-
[TODO] Fix the bug in Hermit the H...
-
[Hold...] Symmetrize the Hamiltonian and other matrices.
-
[DONE!] Speed up the
UtilsPetsc::MatDaggerAccumulate. -
[DONE!] Move the d_factor calculation out of each k loop.
-
[Hold...] Speed up the Berry connection calculation.
-
[Abort.] Write the DeepH interface. (Now we only support
*.petscfor the Matrix size consideration) -
[DONE!] Del the basis transition code and move the code to the old deprecated version
-
[DONE!] Write the correct API for PETSc to PEXSI.
-
[DONE!] Support the multiple task and task relations inside one user inputs.
-
[DONE!] Deal with the spin channel when calc fermi energy.
-
[DONE!] Move all of the the type define to the same file.
-
[DONE!] Solve the big system fail to diagonalize problem.
-
[DONE!] Organized the input and output files.
-
[DONE!] All the interface do read in first, leave the process of the matrix after read into PETSc.
-
Write a special OPENMX (
ReadSr) that only read in the structure and basis, write the S and r(x,y,z). -
Let
DeepHgroup write the*.petscoutput function. -
[DONE!] Write a interface that read in the info of H S r etc (
*.petsc) fromReadSrandDeepH. -
Write a Chern calculation function.
-
[DONE!] Write the band calculation functions.
-
[Abort.] Add a function that can automatically decided the all orbit number when plot band.
-
Make the Fermi function of PEXSI support more active MPI algo.
-
[DONE!] Make the progress bar support the MPI Print.
-
[DONE!] Solve the instability of the eigen solver when ask much bands on large system (1E16 eigen values in some cases).
-
[DONE!] Make the OpenMX Reading needing more less memory (e.g., only root read, and tell the other MPI process).
-
FIX the bug that PEXSI stick when use only one core...
-
FIX the bug that the orbit type cannot print out...