Skip to content

Latest commit

 

History

History
1084 lines (900 loc) · 67.9 KB

CHANGELOG.md

File metadata and controls

1084 lines (900 loc) · 67.9 KB
Raw

Change Log

Notable changes to QMCPACK are documented in this file.

[Unreleased]

  • Support for backflow optimization has been removed as part of refactoring and cleaning the codebase. QMC runs using backflow wavefunctions are still supported. This feature is expected to eventually be reimplemented in v4. Users needing backflow optimization can use previously released versions of QMCPACK or work towards its reimplementation in the modern code. #4688

[3.17.1] - 2023-08-25

This minor release is recommended for all users and includes a couple of build fixes and a NEXUS improvement.

  • Improved HDF5 detection. Fixes cases where HDF5 was not identified by CMake, including on FreeBSD (thanks @yurivict for the report). #4708
  • Fix for building with BUILD_UNIT_TESTS=OFF. #4709
  • Add timer for orbital rotations. #4706

NEXUS

  • NEXUS: Support for spinor inputs. #4707

[3.17.0] - 2023-08-18

This is a recommended release for all users. Thanks to everyone who contributed directly, reported an issue, or suggested an improvement. There are many quality of life improvements, bug fixes throughout the application, and updates to the associated testing. As previously announced, the legacy CUDA support (QMC_CUDA=1) is removed in this version. For GPU support, users should transition to the offload code which is more capable and fully usable in production on NVIDIA GPUs.

This version is intended for long-term support of v3 of QMCPACK. Development effort is now focused towards v4. Contributions of tests, fixes, and features from users and developers are still welcome to v3 for a potential future release. However, these will not be ported towards v4 by the core QMCPACK developers without prior arrangement. Please discuss options with QMCPACK developers.

  • Simplified checkpointing and enabled it in the batched drivers. Users now only need specify checkpoint={-1,0,N} to checkpoint between blocks. #4646
  • NERSC Perlmutter build recipe. #4698
  • qmc-fit: Now supports parameter fitting with jackknife for e.g. DFT+U, EXX scans #4475 and for equation of states and morse fits #4518
  • Improved error checking including NaN checks to protect against potentially unreliable compilers and libraries, #4697, and checks on GPU matrix inversion #4693
  • Significant advances in orbital optimization capability, focusing on LCAO wavefunctions. Development is ongoing for multideterminant support and for spline wavefunctions. See e.g. the Be atom orbital optimization test #4626, #4619, reading and writing of orbital rotation parameters #4580, support for disabled/frozen parameters #4581.
  • Magnetization Density Estimator for non-collinear wavefunctions #4531
  • Pathak-Wagner regularizer for forces #4477
  • The legacy CUDA implementation, the version built with QMC_CUDA=1, has been removed from the codebase, #4431, #4632,#4499, #4442.
  • For increased performance with current AMD GPU support, new QMC_DISABLE_HIP_HOST_REGISTER option is enabled by default for ROCm/HIP builds. #4674
  • Bugfix: J1Spin indexing was wrong #4612
  • Bugfix: 1RDM estimator data written to stat.h5 was incorrect #4568
  • Introduced ENABLE_PPCONVERT option and skip ppconvert compilation when cross compiling. #4601
  • Faster builds compared to v3.16.0 due to code refactoring #4682
  • Many refinements throughout the codebase, cleanup, improved testing.

NEXUS

  • Nexus: Equilibration detection algorithm is now deterministic #4557
  • Nexus: Support for Kagayaki cluster at JAIST #4598
  • Nexus: GPU support fix for NERSC/Perlmutter #4699
  • Nexus: Use simplices in convex_hull to support newer scipy versions #4671
  • Nexus: Add pdos flag for Projwfc #4655
  • Nexus: Adding crowds_serialize_walkers tag to dmc input list #4651
  • Nexus: Qdens handles batched driver input/output #4645
  • Nexus: Fix namelist read for Projwfc input #4644

Known problems

  • When offload builds are compiled with CUDA toolkit versions above 11.2 using LLVM, multideterminant tests and functionality will fail, seemingly due to an issue with the toolkit. This is discussed in llvm/llvm-project#54633 . All other functionality appears to work as expected. As a workaround, the CUDA toolkit 11.2 can be used. The actual NVIDIA drivers can be more recent.

[3.16.0] - 2023-01-31

This release contains important bug fixes as well as feature improvements. It is a recommended release for all users. Thanks to everyone who reported an issue or suggested an improvement. See GitHub for the full list of merged pull requests and closed issues.

This release is expected to be the last including the legacy CUDA implementation, the version built with QMC_CUDA=1. Users should transition to the batched drivers which support greater functionality as well as both CPU and GPU execution. Users should adopt these drivers now and report any issues. The new drivers can be requested with the driver_version input parameter, see https://qmcpack.readthedocs.io/en/develop/performance_portable.html . In a subsequent release, the non-batched CPU drivers will also be removed leaving only the performance portable batched drivers. This will result in a single implementation of most functionality, improving overall usability and maintainability.

