Changelog • fmrihrf

fmrihrf 0.2.1

Fixed hrf_bspline() support handling so values for t > span (and t < 0) are zeroed instead of wrapping to onset-like values.
Fixed block_hrf() block integration to include quadrature step-size scaling, making amplitudes stable across precision.
Fixed hrf_sine() and hrf_fourier() to clamp support to [0, span] and return zero outside the modeled window.
Fixed normalise_hrf() to use fixed normalization constants computed on the HRF support, avoiding data-dependent scaling across evaluation grids.
Fixed evaluate.HRF() block-duration summation to use the same weighted integration scheme as block_hrf().
Fixed evaluate.Reg(normalize = TRUE) to normalize regressor outputs consistently across evaluation methods, including single-trial regressors with different durations.
Fixed block_hrf(summate = FALSE) to return normalized block integration (for both single- and multi-basis HRFs) instead of the legacy pointwise-maximum behavior.

CRAN release: 2026-02-09

New hrf_boxcar() function for simple boxcar (step function) HRFs with optional normalization.
New hrf_weighted() function for arbitrary weighted-window HRFs with constant or linear interpolation.
regressor() now accepts a list of HRF objects for trial-varying HRF designs.
New plot.Reg() method for visualizing regressor objects.
New plot_regressors() for comparing multiple regressors on one plot (ggplot2 or base R).
New plot_hrfs() for comparing multiple HRF shapes.
New print.HRF() method for concise HRF summaries.

Fixed critical bug in as_hrf() where parameters stored in the params attribute were never used at evaluation time. The fix creates a closure that properly captures and applies parameters during evaluation.

CRAN release: 2025-09-16