Evan Weinberg
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Evan Weinberg has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Richard C. Brower
Boston University
8 shared publications - Joel Giedt
Department of Physics, Applied Physics and Astronomy, Rensselaer Polytechnic Institute, 110 8th Street, Troy, New York 12065, USA
5 shared publications - M. A. Clark
NVIDIA Corporation
5 shared publications - Alexei Strelchenko
Fermi National Accelerator Laboratory 60510 Batavia IL USA
3 shared publications - Anna Hasenfratz
University of Colorado, Boulder
3 shared publications - Chung-I Tan
Brown University, Providence, Rhode Island 02912, USA
3 shared publications
Latest Publications
- Open Access
- 15 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 1217-1232
LAMMPS-KOKKOS: Performance Portable Molecular Dynamics Across Exascale Architectures
- 15 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 1217-1232
Open AccessLAMMPS-KOKKOS: Performance Portable Molecular Dynamics Across Exascale Architectures
- Open Access
- 19 February 2025
- Published by Sissa Medialab Srl
- Vol. 466, 046
Improving HISQ propagator solves using deflation
Typically, the conjugate gradient (CG) algorithm employs mixed precision and even-odd preconditioning to compute propagators for highly improved staggered quarks (HISQ). This approach suffers from critical slowing down as the light quark mass is decreased to its physical value. Multigrid is one alternative to combat critical slowing down; however, it involves setup costs that are not always easy to amortize. We consider deflation, which can also remove critical slowing down, but incurs its own setup cost to compute eigenvectors. Results using the MILC and QUDA software libraries to generate eigenvectors and to perform deflated solves on lattices up to $144^3 \times 288$ (with lattice spacing 0.04 fm) and with a range of quark masses from the physical strange down to the physical light quark values will be presented. We compare with CG and comment on deflation versus multigrid.- 19 February 2025
- Published by Sissa Medialab Srl
- Vol. 466, 046
Open AccessImproving HISQ propagator solves using deflation
Typically, the conjugate gradient (CG) algorithm employs mixed precision and even-odd preconditioning to compute propagators for highly improved staggered quarks (HISQ). This approach suffers from critical slowing down as the light quark mass is decreased to its physical value. Multigrid is one alternative to combat critical slowing down; however, it involves setup costs that are not always easy to amortize. We consider deflation, which can also remove critical slowing down, but incurs its own setup cost to compute eigenvectors. Results using the MILC and QUDA software libraries to generate eigenvectors and to perform deflated solves on lattices up to $144^3 \times 288$ (with lattice spacing 0.04 fm) and with a range of quark masses from the physical strange down to the physical light quark values will be presented. We compare with CG and comment on deflation versus multigrid. - Open Access
Stealth dark matter spectrum using LapH and Irreps
We present non-perturbative lattice calculations of the low-lying meson and baryon spectrum of the SU(4) gauge theory with fundamental fermion constituents. This theory is one instance of stealth dark matter, a class of strongly coupled theories, where the lowest mass stable baryon is the dark matter candidate. This work constitutes the first milestone in the program to study stealth dark matter self-interactions. Here, we focus on reducing excited state contamination in the single baryon channel by applying the Laplacian Heaviside method, as well as projecting our baryon operators onto the irreducible representations of the octahedral group. We compare our resulting spectrum to previous work involving Gaussian smeared non-projected operators and find good agreement with reduced statistical uncertainties. We also present the spectrum of the low-lying odd-parity baryons for the first time.
Open AccessStealth dark matter spectrum using LapH and Irreps
We present non-perturbative lattice calculations of the low-lying meson and baryon spectrum of the SU(4) gauge theory with fundamental fermion constituents. This theory is one instance of stealth dark matter, a class of strongly coupled theories, where the lowest mass stable baryon is the dark matter candidate. This work constitutes the first milestone in the program to study stealth dark matter self-interactions. Here, we focus on reducing excited state contamination in the single baryon channel by applying the Laplacian Heaviside method, as well as projecting our baryon operators onto the irreducible representations of the octahedral group. We compare our resulting spectrum to previous work involving Gaussian smeared non-projected operators and find good agreement with reduced statistical uncertainties. We also present the spectrum of the low-lying odd-parity baryons for the first time.
