pme_internal.h File Reference

#include "config.h"
#include <memory>
#include <vector>
#include "gromacs/math/gmxcomplex.h"
#include "gromacs/utility/alignedallocator.h"
#include "gromacs/utility/arrayref.h"
#include "gromacs/utility/defaultinitializationallocator.h"
#include "gromacs/utility/gmxassert.h"
#include "gromacs/utility/gmxmpi.h"
#include "gromacs/utility/mpicomm.h"
#include "gromacs/utility/unique_cptr.h"
#include "gromacs/utility/vectypes.h"

Include dependency graph for pme_internal.h:

This graph shows which files directly or indirectly include this file:

Description

This file contains function declarations necessary for computing energies and forces for the PME long-ranged part (Coulomb and LJ).

Author

Berk Hess hess@.nosp@m.kth..nosp@m.se

Mark Abraham mark..nosp@m.j.ab.nosp@m.raham.nosp@m.@gma.nosp@m.il.co.nosp@m.m

Classes
struct	pme_grid_comm_t
	Data structure for grid communication. More...
struct	pme_overlap_t
	Data structure for grid overlap communication in a single dimension. More...
struct	AtomToThreadMap
	Data structure for organizing particle allocation to threads. More...
class	SplineCoefficients
	Coefficients for theta or dtheta. More...
struct	splinedata_t
	Data structure for beta-spline interpolation. More...
struct	SlabCommSetup
	PME slab MPI communication setup. More...
class	PmeAtomComm
	Data structure for coordinating transfers between PME ranks along one dimension. More...
struct	pmegrid_t
	Data structure for a single PME grid. More...
struct	pmegrids_t
	Data structures for PME grids. More...
struct	PmeAndFftGrids
	The data for PME spread/gather plus FFT grids for one set of coefficients. More...
struct	gmx_pme_t
	Main PME data structure. More...

Macros
#define	PME_ORDER_MAX   12
	We only define a maximum to be able to use local arrays without allocation. An order larger than 12 should never be needed, even for test cases. If needed it can be changed here.

Typedefs
typedef struct gmx_parallel_3dfft *	gmx_parallel_3dfft_t
	A repeat of typedef from parallel_3dfft.h.
template<typename T >
using	AlignedVector = std::vector< T, gmx::AlignedAllocator< T >>
template<typename T >
using	FastVector = std::vector< T, gmx::DefaultInitializationAllocator< T >>

Functions
void	parallel_3dfft_destroy (gmx_parallel_3dfft *pfft_setup)
	Wrapper for gmx_parallel_3dfft_destroy to use as destructor with return type void.

Variables
static constexpr int	sc_numGridsLJLB = 7
	The number of grids for LJ-PME with LB combination rules.
static const real	lb_scale_factor []
	Pascal triangle coefficients scaled with (1/2)^6 for LJ-PME with LB-rules. More...
static const real	lb_scale_factor_symm [] = { 2.0 / 64, 12.0 / 64, 30.0 / 64, 20.0 / 64 }
	Pascal triangle coefficients used in solve_pme_lj_yzx, only need to do 4 calculations due to symmetry.

Variable Documentation

const real lb_scale_factor[]

static

Initial value:

= { 1.0 / 64,  6.0 / 64, 15.0 / 64, 20.0 / 64,
                                        15.0 / 64, 6.0 / 64, 1.0 / 64 }

Pascal triangle coefficients scaled with (1/2)^6 for LJ-PME with LB-rules.