On Linux

It is recommended to install FLAME in a different directory than the source code.

1. First, install futile which is a set of utilities from the BigDFT project. Preferably, use the version provided with FLAME to avoid conflicts. Untar the included futile-suite.tar.gz (tar -zxvf futile-suite.tar.gz), then create a new build directory (e.g., mkdir futile-build ; cd futile-build), and from there run

   for GNU compilers:

   path_to_futile_source/Installer.py build futile -c CC=gcc CXX=g++ FC=gfortran F77=gfortran \ --with-ext-linalg="-llapack -lblas"

   for Intel compilers:

   path_to_futile_source/Installer.py build futile -c FCFLAGS=-O2 \ --with-ext-linalg="-L$MKLROOT/lib/intel64 \ -lmkl_rt -lmkl_scalapack_lp64 -lmkl_blacs_openmpi_lp64 -liomp5 -lm" \ CC=icc CXX=icpc FC=ifort F77=ifort

   for parallel compilation use the corresponding MPI wrappers:

   path_to_futile_source/Installer.py build futile -c FCFLAGS=-O2 \ --with-ext-linalg="-L$MKLROOT/lib/intel64 \ -lmkl_rt -lmkl_scalapack_lp64 -lmkl_blacs_openmpi_lp64 -liomp5 -lm" \ CC=mpicc CXX=mpicxx FC=mpif90 F77=mpif90

   follwed by make build if necessary. Make sure to adapt the library locations and the linking flags appropriately. After the installation of futile, we need to link it with FLAME. To display details on the general linking procedure, run:

   path_to_futile_source/Installer.py link futile

2. To compile FLAME, change into the main FLAME directory and run:

   autoreconf -fi

3. Create a build directory for FLAME (e.g., mkdir build-FLAME ; cd build-FLAME). Explicitly replace $FUTILE with the full path of the futile build-directory during configure, or define it as an environmental variable:

   export FUTILE=path_to_futile_build

   Note that providing the FUTILE variable is required to successfully compile FLAME, and is not optional. Then, run configure. For Intel compilers and MPI parallelization:

   path_to_flame_source/configure FC=mpif90 F77=mpif90 CXX=mpicc CC=mpicc \ FCFLAGS="-I$FUTILE/install/include -shared-intel -mcmodel=large  -mkl=sequential" \ CFLAGS=-mcmodel=large "LIBS=-L$FUTILE/install/lib \ -L/$MKLROOT/lib/intel64 -lfutile-1 -lmkl_rt \ -lmkl_scalapack_lp64 -lmkl_blacs_openmpi_lp64 -liomp5 -lm -lyaml -ldl -lrt -cxxlib"

   For GNU compilers without MKL:

   path_to_flame_source/configure FC=mpif90 F77=mpif90 CXX=mpicc CC=mpicc \ FCFLAGS="-I$FUTILE/install/include -mcmodel=large" \ CFLAGS=-mcmodel=large "LIBS=-L$FUTILE/install/lib \ -lfutile-1 -lm -lyaml -llapack -lfftw3 -ldl -cxxlib"

   To link with SPGLIB, append --with-spglib SPGLIB_ROOT=path_to_spglib

   To link with the BigDFT PSolver, append --with-bps BDIR=path_to_bigdft_root

   To link with LAMMPS, append --with-lammps LAMMPS_ROOT=path_to_lammps_root

4. Run make to compile the code. Upon successful compilation, the executable can be found in src/flame

yaml Format

Keyword-based structural information format. Must start with a block with keyword conf, followed by the following parameters.

bc: (string) Boundary condition.

default: No default value.

options:

free: Free boundary conditions for molecules and clusters

slab: Slab boundary conditions for surfaces, interfaces, and 2D materials

bulk: Periodic boundary conditions for crystals and solids

cell: (list of three lists) Three cell vectors.

default: No default value.

nat: (int) Number of atoms.

default: No default value.

units_length: (string) Length units used in the structure.

default: atomic

options:

angstrom: units in Angstrom

atomic: units in Bohr

coord: (list of nat lists) For every atom in the system, a list of the form [x, y, z, AT, CCC] must be given. x, y, z are the cartesian coordinates of the atom, while AT is the symbol of the particular atomic type, and CCC determines if its x, y, z degree of freedom are allowed to change during the simulation.

