polyTEM.crystal_peaks.crystal_stack_3D.CrystalStack3D

class polyTEM.crystal_peaks.crystal_stack_3D.CrystalStack3D(stack_list, tilt_angles, domain_collections, auto_populate=True, orient_perpendicular=False)

Bases: object

Assumptions: * the sample datacube parameters were used to generate each of the crystal stacks. Thus, the resolution and shape is the same.

Variables:

  • stack_list – list of crystal stacks

  • tilt_angles – list of tilt angles corresponding to each stack in stack_list

  • domain_collections – list of DomainCollections correspond to each stack in stack_list

  • shape – (x,y) shape

  • resolution – resolution of tilt stack (afer sliding fft)

  • vector_field – each row is from one fft peak and contains (x,y,a,b,c) for vector field direction

  • threshold_func – function for peak finding

  • Ks (Kb,) – Frank Elastic Constants in Angstroms

__init__(stack_list, tilt_angles, domain_collections, auto_populate=True, orient_perpendicular=False)

Methods

__init__(stack_list, tilt_angles, ...[, ...])

adjacency_probability_matrix(savedir[, ...])

saves the adjacency probabilities in the directory 'savedir' with the format {savedir}/node{##}.npz, where ## is the starting row index for that submatrix.

align_vector_field(window_size, step_size, ...)

Shifts the (x,y) coordinates in the vector field based on the marker alignment results and convert to (u,v) where u = x/cos(alpha) and v=y

apply_threshold_func([threshold_func])

Sets the threshold function of every stack in the stack list

build_vector_graph(file_dir, node_range)

Build vector graph representation from adjacency probability matrix.

create_vector_field([rotate90_orientation])

Creates vector field (x,y,alpha,a,b,c), where x is the x-axis of the image (across the columns of the data) y is the y-axis of the image (across the rows of the data) alpha is the tilt angle a,b,c represents the unit vector field of chain orientation, centered at (x,y)

estimate_bending_stiffness([flip])

Estimates the bending stiffness by calculating the orientation correlation length Under the Worm-Like-Chain model, the orientation correlation length == bending stiffness

find_peak_threshold(alpha[, inverse])

Use the fft intensities for high tilt angles to estimate noise floor to set threshold function for fft peak finding of lower tilt angles

save(savedir[, prefix])

Saves the stacks based on their individual outdirs, with the prefix inserted in front of crystalstack.pkl Saves the tilt angles at savedir, tilt_angles.json Saves the vector field at savedir, vector_field.json
adjacency_probability_matrix(savedir, resolution=1.15, max_distance_pxl=32, alpha_cutoff=70, threads=8, node_group_size=100, model='nematic', **kwargs)

saves the adjacency probabilities in the directory ‘savedir’ with the format {savedir}/node{##}.npz, where ## is the starting row index for that submatrix. :param savedir: Directory to save adjacency probability sub-matrices :type savedir: str :param Resolution: original imaging resolution of raw TEM image :param max_distance: int, pixel limit to build possible connections :param alpha_cutoff: int, tilt angle cutoff to build possible connections :param threads: int, for multiprocessing :param node_group_size: int, to segment nodes for multiprocessing :param model: “single-chain” or “nematic” :param **kwargs: keyword-argument parameters for bending probability function :param –: :param Lp: float, A, persistence length for “single-chain” model :param Kb: Frank Elastic Constants for nematic, dimensionless :param Ks: Frank Elastic Constants for nematic, dimensionless

align_vector_field(window_size, step_size, alignment_result, alignment_indices, origin_shift=None)

Shifts the (x,y) coordinates in the vector field based on the marker alignment results and convert to (u,v) where u = x/cos(alpha) and v=y

this returns a vector field where (u,v) has the pixel resolution of the original image, before downsampling with the sliding fft

Parameters:

  • window_size – int, parameter used during sliding fft scan

  • step_size – int, parameter used during sliding fft scan

  • alignment_result – AlignmentResult object

  • alignment_indices – the image index in AlignmentResult that corresponds to the images used to create the CrystalStack3D

  • origin_shift – any origin shift applied during alignment

Returns:

vector field containing (u,v,alpha,a,b,c) of each fft peak feature

apply_threshold_func(threshold_func=None)

Sets the threshold function of every stack in the stack list

Parameters:

threshold_func – default None, uses self.threshold_func, otherwise input argument will override

build_vector_graph(file_dir, node_range)

Build vector graph representation from adjacency probability matrix.

The Adjacency Probability Matrix is stored in the format {prob_dir}/node{##}.npz where the ## is the start index of the first node in that numpy array. The graph is built by iterating through the probabilities in {prob_dir}/node{start_index) for start_index in node_range

Parameters:

  • file_dir (str) – Directory

  • node_range (range) – the set of starting indexes for all of the .npz files to iterate through e.g. np.arange(0,total_nodes,node_group_size) where node_group_size was set in CrystalStack3D.adjacency_probability_matrix()

Returns:

networkx graph

create_vector_field(rotate90_orientation=False)

Creates vector field (x,y,alpha,a,b,c), where x is the x-axis of the image (across the columns of the data)

y is the y-axis of the image (across the rows of the data) alpha is the tilt angle a,b,c represents the unit vector field of chain orientation, centered at (x,y)

estimate_bending_stiffness(flip=False)

Estimates the bending stiffness by calculating the orientation correlation length Under the Worm-Like-Chain model, the orientation correlation length == bending stiffness

Parameters:

flip – bool, swaps perpendicular and parallel orientations. i.e. set True if director orientation was obtained as “lamellar” but was actually “backbone”

Returns:

average orientation correlation length stderr: standard error of the mean of Ld

Return type:

Ld

find_peak_threshold(alpha, inverse=False)

Use the fft intensities for high tilt angles to estimate noise floor to set threshold function for fft peak finding of lower tilt angles

Parameters:

  • alpha – the angle cutoff to divide the tilt angles

  • inverse – default False, selects tilt angles greater than alpha. if True, selects tilt angles smaller.

Returns:

lambda function

Return type:

threshold_func

save(savedir, prefix='')

Saves the stacks based on their individual outdirs, with the prefix inserted in front of crystalstack.pkl Saves the tilt angles at savedir, tilt_angles.json Saves the vector field at savedir, vector_field.json