Sivasankaran Rajamanickam
YOU?
Author Swipe
View article: Do AI Models Perform Human-like Abstract Reasoning Across Modalities?
Do AI Models Perform Human-like Abstract Reasoning Across Modalities? Open
OpenAI's o3-preview reasoning model exceeded human accuracy on the ARC-AGI benchmark, but does that mean state-of-the-art models recognize and reason with the abstractions that the task creators intended? We investigate models' abstraction…
View article: Performance Portable Gradient Computations Using Source Transformation
Performance Portable Gradient Computations Using Source Transformation Open
Derivative computation is a key component of optimization, sensitivity analysis, uncertainty quantification, and nonlinear solvers. Automatic differentiation (AD) is a powerful technique for evaluating such derivatives, and in recent years…
View article: ShyLU node: On-node Scalable Solvers and Preconditioners Recent Progresses and Current Performance
ShyLU node: On-node Scalable Solvers and Preconditioners Recent Progresses and Current Performance Open
ShyLU-node is an open-source software package that implements linear solvers and preconditioners on shared-memory multicore CPUs or on a GPU. It is part of the Trilinos software framework and designed to provide a robust and efficient solu…
View article: Breaking the mold: Overcoming the time constraints of molecular dynamics on general-purpose hardware
Breaking the mold: Overcoming the time constraints of molecular dynamics on general-purpose hardware Open
The evolution of molecular dynamics (MD) simulations has been intimately linked to that of computing hardware. For decades following the creation of MD, simulations have improved with computing power along the three principal dimensions of…
View article: Materials Learning Algorithms (MALA): Scalable Machine Learning for Electronic Structure Calculations in Large-Scale Atomistic Simulations
Materials Learning Algorithms (MALA): Scalable Machine Learning for Electronic Structure Calculations in Large-Scale Atomistic Simulations Open
We present the Materials Learning Algorithms (MALA) package, a scalable machine learning framework designed to accelerate density functional theory (DFT) calculations suitable for large-scale atomistic simulations. Using local descriptors …
View article: Breaking the mold: overcoming the time constraints of molecular dynamics on general-purpose hardware
Breaking the mold: overcoming the time constraints of molecular dynamics on general-purpose hardware Open
The evolution of molecular dynamics (MD) simulations has been intimately linked to that of computing hardware. For decades following the creation of MD, simulations have improved with computing power along the three principal dimensions of…
View article: Milestone 49 Report: Batched Sparse LA Phase 5 Implementation
Milestone 49 Report: Batched Sparse LA Phase 5 Implementation Open
Batched sparse linear algebra operations in general, and solvers in particular, have become the major algorithmic development activity and foremost performance engineering effort in the numerical software libraries work on modern hardware …
View article: Breaking the Molecular Dynamics Timescale Barrier Using a Wafer-Scale System
Breaking the Molecular Dynamics Timescale Barrier Using a Wafer-Scale System Open
Molecular dynamics (MD) simulations have transformed our understanding of the nanoscale, driving breakthroughs in materials science, computational chemistry, and several other fields, including biophysics and drug design. Even on exascale …
View article: Towards reverse mode automatic differentiation of Kokkos-based codes
Towards reverse mode automatic differentiation of Kokkos-based codes Open
Derivative computation is a key component of optimization, sensitivity analysis, uncertainty quantification, and the solving of nonlinear problems. Automatic differentiation (AD) is a powerful technique for evaluating such derivatives, and…
View article: An Experimental Study of Two-level Schwarz Domain-Decomposition Preconditioners on GPUs
An Experimental Study of Two-level Schwarz Domain-Decomposition Preconditioners on GPUs Open
The generalized Dryja–Smith–Widlund (GDSW) preconditioner is a two-level overlapping Schwarz domain decomposition (DD) preconditioner that couples a classical one-level overlapping Schwarz preconditioner with an energy-minimizing coarse sp…
View article: Advanced Research Directions on AI for Science, Energy, and Security: Report on Summer 2022 Workshops
Advanced Research Directions on AI for Science, Energy, and Security: Report on Summer 2022 Workshops Open
This is a report about a series of workshops sponsored by the Department of Energy (DOE) to gather input on new and rapidly emerging opportunities and challenges of scientific AI. The members of the workshops believes that AI can have a fo…
View article: Jet: Multilevel Graph Partitioning on Graphics Processing Units
Jet: Multilevel Graph Partitioning on Graphics Processing Units Open
The multilevel heuristic is the dominant strategy for high-quality sequential and parallel graph partitioning. Partition refinement is a key step of multilevel graph partitioning. In this work, we present Jet, a new parallel algorithm for …
View article: An Experimental Study of Two-Level Schwarz Domain Decomposition Preconditioners on GPUs
An Experimental Study of Two-Level Schwarz Domain Decomposition Preconditioners on GPUs Open
The generalized Dryja--Smith--Widlund (GDSW) preconditioner is a two-level overlapping Schwarz domain decomposition (DD) preconditioner that couples a classical one-level overlapping Schwarz preconditioner with an energy-minimizing coarse …
View article: Exploiting Inter-Operation Data Reuse in Scientific Applications using GOGETA
Exploiting Inter-Operation Data Reuse in Scientific Applications using GOGETA Open
HPC applications are critical in various scientific domains ranging from molecular dynamics to chemistry to fluid dynamics. Conjugate Gradient (CG) is a popular application kernel used in iterative linear HPC solvers and has applications i…
View article: Performance Portable Batched Sparse Linear Solvers
Performance Portable Batched Sparse Linear Solvers Open
Solving large number of small linear systems is increasingly becoming a bottleneck in computational science applications. While dense linear solvers for such systems have been studied before, batched sparse linear solvers are just starting…
View article: High-Performance GMRES Mixed-Precision (HPG-MxP) Benchmark
High-Performance GMRES Mixed-Precision (HPG-MxP) Benchmark Open
SAND2024-08539O The High Performance GMRES Mixed-Precision (HPG-MxP) is a benchmark for ranking high-performance supercomputers, allowing use of mixed-precision. Similar to HPCG benchmark, it is designed to profile the computers' capabilit…