- Open Access
- 5 November 2021
- Published by American Physical Society (APS) in Physical Review D
- Vol. 104 (9) , 094502
Radial lattice quantization of 3D field theoryThe quantum extension of classical finite elements, referred to as quantum finite elements (QFE) [R. C. Brower et al., Lattice field theory on Riemann manifolds: Numerical tests for the 2-d Ising CFT on , Phys. Rev. D 98, 014502 (2018). and R. C. Brower et al., Lattice dirac fermions on a simplicial Riemannian manifold, Phys. Rev. D 95, 114510 (2017).], is applied to the radial quantization of 3D theory on a simplicial lattice for the manifold. Explicit counterterms to cancel the one- and two-loop ultraviolet defects are implemented to reach the quantum continuum theory. Using the Brower-Tamayo [Embedded Dynamics for Theory, Phys. Rev. Lett. 62, 1087 (1989).] cluster Monte Carlo algorithm, numerical results support the QFE ansatz that the critical conformal field theory (CFT) is reached in the continuum with the full isometries of restored. The Ricci curvature term, while technically irrelevant in the quantum theory, is shown to dramatically improve the convergence, opening the way for high precision Monte Carlo simulation to determine the CFT data; operator dimensions, trilinear operator product expansion couplings, and the central charge.- 5 November 2021
- Published by American Physical Society (APS) in Physical Review D
- Vol. 104 (9) , 094502
Open AccessRadial lattice quantization of 3D field theoryThe quantum extension of classical finite elements, referred to as quantum finite elements (QFE) [R. C. Brower et al., Lattice field theory on Riemann manifolds: Numerical tests for the 2-d Ising CFT on , Phys. Rev. D 98, 014502 (2018). and R. C. Brower et al., Lattice dirac fermions on a simplicial Riemannian manifold, Phys. Rev. D 95, 114510 (2017).], is applied to the radial quantization of 3D theory on a simplicial lattice for the manifold. Explicit counterterms to cancel the one- and two-loop ultraviolet defects are implemented to reach the quantum continuum theory. Using the Brower-Tamayo [Embedded Dynamics for Theory, Phys. Rev. Lett. 62, 1087 (1989).] cluster Monte Carlo algorithm, numerical results support the QFE ansatz that the critical conformal field theory (CFT) is reached in the continuum with the full isometries of restored. The Ricci curvature term, while technically irrelevant in the quantum theory, is shown to dramatically improve the convergence, opening the way for high precision Monte Carlo simulation to determine the CFT data; operator dimensions, trilinear operator product expansion couplings, and the central charge. - Open Access
Rapid Exploration of Optimization Strategies on Advanced Architectures using TestSNAP and LAMMPS
The exascale race is at an end with the announcement of the Aurora and Frontier machines. This next generation of supercomputers utilize diverse hardware architectures to achieve their compute performance, providing an added onus on the performance portability of applications. An expanding fragmentation of programming models would provide a compounding optimization challenge were it not for the evolution of performance-portable frameworks, providing unified models for mapping abstract hierarchies of parallelism to diverse architectures. A solution to this challenge is the evolution of performance-portable frameworks, providing unified models for mapping abstract hierarchies of parallelism to diverse architectures. Kokkos is one such performance portable programming model for C++ applications, providing back-end implementations for each major HPC platform. Even with a performance portable framework, restructuring algorithms to expose higher degrees of parallelism is non-trivial. The Spectral Neighbor Analysis Potential (SNAP) is a machine-learned inter-atomic potential utilized in cutting-edge molecular dynamics simulations. Previous implementations of the SNAP calculation showed a downward trend in their performance relative to peak on newer-generation CPUs and low performance on GPUs. In this paper we describe the restructuring and optimization of SNAP as implemented in the Kokkos CUDA backend of the LAMMPS molecular dynamics package, benchmarked on NVIDIA GPUs. We identify novel patterns of hierarchical parallelism, facilitating a minimization of memory access overheads and pushing the implementation into a compute-saturated regime. Our implementation via Kokkos enables recompile-and-run efficiency on upcoming architectures. We find a $\sim$22x time-to-solution improvement relative to an existing implementation as measured on an NVIDIA Tesla V100-16GB for an important benchmark.
