Changes in version 1.1.2 (2025-12-11)

This package version requires R >= 4.4

Removed functions and deprecations

• The functions CW2pHMi(), CW2pLM(), CW2pLMi() and CW2pPMi() (deprecated since 1.0.0) have been removed, but their functionality remains in the corresponding convert_CW2*() functions (#992).

• Functions github_commits(), github_branches() and github_issues() are now deprecated and will be removed in a future release (#1026).

• Functions is.RLum(), is.RLum.Data(), is.RLum.Data.Curve(), is.RLum.Data.Spectrum(), is.RLum.Data.Image(), is.RLum.Analysis() and is.RLum.Results(), and are now deprecated and will be removed in a future release. The best way of testing whether an object is of a given type is by using the inherits() function, as in inherits(object, "RLum.Data.Curve") (#1034).

Bugfixes and changes

analyse_Al2O3C_CrossTalk()

• The function no longer crashes if dose_point is misspecified (#1073).

• The function no longer crashes when called on an RLum.Analysis object rather than on a list of them (#1182).

analyse_Al2O3C_Measurement()

• The function no longer crashes when called on RLum.Analysis objects that do not contain records of type "OSL (UVVIS)" or "TL (UVVIS)" (#1232).

analyse_baSAR()

• The function no longer crashes if all objects in the input list are removed because empty (#1041).

• The dose response curves are no longer overplotted when there are fewer than 1000 MCMC iterations (#1058).

analyse_FadingMeasurement()

• The function no longer crashes when called on an object with missing originator (#1130, #1154).

• When called with a list containing object with multiple classes, the message produced reports the correct number of objects removed from the analysis (#1132).

• The function no longer crashes if the normalisation term is zero, which should never happen in real-life analyses, but it may occur if the input data is somewhat malformed (#1144).

• The function no longer crashes when attempting to remove undesired objects from a list-like class that overloads the [<- S3 method (#1236).

analyse_IRSAR.RF()

• The function no longer crashes when n.MC = NULL is used with the FIT method (#1055).

• The function has been optimized to be faster during the bootstrap and sliding phases for the SLIDE and VSLIDE methods, with speed-ups in the region of 45% (#1210, #1230).

analyse_pIRIRSequence()

• The names in the curves legend are no longer cut off (#1206).

• Labels and legend in the rejection criteria plot are better positioned and no longer cut off (#1208).

• The dashed lines in the summarised DRC plot have been extended to reach the outer box instead of leaving small white gaps at the extremes (#1215).

• The space around the plots when using plot_singlePanels = FALSE has been reduced, so that plots can be a bit more detailed (#1219).

analyse_portableOSL()

• The size of the contour labels for mode = "surface" can now be controlled with the ... argument labcex and scale with cex (#1075).

• The size of the interpolation grid can be controlled via the ... arguments nx and ny (#1077).

• The coord argument is now better validated to avoid a crash in case of misspecification (#1097).

analyse_SAR.CWOSL()

• Empty subplots are inserted when onlyLxTxTable = TRUE to preserve the usual plot ordering. This prevents a crash that would otherwise occur if the option was set from analyse_pIRIRSequence() (#1186).

• The function crashed when applied to an object generated by OSLdecomposition::RLum.OSL_decomposition() with plot = TRUE; this was a regression introduced in v1.1.0 (#1188; thanks to @DirkMittelstrass for reporting).

• The visualisation of the ‘Checks’ (former rejection criteria) was not very intuitive. To improve the situation, we now display correct inequality symbols, such as <= or >= instead of <> (which was meant to be understood as a comparator). Furthermore, the recycling ratio now shows better the threshold depending on whether the ratio is smaller or larger than one.

• The function crashed if the rejection.criteria list specified a NULL value for recuperation_reference (#1204).

• The space around the plots when using plot_onePage = TRUE has been reduced, so that plots can be a bit more detailed (#1219).

• The labels in the rejection criteria plot are now shortened only when the available horizontal space is not wide enough (#1222).

bin_RLum.Data()

• Validation of the bin_size (for RLum.Data.Curve), bin_size.row and bin_size.col (for RLum.Data.Spectrum) has been made stricter, so that invalid values produce an error (#1104).

calc_AliquotSize()

• The function no longer hangs when grain.size is set to a small value and its minimum and maximum values are the same (#1114).

• The number of MC iteration is now correctly reported in the plot subtitle (#1115).

