To accelerate the solution of large eigenvalue problems arising from many-body calculations in nuclear physics on distributed-memory parallel systems equipped with general-purpose Graphic Processing Units (GPUs), we modified a previously developed hybrid MPI/OpenMP implementation of an eigensolver written in FORTRAN 90 by using an OpenACC directives based programming model. Such an approach requires making minimal changes to the original code and enables a smooth migration of large-scale nuclear structure simulations from a distributed-memory many-core CPU system to a distributed GPU system. However, in order to make the OpenACC based eigensolver run efficiently on GPUs, we need to take into account the architectural differences between a many-core CPU and a GPU device. Consequently, the optimal way to insert OpenACC directives may be different from the original way of inserting OpenMP directives. We point out these differences in the implementation of sparse matrix-matrix multiplications (SpMM), which constitutes the main cost of the eigensolver, as well as other differences in the preconditioning step and dense linear algebra operations. We compare the performance of the OpenACC based implementation executed on multiple GPUs with the performance on distributed-memory many-core CPUs, and demonstrate significant speedup achieved on GPUs compared to the on-node performance of a many-core CPU. We also show that the overall performance improvement of the eigensolver on multiple GPUs is more modest due to the communication overhead among different MPI ranks.
We present a systematic investigation of few-nucleon systems and light nuclei using the current LENPIC interactions comprising semilocal momentum-space regularized two- and three-nucleon forces up to third chiral order (N$^2$LO). Following our earlier study utilizing the coordinate-space regularized interactions, the two low-energy constants entering the three-body force are determined from the triton binding energy and the differential cross section minimum in elastic nucleon-deuteron scattering. Predictions are made for selected observables in elastic nucleon-deuteron scattering and in the deuteron breakup reactions, for properties of the $A=3$ and $A=4$ nuclei, and for spectra of $p$-shell nuclei up to $A = 16$. A comprehensive error analysis is performed including an estimation of correlated truncation uncertainties for nuclear spectra. The obtained predictions are generally found to agree with experimental data within errors. Similar to the coordinate-space regularized chiral interactions at the same order, a systematic overbinding of heavier nuclei is observed, which sets in for $A sim 10$ and increases with $A$.
The light-front wave functions of hadrons allow us to calculate a wide range of physical observables; however, the wave functions themselves cannot be measured. We discuss recent results for quarkonia obtained in basis light-front quantization using an effective Hamiltonian with a confining model in both the transverse and longitudinal directions and with explicit one-gluon exchange. In particular, we focus on the numerical convergence of the basis expansion, as well as the asymptotic behavior of the light-front wave functions. We also illustrate that, for mesons with unequal quark masses, the maxima of the light-front wave functions depend in a non-trivial way on the valence quark-mass difference.
We investigate selected static and transition properties of $^{12}$C using ab initio No-Core Shell Model (NCSM) methods with chiral two- and three-nucleon interactions. We adopt the Similarity Renormalization Group (SRG) to assist convergence including up to three-nucleon (3N) contributions. We examine the dependences of the $^{12}$C observables on the SRG evolution scale and on the model-space parameters. We obtain nearly converged low-lying excitation spectra. We compare results of the full NCSM with the Importance Truncated NCSM in large model spaces for benchmarking purposes. We highlight the effects of the chiral 3N interaction on several spectroscopic observables. The agreement of some observables with experiment is improved significantly by the inclusion of 3N interactions, e.g., the B(M1) from the first $J^pi T = 1^+ 1$ state to the ground state. However, in some cases the agreement deteriorates, e.g., for the excitation energy of the first $1^+ 0$ state, leaving room for improved next-generation chiral Hamiltonians.
