1. ALGORITHMS:
additions or changes to existing functional ability
2.
Ports to new machines or machine specific changes
5. DEFAULTS in input and throughout the code
7. MINOR or relatively insignificant changes
·
added a real space implementation of the nonlocal pseudopotential
projectors for energy
minization, forces , and stresses.
Even for large systems the G-space forces and stresses is
suggested.
· added option -qmaxmem=-1 to the sysvars.sp3nersc file. This allows unlimited memory
usage by the compiler to optimize. Also added -qhalt=w so it crashes on warnings.
· made workr2_psi in apply_ham_wavefn allocateable. It seems o200 compiler
can't have structure elements as array size.
Made blop_pol in setup_blochl_operator to be of size(*) for same reason.
· added if NL_rspace statements in destructors of hamiltonian and pseudopotential as PARATEC
was crashing at these points on IBMs, due to
the compiler not defining “associated” on undefined ptrs.
· added keywords NLPP_rspace .true. (default = .false.) to enact real space nonlocal PP calc.
· added keyword NLPP_rcut 4.2 4.7 …
to specify cutoff radius for real space. Each number is
for each type of atom.
· added keyword NLPP_rspace_force .true. (default = .false.) to enact real space nonlocal PP
force and stress calculations
· put in new input.pm from DJR in the perllib
· added pwparams%maxitdiag_band with keyword max_iter_diag_band : default is set at 40
· changed default of mxdlqp to 2001 from 1601
· put output of memory usage into routine print_mem_use.f90
· removed the test and inputs directories. A
new and better test directory will be implemented in the future.
· updated and added README files
· roved Archive.txt as it wasn’t used anymore
· removed shared do loop from angular.f90
· changed the name of null(:,:) in gwreal to null_map as null is a fortran intrinsic fuction
· removed sscal in symm_ident.f90. not needed. use vendor BLAS
· removed nspin from create_crystal as it already is defined in inputread