(git:5e7fe52)
Loading...
Searching...
No Matches
nnp_environment_types.F File Reference

Go to the source code of this file.

Data Types

type  nnp_environment_types::nnp_type
 Main data type collecting all relevant data for neural network potentials. More...
type  nnp_environment_types::nnp_symfgrp_type
 Symmetry functions group type. More...
type  nnp_environment_types::nnp_acsf_rad_type
 Set of radial symmetry function type. More...
type  nnp_environment_types::nnp_acsf_ang_type
 Set of angular symmetry function type. More...
type  nnp_environment_types::nnp_neigh_grp_type
 Per-SF-group dense neighbour container. cap is the allocated slab size; the live count is nnp_neighbor_typen_rad/n_ang1/n_ang2 for the matching group. ind(j) is the j-th neighbour's atom index; dist(1:3, j) is its displacement and dist(4, j) its norm. Sized lazily by nnp_neigh_grp_grow. More...
type  nnp_environment_types::nnp_neighbor_type
 Contains neighbors list of an atom (per-group dense layout). More...
type  nnp_environment_types::nnp_cell_list_cache_type
 Linked-cell + Verlet-skin cache for the NNP descriptor neighbour walk. Persisted across MD steps so the image pool, bin geometry, and reference positions survive between force evaluations and the head/next chain can be reused as long as no atom has drifted more than skin/2. Operated on by nnp_cell_list.F. More...
type  nnp_environment_types::nnp_neighbor_pair_map_type
 Species-pair routing table. For one (central element, neighbour element) pair, lists which radial / angular symmetry-function groups need that neighbour and the worst-case relevant cutoff. More...
type  nnp_environment_types::nnp_dgdr_grp_type
 Per-(element, SF-group) dG_k/dr buffer, sized to the group's n_symf (no max_*_symf padding) and the observed peak neighbour count for the group. Grown lazily (1.5x) so it settles at the true peak. More...
type  nnp_environment_types::nnp_neighbor_workspace_type
 Reusable per-element scratch / persistent caches for the ACSF descriptor and force assembly. See nnp_neighbor_interface for the routines that populate and grow these buffers. More...
type  nnp_environment_types::nnp_neighbor_interface_state_type
 Persistent neighbour-interface state. Owned by nnp_type so that the per-nnp pair-routing tables and per-element workspaces survive across MD steps without leaking between independent &NNP force_evals. Operated on by nnp_neighbor_interface.F. More...
type  nnp_environment_types::nnp_arc_type
 Data type for artificial neural networks. More...

Modules

module  nnp_environment_types
 Data types for neural network potentials.

Functions/Subroutines

subroutine, public nnp_environment_types::nnp_env_release (nnp_env)
 Release data structure that holds all the information for neural network potentials.
subroutine, public nnp_environment_types::nnp_cell_list_cache_release (cache)
 Free the allocatable parts of a cell-list cache. Co-located with the type definition to avoid creating a USE-cycle through nnp_cell_list.
subroutine, public nnp_environment_types::nnp_neighbor_interface_state_release (state)
 Free the allocatable parts of a neighbour-interface state. Co-located with the type definition to avoid creating a USE-cycle through nnp_neighbor_interface.
subroutine, public nnp_environment_types::nnp_env_get (nnp_env, nnp_forces, subsys, atomic_kind_set, particle_set, local_particles, molecule_kind_set, molecule_set, local_molecules, nnp_input, force_env_input, cell, cell_ref, use_ref_cell, nnp_potential_energy, virial)
 Returns various attributes of the nnp environment.
subroutine, public nnp_environment_types::nnp_env_set (nnp_env, nnp_forces, subsys, atomic_kind_set, particle_set, local_particles, molecule_kind_set, molecule_set, local_molecules, nnp_input, force_env_input, cell, cell_ref, use_ref_cell, nnp_potential_energy)
 Sets various attributes of the nnp environment.

Variables

integer, parameter, public nnp_environment_types::nnp_cut_cos = 1
 derived data types
integer, parameter, public nnp_environment_types::nnp_cut_tanh = 2
integer, parameter, public nnp_environment_types::nnp_actfnct_tanh = 1
integer, parameter, public nnp_environment_types::nnp_actfnct_gaus = 2
integer, parameter, public nnp_environment_types::nnp_actfnct_lin = 3
integer, parameter, public nnp_environment_types::nnp_actfnct_cos = 4
integer, parameter, public nnp_environment_types::nnp_actfnct_sig = 5
integer, parameter, public nnp_environment_types::nnp_actfnct_invsig = 6
integer, parameter, public nnp_environment_types::nnp_actfnct_exp = 7
integer, parameter, public nnp_environment_types::nnp_actfnct_softplus = 8
integer, parameter, public nnp_environment_types::nnp_actfnct_quad = 9