ChangeLog

----------------
-Date
-Name
-changes


--------------
Oct 01, 2020
Name: Qimen Xu
Changes: (readfiles.c)
1. Modify the function for reading pseudopotentials to support two formats of converted TM pseudopotentials.
2. Memory leak fix in DP (kpt_comm not freed for idle processes).


--------------
Sep 09, 2020
Name: Qimen Xu
Changes: (gradVecRoutinesKpt.c, lapVecNonOrthKpt.c, lapVecOrthKpt.c)
1. Add simd instructions for Laplacian vector routines for k-points (works well for intel/19.0.5 with automatic cpu dispatch).


--------------
Aug 27, 2020
Name: Qimen Xu
Changes: (eigensolver.c, eigensolverKpt.c)
1. Bug fix in DP: when DP is turned on and npdomain > 1, eigenvalues are only correct in the first dmcomm.


--------------
Jul 24, 2020
Name: Qimen Xu, Abhiraj Sharma
Changes: (eigensolver.c, stress.c)
1. Removed repetitive part involving finding global max and min of eigenvalues among spin communicators.
2. Fixed bug related to specification of BCy instead of BCz while calculating cell measure (5 places).


--------------
Jun 22, 2020
Name: Qimen Xu
Changes: 
1. Update makefile-ibm to a generalized and streamlined version.


--------------
Jun 20, 2020
Name: Qimen Xu
Changes: 
1. For MPI_Comm_split calls, change color = -1 to color = INT_MAX.


--------------
Jun 13, 2020
Name: Qimen Xu
Changes: 
1. Remove some unused variables.
2. Enable comments after the "LATVEC:" flag.
3. Update the headings of .ion files in the tests/ folder.


--------------
Jun 10, 2020
Name: Qimen Xu
Changes: 
1. Modify default Chebyshev filter degree for non-orthogonal systems.


--------------
Jun 09, 2020
Name: Qimen Xu
Changes: 
1. Add makefile-IBM for compiling SPARC on IBM machines.
2. Modify scalapack.h for compatibility of ScaLAPACK function declarations on IBM machines.
3. Resolve some abs data type compatibility issues.


--------------
May 26, 2020
Name: Qimen Xu
Changes: 
1. Update README.
2. Modify some simd instructions in gradVecRoutines.c and lapVecNonOrth.c, with some versions of compilers the simd instructions were causing compilation error.


--------------
May 20, 2020
Name: Qimen Xu
Changes: 
1. Update documentation and README.
2. Some small changes to the output headings.
3. Add Examples to tests/ folder, rerun existing tests.


--------------
May 18, 2020
Name: Qimen Xu
Changes: 
1. Fix MD time printing issue. When PRINT_MDOUT is turned off, the MD step time was not correctly printed to .out. 
2. Fix the issue where an empty .static is created for relax and MD simulations.
3. Change default relax method from NLCG to LBFGS.
4. Rerun the tests in the tests/ folder.


--------------
May 11, 2020
Name: Qimen Xu
Changes: 
1. Bug fix: nonlocal force calculation occasionally fails during MPI_Reduce
between kptcomms, switching to MPI_Allreduce fixed the issue.
2. Change input EIG_SCALAPACK_BLKSZ to EIG_PARAL_BLKSZ.
3. Change input EIG_LAPACK_MAXNS to EIG_SERIAL_MAXNS.
4. Remove EIG_USE_LAPACK option.
5. Remove rational function approx. to Kerker preconditioner (truncated_kerker
and resta), move AAR_complex to extras folder.


--------------
May 06, 2020
Name: Qimen Xu
Changes: 
1. Change default preconditioner to density mixing + Kerker preconditioner, with a preconditioner threshold of 0.1.
2. Add a parallelization summary to log file in DEBUG mode.
3. Add an input option to choose whether to use LAPACK for solving the subspace eigenproblem (specify by EIG_USE_LAPACK).
4. Change default block size for solving subspace eigenproblem using ScaLAPACK from 64 to 128.
5. Add an input option to modify the above block size (specify by EIG_SCALAPACK_BLKSZ).
6. Add an input option to change the maximum NSTATES up to which LAPACK will be used to solve the subspace eigenproblem (specify by EIG_LAPACK_MAXNS). 
7. Modify occupation warnings.


