Module optimisation

Module for the optimisation of the frame order models.

Classes

Frame_order_grid_command
Command class for relaxation dispersion optimisation on the slave processor.

Frame_order_memo
The frame order memo class.

Frame_order_minimise_command
Command class for relaxation dispersion optimisation on the slave processor.

Frame_order_result_command
Class for processing the frame order results.

Functions

count_sobol_points(target_fn=None, verbosity=1)
Count the number of Sobol' points for the current parameter values of the model.

list of float grid_row(incs, lower, upper, dist_type=None, end_point=True)
Set up a row of the grid search for a given parameter.

numpy rank-3 array, numpy rank-1 array. minimise_setup_atomic_pos(sim_index=None, verbosity=1)
Set up the atomic position data structures for optimisation using PCSs and PREs as base data sets.

tuple of (numpy rank-2 array, numpy rank-2 array, numpy rank-2 array, numpy rank-1 array, numpy rank-1 array) minimise_setup_pcs(sim_index=None)
Set up the data structures for optimisation using PCSs as base data sets.

tuple of (numpy rank-2 array, numpy rank-2 array, numpy rank-2 array, numpy rank-3 array, numpy rank-2 array, numpy rank-2 array) minimise_setup_rdcs(sim_index=None)
Set up the data structures for optimisation using RDCs as base data sets.

tuple of 3 numpy nx5D, rank-1 arrays minimise_setup_tensors(sim_index=None)
Set up the data structures for optimisation using alignment tensors as base data sets.

bool opt_uses_align_data(align_id=None)
Determine if the PCS or RDC data for the given alignment ID is needed for optimisation.

bool opt_uses_pcs(align_id)
Determine if the PCS data for the given alignment ID is needed for optimisation.

bool opt_uses_rdc(align_id)
Determine if the RDC data for the given alignment ID is needed for optimisation.

store_bc_data(A_5D_bc=None, pcs_theta=None, rdc_theta=None)
Store the back-calculated data.

target_fn_data_setup(sim_index=None, verbosity=1, scaling_matrix=None, unset_fail=False)
Initialise the target function for optimisation or direct calculation.

unpack_opt_results(param_vector=None, func=None, iter_count=None, f_count=None, g_count=None, h_count=None, warning=None, scaling_matrix=None, sim_index=None)
Unpack and store the Frame Order optimisation results.

Variables

__package__ = 'specific_analyses.frame_order'

Imports: cos, pi, generic_minimise, grid_point_array, arccos, array, dot, float64, ndarray, ones, zeros, inv, norm, sys, warn, is_float, isNaN, isInf, RelaxError, RelaxInfError, RelaxNaNError, RelaxNoPCSError, RelaxNoRDCError, tmax_pseudo_ellipse_array, MODEL_DOUBLE_ROTOR, MODEL_FREE_ROTOR, MODEL_ISO_CONE, MODEL_ISO_CONE_FREE_ROTOR, MODEL_ISO_CONE_TORSIONLESS, MODEL_LIST_FREE_ROTORS, MODEL_LIST_PSEUDO_ELLIPSE, MODEL_PSEUDO_ELLIPSE, MODEL_PSEUDO_ELLIPSE_FREE_ROTOR, MODEL_PSEUDO_ELLIPSE_TORSIONLESS, MODEL_RIGID, MODEL_ROTOR, wrap_angles, periodic_table, dipolar_constant, RelaxWarning, Memo, Result_command, Slave_command, interatomic_loop, return_spin, spin_loop, pipe_centre_of_mass, check_domain, check_model, check_parameters, base_data_types, domain_moving, pivot_fixed, tensor_loop, assemble_param_vector, linear_constraints, Frame_order, sobol_data

Function Details

count_sobol_points(target_fn=None, verbosity=1)

Count the number of Sobol' points for the current parameter values of the model.

The count will be stored in the current data pipe and printed out.

Parameters:

• target_fn (target_functions.frame_order.Frame_order instance) - The pre-initialised frame order target function class.
• verbosity (int) - If greater than 0, lots of information will be printed out.

grid_row(incs, lower, upper, dist_type=None, end_point=True)

Set up a row of the grid search for a given parameter.

Parameters:

• incs (int) - The number of increments.
• lower (float) - The lower bounds.
• upper (float) - The upper bounds.
• dist_type (None or str) - The spacing or distribution type between grid nodes. If None, then a linear grid row is returned. If 'acos', then an inverse cos distribution of points is returned (e.g. for uniform sampling in angular space).
• end_point (bool) - A flag which if False will cause the end point to be removed.

Returns: list of float

The row of the grid.

minimise_setup_atomic_pos(sim_index=None, verbosity=1)

Set up the atomic position data structures for optimisation using PCSs and PREs as base data sets.

