Only FPM/Meson/Visual-Studio are supported,
other build systems can copy source files directly,
and ifort/ifx and gfortran compilers are tested.
To use nnps within your fpm project, add the following lines to your fpm.toml file:
[dependencies]
nnps = { git="https://github.com/zoziha/nnps" }Note: Windows-ifort bug, Using static compilation will cause problems with allocatable arrays! Please use the /libs:dll compile option!
For 2D/3D NNPS, OpenMP is used for parallel acceleration, and parallel threads can be set through the OMP_NUM_THREADS environment variable.
> fpm run --example --all # run all examplesprogram example_grid2d
use nnps_module, only: nnps_grid2d, wp
use display_module, only: display
implicit none
type(nnps_grid2d) :: nnps
real(wp), dimension(2, 4) :: loc = reshape([0.0_wp, 1.0_wp, 2.0_wp, 1.5_wp, &
1.0_wp, 1.0_wp, 0.5_wp, 1.0_wp], [2, 4])
integer, pointer :: pairs(:)
real(wp), pointer :: rdxs(:)
call nnps%init(loc, n=4)
call nnps%query(0.6_wp, pairs, rdxs, n=4)
print *, '*** grid find (2D)'
call display(pairs, 'pairs index:', brief=.false.)
call display(loc(:, pairs), 'pairs coordinates:', brief=.false.)
call display(rdxs, 'rdxs:', brief=.false.)
end program example_grid2d
! *** grid find (2D)
! [vector: 4] pairs index:
! 3, 4, 4, 1
! [matrix: 2*4] pairs coordinates:
! 1.000E+00, 5.000E-01, 5.000E-01, 0.000E+00;
! 1.000E+00, 1.000E+00, 1.000E+00, 1.000E+00
! [vector: 6] rdxs:
! 5.000E-01, 5.000E-01, 0.000E+00, 5.000E-01, 5.000E-01, 0.000E+00Originally a study of the NNPS algorithm for SPH particle methods, this library has now been introduced into the SPH code, see zoziha/SPH.