--------------
May 02, 2020
Name: Abhiraj Sharma
Changes: initialization.c
1) Time reversal symmetry edited to account for additional symmetry points on the boundary of IBZ


--------------
Apr 28, 2020
Name: Qimen Xu
Changes: 
--------------
1. Add another MPI_Barrier after MPI_Bcast in Transfer_Veff_loc to avoid slowdown in the MPI_Bcast step.


Apr 27, 2020
Name: Qimen Xu
Changes: 
1. Modify comments in makefile.


--------------
Apr 25, 2020
Name: Qimen Xu
Changes: 
1. Add MPI_Barrier before MPI_Bcast in Transfer_Veff_loc to avoid slowdown in the MPI_Bcast step.
2. Modify occupation warning message.
3. Print relative error in CG linear solver in DEBUG mode.


--------------
Apr 24, 2020
Name: Qimen Xu, Abhiraj Sharma
Changes: 
1. Print stress tensor to .static file for single point calculations instead of .out file, print max stress to .out file.
2. Print energy to .static file and .geopt file as well as .out file.
3. Fix scf# issue for restarted calculations.


--------------
Apr 17, 2020
Name: Qimen Xu, Abhiraj Sharma
Changes: 
1. Add .static file for single point calculations.
2. Memory leak fixes.
3. For relaxation, instead of exiting, print a warning if Brent's method didn't converge.
4. Modify SPIN_TYP option, used to be 1 (no spin) or 2 (with spin). Now changed to 0 (no spin) or 1 (with spin).


--------------
Apr 08, 2020
Name: Qimen Xu
Changes: 
1. Clean some source files.
2. 2D decomposition of proc according to matrix for ScaLAPACK context (for most cases we end up with domain-only parallelization).


--------------
Apr 02, 2020
Name: Qimen Xu
Changes: 
1. Modify occupation warnings (back toocc[Ns * 90%] < 1e-6), also provide a suggested Ns.
2. Add an option to randomize the initial velocities in MD. Set `ION_VEL_DSTR_RAND: 1` to enable it.
3. Move ExtVecAccel/ to extras/.
4. Fix some valgrind warnings.
5. Clean some source files.


--------------
Mar 31, 2020
Name: Qimen Xu
Changes: 
1. Modify occupation warnings (occ[Ns * 95%] < 1e-6).


--------------
Mar 29, 2020
Name: Qimen Xu
Changes: 
1. Simplify paths (Unix style) to canonical paths, e.g., "././foo/../foo/a.txt" => "foo/a.txt".


--------------
Mar 28, 2020
Name: Qimen Xu
Changes: 
1. Allow users to provide both relative path and absolute path to pseudopotentials.
2. Increase max length for file/directory paths to 4096 (default PATH_MAX on linux systems).


--------------
Mar 26, 2020
Name: Qimen Xu
Changes: 
1. LAPACK + GTMatrix option (no MKL or ScaLAPACK).
2. LAPACK + GTMatrix + ScaLAPACK (for subspace eigenproblem only) option.
3. Fix some potential memory leak issue.


--------------
Mar 25, 2020
Name: Abhiraj Sharma
Changes: md.c, relax.c
1. If RestartFlag = 1 and .restart files are absent then dynamics will start from the beginning.


--------------
Mar 19, 2020
Name: Qimen Xu, Hua Huang
Changes: 
1. Remove compiler warnings (mostly "may be uninitialized" warnings).
2. Remove some valgrind warnings.
3. Separate out data distribution routines for projection/rotation and move
them to parallelization.c.
4. Create a framework for using ScaLAPACK only for subspace eigenproblem.
5. Fix bug in MD, allocate correct memory for array storing ":MIND:".
6. Fix bug in MD, initialize mean_TE_ext and std_TE_ext to zero.


--------------
Mar 11, 2020
Name: Qimen Xu
Changes: (readfiles.c, eigenSolver.c, eigenSolverKpt.c)
1. Change warnings regarding occupation numbers.
2. Modify pseudopotential reading so that it reads special pseudopotentials
(lmax = lloc = 0) in psp8 format.
3. Automatically expand bounds in Brent's method.


--------------
Mar 04, 2020
Name: Hua Huang
Changes: (eigenSolver.c, eigenSolverKpt.c, makefile)
1. Replace GTMatrix with MPI_Alltoallv to use domain parallelization (decomposition) only data layout for solving subspace eigenproblem.
2. Update the makefile


--------------
Feb 26, 2020
Name: Qimen Xu
Changes: (electrostatics.c, readfiles.c, initialization.c)
1) Improve speed for spline (around 8x for uniform grid, 2x times for non-uniform grid).
2) Give warning when radial grid in the pseudopotential is not uniform.
3) Read TM pseudopotential in psp8 format.


