Changes in version 0.9.10 (2020-12-10)

Bugfixes and changes

analyse_SAR.CWOSLOSL()

• The recent changes in 0.9.9 improving the self-call had unwanted side effects if the input parameters were itself objects. For example: signal.integral.min = setting led to a missing parameter error; fixed (thanks to Thomas Kolb for reporting).

• Note: While investigating the previous bug, it turned out that some users provide rejection criteria as unnamed list, albeit the manual says that is has to be a named list. This was not a problem until v0.9.8, however, with v0.9.9 parameters in the unnamed list a silently ignored.

calc_FuchsLang2001()

• The function caused a fatal error on the CRAN R-devel resources after recent changes in R and did not work anymore on R-devel. This is fixed now and has no user visible effects.
• Minor code polish and better test coverage.

Changes in version 0.9.9 (2020-11-29)

New functions

• calc_OSLLxTxDecomposed() Calculate Lx/Tx ratio for decomposed CW-OSL signal components by Dirk Mittelstrass

Bugfixes and changes

analyse_SAR.CWOSLOSL()

• The default value for rejection.criteria is now list() instead of NULL. As before, the default value means that internally set rejection criteria are applied and hence this change should not affect the behaviour of the function, but it allowed us to greatly simplify the underlying code.
• The function did not check correctly for all record types, under particular circumstances, some of the non-support curves could slip through; corrected.
• If the input records were not of type OSL, IRSL or POSL the function was supposed to return NULL but crashed instead; fixed.
• As implemented by calc_OSLLxTxRatio() the function now supports NA values for the integrals. If set, sum curves are taken for the calculation. This option enables a bunch of new possibilities to test more experimental protocols without modifying the function.
• Following the NA support, all integral arguments are now preset to NA. To avoid user errors, a warning is thrown each time the function runs with integral limits set to NA.
• The identification of the allowed curves was further improved to cover cases where users may want to provide experimental OSL curves. Allowed are combinations where OSL is combined with additional letters [a-zA-Z]. Everything else is, as before, ignored.
• The function now supports the input from the upcoming R package 'OSLdecomposition' by Dirk Mittelstrass and hence gained a new argument called OSL.component.
• In particular circumstances, when no TL-curves were detected as preheat, the function crashed; fixed (flagged by Dirk Mittelstrass).

calc_OSLLxTxRatio()

• The function now supports NA as input for the signal.integral and/or background.integral. If set to NA the sum of the count values is used. This is useful if the users wants to process component data that were already extracted and do not need integral limits.
• The output of the function is not anymore invisible. This should have no further effect.

plot_GrowthCurve()

• Users sometimes complained about large uncertainties produced by the Monte-Carlo runs to estimate the error on the D_e, which returns the standard deviation of the MC runs. Large standard deviations are a result of large individual uncertainties (e.g., noisy data) provided as input. The function now returns, experimentally, estimates of the highest probability density intervals as an alternative. Beyond we may recall that the results of the MC runs are always returned as a separate object by the function. Means, users ca run any statistic of choice to best estimate the uncertainties on the D_e.

read_Daybreak()

• The function threw a lot of warnings for the *.DAT after R >= 4.0.0 started to be more strict about embedded null bytes within a character. The code was updated to a newer version and no further warnings are shown.
• Under particular circumstances the function left open connections behind; fixed.

Internals

• The automated tests now use consistently testthat >= 3.0.0. Naturally this has no user visible effect, however, we found a couple of minor bugs while updating the tests.

Changes in version 0.9.8 (2020-11-16)

Important user affecting changes

• The version requires R (>= 3.5.0)
• We removed a few of our vignettes from the package. They are now available on https://r-luminescence.org/tutorials/

New functions

• read_RF2R() Add support for importing spatially resolved RF measurement results.

• plot_ROI() Visualisation of ROI data imported by read_RF2R().

• convert_Concentration2DoseRate() by Svenja Riedesel and Martin Autzen. This function converts radionuclide concentrations (K in %, Th and U in ppm) into dose rates (Gy/ka). Beta dose rates are also attenuated for the grain size. Beta and gamma dose rates are corrected for the water content.

• calc_CobbleDoseRate() by Svenja Riedesel and Martin Autzen. The function calculates the dose rate profile through the cobble based on Riedesel and Autzen (2020).

Datesets

• ExampleData.CobbleDate new example dataset to be used in conjunction with the function calc_CobbleDoseRate(). Contributed by Svenja Riedesel and Martin Autzen.

• BaseDateSet.Conversion, Cresswelletal2019 was changed to Cresswelletal2018. The corresponding function scale_GammaDose() was updated.

