NEWS
fastFMM 1.0.1.9000
fastFMM 1.0.1 (2026-05-18)
fui() now only calls get_functional_covariates() for concurrent = TRUE, preventing issues with finding columns with shared prefixes (#8)- Rebuilt under R 4.6.0
- Step 3.1 in
G_estimate_randint() now calls message() instead of print() to be consistent with other messages
- References to Xin et al. (2025) reflect publication of the reviewed preprint on eLife
- Additional comments and spacing changes in
fui.R
fastFMM 1.0.0 (2026-01-08)
- Added concurrent models to
fui(), allowing for fitting data with both functional outcomes and functional covariates.
- Rewrote all standard functions to S3 generics to allow for the same functions to handle both non-concurrent and concurrent functionality. This may need to be refined into S4 and R6 to prevent strange function exports.
- Added documentation for various helpers, which are exported somewhat messily to allow for the main calculation of
fui().
- Added datasets
lick and d2pvt to demonstrate fui() in the vignettes fastFMM and d2pvt, respectively. These datasets replace the previously used synthetic data.
- Updated references to the concurrent model (Xin et al. (2025)) and the data (Jeong et al. (2022), Machen et al. (2025)).
- Minor bug fix in
plot_fui()'s. Adding geom_segment() axes no longer rely on the deprecated ggplot2::aes_string() method.
- Setting
parallel = TRUE now requires n_cores to be manually specified. This avoids problems with asking for too many simultaneous processes on high-performance clusters if the user does not strictly specify the number of available threads.
fastFMM 0.4.0 (2025-03-13)
- Provided pointers to a Python package to call fastFMM from Python.
- Provided pointers to user guides written in Python.
- Updated reference/citations on documentation.
fastFMM 0.3.0 (2024-09-29)
- Fixed bugs.
- Added (optional) parallelization of step 3.2 in analytic inference fui(), leading to substantial speed ups of fui().
- Added in parallelization functionality for PCs.
- Added in code to remove rows with missing functional outcome values and added in option to impute with longitudinal FPCA (experimental feature).
- Changed default method of moments estimator to MoM=1 (appears to perform comparably to MoM=2 but is much faster and less memory intensive).
- Removed some fui() arguments that were not in use.
fastFMM 0.2.0 (2024-02-02)
fastFMM 0.1.0 (2023-11-14)