--------------
Feb 23, 2020
Name: Qimen Xu
Changes: (mixing.c, readfiles.c, initialization.c)
1) Add real-space abinit-like preconditioner. Use by setting
PRECOND_KERKER_THRESH (equivalent to 1/DIEMAC).
2) Give warning when pspxc in the pseudopotential is not equal to input XC.
3) Exit if nonlinear core correction is present in the pseudopotential file.


--------------
Feb 20, 2020
Name: Qimen Xu
Changes: (included headers, makefile)
1) Add option to use ScaLAPACK instead of MKL.
2) Add strcmpi function for comparing strings igoring case.


--------------
Feb 19, 2020
Name: Abhiraj Sharma
Changes: (relax.c, initialization.c, isddft.h, orbitalelecDens.c)
1) Added full relaxation (atom + stress) option in the code.
2) Changed the units of stress for different BCs.
3) Stress for different BCs is printed differently.


--------------
Feb 13, 2020
Name: Qimen Xu
Changes: (initialization.c, readfiles.c, relax.c)
1) Add cell volume relaxation.
2) Print memory estimate.


--------------
Feb 10, 2020
Name: Abhiraj Sharma
Changes: (initialization.c, readfiles.c, finalization.c)
1. Removed ion_elec_eqT as input option.
2. Changed default electronic smearing and its type for MD depending on different user inputs.
3. Memory bug regarding release of pSPARC->forces fixed.


---------------
Feb 07, 2020
Name: Qimen Xu
* Add MESH_SPACING, ECUT options (alternative ways to define mesh spacing).


---------------
Feb 05, 2020
Name: Abhiraj Sharma, Qimen Xu
* Modify k-point grid definition. K-point shift is defined wrt the gamma point.
* Replace "strcpy" function with safer function "snprintf".


---------------
Feb 02, 2020
Name: Abhiraj Sharma
Fixed compilation bug in non-debug mode

----------------
Feb 01, 2020
Name: Abhiraj Sharma
Changes: (isddft.h, md.c, readfiles.c, initialization.c, energy.c, electronicgroundstate.c)
1. Pressure bug for the first step of MD fixed (ionic pressure was calculated twice)
2. ion_T used instead of thermos_T in Ideal gas pressure implementation.
3. potential energy in MD corrected.
4. occ in .eigen file factored by 2 for non-spin cases.
5. MD_Nstep made to take input in exponential forms like 1e9.
6. TWtime or MD_Nstep or both can be used for MD simulation. Default MD_Nstep made a very high number.
7. Minimum atomic distance uncommented and printed in .aimd file.
8. All constants put in isddft.h file and values are taken from NIST website.
9. Default values of ion_elec_eqT made 0.
10. :MDTM: in .aimd file shows wall time rather than MD simulation time.  

----------------
Jan 30, 2020
Name: Xin Jing
Changes: (electrostatics.c, isddft.h, linearSolver.h, initialization.c, linearSolver.c, readfiles.c)
1. Add Conjugate Gradient Solver
2. Add control in input file and output


----------------
Jan 02, 2020
Name: Qimen Xu
Changes: (electrostatics.c)
1. Add wire boundary condition for electrostatics.


----------------
Jan 01, 2020
Name: Qimen Xu
Changes:(readfiles.c, electrostatics.c)
1. Add BC input option.
2. Add surface boundary condition in electrostatics.


----------------
Dec 20, 2019
Name: Qimen Xu
Changes:(parallelization.c, readfiles.c)
1. Fix bug when ONCV pseudopotential has lloc < lmax.
2. Fix bug in parallelization when kptgroup is larger than kpt_topo.


----------------
Dec 19, 2019
Name: Abhiraj Sharma
Changes:(exchangeCorrelation.c, parallelization.c, stress.c, pressure.c,finalization.c)
1. Put in RPBE and PBEsol GGAs in the code.


----------------
Dec 18, 2019
Name: Abhiraj Sharma
Changes:
1. Small changes in files to make code compatible with different boundary conditions.(Cyclix)


----------------
Nov 16, 2019
Name: Qimen Xu
Changes: (eigenSolver.c)
1. Remove Jacobi eigensolver header inclusion in eigenSolver.c when using GTMatrix.


----------------
Nov 05, 2019
Name: Qimen Xu
Changes: (md.c, electronicGroundstate.c)
1. MD walltime limit, using last MD walltime to decide whether to continue MD or not.
2. Throw error if Etot becomes NaN.


