j-iwata/RSDFT_DEVELOP

RSDFT

NOTICE

The format of the input parameters for atomic structure optimization have been changed. If you use the old one, for example,

ATOMOPT1  50  6  5                     / ncycle  most  nrfr
ATOMOPT2  0.5d0  1.d-10  5.d-4  1.d-1  / okatom  eeps  feps  decr
ATOMOPT3  100                          / diter_opt

please change as follows

ATOMOPT1  50  5.d-4  .f.        / ncycle  feps  continue_flag
ATOMOPT2  6  5  100             / most  nrfr  diter_opt
ATOMOPT3  0.5d0  1.d-1  1.d-10  / okatom  decr  eeps

If you use CG optimization (SWOPT=1 or 2), continue_flag is not used.

You can also omit the other parameters. A minimal example is,

ATOMOPT1  50  5.d-4  / ncycle  feps

In this case the omitted parameters are set by default values.

ITO (Kyusyu university)

Compile

module load intel fftw
cd RSDFT/src
cp make.inc.ITO.complex16 make.inc
make

after a few minutes, you get rsdft.x which is the execution file for the calculation with double-complex wavefunctions. The double-complex rsdft.x is necessary for the calculaiotns with BZ sampling for k/=(0,0,0) and band-structure calculations.

If your k-point sampling is only Γ point, you can use rsdft.x for the calculation with double-real wavefunctions. The compilation procedure is

module load intel fftw
cd RSDFT/src
cp make.inc.ITO.real8 make.inc
make

With the doube-real rsdft.x, you can also perform RS-CPMD calculations.

Samples

Please see samples_for_supercomputers/ITO_KYUSHU-U/ .

OHTAKA (System B, ISSP)

Preparation of MKL's FFTW

First we need to prepare the wrapper of FFTW for MPI-Fortran. Execute following commands:

mkdir -p $HOME/intel/mkl
cp -r $MKLROOT/interfaces/fftw3x_cdft/ $HOME/intel/mkl/
cp -r $MKLROOT/include $HOME/intel/
cd intel/mkl/fftw3x_cdft
make libintel64

Then you can find libfftw3x_cdft_lp64.a in $HOME/intel/lib/intel64/.

Compile

The complex16 version of rsdft.x can be compiled as follows:

cd RSDFT/src
cp make.inc.complex16.OHTAKA make.inc
make

and the real8 version is

cd RSDFT/src
cp make.inc.real8.OHTAKA make.inc
make

FUGAKU (RIKEN)

FLOW-FX (Nagoya university)

The complex16 version of rsdft.x can be compiled as follows:

module load fftw
cd RSDFT/src
cp make.inc.FLOW-FX.complex16 make.inc
make

and the real8 version is

module load fftw
cd RSDFT/src
cp make.inc.FLOW-FX.real8 make.inc
make

Samples

Please see samples_for_supercomputers/FLOW-FX_NAGOYA-U/ .

REEDBUSH-H (The university of Tokyo)

The complex16 version of rsdft.x can be compiled as follows:

cd RSDFT/src
cp make.inc.REEDBUSH-H.complex16 make.inc
make

and the real8 version is

cd RSDFT/src
cp make.inc.REEDBUSH-H.real8 make.inc
make

Samples

Please see samples_for_supercomputers/REEDBUSH-H_U-TOKYO/ .