  • Important bugfix to NLPP integration grid rotations and update to all relevant deterministic test values. See issue #4362 for full discussion and visualization. Found and corrected by @markdewing, this bug has existed since the earliest days of QMCPACK. The stochastic rotations used to randomly reorient the integration grids for the non-local pseudopoptentials would not cover the full sphere unless they had many points and sufficient symmetry, as was the case for the QMCPACK default. However, calculations with custom integration grids with only a few points (small nrule) could show error or excess statistical noise in the non-local part of the pseudopotential energy. Standard calculations and tests on carbon diamond, lithium hydride, and hydrocarbon molecules were not affected due to QMCPACK's conservative defaults. Tests updated in #4383
  • NLPP grid randomization can be disabled for debugging and greater reproducibility #4394
  • Two-body Jastrow support for true 2D calculations #4289 (contributed by @Paul-St-Young)
  • Fix for very large calculations requesting too large grids in CUDA spline implementation #4421 (contributed by @pwang234)
  • Bugfix in the batched OpenMP offload implementation memory errors #4408 when the number of splines is not a perfectly aligned size (multiple of 8 single precision or 4 double precision).
  • Updates to test tolerances for many build types and platforms to improve reliability of deterministic tests. Goal: ctest -L deterministic should pass on all platforms. Please report any failures.
  • Improved CMake configuration including detecting use of parallel HDF5 in non-MPI builds #4420 and detection of missing OpenMP support #4422
  • Optimization of spinor wavefunctions with spin-orbit and pseudopotentials re-enabled #4418
  • QMCPACK output now indicates status of QMC_COMPLEX #4412
  • Initial work for eventual GPU offloading of Gaussian basis wavefunctions for molecules and solids #4407
  • Bugfix to support one-body Jastrow functions where only a subset of elements is given #4405
  • Electron coordinates are printed in case a NaN is detected #4401
  • To evade support problems for complex reductions in OpenMP offload compilers, real builds no longer reference any complex reductions #4379
  • Enabled HIP as language in CMake (requires >= 3.21) #3646. When using HIP targeting AMD GPUs, replace HIP_ARCH with CMAKE_HIP_ARCHITECTURES if HIP_ARCH was used to specify the GPU architectures.
  • Refinements to SYCL usage, e.g., #4384, #4382, #4380
  • Many expanded tests including for NLPP parameter derivatives #4394, more boundary conditions in distance tables #4374, for reptation Monte Carlo observables #4327, and orbital rotations #4304
  • Many updates to HDF5 usage including adoption of HDF5 1.10 API #4352 and related cleanup, e.g. #4300
  • Initial Perlmutter CPU build recipe #4398
  • Initial ALCF Sunspot build recipe including offloading to Intel Ponte Vecchio/Xe HPC GPU #4391
  • Better support for FreeBSD #4416
  • Minimum supported Intel classic compiler version is 2021.1. #4389
  • Ongoing improvement to orbital optimization and rotation, e.g. #4288, #4402
  • Ongoing code cleanup, e.g. #4276, #4275, #4273
  • Updated bmpi3 MPI "wrapper"
  • Various other small bug fixes and quality of life improvements. See the full list of merged PRs on GitHub for details.

Known problems

  • When offload builds are compiled with CUDA toolkit versions above 11.2 (tested 11.3-11.8) using LLVM15, multideterminant tests and functionality will fail, seemingly due to an issue with the toolkit. This is discussed in llvm/llvm-project#54633 . All other functionality appears to work as expected. As a workaround, the CUDA toolkit 11.2 can be used. The actual NVIDIA drivers can be more recent.
  • CUDA toolkit version 12.0 is not compatible with LLVM OpenMP offload llvm/llvm-project#60296

NEXUS

  • Nexus: Support for use of templates for job submission scripts #4344
  • Nexus: twist_info.dat files now added to results directory for easier analysis of twist average quantities #4302
  • Nexus: Initial support for Polaris at ALCF #4354
  • Nexus: Initial support for Perlmutter at NERSC #4356
  • Nexus: Support for gpusharing keyword for legacy CUDA #4403
  • Nexus: Support for handling multiple pickle protocols #4385
  • Nexus: CPU/GPU flags for batched code #4341
  • Nexus: Jastrow factors can be read from existing files #4339
  • Nexus: Fix VASP POSCAR write #4331
  • Nexus: Better handling of VASP pseudopotentials #4330

Known problems

  • The new QE7.1 DFT+U input style is not yet supported #4100

[3.15.0] - 2022-09-29

This is a recommended release for all users. There are many quality of life improvements, bugfixes throughout the application, and updates to the associated testing. Thanks to everyone who reported an issue or suggested an improvement.

We are working to make the performance portable "batched drivers" the default in an upcoming version. These support execution on CPUs and multiple GPU architectures with high performance. Most standard QMC calculations and many observables are already supported. Because some changes to the input files will be required, we recommend trying these drivers now and reporting any issues.

  • Important bug fix to excited states in splines when spin-up/down sets are built from the same spin species and occupation is specified on the first sposet #4158
  • The Quantum ESPRESSO converter, pw2qmcpack is now supported via a plugin activated via -DQE_ENABLE_PLUGINS=pw2qmcpack on the QE CMake configure line, see https://qmcpack.readthedocs.io/en/develop/installation.html#quantum-espresso-7-0 . The latest QE 7.1 and earlier 7.0 are supported, and new versions should be automatically compatible.
  • Substantial improvements to the performance portable / batched implementation. Using LLVM 15.0, high performance production calculations can be performed on NVIDIA GPUs for several wavefunction types, in addition to running on all CPU systems.
  • As introduced in v3.14.0, the optional project data input parameter driver_version specifies whether legacy or batched drivers are used. In future versions of QMCPACK this tag will be required to avoid ambiguity and allow e.g. the batched VMC driver to be obtained via vmc in addition to vmc_batch. See https://qmcpack.readthedocs.io/en/develop/methods.html#transition-from-classic-drivers
  • Non-local pseudopotential energy contributions are consistently included in the objective function used for optimization, improving convergence and achievable wavefunction quality e.g. #4177
  • Support for multistep wavefunction optimization, specifying different parameter sets to be frozen at each step. #4169
  • Parameter filtration during optimization based on statistical uncertainties #4126
  • Support for 2D HEG calculations (e.g. #4084 )
  • Pseudopotential non-local channel can be specified in input #4032
  • Use of deprecated CUDA texture API removed for greater compatibility #4022
  • Optimization has been removed from the legacy CUDA code. New calculations needing GPU support should use the batched drivers and their GPU capabilities for optimization. #4138
  • Initial version of determinant update in SYCL for Intel architectures (e.g. #4118 )
  • Updated walker counts in several of the performance tests. Due to the changed but more representative workloads, new performance timings should not be compared with older runs (e.g. #4112 )
  • Maximum system sizes run in the performance tests can be specified in CMake via QMC_PERFORMANCE_NIO_MAX_ATOMS, QMC_PERFORMANCE_C_GRAPHITE_MAX_ATOMS, and QMC_PERFORMANCE_C_MOLECULE_MAX_ATOMS (e.g. #4134 )
  • Readability refinements in the output, e.g. #4149
  • UHF/UKS support in PySCF converter #4089
  • Example installation scripts for more machines placed in config directory, including Archer2, and Polaris.
  • Many improvements in testing including additional tests, better reliability, and bug fixes.
  • Minimum version of CMake is now v3.17.0 for CPU builds. For GPU builds, more recent versions may be required. Use of the latest CMake version is generally recommended.
  • Minimum CUDA version is 11.0 #3957
  • Minimum version of GCC is now v9.