E.g., [0., 0., 0., Si, TTT] is a silicon atom at the origin, free to move in all dimensions, while [1., 1., 1., C, TFF] is a carbon atom at (1, 1, 1), only allowed to move along the x direction.

conf:
  nat                                 : 12
  bc                                  : bulk
  units_length                        : angstrom
  cell:
  -  [5.462, 0.0, 0.0]
  -  [0.0, 5.462, 0.0]
  -  [0.0, 0.0, 5.462]
  coord:
  -  [0.0, 0.0, 0.0, Ca, TTT]
  -  [2.731, 2.731, 0.0, Ca, TTT]
  -  [2.731, 0.0, 2.731, Ca, TTT]
  -  [0.0, 2.731, 2.731, Ca, TTT]
  -  [1.3655, 1.3655, 1.3655, F, TTT]
  -  [4.0965, 1.3655, 1.3655, F, TTT]
  -  [1.3655, 4.0965, 1.3655, F, TTT]
  -  [4.0965, 4.0965, 1.3655, F, TTT]
  -  [1.3655, 1.3655, 4.0965, F, TTT]
  -  [4.0965, 1.3655, 4.0965, F, TTT]
  -  [1.3655, 4.0965, 4.0965, F, TTT]
  -  [4.0965, 4.0965, 4.0965, F, TTT]

ascii Format

The ascii format is described here in detail: http://inac.cea.fr/sp2m/L_Sim/V_Sim/sample.html#sample_ascii It is the native format of the task minhocao and the only format that currently supports the constraint of individual lattice parameters. Periodic cell vectors have to be provided to describe the simulation cell. The cell consisting of the cell vectors a, b, c has to be rotated such that a points along the x direction, and b lies within the x-y plane. See below figure for the orientation of the cell and the projections dxx, dyx, dyy, dzx, dzy, and dzz.

line 1: (integer) Number of atoms in the system, Nat.

line 2: (real, real, real) dxx dyx dyy values

line 3: (real, real, real) dzx dzy dzz values

lines 4 – n: Only lines starting with #keyword: are allowed and read/interpreted accordingly. The #keyword: tag must be followed by the following options.

reduced: the atomic coordinates are treated as reduced coordinates, i.e., with respect to the cell vectors.

fixlat a b c alpha beta gamma vol: specific components of the lattice vectors can be fixed. A True (or T) value fixes the degree of freedom, while False (or F) does not fix it. The last parameter (vol) is used for fixed-cell-shape-variable-volume simulations. I.e., #keyword: fixlat  F F T T T F F allows the length of the a and b vectors to change, while keeping the length of the c vector constant. At the same time, only the angle between a and b is allowed to change. This setting is particularly useful for simulations of 2D materials or surfaces.

lines n+1 – Nat: Nat lines with the coordinates of every atom of the form x, y, z, AT, c. The x, y, z coordintes, followed by the chemical symbol AT, and optionally f for fixed atom. Note that the reduced coordinates will be fixed instead of the Cartesian one if the system is periodic

12
5.4620E+00   0.0000E+00   5.4620E+00
0.0000E+00   0.0000E+00   5.4620E+00
#keywords: fixlat F F T T T F F
0.0000E+00   0.0000E+00   0.0000E+00   Ca f
2.7310E+00   2.7310E+00   0.0000E+00   Ca f
2.7310E+00   0.0000E+00   2.7310E+00   Ca f
0.0000E+00   2.7310E+00   2.7310E+00   Ca f
1.3655E+00   1.3655E+00   1.3655E+00    F
4.0965E+00   1.3655E+00   1.3655E+00    F
1.3655E+00   4.0965E+00   1.3655E+00    F
4.0965E+00   4.0965E+00   1.3655E+00    F
1.3655E+00   1.3655E+00   4.0965E+00    F
4.0965E+00   1.3655E+00   4.0965E+00    F
1.3655E+00   4.0965E+00   4.0965E+00    F
4.0965E+00   4.0965E+00   4.0965E+00    F