• BaseDateSet.GrainSizeAttenuation.rda: Grain size correction data for beta-dose rates published by Guérin et al. (2012). Dataset contributed by Svenja Riedesel and Martin Autzen.

Bugfixes and changes

analyse_SAR.CWOSLOSL()

• The function gained a new argument plot_onePage. If set to TRUE, the plot output is gathered in one single plot window instead of three.
• If the user decided for odd reasons to mix preheat steps recorded as TL (signal recorded) and simple preheat (no signal recorded) the function stopped with an uncontrolled error. Now the function is more flexible. However, this comes at a cost, e.g., if the user accidentally swaps the sequence order (such as OSL-TL, instead of TL-OSL) this may remain unnoticed.
• For zero count data (e.g., observed in single grain curves), the function crashed with a non-indicative error message because NaN are usually produced if the recycling ratio is calculated; fixed (thanks to Brice Lebrun for reporting the issue).

analyse_FadingMeasurements()

• Typo correction in the manual and a better documentation of supported ... arguments
• The ylim argument can now be used to control the ‘Signal Fading’ plot output.
• While the x-axis of the fading curve output was correct, it probably caused too much confusion, and now the values are shown with superscripts, e.g., 10^0, 10^1 etc. If the number of data points is too small, the function will fall back to the old representation.
• The function did not work properly for less than five data points and stopped with Error in rep(rgb(0, 0, 0, 0.3), length(TIMESINCEIRR) - 5) (thanks to Annette Kadereit for flagging this issue)
• The function now shows also the normalised log-values under the time-scale (#97)
• As it turned out, t_star did not work if the input was produced by read_BIN2R(); fixed along with (#97)
• t_star now can be a function, e.g., function(x) x+100, which would add a constant value to all TIMESINCEIRR values (#97).

analyse_IRSAR.RF()

• The function learned to recognise input from read_RF2R and changes the header labelling accordingly.
• An edge case crash that might occur in combination with the vertical sliding has be sanitized.
• The text progress bar was not closed correctly, this could lead to artefacts in the terminal output; fixed.
• The internal C++ function used for the curve sliding now uses Armadillo vectors via RcppArmadillo. Depending on the platform this gives a performance plus free house by at least 50%. However, it comes at the cost of a reduced maximum vertical sliding range of 1e+07 (before 1e+08).

plot_RLum.Data.Curve()

• The argument norm now supports the options "max" for a normalisation to the highest point (the default), "min" for a normalisation to the lowest point and huot for a normalisation suggested by Sébastien Huot (#87) via GitHub

plot_RLum.Analysis()

• The argument norm now supports the options "max" for a normalisation to the highest point (the default), "min" for a normalisation to the lowest point and huot for a normalisation suggested by Sébastien Huot (#87) via GitHub. Please note that norm only works if combine = TRUE.

report_RLum()

• Function complained about missing header; fixed.
• Function threw warnings Duplicate identifier 'root'; fixed.

read_SPE2R()

• The function threw a warning about truncated strings; the warning was not relevant and is no suppressed.

verify_SingleGrainData()

• The function did not always return a plot: Fixed.

Internals

• Risoe.BINfileData-class objects got a as.data.frame() S3-method to allow a direct conversion into a single data.frames. Count values are added in columns for each record (thanks to Annette Kadereit for the idea).

• structure_RLum() for RLum.Analysis-class recycled rows in the output data.frame if objects had more than one parent id (.pid). This led to odd behaviour if functions expected (correctly) that the number of entries in the data.frame returned by structure_RLum(), cannot not have more rows than number of records in the object. This could happen, e.g., after merging RLum.Data.Curve-class objects. ow multiple .pid entries still create only one string and the number of rows matches the number of records in the object.

Changes in version 0.9.7 (2020-01-04)

Bugfixes and changes

analyse_Al2O3C_Measurements()

• Supplementary argument main never worked; fixed.
• New supplementary argument title which allows to pass further names of a sample to the plot output.

plot_AbanicoPlot()

• The plot now can handle negative values, fixed in combination with #88 (reported by Sébastien Huot)

plot_RadialPlot()

• Fix display problem when log.z = FALSE combined with negative values (#88, reported by Sébastien Huot)

read_BIN2R()

• Due to a typo, some bytes were read in the wrong order (big-endian instead of little-endian). While the fix does not affect the behaviour of the function, the non-fixed function would have soon crashed on CRAN servers. A big thanks to Tomas Kalibera from the CRAN team for his support and for pointing out this issue.

internal_as.latex.table()

• The function gained a new argument rm.zero which is preset to TRUE. If set, all columns containing zero values only are removed. The new argument enables a better fine tuning of the latex output.

