- added missing char array length arguments to Fortran calls.
- fix an issue with the
group
argument when the first group(s) is ignored there is a "hole" in the subsequent window. This previously caused a crash.
- added ORCID and used
rng = false
inRcpp::export
attributes will have reduced the computation time.
- the c++ code has been re-written. It now does essentially as described in the LINPACK user guide chapter 8 and 9. This is faster and more precise.
- handle data sets with gaps
- add the option to require a minimum number of observations in a window
- output with groups have changed such that the first window max
grp
value will have at leastwidth - 1L
distance frommin(grp)
inroll_regres
androll_regres.fit