RMG v6.1.0

RMG v6.1.0 is a bug fix and feature enhancement release.

Bug fixes:
An unnecessary symmetrization was inadvertently enabled for gamma point calculations. While this had
no impact on results or on performance for small problems it caused a significant initialization slowdown for
large gamma point calculations.

New features:
The adaptive finite difference code has been rewritten and uses a new polynomial curve fitting algorithm.
An initial implementation of adaptive interpolation is now enabled by default. This improves the accuracy of stress calculations at lower grid densities/cutoffs. It can be disabled via an input file option if the previous behavior is desired.

Test cases have been updated since the new features caused small changes in energies/forces/stress.

RMG v6.0.0

The RMG developers are pleased to announce the release of version 6.0.0 of RMG, which is an open source DFT based code that uses real-space grids to represent electronic orbitals, charge densities and potentials. RMG is designed for high performance on clusters or supercomputers but works well for workstation-class problems as well. It uses extensive multi-level parallelization and provides full GPU support for Nvidia, AMD and Intel hardware. RMG v6.0.0 is a major update to the version 5.x releases. New features include Intel GPU support, major updates to the TDDFT module which enable larger calculations, improved accuracy for non-orthogonal lattice types and improved stress calculations.

RMG v5.5.0

RMG v5.5.0 is a bug fix and feature enhancement release.

In addition to various bug fixes this release provides support for Intel GPUs using SYCL and the Intel OneAPI compilers and libraries. The included version of ScaLAPACK has also been updated to 2.2.0 and the RMG python gui has been removed and placed in it's own separate repository.

RMG v5.4.0

RMG v5.4.0 is a bug fix and feature enhancement release.

The new inter grid interpolation introduced in v5.3.1 has been made
the default. While slightly slower it improves accuracy/stability.
Potential acceleration is disabled by default for TDDFT.
Performance of cube file output for the charge density and
hartree potential has been improved for large systems.
Initialization time for rocfft has been improved.

RMG v5.3.1

RMG v5.3.1 is a bug fix and feature enhancement release.
New features: Additional inter grid interpolation option which can improve accuracy.
Build updates and a bug fix for certain hybrid calculations when using scalapack.

RMG v5.2.0

RMG v5.2.0 is a bug fix and feature enhancement release.
New features:
Support for additional Brillouin zone integration methods.
Compute and print out additional energy components for use by QMCPACK.
Bug fixes:
A change introduced in v5.1.0 that improved total energies at low cutoffs
increased grid sensitivity and degraded forces so it was backed out.
A bug that caused crashes in certain cases for monoclinic or triclinic lattices was fixed.

RMG v5.1.0

RMG 5.1.0 includes bug fixes and updates to improve accuracy.

Low cutoff accuracy improved.
Instability in adaptive finite differencing for very high cutoffs fixed.
A crash when 12th order finite differences were selected for tricilinc lattices fixed.
Updates to test cases as there are shifts in energy for lower cutoffs.

RMG v5.0.5

This is a bug fix release that corrects hangs that occurred for certain specific grid layouts.

RMG v5.0.4

Build improvements and test updates.

RMG v5.0.3

RMG v5.0.3 is a bug fix release for the 5.x series which improves build compatibility on additional platforms.