Changes in version 0.9.6 (2019-12-05)

Bugfixes and changes

analyse_SAR_CWOSL()

• Setting the argument plot.single = TRUE does not any longer interferes with the global par settings for the rejection criteria plot.

fit_OSLLifeTimes()

• The function gained a new argument plot_simple to produce plots without residuals that can be combined using standard R layout plot functionality, e.g., par(mfrow = c(2,1)).
• Function now handles the log = "x", log = "y" and log = "xy" options better.
• In the self-call mode (input is RLum.Analysis-class object), all ... objects are supported as list and get recycled if the list does not equal the number of objects

plot_GrowthCurve()

• The y-axis labelling of the test-dose response plot finally shows what was done since the beginning: the Tx values are normalised by the Tn values, i.e. Tn/Tx and not Tn/Tx. Thanks to Andrea Junge for spotting this flaw.

read_XSYG2R()

• The function now supports the temperature recalculation of TL curves even they consist of only a single point value.

Internals

• Changes in the development version of R caused error messages on CRAN and would have stopped the package from function properly in the future. This was corrected; the changes do not have any user-visible effect.
• The internal function for the DRAC output .as.latex.table() gained a new argument tabular_only which gives a little bit more flexibility at hand of the output the DRAC is combined with LaTeX.

Changes in version 0.9.4/0.9.5 (2019-09-20/2019-09-21)

Important user visible changes

1. plot_KDE() wrong error bars As it turned out, unfortunately, the error bars produced by plot_KDE() had been wrong since 2014. The input error values (2nd column in the data.frame) had been, for reasons we cannot recall anymore, divided by two leading to an ‘improved’ visualisation of uncertainties. A big thank goes to Andrea Junge (#93) who made us aware of this bug.

2. The way the x-axis channels from BIN/BINX-files are generated changed So far, during the file import, minimum and maximum values were taken, and channels were distributed equally according to the number of available data points. Thanks to Sébastien Huot for a constructive discussion.

• Example old behaviour: For an OSL curve recorded over 40 s with 100 channels, the first channel was set at 0 s, the last channel to at 40 s. Thus, the width of each channel was 0.404 s.

• Example new behaviour: For an OSL curve recorded over 40 s with 100 channels, the first channel is set to 0.4 s, the last channel to at 40 s. Thus, the width of each channel is 0.4 s.

• In other words, the first channel does not start anymore at 0 s, as it is the case, e.g., in the Analyst.

Please note that these changes will impact your analysis results!

Difference between version 0.9.4 and 0.9.5

Non-used dependency to ‘interp’ removed.

Bugfixes and changes

analyse_SAR.CWOSL()

• The automated curve type recognition introduced with the last release broke the vignette of the R package ‘RLum.Model’; now refactored code prevent other people’s code from breaking.

calc_CentralDose()

• The function does not crash anymore with an odd error if NA values are provided (#82 reported by Sébastien Huot).
• The function gained a new argument na.rm which is set to FALSE by default.

plot_DetPlot()

• The argument analyse_function.control now understands the arguments fit.method and fit.force_through_origin.

calc_MinDose()

• The final parameter estimates and their confidence intervals in the console output are now always reported in their absolute values, independent of whether the logged or non-logged model was used (#84 suggested by Sébastien Huot). This is a purely cosmetic change, and no changes to the calculations were made.
• The values in the profile log-likelihood plots are now always given in their absolute values, independent of whether the logged or non-logged model was used (#84 suggested by Sébastien Huot). This is a purely cosmetic change, and no changes to the calculations were made. The output object still returns the original profile objects with the logged values.
• Improved handling of values close to zero if bootstrap = TRUE (#91, reported by Andrea Junge).

plot_GrowthCurve()

• Remove rounding from numerical output values (#81 suggested by Sébastien Huot); this affects all subsequent functions. However, terminal and graphical output do not change.

Internals

• The package data/datalist file listing the example data shipped with the package was not correct (#85, CRAN comment); now corrected.

Changes in version 0.9.3 (2019-07-31)

• This version takes care of a CRAN note complaining about a non-used package declared in the Import fields.

Changes in version 0.9.2 (2019-07-30)

• This version fixes an URL problem discovered by the CRAN precheck

Changes in version 0.9.1 (2019-07-30)

Removed functions

• model_LuminescenceSignals(): This wrapper function for the R package 'RLumModel' was removed due to changes on CRAN inflicting a warning because the object is masked from 'package:Luminescence' each time 'RLumModel' was loaded. In 'Luminescence' the function had no real scope, but provided a place holder to point the users on the existence of the package 'RLumModel'.