----------------
Oct 24, 2019
Name: Qimen Xu
Changes: (electronicGroundstate.c)
1. Bug fix: fix failure problem for Dirichlet BC.


----------------
Oct 22, 2019
Name: Qimen Xu, Abhiraj Sharma
Changes: (GTMatrix/*, extras/Jacobi_dsyev.{c,h}, exchangeCorrelation.c)
1. remove non-unix text in GTMatrix/ directory.
2. Bug fix: spin + extrapolation gives NaN V_xc values.


----------------
Oct 16, 2019
Name: Qimen Xu
Changes: (parallelization.c, energy.c, include/scalapack.h, eigenSolver.c, eigenSolverKpt.c)
1. Bug fix: in Exc calculation, sendbuf and recvbuf are swapped in MPI_Allreduce.
2. Bug fix: for idle process not in kptcomm_topo, initialize Nd_d_kptcomm to 0 to avoid memory allocation failure.
3. Modify Chefsi bound calculation, only use Lanczos once in the first SCF of each MD/relax step.
4. Remove explicit declaration of numroc_, descinit_, pdgemr2d_ and pzgemr2d_, include "mkl_scalpack.h" instead.


----------------
Oct 12, 2019
Name: Qimen Xu
Changes: (across all sources files)
1. Remove unused variables.


----------------
Oct 09, 2019
Name: Qimen Xu
Changes: (electronicGroundState.c, parallelization.c, etc)
1. Add QE scf error option. Meant for spin-unpolarized systems with density
mixing. For other cases, it might not be equivalent to QE's definition.
2. Modify current scf error definition for potential mixing: denominator
||veff|| is also the shifted veff.


----------------
Oct 07, 2019
Name: Qimen Xu
Changes: (extras/Jacobi_dsyev.{c,h})
1. Move jacobi eigensolver to extras/.


----------------
Oct 05, 2019
Name: Hua Huang
Changes: (src/makefile, src/parallelization.c, src/eigenSolver.c, src/eigenSolverKpt.c)
1. Add GTMatrix (a git submodule in this git repo) as an alternative to ScaLAPACK.


----------------
Oct 04, 2019
Name: Qimen Xu
Changes: (finialization.c, nlocVecRoutines.c, parallelization.c)
1. Bug fix: various memory leak fix.


----------------
Oct 02, 2019
Name: Qimen Xu
Changes: (initialization.c,electronicGroundState.c,parallelization.c)
1. Bug fix: failure in D2D when poisson domain is larger than psi domain.
2. Bug fix: failure when chefsi_bound_flag is off.
4. Bug fix: in a few places, abs() was used for double type variables.
5. Add printing eigenvalues and occupations (specify by 'PRINT_EIGEN').
4. Move warnings to .out file.

----------------
Sep 23, 2019
Name: Qimen Xu
Changes: (initialization.c,electronicGroundState.c,md.c)
1. Change .out format.
2. Change .aimd format.


----------------
Sep 19, 2019
Name: Qimen Xu
Changes: (electronicGroundState.c)
1. Add minimum SCF iterations (variable name MINIT_SCF).


----------------
Sep 13, 2019
Name: Qimen Xu
Changes: (electronicGroundState.c)
1. change output format.


----------------
Sep 12, 2019
Name: Qimen Xu
Changes: (stress_new.c, psd_oncv_Cu.pot, pressure.h, stress.h, ExtVecAccel/,
Cu.psp8)
1. Convert non-unix text to unix text.


----------------
Sep 10, 2019
Name: Qimen Xu
Changes: (doc/ folder)
1. Add documentation.


----------------
Sep 09, 2019
Name: Qimen Xu
Changes: (initialization.c, electronicGroundState.c)
1. Print density for spin polarized or spin unpolarized systems.


----------------
Sep 06, 2019
Name: Qimen Xu
Changes: (initialization.c, mixing.c)
1. Add MAXIT_POISSON: maximum number of iterations for Poisson solver.
2. Add Mixing parameter for simple mixing step.
3. Fix bug in k-point parallelization (with symmetry reduction).
4. Fix bug in scaling density (when density becomes negative) in density mixing.


----------------
Sep 04, 2019
Name: Qimen Xu
Changes: (initialization.c, mixing.c)
1. Add Resta preconditioner and truncated Kerker preconditioner.