NEXUS

  • Nexus: support to current batched driver style. Example inputs for batched runs using trial wavefunctions from QE are included in examples/qmcpack/rsqmc_quantum_espresso #4246
  • Nexus: add override_vp_parameters element #4245
  • Nexus: fix convert4qmc hdf5 issue #4243
  • Nexus: extend angular channels for pseudopotentials up to l_max=21 #4148
  • Nexus: Pass PYTHONPATH recorded at cmake step to nxs-test to ensure tests run #3935
  • Nexus: Support for VASP keywords to version 6.3 #4056
  • Nexus: Adding docs for limiting the number of simultaneously submitted jobs to a queue #4133

[3.14.0] - 2022-04-06

This release focuses on performance improvements to the OpenMP target offload version for GPUs as well as ongoing minor improvements. The new GPU implementation rivals the legacy CUDA version for performance for broad range of problems while offering more functionality, such as three body Jastrow functions. Developers are very interested in feedback from users about the new version and will prioritize developments based on comments received. A new driver_version switch is introduced, currently optional, to disambiguate between the versions and their inputs.

  • New global driver_version switch to select between batched and legacy codes. This will become a required input tag in the next major release series of QMCPACK, but remains optional in 3.x versions #3897
  • Optimization of block sizes in GPU offload kernels #3910
  • GPU Offload of one-body Jastrow ratio calculation in pseudopotential evaluation #3905
  • GPU Offload of some Coulomb potential evaluations #3842
  • Partial GPU offload of multideterminant evaluation e.g. #3892
  • Increased performance via more selective distance table computation #3846
  • Improved performance on AMD GPUs via rocSOLVER integration #3756
  • HIP build options shown in output #3919
  • Documentation improvements, particularly relating to installation.
  • Various bug fixes and ongoing cleanup.

NEXUS

  • Nexus: proper use of max_seconds in legacy drivers #3877

[3.13.0] - 2022-02-16

Notes

This release incorporates support for trial wavefunctions from Quantum ESPRESSO 7.0 and adds GPAW support for the first time. Non-local pseudopotential derivatives are fully supported in the optimizer and recommended in standard calculations.
Numerous minor bug fixes, test and installation improvements have been made. Behind the scenes updates include maturation of the OpenMP target offload implementation and the batched drivers, a partial implementation of fast force calculations, and ongoing modernization of the code. This is a recommended release for all users.

  • Support for Quantum ESPRESSO (QE) 7.0 #3683
  • Support for GPAW and GPAW to QMCPACK converter #3490
  • use_nonlocalpp_deriv is fully supported and preferred in optimization #3785 and others.
  • Save and restore of variational parameters during optimization #3640
  • Twist attribute takes precedence over twistnum. Twist is preferred specification. #3799
  • Fixed inconsistent twist directions between electron gas and spline wavefunctions #1386
  • Fixed reported Madelung constant in CoulombPBCAA #3806
  • More robust computation of reference ion-ion Coulomb energy #3763
  • Expanded test set, including more coverage of plane-wave basis sets and complex molecules #3105, #3822
  • More consistent python invocations #3680
  • Builds with OpenMP disabled (QMC_OMP=0) again supported #3723
  • Modernization of HDF5 usage e.g. #3705
  • Minimum supported Intel classic compiler version is 19.1. #3747
  • Various minor bug fixes and ongoing code cleanup.

NEXUS

  • Nexus: Add --user $USER to squeue command #3796
  • Nexus: Add Example and tests for qdens-radial tool #3676
  • Nexus: Add Lowdin example #3666
  • Nexus: Fixed Nexus 'install' target #3720
  • Nexus: Harden Nexus excitation checks #3729
  • Nexus: Small fix to excitation checks #3701
  • Nexus: Faster configuration time #3706

[3.12.0] - 2021-12-08

Notes

This release incorporates several hundred changes to QMCPACK and the supporting ecosystem. It is a recommended release for all users. Note that compilers supporting C++17 and CMake version 3.15 or newer are now required. Changes include newly added support for the DIRAC quantum chemistry code, the RMG-DFT code, and updates for the latest version of Quantum ESPRESSO. Through DIRAC it is now possible to perform highly accurate molecular calculations incorporating spin-orbit with multideterminant trial wavefunctions. Behind the scenes updates include increased checking of inputs, fixes to many edge case bugs, and removal of memory leaks in both QMCPACK and the various converters. In readiness for transition to the new batched drivers that support both CPU and GPU execution, more features are supported and performance improved. Test coverage and robustness is improved in all areas. For developers, tests, sanitizers, and code coverage are now run on Pull Requests using GitHub Actions.

  • To aid coexistence of real and complex builds, the qmcpack executable is now named qmcpack_complex for builds with QMC_COMPLEX=1
  • Added DIRAC converter and support for MSD wave functions #3510
  • Spin-Orbit implementation completed #1770
  • Quantum ESPRESSO (QE) v6.8 support #3301
  • Support for RMG DFT code #3351
  • CMake 3.15 minimum required #3492
  • C++17 is required #3348
  • CMake CUDA support uses modern FindCUDAToolkit #3460
  • Support latest Sphinx-contrib BibTeX 2.x #3176
  • One Body Density Matrices supported in batched drivers #3622
  • Batched performant Slater matrix inverses #3470
  • Safeguards for requesting more orbitals than the input h5 provide #2341
  • Implemented One-body spin-dependent Jastrow #3257
  • Fixes for low particle counts, such as using a two body Jastrow with more than 2 particle types but only one particle of each type #3137
  • ppconvert is built by default #3143
  • Documentation on revised input format where SPO sets are created outside the determinant #3456

NEXUS

  • Add Density functionality to qdens tool #3541
  • Add new qdens-radial tool for radial analysis of densities #3587
  • Radial density of requested species only #3099
  • Extend structure plotting capabilities for 2D materials #3220
  • Support grand-canonical twist averaging #3153
  • Extend excitations to allow 'lowest' gap #3628
  • Allow singlet/triplet excitation types #2290
  • Allow bandstructure plotting with custom k-path #3293
  • Generate PySCF inputs without a template #3550
  • Add punch extension for GAMESS analysis #3433
  • Read pseduopotentials in numhf format (Eric Shirley's numerical HF code) #3097
  • Add L2 generation functionality #3079
  • Support QMCPACK batched drivers #2901
  • Make qdens test more informative #3593
  • Resource lock Nexus examples for reliable parallel execution #3585
  • Support running tests without mpirun available #3584
  • Small fix for custom band plotting #3566
  • Improve error handling for bad Jastrow requests #3554
  • Fix sizing problem in some single atom workflows #3553
  • Fix syntax warnings #3497
  • Fix convert4qmc usage #3495
  • Verify cif2cell is available before running ntest_nexus_structure #3511
  • Fix to add_L2 function in pseudopotential.py #3386
  • Expand eshdf features #3334
  • Add delay_rank input #3218
  • Add max_seconds input #3159
  • Add Tref (initial tilematrix) argument to optimal_tilematrix #3141
  • Use OS environment by default #3108

[3.11.0] - 2021-04-09

Notes

This release includes a large number of refinements to QMCPACK and the supporting ecosystem. These include support for the latest version of Quantum ESPRESSO, new capabilities in AFQMC, space-warp transformation for forces, numerous bug fixes, user-requested feature improvements, and further upgrades to the test system.

