molgri.molecules.transitions

Combine a ParsedTrajectory and a FullGrid to generate a MSM or SqRA model.

In this module, the two methods of evaluating transitions between states - the MSM and the SqRA approach - are implemented.

Classes

MSM(sim_hist[, tau_array, use_saved])	Markov state model (MSM) works on simulation trajectories by discretising the accessed space and counting transitions between states in the time span of tau.
SQRA(sim_hist, **kwargs)	As opposed to MSM, this object works with a pseudo-trajectory that evaluates energy at each grid point and with geometric parameters of position space division.
SimulationHistogram(parsed_trajectory, full_grid)	This class takes the data from ParsedTrajectory and combines it with a specific FullGrid so that each simulation step can be assigned to a cell and the occupancy of those cells evaluated.
TransitionModel(sim_hist[, tau_array, use_saved])	A class that contains both the MSM and SQRA models.

class molgri.molecules.transitions.SimulationHistogram(parsed_trajectory: ParsedTrajectory, full_grid: FullGrid)

This class takes the data from ParsedTrajectory and combines it with a specific FullGrid so that each simulation step can be assigned to a cell and the occupancy of those cells evaluated.

__init__(parsed_trajectory: ParsedTrajectory, full_grid: FullGrid)

get_name() → str

get_all_assignments() → Tuple[ndarray[Any, dtype[ScalarType]], ndarray[Any, dtype[ScalarType]]]

For each step in the trajectory assign which cell it belongs to. Uses the default atom selection of the ParsedTrajectory object.

Returns:

(all centers of mass within grid, all assignments to grid cells)

class molgri.molecules.transitions.TransitionModel(sim_hist: SimulationHistogram, tau_array: ndarray[Any, dtype[ScalarType]] | None = None, use_saved: bool = False)

A class that contains both the MSM and SQRA models.

__init__(sim_hist: SimulationHistogram, tau_array: ndarray[Any, dtype[ScalarType]] | None = None, use_saved: bool = False)

get_name() → str

abstract get_transitions_matrix() → ndarray[Any, dtype[ScalarType]]

For MSM, generate the transition matrix from simulation data. For SQRA, generate rate matrix from the point energy calculations.

Returns:

an array of shape (num_tau, num_cells, num_cells) for MSM or (n1, num_cells, num_cells) for SQRA

get_eigenval_eigenvec(num_eigenv: int = 15, **kwargs) → Tuple[ndarray[Any, dtype[ScalarType]], ndarray[Any, dtype[ScalarType]]]

Obtain eigenvectors and eigenvalues of the transition matrices.

Args:

num_eigenv: how many eigenvalues/vectors pairs to return (too many may give inaccurate results) **kwargs: named arguments to forward to eigs()

Returns:

(eigenval, eigenvec) a tuple of eigenvalues and eigenvectors, first num_eigv given for all tau-s Eigenval is of shape (num_tau, num_eigenv), eigenvec of shape (num_tau, num_cells, num_eigenv)

class molgri.molecules.transitions.MSM(sim_hist: SimulationHistogram, tau_array: ndarray[Any, dtype[ScalarType]] | None = None, use_saved: bool = False)

Markov state model (MSM) works on simulation trajectories by discretising the accessed space and counting transitions between states in the time span of tau.

get_transitions_matrix(noncorr: bool = False) → ndarray[Any, dtype[ScalarType]]

Obtain a set of transition matrices for different tau-s specified in self.tau_array.

Args:

noncorr: bool, should only every tau-th frame be used for MSM construction

(if False, use sliding window - much more expensive but throws away less data)

Returns:

an array of transition matrices of shape (self.num_tau, self.num_cells, self.num_cells)

class molgri.molecules.transitions.SQRA(sim_hist: SimulationHistogram, **kwargs)

As opposed to MSM, this object works with a pseudo-trajectory that evaluates energy at each grid point and with geometric parameters of position space division.

__init__(sim_hist: SimulationHistogram, **kwargs)

get_transitions_matrix(D: float = 1, energy_type: str = 'Potential') → ndarray[Any, dtype[ScalarType]]

Return the rate matrix as calculated by the SqRA formula:

Q_ij = np.sqrt(pi_j/pi_i) * D * S_ij / (h_ij * V_i)

Args:

D: diffusion constant energy_type: the keyword to pass to self.sim_hist.parsed_trajectory.get_all_energies to obtain energy information at centers of cells

Returns:

a (1, self.num_cells, self.num_cells) array that is a rate matrix estimated with SqRA (first dimension is expanded to be compatible with the MSM model)