Bugfix: Whether an HMR fit was successful with difficult input data was platform dependent. Fits are now reported as failed with error message "Dubious fit (extreme standard error)" if the p value of the flux is > (1 - .Machine$double.neg.eps), which ensures that fits with extreme standard error of the flux estimate (in particular SE = Inf) are not reported anymore. This impacts the result of the recommended simulation to determine f.detect and thus the flux selection in selectfluxes on some platforms, especially on Windows systems. fdetect will be lower leading to successful HMR fits being more likely to be selected but there will also be less successful but dubious HMR fits.
Fixed for CRAN note (regarding a documentation link).
Fixed an embarrassing maths error in the documentation.
Following a suggestion by Thomas Gremmen, diagnostic plots are now in a single PDF file. The (potentially numerous) PNG files created by previous versions caused issues on some network drives.
The old deprecated original HMR algorithm has been removed. Old deprecated flux selection algorithms have been removed. Removed dependency on package AICcmodavg.
Warning handlers did not work correctly.
There was a scoping issue with the plot parameter in gasfluxes if one of the ID columns was called "plot".
selectfluxes has now gained a tolerance parameter to enable fine-tuning the check introduced in the previous version.
CRAN has asked for a package rebuild with the newest knitr version due to encoding issues in the vignette.
Added Pearson's correlation coefficient to output of linear fit.
Breaking change: Flux selection has now its own function selectfluxes. This makes implementing new algorithms easier. Please consult help("selectfluxes") for usage examples.
Implemented Roman Hueppi's flux selection algorithm.
New Vignette included.