| QMCPACK | |
|---|---|
| Name | QMCPACK |
| Developer | Oak Ridge National Laboratory; Lawrence Livermore National Laboratory; Argonne National Laboratory; academic contributors |
| Released | 2008 |
| Programming language | Fortran, C++ |
| Operating system | Linux, macOS |
| Genre | Scientific simulation software, Quantum Monte Carlo |
| License | Open-source |
QMCPACK
QMCPACK is an open‑source software package for performing high‑accuracy electronic structure calculations using Quantum Monte Carlo (QMC) methods. It is designed to compute ground‑state and excited‑state properties of atoms, molecules and solids with explicit many‑body wavefunctions, providing benchmark‑quality results that complement methods such as Density functional theory and Coupled cluster theory. QMCPACK matters in quantum physics and materials science because it enables controlled, systematically improvable treatment of electron correlation for problems of fundamental and technological importance, supporting research at institutions like Oak Ridge National Laboratory and Lawrence Livermore National Laboratory.
QMCPACK was created to deliver scalable, production‑grade QMC capabilities targeting high‑performance computing facilities such as Oak Ridge Leadership Computing Facility and Argonne Leadership Computing Facility. Its purpose is to provide reliable many‑body calculations for electronic structure, addressing limitations of mean‑field and perturbative approaches when strong correlation, van der Waals forces, or fixed‑node errors are significant. The code is used by researchers in condensed matter physics, quantum chemistry, and materials engineering to obtain reference energies, excitation gaps, and properties relevant to emergent phenomena and materials design.
QMCPACK implements a suite of QMC algorithms rooted in the variational and projector formulations of many‑body quantum mechanics. Primary methods include Variational Monte Carlo (VMC) and Diffusion Monte Carlo (DMC), which sample the high‑dimensional electronic wavefunction to evaluate expectation values. The code supports trial wavefunctions built from Slater determinants, Jastrow factors, multi‑determinant expansions, and backflow transformations to reduce fixed‑node errors. QMCPACK interfaces with basis sets and pseudopotentials, leveraging norm‑conserving pseudopotentials and effective core potentials developed in quantum chemistry. Its theoretical framework connects to fundamental concepts such as the Born–Oppenheimer approximation, electron correlation, and stochastic evaluation of integrals, providing a bridge between many‑body theory and practical computation.
QMCPACK is implemented in a combination of modern C++ and Fortran, optimized for distributed memory parallelism using Message Passing Interface (MPI) and thread parallelism with OpenMP. The architecture separates trial wavefunction evaluation, propagators, Hamiltonian operators, and accumulation of observables to facilitate modular development and testing. QMCPACK includes interfaces to electronic structure pre‑processors like Quantum ESPRESSO and PySCF for converting mean‑field orbitals into QMC trial wavefunctions, and to libraries such as Libxc for exchange‑correlation functionals where needed. The code supports heterogeneous architectures with GPU acceleration using CUDA and performance portability efforts through projects like Kokkos. Test suites, continuous integration, and code review processes are organized via repositories hosted on platforms used by the scientific community.
QMCPACK is applied to benchmark cohesive energies, reaction barriers, band gaps, and quasiparticle properties for systems where correlation is pivotal. Notable application areas include transition metal oxides, low‑dimensional materials such as graphene and transition metal dichalcogenides, actinide and lanthanide compounds, and catalytic surfaces. It provides reference data for developing improved exchange–correlation functionals in DFT and for validating quantum embedding methods such as dynamical mean field theory. Community projects use QMCPACK results to inform materials design efforts in energy storage, superconductivity research, and semiconductor device modeling at institutions including University of Illinois Urbana–Champaign and Princeton University.
Performance engineering for QMCPACK targets scalability to tens of thousands of cores on leadership‑class supercomputers; published scaling studies demonstrate efficient weak and strong scaling on systems such as Summit and Titan. Validation protocols compare QMCPACK results against high‑level quantum chemistry methods (e.g., CCSD(T)) and experimental data where available. Benchmarks routinely include molecules from the G2 and S22 test sets, solid‑state cohesive energies, and excitation energies for prototypical materials. Error analysis emphasizes fixed‑node bias, time‑step errors, pseudopotential accuracy, and statistical uncertainty, with community standards for reproducibility and data provenance.
QMCPACK development is collaborative, involving national laboratories, universities, and user contributors. Governance follows an open development model with steering committees and contributor agreements that balance stable releases with research‑driven features. The project coordinates with related initiatives such as the Quantum Monte Carlo Working Group and contributes to community workshops and conferences like the ACM/IEEE Supercomputing Conference and domain‑specific meetings. Training materials, tutorials, and example input sets are maintained to support adoption by practitioners and students, fostering continuity between academic research and national research priorities.
QMCPACK is distributed under an open‑source license that permits academic and industrial use while encouraging citation of foundational publications. Source code, issue trackers, and release artifacts are hosted on public repositories to maximize transparency and reproducibility. Binary builds and container images are provided for common HPC environments, and users are encouraged to cite key papers and datasets when publishing results obtained with QMCPACK.
Category:Quantum Monte Carlo Category:Computational physics software Category:Scientific simulation software