  • Quantum ESPRESSO (QE) v6.7 support. #2927.
  • Detect and automatically use patched version of QE found on the PATH. #2974.
  • Support for global max_seconds and STOP file to cleanly halt QMCPACK during a run. #3028.
  • Freezing of two-body Jastrow parameters in optimization works. #2814.
  • Multideterminant code now works with only alpha determinants (no down electrons). #2698.
  • High l-momentum channels as local channels in ECPs work. #2920.
  • Space Warp Transformation for ZVZB Forces. #2828.
  • Important bug fixes in legacy CUDA implementation causing incorrect energies. #2883.
  • Implemented DLA in legacy CUDA. #2887.
  • Updates to support CUDA 11.2.1 e.g. #2950.
  • AFQMC supports energy estimator with different Hamiltonian (from propagation). #2795.
  • Trial wavefunction optimization with spin-orbit supported. #3034.
  • ppconvert executable automatically built when configured. #2904.
  • Tests added for ppconvert. #2929.
  • Fixed SIMD alignment for AVX512 on some systems. #2981.
  • Improved wavefunction restart logic in AFQMC. #2942.
  • Spin-density supported in batched code. #2840.
  • Reduced I/O operations during cmake. #2808.
  • Improved detection of unsupported-by-Intel combinations of Intel compilers and libstdc++. #2794.
  • Initial support for Andes at OLCF. #3073.
  • Deterministic tests expanded in scope and made reliable for more build types and compilers.
  • Various minor bug fixes and feature improvements based on user requests for both real-space and AFQMC.
  • Improved error handling throughout.
  • Numerous performance improvements, expansion of tests, and bug fixes to the batched VMC and DMC codes. Reasonable but not optimal GPU acceleration can now be achieved for spline-based wavefunctions.

NEXUS

  • Support AMD nodes on Cori. #2809.
  • Interface for RMG code. #2932.
  • Added h-channel to list of possible local channels in pseudopotential. #2915.
  • Allow non spin-specific occupations in case of noncollinear. #2957.
  • More robust handling of QE output when printed eigenvalues touch. #3042.
  • Fixed type check for reblock_factors in qmc-fit. #2830.
  • Fixed a Jastrow read error/warning, add several QE inputs. #2819.
  • Fixed tests on Summit. #2983.
  • Fixed module overwrite bug in qmca. #2802.

[3.10.0] - 2020-11-10

Notes

This release contains multiple feature improvements and bug fixes. The AFQMC implementation has been significantly enhanced, and an important wavefunction optimization bug fixed in real space QMC.

  • The QMCPACK manual is now available at https://qmcpack.readthedocs.io, having been converted to use reStructuredText and the sphinx documentation system.
  • Significant improvements to the AFQMC code including HIP support for AMD GPUs, updated documentation, and support for non-collinear calculations and spin-orbit k-point Hamiltonians #2734.
  • Improved support for spin-orbit in real-space QMC including documentation #2733.
  • Important bug fix for wavefunction optimization in few electron systems such as isolated atoms. The bug would result in slow or no convergence. Thanks to Jaron Krogel and Matus Dubecky for reports and reproducers. #2496.
  • Implementation of L2 potentials and evaluation in DMC #1948.
  • Consistent with our two year support policy for open source compilers, libraries, and tooling, several version minimums have been increased to either avoid bugs or to utilize new features.
  • Clang 7 is the earliest supported Clang compiler. The latest release is recommended.
  • Intel 2019 is the earliest supported Intel compiler. The latest release is recommended.
  • Future releases of QMCPACK will require C++17. The current minimum is C++14.
  • AoS builds are no longer supported. The code has been removed now that the default structures-of-arrays (SoA) build has sufficiently broad capability.
  • The default CUDA architecture is set to sm_70 (Volta).
  • QMCPACK is built with ENABLE_TIMERS=ON by default #2663
  • Various bug fixes to complete the transition to Python 3.
  • Ongoing improvements to the OpenMP offload implementation.

NEXUS

  • NEXUS manual is now available at https://nexus-workflows.readthedocs.io, having been converted to use the reStructuredText and sphinx documentation system.
  • Various small fixes and improvements.

[3.9.2] - 2020-04-29

Notes

This is an important bug fix release. As described in #2330, since v3.8.0 the timestep was not correctly changed between different DMC blocks if the time step was changed in the input, biasing the results. Runs using a single time step were not affected. Thanks to Chandler Bennett for identifying the problem.

  • Bug fix: timestep was not correctly changed between DMC blocks if it was changed in the input #2330.
  • qmcfinitesize tool added #2329.
  • QMCPACK spack package now supports AFQMC #2237.
  • Improvements to deterministic tests: these are now fully reliable other than for some CUDA builds and some mixed precision CPU configurations.
  • Many improvements to cmake configuration for faster builds, e.g. #2389.
  • Ongoing source cleanup and fewer compile-time warnings, e.g. #2375.

[3.9.1] - 2020-02-11

Notes

This release is the same as v3.9.0 except that the version number of QMCPACK is reported correctly. See the v3.9.0 part of the CHANGELOG for important changes compared to v3.8.0.

[3.9.0] - 2020-02-11

Notes

This release includes a large number of refinements to improve or extend the functionality of QMCPACK and NEXUS. Importantly, this release supports and requires Python 3. After this release we plan to remove the array-of-structures build configuration and also the legacy CUDA implementation for GPUs. If any needed functionality is not supported by the now-default structures-of-arrays configuration, users should contact the developers via the QMCPACK Google Groups or via an issue on the QMCPACK GitHub repository. Work is ongoing to support dynamical spin variables, implement spin-orbit, and to develop new support for accelerators via a new framework that will consistently support CPUs and GPUs from the same codebase.