Bugfixes and changes

apply_EfficiencyCorrection()

• The function now supports RLum.Analysis objects as input
• All allowed input objects can now be provided as list (self-call support)
• Minor code polish

analyse_baSAR()

• The manual stated that RLum.Analysis-class are supported, but actually this is not true, it supports a list of RLum.Analysis objects only.
• Some minor manual corrections.

analyse_SAR.CWOLS()

• Add support for POSL data,before the function crashed (reported by Alice Versendaal). POSL curves had been not listed as supported curve types though.
• Improve dataset recognition (OSL, IRSL, POSL).

calc_Lamothe2003()

• The documentation wrongly stated that an input data.frame should have the columns ‘dose’, ‘De’ and ‘De error’, indeed ‘dose’, ‘LxTx’ and ‘LxTx error’ are expected (spotted by Jeong-Heon Choi)

calc_TLLxTxRatio()

• The LxTx.Error is no longer negative, even if the subtration lead to negative count values (#78, reported by Sébastian Huot)

calc_SourceDoseRate()

• The function learned about the half-life of Cs-137

fit_OSLLifeTimes()

• 0 count values crashed the function for obvious reasons (a / 0 = Inf), this can happen in particular for artificial datasets. Now, if 0 is detected, count values are increased by 0.1 (reported by Dirk Mittelstrass)

plot_AbanicoPlot()

• If summary = "se.abs" was chosen, the abanico plot showed se.rel instead; corrected (reported by Maryam Heydari)

plot_DRTResults()

• Correct documentation, where it was erroneously implicated that the ... argument pipes all arguments,
• add support for argument las.

plot_GrowthCurve()

• If the dose points included NA values the function crashed unexpectedly.

S4-classes and methods

• RLum.Analysis-class
  • Code polish (no visible changes)
• RLum.Data.Curve-class
  • Code polish and clarifying the documentation
• RLum.Data.Image-class
  • Conversion methods to a matrix and from and to data.frame never worked as intended.
  • Code housekeeping

Internals

• The ‘DESCRIPTION’ file contained the string “[upcoming]”. This was a left-over from the development version and is now removed.

• Version numbering is now again more R canonical consisting only of three numbers (‘stable.major.minor’). Development versions are only indicated by .9000 in the version number. Numbers after 9000 (e.g., 9000-1) indicate build numbers. Both is used for GitHub build versions only.

• The file name RisoeBINfileData-class.R was changed to Risoe.BINfileData-class.R for consistency reasons, however, this has no further visible effect (#76, spotted by Sébastien Huot)

Changes in version 0.9.0.110 (2019-04-21)

'Luminescence' 0.9.X releases will be the last versions supporting R < 3.5.0.

Important R related changes (R >= 3.6.0)

The behaviour of the base R function sample() was changed / corrected (see news on https://www.r-project.org for details). This function is used heavily within functions by ‘Luminescence’ (e.g., plot_GrowthCurve()). That being said, it means that old data re-analysed by ‘Luminescence’ using R versions >= 3.6.0 will not give the same results. Differences are small, however, they may significant. If you want to reproduce your data, we recommend to use one of our Docker images which include the R version from the package release date.

New functions

• convert_Wavelength2Energy() Smooth and quick emission spectra conversion from wavelength to energy scales
• fit_OSLLifeTimes() Fit and deconvolution of OSL lifetimes using off-time measurements, i.e. after the stimulation was switched off.
• plot_DRCSummary() Summarise all dose-response curves from your SAR OSL analysis in one single plot.
• scale_GammaDose() Scale the gamma dose rate considering layer-to-layer variations in soil radioactivity. Contributed by S. Riedesel and M. Autzen.
• fit_ThermalQuenching() Provide an easy option to fit thermal quenching data and determine the activation energy.

