Simulation

#include "Spirit/Simulation.h"

This API of Spirit is used to run and monitor iterative calculation methods.

If many iterations are called individually, one should use the single shot simulation functionality. It avoids the allocations etc. involved when a simulation is started and ended and behaves like a regular simulation, except that the iterations have to be triggered manually.

Definition of solvers

Note that the VP and LBFGS Solvers only do direct minimization and not suitable for dynamics.

Solver_VP 0: VP: Verlet-like velocity projection

Solver_SIB 1: SIB: Semi-implicit midpoint method B

Solver_Depondt 2: Depondt: Heun method using rotations

Solver_Heun 3: Heun: second-order midpoint

Solver_RungeKutta4 4: RK4: 4th order Runge-Kutta

Solver_LBFGS_OSO 5: LBFGS_OSO: Limited memory Broyden-Fletcher-Goldfarb-Shanno, exponential transform

Solver_LBFGS_Atlas 6: LBFGS_Atlas: Limited memory Broyden-Fletcher-Goldfarb-Shannon, stereographic projection

Solver_VP_OSO 7: Solver_VP_OSO: Verlet-like velocity projection, exponential transform

Definition of Monte Carlo Algorithms

MC_Algorithm_Metropolis 0: Metropolis: Metropolis algorithm

MC_Algorithm_Metropolis_MDC 1: Metropolis_MDC: Magnetization Direction Constrained Metropolis algorithm

Run info

struct Simulation_Run_Info

A struct that can be passed as an additional argument to the Simulation_XXX_Start methods to gather some basic information about the simulation run

Public Members

int total_iterations = 0: total number of iterations

int total_walltime = 0: total walltime passed in miliseconds

scalar total_ips = 0: total iterations per second

scalar max_torque = 0: maximum torque at the end of the simulation

int n_history_iteration = 0: number of iteration entries in history

int *history_iteration = nullptr: history of iteration counts

int n_history_max_torque = 0: number of max torque entries in history

scalar *history_max_torque = nullptr: history of max torque values

int n_history_energy = 0: number of energy entries in history

scalar *history_energy = nullptr: history of total energy values

void free_run_info(Simulation_Run_Info info): Free the memory of an acquired Simulation_Run_Info struct. This struct is allocated when a legal address is passed to the simulation on start for the info parameter.

Start or stop a simulation

void Simulation_MC_Start(State *state, int mc_algorithm = MC_Algorithm_Metropolis, int n_iterations = -1, int n_iterations_log = -1, bool singleshot = false, Simulation_Run_Info *info = nullptr, int idx_image = -1, int idx_chain = -1): Monte Carlo.

void Simulation_LLG_Start(State *state, int solver_type, int n_iterations = -1, int n_iterations_log = -1, bool singleshot = false, Simulation_Run_Info *info = nullptr, int idx_image = -1, int idx_chain = -1): Landau-Lifshitz-Gilbert dynamics and energy minimisation.

void Simulation_GNEB_Start(State *state, int solver_type, int n_iterations = -1, int n_iterations_log = -1, bool singleshot = false, Simulation_Run_Info *info = nullptr, int idx_chain = -1): Geodesic nudged elastic band method.

void Simulation_MMF_Start(State *state, int solver_type, int n_iterations = -1, int n_iterations_log = -1, bool singleshot = false, Simulation_Run_Info *info = nullptr, int idx_image = -1, int idx_chain = -1): Minimum mode following method.

void Simulation_EMA_Start(State *state, int n_iterations = -1, int n_iterations_log = -1, bool singleshot = false, Simulation_Run_Info *info = nullptr, int idx_image = -1, int idx_chain = -1): Eigenmode analysis.

void Simulation_SingleShot(State *state, int idx_image = -1, int idx_chain = -1)

Single iteration of a Method

If singleshot=true was passed to Simulation_..._Start before, this will perform one iteration. Otherwise, nothing will happen.

void Simulation_N_Shot(State *state, int N, int idx_image = -1, int idx_chain = -1)

N iterations of a Method

If singleshot=true was passed to Simulation_..._Start before, this will perform N iterations. Otherwise, nothing will happen.

void Simulation_Stop(State *state, int idx_image = -1, int idx_chain = -1): Stop a simulation running on an image or chain.

void Simulation_Stop_All(State *state): Stop all simulations.

Get information

scalar Simulation_Get_MaxTorqueNorm(State *state, int idx_image = -1, int idx_chain = -1)

Get maximum torque norm.

If a MC, LLG, MMF or EMA simulation is running this returns the max. torque on the current image.

If a GNEB simulation is running this returns the max. torque on the current chain.

void Simulation_Get_Chain_MaxTorqueNorms(State *state, scalar *torques, int idx_chain = -1)

Get maximum torque norms on the images of a chain.

Will only work if a GNEB simulation is running.

scalar Simulation_Get_IterationsPerSecond(State *state, int idx_image = -1, int idx_chain = -1)

If a MC, LLG, MMF or EMA simulation is running this returns the IPS on the current image.

If a GNEB simulation is running this returns the IPS on the current chain.

Returns:: the iterations per second (IPS).

int Simulation_Get_Iteration(State *state, int idx_image = -1, int idx_chain = -1)

Returns:: the number of iterations performed by the current simulation so far.

scalar Simulation_Get_Time(State *state, int idx_image = -1, int idx_chain = -1)

int Simulation_Get_Wall_Time(State *state, int idx_image = -1, int idx_chain = -1)

Whether a simulation is running

bool Simulation_Running_On_Image(State *state, int idx_image = -1, int idx_chain = -1): Check if a simulation is running on specific image of specific chain.

bool Simulation_Running_On_Chain(State *state, int idx_chain = -1): Check if a simulation is running across a specific chain.

bool Simulation_Running_Anywhere_On_Chain(State *state, int idx_chain = -1): Check if a simulation is running on any or all images of a chain.