  • All uses of Python updated to Python 3, which is now required. Python 2 was retired at the end of 2019, and many packages already only support Python 3.
  • A greatly expanded selection of effective core potentials is available at https://pseudopotentiallibrary.org/ in formats suitable for QMCPACK and common DFT and quantum chemistry codes.
  • All major functionality is now supported by the default structures-of-arrays (SoA) build. This release is the last to support the legacy array-of-structures (AoS) build. See #861.
  • Major bug identified and fixed in the periodic Coulomb evaluation (Optimized breakup method of Natoli-Ceperley). Many thanks to Jan Brndiar and coworkers for reporting this. For large anisotropic supercells such as a graphene layer with substantial vacuum, the ion-ion potential was incorrectly computed. Results in all bulk-like supercells tested so far have been accurate. An independent Ewald check of the ion-ion potential evaluation has been added. See #2137. The Coulomb potential evaluation has also been found to converge very slowly for certain anisotropic supercells, particularly for quasi-2D cells where huge errors can result. The new independent Coulomb check will abort if a tolerance is not reached and provide guidance. Research is ongoing to develop an improved methodology #2185.
  • Support for periodic gaussian-based trial wavefunctions at complex k-points #1988.
  • Determinant-localization approximation (DLA) of Zen et al. J. Chem. Phys. 151, 134105 (2019) for DMC non-local pseudopotential evaluation implemented.
  • Improved force implementation #1769, #1768.
  • Non-local pseudopotential derivatives are supported in the SoA build and recommended for all optimizations #2083.
  • Above 192 electrons in a spin determinant, delayed updating with delay 32 is enabled by default for higher performance, #2027. Rank-1 updating is used by default for smaller determinants.
  • Improved configuration and detection of Intel MKL and vector MKL when used with non-Intel compilers.
  • QMCPACK will now run with wavefunctions where only electrons of a single spin are specified. #2148.
  • AFQMC estimators now include 1 and 2 body reduced density matrices (1RRM, 2RDM) and on-top pair density. #2097.
  • Dense real hamiltonian added for AFQMC allowing for GPU acceleration for chemistry applications. #2131.
  • QMCPACK spack package supports the latest release as well as the development version. This package can also install and patch Quantum Espresso.
  • Support for Blue Gene removed due to retirement of this architecture.
  • Many minor bug fixes, expanded testing, and small feature improvements.

Known bugs

See list of open bugs.

  • Use of reconfiguration in DMC is disabled since it is incorrect. #2254

NEXUS

  • NEXUS version is increased to 2.0.0 due to major updates in this release.
  • NEXUS has been transitioned to Python 3 and now requires it.
  • Significantly expanded test system to cover all major functionality.
  • Full support for PySCF to QMCPACK and AFQMC workflows #1970.
  • Support for DLA #2061.
  • VMC optimization performed with NLPP derivatives by default #2128.
  • Many minor bugfixes and feature improvements.

[3.8.0] - 2019-07-23

Notes

This release includes Quantum Espresso v6.4.1 support, new examples for adding wavefunctions and Jastrow functions, and many updates to the AFQMC code functionality. Additionally, all the updated scripts and functionality utilized during the 2019 QMCPACK workshop are provided; this link also includes several new tutorials. A large number of feature refinements, bugfixes, testing improvements and source code cleanup have also been performed.

  • Quantum Espresso v6.4.1 support #1732.
  • New tutorial for adding a simple wavefunction (He) #1621.
  • New tutorial and capability for adding Jastrow functors from symbolic expressions written in Python #1557.
  • Updated compiler and library support policy, and matching testing. We aim to support open source compilers and libraries within two years of release. Use of older software is discouraged and untested. Support for closed source compilers over the same period may require use of an exact version.
  • Many updates to AFQMC code to support more compilers and libraries.
  • Newly expanded deterministic test set should now pass on all platforms and be usable as an. installation check. Recommend to run "ctest -L deterministic" after building QMCPACK.
  • AFQMC code now only reads HDF5 format data to improve I/O performance and storage utilization.
  • K-point AFQMC code usable in production (e.g. bug fix #1524).
  • Updated AFQMC workflow scripts for interfacing with PySCF.
  • Faster initial cusp correction calculation for all-electron calculations, e.g. #1643.
  • Improved stability of cusp correction calculation #1594.
  • New short-ranged e-n Jastrow #1680.
  • Substantially faster 1-body reduced density matrix (1DRDM) estimator #1672.
  • Performance tests added for LCAO code and Gaussian basis sets #1639.
  • Reduced configuration output by default. Use -DQMC_VERBOSE_CONFIGURATION=1 on CMake line for greater detail.
  • Partial support for forces in LCAO basis e.g. #1559. See details given at 2019 QMCPACK Workshop and in manual.
  • Improved human-readable Jastrow output #1525.
  • Improved MPI implementation. QMCPACK is now compatible with OpenMPI v4.
  • Majority of the manual has been professionally edited.

Known bugs

See list of open bugs.

  • There is a general problem with MVAPICH2 involving aligned memory allocations that will cause QMCPACK to crash if MVAPICH is compiled using defaults. See #1703 for details and workaround.

NEXUS

  • Examples added of PySCF molecular and solid-state workflows #1552.
  • Update support for Quantum Package 2.0 #1538.
  • Support for additional machines including SuperMUC-NG #1665.
  • Support for ghost atoms #1653.
  • Update outdated cubic specifier to alat for QE #1642.
  • K-point grids are symmetrized with spglib #1544.
  • Bundling of jobs at NERSC #1748.

[3.7.0] - 2019-03-29

Notes

This release includes GPU support for the AFQMC implementation, Quantum Espresso v6.4 support, and in the real-space code makes the structure-of-arrays (SoA) code path the default. A large number of feature refinements, bugfixes, testing improvements and source code cleanup have been performed.