Bugfixes and changes

apply_CosmicRayRemoval()

• Thanks to black magic, this function now runs also over a list of RLum.Data.Spectrum objects and and RLum.Analysis objects (also nested in a list)

apply_EfficiencyCorrection()

• The function produced a warning (“collapsing to unique x values”) on R-devel; fixed.

analyse_SAR.CWOSL()

• If the function was used in a self-call mode (object of type list), the arguments contrary to what is written in the manual, the arguments such as signal.integral.min had not been expanded properly; no one complained, however, corrected.
• The argument main no also supports the input list in the self-call mode.

analyse_SAR.TL()

• The plot legend was not correct, if no dose points had been provided; fixed
• The function returned a confusion warning; fixed.
• If background signals are provided, now the background subtracted signal is shown automatically instead of the uncorrected curves (suggestion by Tilmann Wolpert)
• The function now supports a list of RLum.Analysis objects as input
• The plot settings were sufficiently reset; fixed.
• Code polish

analyse_FadingMeasurements()

• Enable support for data imported from a BIN/BINX-file
• If the g-value is negative, now it also returns a negative g-value; before always the absolute values was taken, which was not meaningful
• If rho was negative, the log10 value could not be calculated, this caused an unwanted warning; now suppressed
• It showed a warning on R-devel due to changes in stats::approx(); fixed
• Records with negative ‘time since irradiation’ are now automatically removed (thanks to input by Sébastien Huot)
• Function is now more relaxed when odd data a provided and it is less talkative

calc_Huntley2006()

• The function now checks if rho’ (rhop) is a sensible non-zero positive value and, if otherwise, stops with a more meaningful error message. (Issue #74)
• Now the function is a little bit less talkative if extreme dataset are provided as input

read_BIN2R()

• Add support for missing BINX-file version 5 (pointed out by Harrison Gray)
• In the auto-file recognition (only a path instead of a file is provided), sometimes the found BIN/BINX-files were imported twice. The corresponding code was simplified to avoid such errors in future
• Sometimes the time stamp is invalid since it is missing a leading 0. For BINX-files >= version 6 such timestamps now get corrected during the import

read_XSYG2R()

• The function failed for R-devel importing spectrometer data (approx(..., ties = -2)); fixed.

plot_GrowthCurve()

• The function crashed with Error in data.MC[, i] : incorrect number of dimensions under very particular circumstances (reported by Alice Versendaal). This was due to unwanted internal object conversion; fixed.

plot_RLum.Data.Spectrum()

• The wavelength to energy scale conversion was not correct. This was corrected and internally, the function now calls consistently the new function convert_Wavelength2Enegy() (the corresponding code and documentation have been updated or removed).
  
• The function gained a new argument norm for normalising the data to the highest or the lowest count value.
• The function gained a new argument bg.spectrum allow to pass a background spectrum that can be used for the subtraction
• The function returned a warning if a single matrix was used but plot.type was already set to single; fixed.
• Wrong settings in bg.channels crashed the function. Now it auto-corrects the wrong values.
• Channel binning now consistently calls the internal function Luminescence:::.matrix_binning(), which is more efficient cleaner than the code used before
• The function now longer crashes for plot.type = 'persp' if 'xaxis.energy = TRUE.
• The colour picking option was not always working; fixed.
• Warnings are formatted more nicely.

verify_SingeGrainData()

• The function crashed for crazy datasets; fixed
• Cleaning-up RLum.Analysis-class objects did not worked; fixed

write_RLum2CSV()

• The function gained a new argument compact which is set TRUE be default, and keeps the element output as simple as possible, which is in particular helpful for RLum.Results objects
• The function now behaves more friendly to RLum.Results objects, before the output was hard to understand.

write_RLum2BIN()

• The export failed if the time stamp was invalid (NA); now NA values are consistently replaced by 000000 (other values are not possible, otherwise the re-import of such file would fail) (reported by Alice Versendaal via email, 2019-03-06)

Changes in S4-objects and methods

get_RLum

• The method working on list elements got a new argument class to remove unwanted RLum objects in from list of RLum objects

RLum.Analysis-class

• The show-method crashed under very rare circumstances if inconsistent objects had been combined; fixed.

RLum.Data.Spectrum-class

• Support added for bin_RLum.Data()

New example datasets

• ExampleData.TR_OSL provides a single fast-pulsing OSL curve that can be used to test the function fit_OSLLifeTimes().

Internals

• Package news are now provided as markdown document (NEWS.md) which is supported by CRAN
• The package version numbering scheme has been modified to simplify the distinction between different developer versions. While relevant numbers for CRAN remain X.X.X, which reads ‘stable.major.minor’ version, a 4th number indicates the running package build number, e.g., 0.9.0.7.
• New internal function Luminescence:::.matrix_binning() to get the matrix (and curve) binning consistent; so far used only by the the bin_RLum.Data() method for RLum.Data.Spectrum-class
• Vignette S4-class object structure updated
• Internal changes to address changes in R-devel checking logical arguments with length more than one (e.g., if(is.na(x)) returned an error if x was not a vector of length 1)
• src/Makevars file removed to prevent unconditional stripping in response to a recent CRAN request; in return the installation size of the package increased in size on some platforms.