Optimise subject-specific cvxEDA slow time constant — optimize_gazepoint_cvxeda

Performs a dependency-light grid search over the slow Bateman impulse-response time constant used in cvxEDA-style EDA decomposition workflows. The default fast time constant is fixed at 0.7 seconds and the slow time constant is searched between 2 and 4 seconds. This function does not run the original cvxEDA optimisation; it provides a subject-specific tau-selection bridge for downstream cvxEDA-style workflows.

Usage

optimize_gazepoint_cvxeda_tau(
  dat,
  eda_col = "GSR_US",
  time_col = "CNT",
  group_cols = NULL,
  tau0_grid = seq(2, 4, by = 0.25),
  tau1 = 0.7,
  sampling_rate = NULL,
  ridge_lambda = 0.01,
  max_irf_seconds = 20
)

Arguments

dat: A data frame containing EDA data.
eda_col: Numeric EDA/conductance column.
time_col: Numeric time column.
group_cols: Optional grouping columns, usually participant/session.
tau0_grid: Candidate slow time constants.
tau1: Fixed fast time constant.
sampling_rate: Optional sampling rate in Hz. If NULL, estimated from time_col.
ridge_lambda: Small ridge penalty used in frequency-domain deconvolution.
max_irf_seconds: Maximum impulse-response duration.

Value

A list with overview, best_tau, optimization_table, and settings.