Open AccessRapid Exploration of Optimization Strategies on Advanced Architectures using TestSNAP and LAMMPS
The exascale race is at an end with the announcement of the Aurora and Frontier machines. This next generation of supercomputers utilize diverse hardware architectures to achieve their compute performance, providing an added onus on the performance portability of applications. An expanding fragmentation of programming models would provide a compounding optimization challenge were it not for the evolution of performance-portable frameworks, providing unified models for mapping abstract hierarchies of parallelism to diverse architectures. A solution to this challenge is the evolution of performance-portable frameworks, providing unified models for mapping abstract hierarchies of parallelism to diverse architectures. Kokkos is one such performance portable programming model for C++ applications, providing back-end implementations for each major HPC platform. Even with a performance portable framework, restructuring algorithms to expose higher degrees of parallelism is non-trivial. The Spectral Neighbor Analysis Potential (SNAP) is a machine-learned inter-atomic potential utilized in cutting-edge molecular dynamics simulations. Previous implementations of the SNAP calculation showed a downward trend in their performance relative to peak on newer-generation CPUs and low performance on GPUs. In this paper we describe the restructuring and optimization of SNAP as implemented in the Kokkos CUDA backend of the LAMMPS molecular dynamics package, benchmarked on NVIDIA GPUs. We identify novel patterns of hierarchical parallelism, facilitating a minimization of memory access overheads and pushing the implementation into a compute-saturated regime. Our implementation via Kokkos enables recompile-and-run efficiency on upcoming architectures. We find a $\sim$22x time-to-solution improvement relative to an existing implementation as measured on an NVIDIA Tesla V100-16GB for an important benchmark.
- Open Access
- 24 November 2020
- Published by American Physical Society (APS) in Physical Review D
- Vol. 102 (9) , 094517
Multigrid for chiral lattice fermions: Domain wall
The phenomena of critical slowing down in the iterative solution of the Dirac equation presents a major challenge to further applications of lattice field theory in the approach to the continuum solution. We propose a new multigrid approach for chiral fermions, applicable to both the 5D domain wall or 4D overlap operator. The central idea is to directly coarsen the 4D Wilson kernel, giving an effective domain wall or overlap operator on each level. We provide here an explicit construction for the Shamir domain wall formulation with numerical tests for the 2D Schwinger prototype, demonstrating near ideal multigrid scaling. The framework is designed for a natural extension to 4D lattice QCD chiral fermions, such as the Möbius, Zolotarev or Borici domain wall discretizations or directly to a rational expansion of the 4D overlap operator. For the Shamir operator, the effective overlap operator is isolated by the use of a Pauli-Villars preconditioner in the spirit of the Kähler-Dirac spectral map used in a recent staggered multigrid algorithm [R. C. Brower, E. Weinberg, M. A. Clark, and A. Strelchenko, Phys. Rev. D 97, 114513 (2018)].