----------------
Sep 01, 2019
Name: Qimen Xu
Changes: (initialization.c, parallelization.c)
1. Add k point shift from Monkhorst-Pack grid.


----------------
Aug 27, 2019
Name: Qimen Xu
Changes: (stress.c, initialization.c, electronicGroundState.c)
1. Remove non-unix text in stress.c.
2. Fix bug when TOL_SCF is not set, TOL_SCF becomes negative.
3. Set default preconditioner to 'none'.
4. For large peridoc systems, print warning if preconditioner is off.
5. Change .out heading for Energy section to 'Energy and force calculation'.


----------------
July 29, 2019
Name: Abhiraj Sharma
Changes: (md.c, relax.c, initialization.c, isddft.h)
1. Restart for MD changed to incorporate atleast two histories to be stored.

----------------
July 27, 2019
Name: Abhiraj Sharma
Changes:
1. Spin polarized calculation added.
2. Spin parallelization added.


----------------
June 27, 2019
Name: Abhiraj Sharma
Changes: (initialization.c, readfiles.c, md.c, relax.c, isddft.h)
1. Formatted the .aimd and .out file.


----------------
June 18, 2019
Name: Abhiraj Sharma
Changes: (hamiltonianVecRoutines.c, eigenSolver.c, eigenSolverKpt.c)
1. Fixed the bug associated with nonorthogonal cell when chefsibound_flag = 1.
2. Changed the filter cutoff criteria to be based on max eigval rather than fermi-level.


----------------
May 30, 2019
Name: Abhiraj Sharma
Changes: (parallelization.c, eigenSolverKpt.c, finalization.c, isddft.h)
1. Use D2D instead of DD2DD for finding eigen bounds.
2. Bug fixed for initial guess to lanczos.


----------------
May 29, 2019
Name: Qimen Xu
Changes: (readfiles.c, parallelization.c, electronicGroundState.c, mixing.c, tools.c)
1. Add SCF_FORCE_ACC, SCF_ENERGY_ACC, CHEFSI_BOUND_FLAG options.
2. Precondioned SCF: minimize the residual instead of minimzing the preconditioned residual.
3. Modify the seed for random number generator back to time independent seed. Use rank*100 + 1 as seed.
4. Bug fix: fisrt argument of descinit_ should be a pointer instead of a pointer to a pointer.
5. Bug fix: initial guess for Lanczos was deallocated after each SCF, which causes Lanczos to fail once chefsi_bound_flag is turned on.


----------------
May 20, 2019
Name: Abhiraj Sharma
Changes:
1) Added constrained relaxation and constrained MD features.
2) Changed default relaxation tolerance to 5e-4.


----------------
May 15, 2019
Name: Abhiraj Sharma
Changes:
1) Added kpts in the code
2) Created separate file for electron density (electronDensity.c)


----------------
May 10, 2019
Name: Qimen Xu
Changes: (mixing.c, tools.c)
1. Precondioned SCF: minimize the preconditioned residual instead of minimzing the residual.
2. Modify the seed for random number generator. Use MPI_Wtime() + rank + 1 as seed.


----------------
May 06, 2019
Name: Qimen Xu
Changes: (initialization.c, forces.c, stress.c, nlocVecRoutines.c)
1. Bug fix: fix undefined behavior in nonlocal under -O2, -O3 optimization. (By creating temporary variable int lloc = localPsd[ityp].)

----------------
May 02, 2019
Name: Qimen Xu
Changes: (initialization.c, parallelization.c, orbitalElecDensInit.c)
1. Fix the random number seeds so that all random numbers generated in
   parallel under MPI are the same as those generated in sequential execution,
   this will ensure that SPARC gives the same results when excecuted with
   any number of processors.

----------------
April 30, 2019
Name: Qimen Xu
Changes: (initialization.c, scalapack.h)
1. Adjust default Kerker tolerance and rhoTriger.
2. Modify declaration of pdgemr2d_ and descinit_.

----------------
April 26, 2019
Name: Hua Huang
Changes: ExtVecAccel module
1. Support non-orthogonal cell types in ExtVecAccel module (OpenMP & CUDA)
2. Remove EVA module entrance in poisson_residual() (lapVecRoutines.c)

----------------
April 24, 2019
Name: Qimen Xu
Changes: (initialization.c)
1. Add accuracy level (specify by ACCURACY in .inpt file).


