DFT-FE v1.0.2

This is a patch release which has minor fixes

DFT-FE v1.0.0

This release has the following changes:

• Improved electrostatics methodology using smeared nuclear charges. This allows for efficient finite-element mesh generation with floating nuclear charges for pseudopotential calculations
• GPU porting of all key computational steps of the Chebyshev filtered eigensolver, and computation of ionic forces and cell stresses. Both Gamma point and multiple k-point calculations are GPU ported. Currently only NVIDIA GPU's are supported using the CUDA platform.
• Enable usage of NVIDA Collective Communications Library (NCCL) and GPU kernel of ELPA library (relevant for calculations using GPUs)
• Support for ONCV pseudopotentials with non-linear core correction
• ScaLAPACK support for complex datatype required in multiple k-point case
• Optimizations and bug fixes
• Optimizations to reduce peak memory on GPUs
• Refactoring of the code base to templatize usage of real and complex datatypes
• Extension and cleanups of restart capability in geometry relaxation and SCF restart

DFT-FE v0.6.0

This release has the following changes:

• Capability to do Broyden mixing added.
• Mixed precision capability added.
• Minor optimizations in orthogonalization routine.
• Minor bug fixes.

DFT-FE v0.5.2

This patch release has the following changes:

• BugFix in handling semi-periodic boundary conditions which fails in only some cases of semi-periodic boundary conditions.
• Incorporated reuse of wave-functions from previous step during geometry optimization.
• Minor optimizations during orthogonalization procedure.

DFT-FE v0.5.1

This release has a patch for fixing a compilation error when dealii is not compiled with ScaLAPACK.

DFT-FE v0.5.0

First open-source beta release