- 24 November 2020
- Published by American Physical Society (APS) in Physical Review D
- Vol. 102 (9) , 094517
Open AccessMultigrid for chiral lattice fermions: Domain wall
The phenomena of critical slowing down in the iterative solution of the Dirac equation presents a major challenge to further applications of lattice field theory in the approach to the continuum solution. We propose a new multigrid approach for chiral fermions, applicable to both the 5D domain wall or 4D overlap operator. The central idea is to directly coarsen the 4D Wilson kernel, giving an effective domain wall or overlap operator on each level. We provide here an explicit construction for the Shamir domain wall formulation with numerical tests for the 2D Schwinger prototype, demonstrating near ideal multigrid scaling. The framework is designed for a natural extension to 4D lattice QCD chiral fermions, such as the Möbius, Zolotarev or Borici domain wall discretizations or directly to a rational expansion of the 4D overlap operator. For the Shamir operator, the effective overlap operator is isolated by the use of a Pauli-Villars preconditioner in the spirit of the Kähler-Dirac spectral map used in a recent staggered multigrid algorithm [R. C. Brower, E. Weinberg, M. A. Clark, and A. Strelchenko, Phys. Rev. D 97, 114513 (2018)].
- Open Access
Pushing Memory Bandwidth Limitations Through Efficient Implementations of Block-Krylov Space Solvers on GPUs
Lattice quantum chromodynamics simulations in nuclear physics have benefited from a tremendous number of algorithmic advances such as multigrid and eigenvector deflation. These improve the time to solution but do not alleviate the intrinsic memory-bandwidth constraints of the matrix-vector operation dominating iterative solvers. Batching this operation for multiple vectors and exploiting cache and register blocking can yield a super-linear speed up. Block-Krylov solvers can naturally take advantage of such batched matrix-vector operations, further reducing the iterations to solution by sharing the Krylov space between solves. However, practical implementations typically suffer from the quadratic scaling in the number of vector-vector operations. Using the QUDA library, we present an implementation of a block-CG solver on NVIDIA GPUs which reduces the memory-bandwidth complexity of vector-vector operations from quadratic to linear. We present results for the HISQ discretization, showing a 5x speedup compared to highly-optimized independent Krylov solves on NVIDIA's SaturnV cluster.
Open AccessPushing Memory Bandwidth Limitations Through Efficient Implementations of Block-Krylov Space Solvers on GPUs
Lattice quantum chromodynamics simulations in nuclear physics have benefited from a tremendous number of algorithmic advances such as multigrid and eigenvector deflation. These improve the time to solution but do not alleviate the intrinsic memory-bandwidth constraints of the matrix-vector operation dominating iterative solvers. Batching this operation for multiple vectors and exploiting cache and register blocking can yield a super-linear speed up. Block-Krylov solvers can naturally take advantage of such batched matrix-vector operations, further reducing the iterations to solution by sharing the Krylov space between solves. However, practical implementations typically suffer from the quadratic scaling in the number of vector-vector operations. Using the QUDA library, we present an implementation of a block-CG solver on NVIDIA GPUs which reduces the memory-bandwidth complexity of vector-vector operations from quadratic to linear. We present results for the HISQ discretization, showing a 5x speedup compared to highly-optimized independent Krylov solves on NVIDIA's SaturnV cluster.