----------------
April 23, 2019
Name: Abhiraj Sharma
Changes: (electronicGroundState.c, eigenSolver.c)
1. Restructured the eigenSolver file


----------------
April 20, 2019
Name: Qimen Xu
Changes: (electronicGroundState.c)
1. Potential mixing with kerker preconditioner.


----------------
April 19, 2019
Name: Abhiraj Sharma
Changes: (exchangeCorrelation.c)
1. Bug fixed in GGA exchange-correlation corresponding to zero e- density at a node.


----------------
April 15, 2019
Name: Qimen Xu
Changes: (electronicGroundState.c, tools.c, initialization.c)
1. Modify energy evaluation for density mixing.
2. Set default poisson tol and pseudocharge tol based on SCF tol.
3. Modify rb tol definition to abs tol instead of charge normalized tol.
4. Modify rhoTriger option so that it's consistent with previous settings.
5. Create function for printing and reading distributed vectors.


----------------
April 12, 2019
Name: Abhiraj Sharma
Changes: (eigensolver.c, electronicgroundstate.c, isddft.h)
1. Formatted the layout of SCF
2. Added option to perform many Chebyshevs <= rhoTriger in each SCF

----------------
April 11, 2019
Name: Qimen Xu, Abhiraj Sharma
Changes: (initialization.c, readfiles.c)
1. Change default smearing for Fermi-Dirac.
2. Change default nstates calculation (potential bug if npkpt is not specified).
3. Default smearing and smearing type for MD included.


 ----------------
April 10, 2019
Name: Qimen Xu
Changes: (stess.c)
1. update stess.c, include lapVecNonOrth.h instead of lapVecnonorth.h.


----------------
April 09, 2019
Name: Qimen Xu
Changes: (electronGroundState.c, energy.c, initialization.c, readfiles.c)
1. Add gaussian smearing and set as default. To specify smearing type, use "ELEC_TEMP_TYPE" flag, the options are "fd" (or "fermi-dirac") and "gaussian".
2. Add smearing option "SMEARING" in input file, set default smearing to 0.2 eV = 0.00735 Ha.
3. Modify default number of bands.
4. Update bug fix on input options "BETA" and "ELEC_TEMP", which still cause failure if they are specified at the end of the file.


---------------
April 05, 2019
Name: Abhiraj Sharma
Changes: md.c
Added Isokinetic ensemble with Gaussian Thermostat


---------------
Mar 30, 2019
Name: Abhiraj Sharma
Changes: (electrostatics.c, forces.c, stress.c)
Merged orthgonal and non orthogonal components into one function


---------------
Mar 29, 2019
Name: Abhiraj Sharma
Changes: (lapVecnonorth.c, electrostatics.c, forces.c, stress.c)
1. Stencil for non orthogonal laplacian created and used in Poisson, Chebyshev, electrostatics, forces, stress
2. For non orth lap_VJ, a separate function created.


----------------
Mar 28, 2019
Name: Qimen Xu
Changes: (lapVecOrth.c, electrostatics.c, forces.c)
1. Modify free energy functional evaluation.
    (1.a) Add self consistent correction to free energy for potential mixing.
    (1.b) Evaluate energy before density update for density mixing.
2. Bug fix: stencil kernel fails unpredictably when overlapping of communication
and computaion happens. This might be due to the fact that there are two indices
inside the for loop under SIMD instruction (need to understand this behavior).
3. Use stencil kernel in b and force calculation.

 ----------------
Mar 21, 2019
Name: Qimen Xu
Changes: (lapVecOrth.c, readfiles.c)
1. Start using EVA stencil kernel (with a few modifications) in orth Laplacian
   mat-vec, which will speed up the mat-vec routines.
2. Bug fix: input options "BETA" and "ELEC_TEMP" will cause program failure if
            there's an extra blank line after them, this is fixed now.

-------------------
Mar 17, 2019
Name: Abhiraj Sharma
Changes: Added GGA_PBE exchange correlation

Mar 13, 2019
Name: Qimen Xu
Changes: (add lapVecOrth.c, scalapack.h)
1. Create general Laplacian (orthogonal case) function (a*Laplacian+b*v+c)*x,
   remove all the subroutines.
2. Add scalapack.h for explicit declaration of scalapack functions.
3. Bug fix: in (Lap + v + c)*x, v and x are sharing the index, this is wrong
            if ncol > 1.