  • The improved structures of arrays (SoA) build is now the default. This is generally significantly faster and uses less memory than the AoS build due to better algorithms, but does not yet have the full range of functionality. The older AoS build can be selected with -DENABLE_SOA=0.
  • AFQMC code fully supports GPU acceleration via NVIDIA CUDA. Use -DENABLE_CUDA=1.
  • Quantum Espresso v6.4 is supported. #1457
  • Better error handling e.g. #1423
  • Workarounds for MPI support on Summit. #1479
  • ppconvert should be more reliable. #891
  • Delayed update implementation on GPUs. #1279
  • Continued improvements to the testing system and test coverage. While still under development, a new set of deterministic tests is intended to rapidly and reliably test the code, with good coverage. Tests pass for real and complex, but not yet mixed-precision or GPU builds.
  • Source code has been formatted with clang-format for consistency throughout.

Known Bugs

See list of open bugs.

  • Theres is a bug that could result in an incorrect local electron-ion pseudopotential energy with CUDA v9.1 and Kepler GPUs. This is still being investigated. #1440

  • QMCPACK will not build with OpenMPI v4 due to use of deprecated functions. This will be addressed when the new MPI wrappers are fully adopted. Older OpenMPI libraries are fully capable.

NEXUS

[3.6.0] - 2018-12-19

Notes

This release includes a completely new AFQMC implementation, significant performance improvements for large runs, greater functionality in the structure-of-arrays (SoA) code path, support for larger spline data on multiple GPUs, and support for new machines and compilers. The manual has been improved, bugs have been fixed, and source code cleanup continued.

A C++14 and C99 capable compiler, Boost 1.61.0, and CMake 3.6 or greater are now required.

  • Completely updated AFQMC implementation including reduced scaling separable density fitting https://arxiv.org/abs/1810.00284 Documentation and examples will be added in v3.7.0. Contact the developers for use instructions in the interim. #1245

  • Implementation of delayed updates for CPU. Substantial speedups for runs with 100s of electrons, with increasing gains at larger electron counts. See manual for details. #1170

  • Initial support for nested OpenMP to further reduce time-to-solution for large problems. #1082

  • Support for splitting/distributing spline orbital data across multiple GPUs on a single node. #1101

  • Cusp correction for all electron calculations is implemented in the SoA version. #1172

  • Backflow is implemented in the SoA version. #1225

  • K-points with real coefficients are supported in periodic LCAO. #1006

  • Initial support for Summit at OLCF. Revisions may be needed in January 2019 as the software stack is updated. This will be addressed in a new version as required.

  • Initial support for PGI compiler.

  • Build instructions for ARM-based systems. #1148

  • Setup scripts are python 2 and 3 compatible. #1261

  • QMCPACK and NEXUS can now be installed by "make install" after configuring CMake with CMAKE_PREFIX_PATH. #1020

  • Significantly reworked test labeling and categorization system. #1155

  • Partial transition to a new MPI wrapper implementation for greater compatibility.

  • Utilities have been renamed for clarity and to avoid name collisions with other applications. getSupercell is renamed qmc-get-supercell. extract-eshdf-kvectors is renamed qmc-extract-eshdf-kvectors.

Known bugs

Several potentially significant bugs are outstanding and will be addressed in the next release. See list of open bugs.

  • LCAO (Gaussian basis) molecular calculations are incorrect with certain diffusion functions. The reason for this bug is currently unclear. #1145

  • On NVIDIA Volta GPUs some runs show inconsistencies with the CPU version. Standard carbon diamond and LiH tests pass with good agreement with the CPU implementation. #1054

  • QMCPACK will not build with OpenMPI v4.0.0 due to use of deprecated functions. This will be addressed in the next version as the new MPI wrappers are fully adopted. Older OpenMPI libraries are fully capable.

NEXUS

  • Interface to and support for PySCF. #1220
  • Interface to and support for Quantum Package (QP). #1093
  • Support for excited state calculations. #1200
  • qfit is renamed qmc-fit.
  • ntest, sim, redo are renamed nxs-test, nxs-sim, nxs-redo.
  • Many smaller improvements.

[3.5.0] - 2018-08-02

Notes

This release includes support for the latest Quantum Espresso version 6.3, an initial implementation of periodic Gaussian support via PySCF, and a new version of the hybrid or "APW" representation of orbitals. Many minor bugs have been fixed, configuration and documentation improved.

Note that the PDF manuals are no longer included with the source. Versions are available online via https://qmcpack.org . The PDFs can be built using manual/build_manual.sh and nexus/documentation/user_guide_source/build_nexus_user_guide.sh

Attention developers: This version contains substantially fewer source lines than previous versions due to clean out of old code and unused execution paths. Refactoring to improve the internal structure of QMCPACK is ongoing. Track the develop branch and follow discussion on GitHub closely to avoid difficult merges.

  • Support for Quantum Espresso 6.3 and 6.2.1. Check documentation to ensure compiled with required HDF5 support.
  • Support for periodic gaussians and PySCF generated wavefunctions. Initial version is limited to Gamma-point.
  • Improved hybrid representation of single particle orbitals (APW-like) for significantly reduced memory usage and possible accuracy increase compared to conventional spline representation. https://arxiv.org/abs/1805.07406
  • Norms of orbitals are checked inside QMCPACK to catch conversion errors.
  • Added verbosity setting to QMCPACK output.
  • CUDA can now be enabled with SoA builds.
  • Many improvements to QMCPACK manual, including all new features, CIPSI, 3-body jastrow factor description, spack package, and enabling HTML generation.
  • CMake configuration improvements, particularly around MKL handling.
  • Extensive cleanup of unused source files and unused code paths removed, reducing the number of source lines by over 30 percent.

Known bugs

  • Weight of first block of DMC density is incorrect in CPU code. DMC densities in CUDA GPU code are incorrect for all blocks. #934 and #925
  • Runs with only a single electron may crash. #945

NEXUS

  • Support for GAMESS HDF5 workflows.
  • Nexus accepts command line inputs.
  • Nexus testing via ntest executable.
  • Added GAMESS-NEXUS examples for RHF, CISD, and CASSCF wavefunction.
  • Added support for -nojastrow workflows.
  • Added support for Stampede supercomputer.
  • Added script to build NEXUS user guide.
  • Various bugfixes including to GAMESS input parsing.

[3.4.0] - 2018-01-29

Notes

This release includes size-consistent t-moves, and improvements to load balancing and memory usage that will be visible in large runs. Significant revisions have been made to the gaussian wavefunction reader and a PySCF interface is in progress. A bug affecting non-git installs (from release tarballs) is fixed. Feedback is particularly welcome on the new features.