Parameters:

• sim_index (None or int) - The index of the simulation to optimise. This should be None if normal optimisation is desired.
• verbosity (int) - If set to 1 or higher, then printouts and warnings will be active.

Returns: numpy rank-3 array, numpy rank-1 array.

The atomic positions (the first index is the spins, the second is the structures, and the third is the atomic coordinates) and the paramagnetic centre.

minimise_setup_pcs(sim_index=None)

Set up the data structures for optimisation using PCSs as base data sets.

Parameters:

• sim_index (None or int) - The index of the simulation to optimise. This should be None if normal optimisation is desired.

Returns: tuple of (numpy rank-2 array, numpy rank-2 array, numpy rank-2 array, numpy rank-1 array, numpy rank-1 array)

The assembled data structures for using PCSs as the base data for optimisation. These include:

• the PCS values.
• the unit vectors connecting the paramagnetic centre (the electron spin) to
• the PCS weight.
• the nuclear spin.
• the pseudocontact shift constants.

minimise_setup_rdcs(sim_index=None)

Set up the data structures for optimisation using RDCs as base data sets.

Parameters:

• sim_index (None or int) - The index of the simulation to optimise. This should be None if normal optimisation is desired.

Returns: tuple of (numpy rank-2 array, numpy rank-2 array, numpy rank-2 array, numpy rank-3 array, numpy rank-2 array, numpy rank-2 array)

The assembled data structures for using RDCs as the base data for optimisation. These include:

• rdc, the RDC values.
• rdc_err, the RDC errors.
• rdc_weight, the RDC weights.
• vectors, the interatomic vectors.
• rdc_const, the dipolar constants.
• absolute, the absolute value flags (as 1's and 0's).

minimise_setup_tensors(sim_index=None)

Set up the data structures for optimisation using alignment tensors as base data sets.

Parameters:

• sim_index (None or int) - The simulation index. This should be None if normal optimisation is desired.

Returns: tuple of 3 numpy nx5D, rank-1 arrays

The assembled data structures for using alignment tensors as the base data for optimisation. These include:

• full_tensors, the full tensors as concatenated arrays.
• full_err, the full tensor errors as concatenated arrays.
• full_in_ref_frame, the flags specifying if the tensor is the full or reduced tensor in the non-moving reference domain.

opt_uses_align_data(align_id=None)

Determine if the PCS or RDC data for the given alignment ID is needed for optimisation.

Parameters:

• align_id (str) - The optional alignment ID string.

Returns: bool

True if alignment data is to be used for optimisation, False otherwise.

opt_uses_pcs(align_id)

Determine if the PCS data for the given alignment ID is needed for optimisation.

Parameters:

• align_id (str) - The alignment ID string.

Returns: bool

True if the PCS data is to be used for optimisation, False otherwise.

opt_uses_rdc(align_id)

Determine if the RDC data for the given alignment ID is needed for optimisation.

Parameters:

• align_id (str) - The alignment ID string.

Returns: bool

True if the RDC data is to be used for optimisation, False otherwise.

store_bc_data(A_5D_bc=None, pcs_theta=None, rdc_theta=None)

Store the back-calculated data.

Parameters:

• A_5D_bc (numpy float64 array) - The reduced back-calculated alignment tensors from the target function.
• pcs_theta (numpy float64 array) - The back calculated PCS values from the target function.
• rdc_theta (numpy float64 array) - The back calculated RDC values from the target function.

target_fn_data_setup(sim_index=None, verbosity=1, scaling_matrix=None, unset_fail=False)

Initialise the target function for optimisation or direct calculation.

Parameters:

• sim_index (None or int) - The index of the simulation to optimise. This should be None if normal optimisation is desired.
• verbosity (int) - The amount of information to print. The higher the value, the greater the verbosity.
• scaling_matrix (numpy rank-2, float64 array or None) - The diagonal and square scaling matrices.
• unset_fail (bool) - A flag which if True will cause a RelaxError to be raised if the parameter is not set yet.

unpack_opt_results(param_vector=None, func=None, iter_count=None, f_count=None, g_count=None, h_count=None, warning=None, scaling_matrix=None, sim_index=None)

Unpack and store the Frame Order optimisation results.

Parameters:

• param_vector (numpy array) - The model-free parameter vector.
• func (float) - The optimised chi-squared value.
• iter_count (int) - The number of optimisation steps required to find the minimum.
• f_count (int) - The function count.
• g_count (int) - The gradient count.
• h_count (int) - The Hessian count.
• warning (str or None) - Any optimisation warnings.
• scaling_matrix (numpy rank-2, float64 array or None) - The diagonal and square scaling matrices.
• sim_index (None or int) - The index of the simulation to optimise. This should be None for normal optimisation.