----------------
Mar 11, 2019
Name: Abhiraj Sharma
Changes:
1) Add stress and pressure
2) Fixed bug in lapvecroutines.c

Mar 07, 2019
Name: Qimen Xu
Changes:
1. Add density mixing
2. Add Kerker preconditioner

----------------
Feb 27, 2019
Name: Hua Huang
Changes:
1. Add CUDA support for Chebyshev Filtering procedure.
2. Replace MKL sparse BLAS SpMV & SpMM functions with an OpenMP implementation for portability.
3. Update EVA module README file.

----------------
Feb 14, 2019
Name: Hua Huang
Changes: EVA entry and related changes

----------------
Feb 12, 2019
Name: Abhiraj Sharma
Changes: Commented extra(hypothetical) cases of non orthogonal cells from the code for faster compilation.

----------------
Feb 11, 2019
Name: Abhiraj Sharma
Changes: 1) Added feature to perform DFT calculations (energy, force, MD, relaxation) on non orthogonal cells.
         2) Removed error in Relaxation restart

----------------
Feb 09, 2019
Name: Abhiraj Sharma
Changes: (md.c, relax.c, readfiles.c, initialization.c)
1. Changed MD_maxIter -> MD_Niter and Relax_maxIter to Relax_Niter.
2. Added restart options for continuation from where the MD stopped and to do a fresh restart using last step's atomic positions and velocities only.
3. Added the fractional atomic coordinate option for the user input of atomic positions.


-----------------
Feb 08, 2019
Name: Qimen Xu
Changes: (electronicGroundState.c)
1. Use eigenvector of periodic discrete Laplacian as initial guess for Lanczos.


-----------------
Feb 05, 2019
Name: Qimen Xu
Changes: (initialization.c, readfiles.c)
1. Change format of input files:
   1.1) NTYPES is removed from .inpt file
   1.2) PSEUDOPOTENTIAL_LOCAL is removed from .inpt file, the local component (lloc) will be read from the pseudopotential file.
   1.3) PSEUDOPOTENTIAL_FILES is moved from .inpt file to .ion file, and the flag name is changed to PSEUDO_POT.
   1.4) ATOMIC_MASS is moved from .inpt file to .ion file.
   1.5) Valence charge is not needed in .ion file any more. It will be read from the pseudopotential file.
2. The .out file format is changed according to the input file format.
3. Check if position goes out the domain for Dirichlet BC.
4. Print first and last occupation in the warning if nstates is not large enough.


----------------
Jan 24, 2019
Name: Qimen Xu
Changes: (most source files and a few header files)
1. Reorganize the functions and files in the code.
	1.01) Move min_eig_solver() from eigenSolver.c to extras.c
	1.02) electronDensInit.c --> orbitalElecDensInit.c (rename file)
		  ChargeExtrapolation() -> elecDensExtrapolation() (rename function)
	1.03) Calculate_EGS_Energy() -> Calculate_EGS_elecDensEnergy()
	1.04) Move Calculate_Eband and Calculate_Entropy into Calculate_Free_Energy()
	1.05) Create parallelization.c, include functions  DD2DD(), Setup_Comms(),
	      SPARC_Dims_Create(), ScaLAPACK_Dims_2D_BLCYC(), block_decompose(),
	      block_decompose_nstart()
    1.06) Move Mesh2ChebDegree() from initialization.c to eigenSolver.c
    1.07) Separate matVecRoutines.c into three files: lapVecRoutines.c, gradVecRoutines.c
    	  and hamiltonianVecRoutines.c
    1.08) Create Mixing_Pulay(), call this from Mix_Veff_loc()
    1.09) Move RealSphericalHarmonic() from nonlocal.c to tools.c
	1.10) nonlocal.c --> nlocVecRoutines.c
	1.11) Remove poisson.h
	1.12) Remove Read_psd_TM()
	1.13) Remove sddft.h
	1.14) Change format of output file (Developers' info. section)
	1.15) Print warning when SCF is not converged (both log file and output file)
	1.16) Print time for force calculation

----------------
Jan 19, 2019
Name: Qimen Xu
Changes: (eigenSolver.c)
1. Use eigmin = -2.0 and eigmax = max eigval of -1/2*Lap as initial lower/upper bounds
   for Chebyshev Filtering in the very first SCF step.
2. Created a function for estimating max eigval of the Laplacian operator
   using Lanczos for Dirichlet B.C..

----------------
Jan 11, 2019
Name: Hua Huang
Changes: (electronGroundState.c, eigenSolver.c, matVecRoutines.c, makefile, ./ExtVecAccel/)
1. Add EVA module entrance points (see ./ExtVecAccel/README.md for detail information).