  • Size consistent t-moves implemented (Casula 2010 algorithm). Enabled via nonlocalmoves parameter, see manual.
  • Bugfix: For non-git builds, build process failed on some systems due to git-rev.h handling.
  • Optimized load balancing in DMC. Command line option async_swap removed. Parameter use_nonblocking now disables non-blocking MPI load balancing. Non-blocking MPI is now enabled by default.
  • Improved memory handling and usage in SoA code, increases performance.
  • Improved stability of GPU matrix inversion for large runs.
  • Ongoing improvements to output to improve readability.
  • Initial interface to PySCF for real space QMC trial wavefunctions.
  • Enabled use of HDF5 files for Gaussian based wavefunctions with SoA implementation.
  • Added Appendix to manual listing all known QMCPACK publications. This will be updated on an ongoing basis. Please advise of any missing publications.
  • Optimized momentum distribution estimator. Supported by SoA and 1,2,3-body Jastrow functions.
  • Support for labeled timers in Intel VTune based profiling.

NEXUS

  • Minor bugfixes and improvements.

Known limitations

  • PySCF interface is preliminary. convert4qmc is updated, but manual entries are not yet provided. This will be improved in later versions. The interface is currently only for isolated molecular systems. A full periodic implementation is in progress.

  • Documentation, examples and tutorials are not yet consistent with the updated converter convert4qmc.

[3.3.0] - 2017-12-18

Notes

This release includes new methods, converter updates, and many optimizations, feature improvements, and bug fixes. It is a recommended update for all users.

QMCPACK updates

  • Support for finite difference linear response (FDLR) method and wavefunctions, developed and contributed by Nick Blunt and Eric Neuscamman, see Journal of Chemical Physics 147, 194101 (2017), https://doi.org/10.1063/1.4998197 and https://arxiv.org/abs/1707.09439 .
  • Major update to convert4qmc, conversion from GAMESS and other gaussian basis set codes. HDF5 output is now supported for large wavefunctions with -hdf5 option. Significantly improved example inputs *.qmc.in.xml.
  • Gaussian based trial wavefunctions now supported by structure of arrays implementation (ENABLE_SOA=1). A full reimplementation that will also support gaussians in periodic boundary conditions, e.g. from pyscf, is in progress.
  • Initialization of multideterminant wavefunctions improved for faster startup and lower memory usage. In practice this significantly raises the usable maximum number of determinants.
  • Maximum CPU time setting (maxcpusecs): QMC drivers will not start a new block if there is not enough estimated time remaining to complete the next block and gracefully shut down.
  • Homogeneous electron gas wavefunction support and tests.
  • New command line verbosity command line flag -verbosity. Output of QMCPACK will be overhauled over the next few releases to support low, high, and debug options, and also to significantly improve readability and utility.
  • Bugfix: Umrigar drift diffusion term is now consistent with the Umrigar small time step error algorithm with complex wavefunctions.
  • Bugfix: Momentum distribution is now correctly weighted and also correctly signed for twist averaging.
  • Renamed performance tests with atom and electron count.
  • Removed support for "buffering" of non-local pseudopotential wavefunction components during optimization (useBuffer setting) to reduce memory usage and for simplicity.
  • doxygen documentation for developer-level documentation of the code and file structure. Produced via make in qmcpack/doxygen. HTML currently published at http://docs.qmcpack.org/doxygen/doxy/
  • Many minor bug fixes and improved tests.

NEXUS

  • Improved postprocessing support for Quantum Espresso.
  • Various minor bug fixes.

Known issues and limitations

  • Documentation, examples and tutorials are not yet consistent with the updated converter convert4qmc.
  • Core functionality is largely compatible with ENABLE_SOA but some specialized wavefunctions and observables are not.
  • Use of GNU compilers with glibc 2.23 builds will crash due to a bug in libmvec of glibc. The glibc version can be verified by "ldd --version".

[3.2.0] - 2017-09-21

Notes

This release provides a significant speed increase for many calculations. A C++11 compiler is now required. It is a recommended update.

QMCPACK updates

  • Major speedup for calculations using spline wavefunctions via initial implementation of "Structure of Arrays" data layout and improved algorithms. Enabled via -DENABLE_SOA=1. Benefits all CPU architectures. Many runs are doubled in speed. Not yet available for Gaussian-basis sets or for all observables and QMC methods. See writeup in manual for guidance.
  • A compiler supporting C++11 is now required.
  • DMC respects MaxCPUSecs parameter and will gracefully shut down and not start a new block if there is not sufficient estimated time to complete it.
  • Checkpointing code rewritten for robustness and performance at scale. Parallel as well as serial HDF5 supported and autodetected.
  • Improved beta-release of AFQMC code and documentation.
  • Backflow documentation and optimization tips added.
  • Correlated sampling VMC drivers reactivated.
  • Added carbon graphite performance test similar to CORAL benchmark.
  • Improvements to CMake and CTest usage.
  • Build instructions for NERSC, ALCF, and OLCF machines updated.
  • Latest manual PDF now available at http://docs.qmcpack.org

NEXUS

  • Significantly improved manual entry for "qmca" analysis tool, the main recommended tool for statistical analysis of QMCPACK data.
  • Added time step fitting tool "qfit" for timestep extrapolation. Uses jack-knife statistical technique.
  • Improved density file postprocessing.
  • Support for Makov-Payne corrections.

[3.1.1] - 2017-08-01

Notes

This is a bugfix release and recommended update.

QMCPACK updates

  • Added numerical tolerance to check of jastrow cutoff and Wigner Seitz radius.
  • CMake correctly configures when MPI is not present.
  • Improved support for test coverage measurements.
  • Added unit tests for some estimators.

NEXUS

  • IPython compatible exit handling (from Duy Le)

[3.1.0] - 2017-06-21

Notes

This release incorporates an improved DMC equilibration scheme, numerous bugfixes, small improvements, and significantly improved testing. It is a recommended update.

