Skip to contents

Extracts or computes various types of Symmetric Positive-Definite (SPD) matrix representations for subjects from a `hatsa_projector` object.

Usage

get_spd_representations(object, ...)

# S3 method for class 'hatsa_projector'
get_spd_representations(
  object,
  type = c("cov_coeffs", "fc_conn"),
  subject_idx = NULL,
  regularize_epsilon = 1e-06,
  subject_data_list_for_fc = NULL,
  k_conn_params_for_fc = list(),
  ...
)

# S3 method for class 'task_hatsa_projector'
get_spd_representations(
  object,
  type = c("cov_coeffs", "fc_conn", "fc_task", "fc_hybrid"),
  subject_idx = NULL,
  regularize_epsilon = 1e-06,
  subject_data_list_for_fc = NULL,
  k_conn_params_for_fc = list(),
  lambda_blend_for_hybrid = NULL,
  ...
)

Arguments

object

A `hatsa_projector` or `task_hatsa_projector` object.

...

Additional arguments, potentially passed to specific computation functions.

type

Character string indicating the type of SPD representation to use. Currently supported for `hatsa_projector`: `"cov_coeffs"` (covariance of aligned spectral coefficients). Other types like `"fc_conn"` might be added or supported by specific methods.

subject_idx

Optional integer or vector of integers. If provided, returns SPD matrices only for these subjects. If NULL (default), for all subjects.

regularize_epsilon

Small positive value for SPD regularization. Default from RGEOM-001.

subject_data_list_for_fc

Optional list of subject time-series matrices (T_i x V_p), needed for `type = "fc_conn"`.

k_conn_params_for_fc

List of parameters for `compute_subject_connectivity_graph_sparse`, e.g., `list(k_conn_pos = 10, k_conn_neg = 10, zscore_type = "abs", ...)`. Needed for `type = "fc_conn"`.

Value

If `subject_idx` has length 1, a single SPD matrix is returned. Otherwise a list of matrices is returned with `NULL` for subjects whose SPD representation could not be computed.