----------------
Jan 10, 2019
Name: Qimen Xu
Changes: (matVecRoutines.c)
1. Fix bugs caused by undefined behavior and sequence points.

----------------
Jan 08, 2019
Name: Qimen Xu
Changes: (matVecRoutines.c, initialization.c)
1. Separate out Vnl * vector from Hamiltonian * vector.
2. Use LAPACK for subspace eigenproblem when Nstates <= 2000.

----------------
Jan 02, 2019
Name: Qimen Xu
Changes: (nonlocal.c, nonlocal.h)
1. Convert dos/mac newlines to unix format.


----------------
Dec 22, 2018
Name: Qimen Xu
Changes: (eigensolver.c, initialization.c, finalization.c)
1. Provide more workspace for ScaLAPACK eigensolver.
2. Use LAPACK for solving generalized eigenvalue problems
   when Nstates <= 3000.
3. Free D2D targets.


----------------
Dec 20, 2018
Name: Qimen Xu
Changes: (readfiles.c)
1. Correct misspelled flag "EXCHAGE_CORRELATION" to
   "EXCHANGE_CORRELATION". Modify corresponding input test files.


----------------
Dec 16, 2018
Name: Qimen Xu
Changes: (initialization.c, eigensolver)
1. Switch to LAPACK for subspace eigenproblem when Nstates <= 1000.

----------------
Dec 08, 2018
Name: Abhiraj Sharma
Changes: (md.c)
1. Added thermostat temperature variation option in md


----------------
Dec 07, 2018
Name: Qimen Xu, Abhiraj Sharma
Changes: (electronicGroundState.c, eigensolver.c, initialization.c, relax.c)
1. Improve DD2DD. Split DD2DD into three steps: find target, transfer data,
free target object, where finding target is only done once. (by Qimen)
2. Add FIRE algorithm. (by Abhiraj)

----------------
Dec 06, 2018
Name: Abhiraj Sharma
Changes: (relax.c)
1. Added LBFGS and constraint on atoms for relaxation


----------------
Nov 27, 2018
Name: Qimen Xu
Changes: (electronicGroundState.c)
1. Improve bounds for calculating fermi level. TODO: change accordingly when
   k-points are added.


----------------
Nov 25, 2018
Name: Qimen Xu
Changes: (electronicGroundState.c, eigensolver.c, initialization.c)
1. Remove scaling rho to impose the condition int_rho + int_b = 0. This will
   reduce unnecessary global communication time in rho calculation.
2. Use previous minimum eigenvalue as lowerbound for Chebyshev filtering,
   instead of calculating lowerbound using Lanczos in each scf iteration.
3. Disable default pseudopotential filenames and default path.


----------------
Nov 06, 2018
Name: Qimen Xu
Changes: (finalization.c)
1. Deallocate local and nonlocal influencing atom info. and nonlocal projectors
   in the end of each electronic ground state calculation.

----------------
-Date:  05 Nov 2018
-Name:   Qimen, Abhiraj
-changes: Added Dirichlet boundary condition (Qimen) and Molecular dynamics (Abhiraj)



----------------
Oct 30, 2018
Name: Qimen Xu
Changes: (finalization.c)
1. Bug fix: in the case nonlocal influencing atom = 0, the corresponding
			Atom_Influence_nloc arrays are not allocated, thus
			deallocating those arrays will cause error.


----------------
Oct 27, 2018
Name: Qimen Xu
Changes: (electrostatics.c, nonlocal.c)
1. Spherical harmonics formula l = 6, m = 3 fixed, const C6p3 changed.
2. Multiple moment expansion bug: radius (y2) not calculated correctly
   during calculation of the multipole moments Qlm.


----------------
Oct 18, 2018
Name: Qimen Xu
Changes: (electrostatics.c, tools.c)
1. Bug fix: interpolation error when len = 0.
   Sol: return directly if len = 0. (tools.c)
2. Bug fix: error when there's no nonlocal influencing atom in current
            process domain, which will cause a segment fault error.
   Sol: specify influencing n_atom = 0 in such cases. (electrostatics.c)


----------------
Oct 16, 2018
Name: Qimen Xu
Changes: (nonlocal.c, force.c)
1. Bug fix: MKL cblacs_pdgemm error for systems with hydrogen atoms.
   Sol: skip nonlocal component if nproj = 0.


----------------
Oct 12, 2018
Name: Qimen Xu
Changes: (initialization.c)
1. Print last modified date of the code to output file.
2. Create ChangeLog