- Open Access
- 6 July 2018
- Published by American Physical Society (APS) in Physical Review D
- Vol. 98 (1) , 014502
Latticefield theory on Riemann manifolds: Numerical tests for the 2D Ising CFT onWe present a method for defining a lattice realization of the quantum field theory on a simplicial complex in order to enable numerical computation on a general Riemann manifold. The procedure begins with adopting methods from traditional Regge calculus (RC) and finite element methods (FEM) plus the addition of ultraviolet counterterms required to reach the renormalized field theory in the continuum limit. The construction is tested numerically for the two-dimensional scalar field theory on the Riemann two-sphere, , in comparison with the exact solutions to the two-dimensional Ising conformal field theory (CFT). Numerical results for the Binder cumulants (up to 12th order) and the two- and four-point correlation functions are in agreement with the exact CFT solutions.- 6 July 2018
- Published by American Physical Society (APS) in Physical Review D
- Vol. 98 (1) , 014502
Open AccessLatticefield theory on Riemann manifolds: Numerical tests for the 2D Ising CFT onWe present a method for defining a lattice realization of the quantum field theory on a simplicial complex in order to enable numerical computation on a general Riemann manifold. The procedure begins with adopting methods from traditional Regge calculus (RC) and finite element methods (FEM) plus the addition of ultraviolet counterterms required to reach the renormalized field theory in the continuum limit. The construction is tested numerically for the two-dimensional scalar field theory on the Riemann two-sphere, , in comparison with the exact solutions to the two-dimensional Ising conformal field theory (CFT). Numerical results for the Binder cumulants (up to 12th order) and the two- and four-point correlation functions are in agreement with the exact CFT solutions. - Open Access
- 28 June 2018
- Published by American Physical Society (APS) in Physical Review D
- Vol. 97 (11) , 114513
Multigrid algorithm for staggered lattice fermions
Critical slowing down in Krylov methods for the Dirac operator presents a major obstacle to further advances in lattice field theory as it approaches the continuum solution. Here we formulate a multigrid algorithm for the Kogut-Susskind (or staggered) fermion discretization which has proven difficult relative to Wilson multigrid due to its first-order anti-Hermitian structure. The solution is to introduce a novel spectral transformation by the Kähler-Dirac spin structure prior to the Galerkin projection. We present numerical results for the two-dimensional, two-flavor Schwinger model; however, the general formalism is agnostic to dimension and is directly applicable to four-dimensional lattice QCD.
- 28 June 2018
- Published by American Physical Society (APS) in Physical Review D
- Vol. 97 (11) , 114513
Open AccessMultigrid algorithm for staggered lattice fermions
Critical slowing down in Krylov methods for the Dirac operator presents a major obstacle to further advances in lattice field theory as it approaches the continuum solution. Here we formulate a multigrid algorithm for the Kogut-Susskind (or staggered) fermion discretization which has proven difficult relative to Wilson multigrid due to its first-order anti-Hermitian structure. The solution is to introduce a novel spectral transformation by the Kähler-Dirac spin structure prior to the Galerkin projection. We present numerical results for the two-dimensional, two-flavor Schwinger model; however, the general formalism is agnostic to dimension and is directly applicable to four-dimensional lattice QCD.
- Open Access
- 23 June 2017
- Published by American Physical Society (APS) in Physical Review D
- Vol. 95 (11) , 114510
Lattice Dirac fermions on a simplicial Riemannian manifold
The lattice Dirac equation is formulated on a simplicial complex which approximates a smooth Riemann manifold by introducing a lattice vierbein on each site and a lattice spin connection on each link. Care is taken so the construction applies to any smooth D-dimensional Riemannian manifold that permits a spin connection. It is tested numerically in 2D for the projective sphere in the limit of an increasingly refined sequence of triangles. The eigenspectrum and eigenvectors are shown to converge rapidly to the exact result in the continuum limit. In addition comparison is made with the continuum Ising conformal field theory on . Convergence is tested for the two point, , and the four point, , correlators for the energy, , and twist operators, , respectively.- 23 June 2017
- Published by American Physical Society (APS) in Physical Review D
- Vol. 95 (11) , 114510
Open AccessLattice Dirac fermions on a simplicial Riemannian manifold
The lattice Dirac equation is formulated on a simplicial complex which approximates a smooth Riemann manifold by introducing a lattice vierbein on each site and a lattice spin connection on each link. Care is taken so the construction applies to any smooth D-dimensional Riemannian manifold that permits a spin connection. It is tested numerically in 2D for the projective sphere in the limit of an increasingly refined sequence of triangles. The eigenspectrum and eigenvectors are shown to converge rapidly to the exact result in the continuum limit. In addition comparison is made with the continuum Ising conformal field theory on . Convergence is tested for the two point, , and the four point, , correlators for the energy, , and twist operators, , respectively.