QMCPACK updates

  • Improved population control during DMC equilibration. Reduces variance on larger runs.
  • Bugfix: Real valued wavefunction GPU code gave incorrect result for some non-gamma twists that could be made real, e.g. X point. Complex code (QMC_COMPLEX=1) was always correct.
  • All particle move VMC and DMC algorithms enabled, tests added.
  • Reptation Monte Carlo (RMC) enabled, tests added.
  • Significantly improved AFQMC implementation.
  • Added NiO based VMC and DMC performance tests and description in manual. Wavefunction files accessed via QMC_DATA.
  • Added DMC tests with locality and t-moves approximations.
  • Added AFQMC tests.
  • Added test of real space QMC restart capabilities.
  • Added tests for several estimators.
  • Added unit test for DMC walker propagation, effective core potentials, and OhmmsPETE.
  • To avoid filesystem limitations, QMC_SYMLINK_TEST_FILES can be set to symlink (1) or copy test files (0).
  • Fixed mixed precision Ceperley force evaluation.
  • Many updated tests to improve statistical reliability. Removed flux estimator from short tests because they were not reliable enough.
  • Tests that rely on non-standard python modules that are not available are skipped.
  • Error trap jastrow factors with cutoff radii larger than Wigner Seitz radius.
  • Bugfix: Prevent users from adding correlation terms on non-existing electron pairs, e.g. up-down correlation terms when only up-spin particles are present.
  • Support for measuring test coverage and performing coverage runs with cmake and ctest.
  • Support for GCC7 and IBM XL (non Blue Gene) compiler.
  • Support selecting GPU microarchitecture via -DCUDA_ARCH=sm_35(default).
  • SummitDev IBM Minsky build recipe (Power8 + NVIDIA Pascal P100 GPUs).
  • Significantly updated optimizer description in manual, including excited state optimization.
  • Added description of using Intel MKL with non-Intel compilers in manual.
  • Added description of MPIEXEC and MPIEXEC_NUMPROCS_FLAG to manual for systems where MPI runner is non-standard.
  • Updated labs with correct pseudopotentials, basis set files.
  • Many updated error messages and warnings.

Known problems

  • AFQMC without MKL will fail, e.g. short-afqmc-N2_vdz-4-1 test fails.

NEXUS updates

  • Improved selection algorithm to obtain optimally tiled supercells.
  • Support for parallel pw2qmcpack workflows.
  • Support for HPC resources at the Leibniz Supercomputing Center.
  • Better consistency checks for the Structure class.
  • Bugfix: forbid job bundling for simulations that depend on each other.
  • Bugfix: correctly select low spin polarization in primitive and tiled (net_spin="low" option).

[3.0.0] - 2017-01-30

Notes

We are adopting Semantic Versioning with this release. It is the first to be made from the git repository on GitHub, and the first named release since 2016-06-02 and subversion revision 6964.

A potentially severe bug is fixed for periodic wavefunctions in this version, in addition to many usability improvements and bugfixes. All users are strongly recommended to upgrade.

NEXUS updates are listed after QMCPACK updates.

QMCPACK updates

  • IMPORTANT BUGFIX: Real-valued wavefunction code would occasionally make a numerically unstable choice for constructing real-valued periodic wavefunctions, leading to large variances and poor energies. Algorithm for constructing wavefunctions improved.
  • Fully parallel pw2qmcpack.x for QE 5.3, enables conversion of large wavefunctions and use of same parallel setup as pw.x runs.
  • Full testing of Quantum Espresso workflows (pw.x -> pw2qmcpack.x -> qmcpack). Specify directory containing QE binaries via QE_BIN during configuration.
  • Added open boundary conditions tests using QE wavefunctions, as might be used for molecular work. Requires QE_BIN and computes trial wavefunction on the fly.
  • Added DMC, optimizer and additional system tests.
  • Added unit tests using the Catch framework.
  • Plane wave wavefunctions can be evaluated in plane waves, use "pw" as determinantset type. Slow, but useful for checking spline accuracy. Tests added.
  • Complex implementation on GPUs, supports arbitrary twists and complex phase wavefunctions as per CPU code.
  • Flux estimator correct for complex wavefunctions.
  • Mixed precision CPU implementation, activated via -DQMC_MIXED_PRECISION=1.
  • Double precision GPU implementation, complementing existing mixed precision implementation, activated via -DQMC_MIXED_PRECISION=0.
  • GAMESS CI converter improved.
  • C++11 detection and support.
  • Initial release of new optimizer, requires C++11 (contact Eric Neuscamman).
  • Initial release of orbital-based AFQMC code, requires C+11 and MKL (contact Miguel Morales).
  • Fine grained timers implemented, activated via -DENABLE_TIMERS=1.
  • Improved Intel math and vector math library support. MKL and MKL VML more easily supported with GCC as well as Intel compilers.
  • Many code updates to eliminate CLANG warnings.
  • Configure scripts, printed headers, manual updated for git. Git version printed during configure and on standard output.
  • Source files headers updated to consistently show UIUC/NCSA open source license and list development history.
  • Numerous manual updates.
  • Updated QMCPACK tutorial laboratories.
  • Many small bug fixes, improvements and optimizations.

NEXUS updates

  • General
    • Nexus output now tracks time instead of poll number.
    • Reported memory use now includes child processes.
  • Workflow generator
    • Major new capability to generate simple to complex workflows involving QE, VASP, and QMCPACK.
    • Aim is to allow single notebook/worksheet describing all simulation workflows needed in a project.
    • Users can succinctly create any subchain of the workflow: relax->scf->nscf->orbital_conv->qmc.
    • Additional elements can be added to workflow chains over time as needed.
    • Scans of structural parameters and input parameters at any level of the chain are possible.
    • No programming constructs are required (for/if, etc).
    • Directory substructure is automatically generated in the case of scans.
    • Native support for visualizing workflows via pydot is provided.
    • Documentation for this feature is pending.
  • Quantum Espresso workflows
    • Support for vdW functional input.
    • Fixes to SCF->NSCF workflows for QE 5.3.0+.
    • Support for automatic restarts of SCF runs.
    • Native support for workflows involving post-processing tools
    • pp.x, dos.x, bands.x, projwfc.x, cppp.x, pw_export.x supported.
    • Postprocessing and summary of Lowdin charge data from projwfc.x.
  • QMCPACK workflows
    • Fixes for QE/VASP structural relaxation -> QMCPACK workflows.
    • Fixed job bundling of twist averaged runs.
    • Support for partitioned sposet input.
  • Supercomputing environments
    • Native support for several supercomputing environments located at Sandia Nat. Labs.
  • Atomic structure manipulation
    • Ability to find optimal supercells, similar to getSupercell tool.
    • Robustness fixes to tiling operations.
  • Tools
  • qmca * Fix for twist averaging with user-provided weights.
  • qmcfit * New command line tool for jack-knife fitting of QMCPACK data. * Timestep extrapolation currently supported. * General binding/equation of state fitting pending.