarcity. Post-variational methods are ...\n
 \n\nMaxence Vandromme (RIKEN Center for Computational Science (R-CCS)) and
  Miwako Tsuji (RIKEN Center for Computational Science (R-CCS); Center for 
 Computational Sciences, University of Tsukuba)\n\nRecording: Livestreamed,
  Recorded\n\nRegistration Category: Technical Program Reg Pass, Workshop R
 eg Pass\n\nSession Chairs: Vassil Alexandrov (Hartree Centre, STFC); Jack 
 Dongarra (University of Tennessee, Knoxville; Oak Ridge National Laborator
 y (ORNL)); Erik Draeger (Lawrence Livermore National Laboratory (LLNL), Ce
 nter for Applied Scientific Computing); Philippa Rubin (STFC Hartree Centr
 e); Dieter A. Kranzlmueller (Ludwig-Maxmilians-Universität München, Leibni
 z Supercomputing Centre (LRZ)); and Christian Engelmann (Oak Ridge Nationa
 l Laboratory (ORNL))
END:VEVENT
END:VCALENDAR