Activity for JDFTx
-
Ravishankar Sundararaman committed [040df6] on Git
Added separate output of forces due to external potential
-
Added option to skip alignment potential in defect-supercell
-
Formatting issue in docs from previous commit.
-
Implemented external Gaussian potentials in planar/cylindrical/spherical geometries for ionic dynamics
-
Updated NERSC instructions to use pre-built module.
-
Fixed crash in classicalDFT vdwCoupling (bug introduced in D3 implementation)
-
Added a few vector and tensor operators needed for CD implementation
-
Elevated CANDLE and SaLSA parameterization warnings to errors, as they lead to failures later if allowed to proceed.
-
Added reduction of symmetries by mismatched constaints
-
Updated perlmutter instructions
-
Support CUDA >= 11.5, replacing buggy cusolver*trtri with cublasZtrsm
-
Fixed hermiticity error in DumpCprime (due to round off errors + lack of absolute threshold in check)
-
Merge branch 'master' of https://github.com/shankar1729/jdftx
-
Fixed EXX non-hermiticity bug introduced in previous commit.
-
Applying RPA occupation correction to occupation product instead in EXX.
-
Switched RPA EXX to use energy eigenvalues to switch cases (to handle nonmonotonic smearing schemes)
-
Replaced Fk -> 1 in rpa-mode EXX
-
Added magentic units
-
Fixed missing overlap operator and lightly cleaned-up Berry code.
-
GitHub committed [828dc9] on Git
Merge pull request #229 from jxuucsc/master
-
jxuucsc committed [8b8c56] on Git
fix bugs for dump Berry curvature
-
jxuucsc committed [09a4f0] on Git
fix bugs for dump Berry curvature
-
jxuucsc committed [79ae31] on Git
dump Berry Curvature
-
Updated perlmutter compilation instructions to use GSL module
-
More doc updates
-
Documentation updates on compilation
-
Fixed g++ 11 warnings
-
Separated out triangular solve and cholesky into separate functions from
-
Fixed uninitialized saveRP in Wannier
-
Added mechanism to select BLACS layer
-
Fixed read-in of fillings with MP1 smearing. Fine tuned Scalapack,
-
Removed unused code from previous changes
-
Stripped out wannierization of L and Q, which does not work.
-
Implemented Wannierized R*P matrix elements using finite difference mesh
-
Added separate velocity degeneracy threshold to Cprime
-
Prototype version of Cprime using momentum matrix elements
-
Fixed off-by-one indexing error leading to zero Cprime for highest band.
-
Fixed off-by-one indexing error leading to zero Cprime for highest band.
-
Fixed trace term in quadrupole matrix element output
-
Added real space version of r in L, Q for truncated directions.
-
Removed defunct L and Q wannier code based on k-mesh finite differences.
-
Added tensor transformations for Q wannier output.
-
Fixed transformation of L
-
Tentative wannierization of L matrix elements.
-
Added hermitian symmetrization to the outputs
-
Fixed calculation of Q
-
Added tentative implementation of ultrasoft augmentation to [r,H] cross terms.
-
Added L and Q output from DFT.
-
Switched Cprime in wannier to actuallyc ompute dC/dk instead of d(CU)/dk * dagger(U).
-
Added wannierized L and Q output.
-
Switched <rP> implementation to use [r,H] operator instead of H-eps version
-
Initial draft of r*p calculation for the Bloch states.
-
Preliminary implementation of dC/dk in wannier (not yet incorporated into useful output)
-
Skeleton of wavefunction k-derivative in Wannier
-
Added RPA mode EXX calculation in BGW output (selected by key rpaExx in bgw-params).
-
Updated perlmutter compile docs
-
BGW dense wavefunction switched from block-cyclic to column-split for contiguous HDF5 writes
-
Added saveVxx support to dense diagonalization in BGW output
-
Added option to output Vxx matrix elements from BGW
-
Version change to 1.7.0
-
Merge branch 'master' of https://github.com/shankar1729/jdftx
-
Fixed cholesky bug on perlmutter cuda version
-
Added Perlmutter compilation instructions
-
Added inner-loop Davidson termination once occupied eigenvalues are converged.
-
Made DumpBGW explicitly select independent IO.
-
Updated changelog
-
Replacing invsqrt with orthoMatrix (Cholesky) wherever applicable for orthonormalization
-
Removed redundant Hsub.diagonalize in applyHamiltonian calls from Davidson
-
Removed unnecessary setEigenvectors() happening after every Davidson sequence within SCF
-
Added support for CUDA newer than 11.2 which replaces cusolverDnZtrtri
-
Added Cholesky orthonormalization option to Davidson
-
Replaced cusolver routine zheevj with zheevd for improved diagonalize performance.
-
Added optional bypass to MPI_Finalize controlled by compile time flag DONT_FINALIZE_MPI.
-
Fixed underflow issue in DFT-D3 interpolation of C6 between CNs.
-
Renamed WignerSeitz::restrict to reduce to avoid errors in C99-extended C++ compilers.
-
Merge branch 'master' of https://github.com/shankar1729/jdftx
-
GitHub committed [6c2851] on Git
Merge pull request #189 from ngaro/dockerfile
-
Nikolas Garofil committed [7d45b4] on Git
Added a Dockerfile
-
Merged RPA frequency grid fix from Jacob.
-
Added timer to DFT-D3 calculator, and confirmed implementation correctness.
-
Added LJ override support to DFT-D3
-
Updated LJ override in DFT-D3 to keep original damping form instead of true LJ
-
Gradient propagation of DFT-D3 to forces and stresses.
-
First draft of DFT-D3 with seemingly correct energy; no gradients yet.
-
Added coordination number and C6 interpolation logic in DFT-D3
-
Added coordination number calculation
-
Separated D3 data into its own header file, and added large R0AB data array
-
Added framework for parameters of D3, with partial data tables and XC parameter selection
-
Refactored VanDerWaals to be an abstract base class supporting DFT-D2 and DFT-D3 in derived classes.
-
Added support to use libxc functionals without exchange and/or correlation
-
Restored fixed-H support for potential-only functionals
-
Fixed crash on out-of-range OrthoOrbital in DOS, by including it in the initialization check along with DOS.
-
Added HDF5 flush for partial consistency of BGW dump during dense diagonalization
-
Fixed corner case in DOS of combined s orbital projections of relativistic pseudopotentials.
-
Added rALDA kernel option to BGW output
-
Added option to sum over all atoms of species in orbital-projected DOS
-
Revert "Fixed behavior of SIC for spin-unpolarized case"
-
Added check for null projection matrices in SIC
-
Fixed behavior of SIC for spin-unpolarized case
-
Minor cleanup of DumpSIC.cpp; added augmentation to density calculation therein
1 >