• The function now validates the MC.iter argument to avoid an ugly crash if a non-positive value is provided (#1124).

• The function now better validates its grain.size argument (#1152).

calc_CentralDose()

• The profile log-likelihood plot is now cut at a log-likelihood of -100, so that for profiles with log-likelihoods more negative than that don’t get too squashed in the region of interest (#1227; thanks to @DirkMittelstrass for reporting).

calc_CosmicDoseRate()

• The function is now more robust against NA values in its input arguments (#1101).

calc_EED_Model()

• A crash that occurred with a small number of simulation was fixed (#1051).

• The error message in case of failed surface interpolation is now clearer and more complete (#1053).

calc_FadingCorr()

• The function produces an error if g_value is an RLum.Results object with an unsupported originator instead of returning NULL (#1136).

calc_FiniteMixture()

• The function no longer crashes if the De column contains zeros or missing values (#1146, #1148).

• The legend for the components is positioned consistently also at cex values other than 1 and is drawn also when pdf.colors = "gray" (#1175).

calc_HomogeneityTest()

• The function ignores all columns after the first two, as they could lead to crashes if they contained non-numeric values (#1224).

calc_Huntley()

• A warning raised in rare occasions if the number of Monte Carlo iterations is very small has been fixed (#1048).

calc_OSLLxTxRatio()

• The Lx.data and Tx.data arguments are now better validated (#1178).

calc_SourceDoseRate()

• The function returns a clearer error message if dates are specified in an unexpected or ambiguous format (#1162).

calc_Statistics()

• The function could crash if an error was exactly zero. This affected also the plotting functions that use calc_Statistics() (#1160).

combine_De_Dr()

• It is now possible to control the random seeds used by the JAGS MCMC chains via the method_control argument (#1038).

fit_CWCurve()

• If model fitting failed but an object named fit was present in the workspace, the function tried to use that one, which would lead to a crash or to unexpected results (#1081).

• The output.table field of the RLum.Results object returned now contains only columns corresponding to the components effectively fitted (#1083).

fit_EmissionSpectra()

• The function makes better use of the plot area, leaving smaller margins around the plot (#1011).

fit_OSLLifeTimes()

• The function throws a warning when the fit fails (#1005).

• The function makes better use of the plot area, leaving smaller margins around the plot (#1011).

install_DevelopmentVersion()

• The function has been refactored so that it no longer depends on the deprecated github_branches() function (#1079).

plot_AbanicoPlot()

• A warning was raised if the function returned early (for example due to invalid inputs) and the graphical device was off (#1001).

• The function now validates the frame argument in order to avoid crashing on misspecified values (#1036).

• The function no longer crashes when a small plot.ratio value is specified (#1062).

• The function now validates the zlim argument to avoid a crash if negative values are used with log.z = TRUE (#1063).

• The legend text now scales better at high cex values (#1066).

• If summary.pos contains multiple valid positions, the first one is used rather than generating warnings and not showing the summary (#1093).

plot_DoseResponseCurve()

• The plot symbols are now correctly assigned for mode = "extrapolation" and in the legend. If reg_points_pch is specified, it must now be a vector of 3 elements, corresponding to the symbols to use for normal points, point 0 and repeated points, respectively (#1072).

• The baseline line in the sensitivity plot has been extended to reach the outer box instead of leaving small white gaps at the extremes (#1213).

• The space around the plots has been reduced, so that plots can be a bit more detailed (#1219).

plot_DRTResponse()

• The function now suppresses the warnings generated by graphics::arrows() when datasets having wildly different distributions are plotted (#1184).

plot_Histogram()

• Since version 1.0.0, setting summary.pos to one of “left”, “center” or “right” and normal_curve = TRUE resulted in the summary table not being visible (#1118).

plot_KDE()

• Space at the bottom of the plot is no longer added when the boxplot is not plotted because values.cumulative = FALSE (#1110).

• The plot title and subtitle now respect the cex option (#1112).

• The function now better validates its input arguments (#1158).

plot_RadialPlot()

• The function avoids some possible overprinting at the extremes of the z-axis labels and draws lines at the extremes of the ellipse (#1013).

• The appearance of the y-axis ticks has been improved in the case when the y-axis is particularly narrow (#1060).

• If summary.pos contains multiple valid positions, use the first one rather than generating warnings and not showing the summary (#1093).

• The functionality of the stats argument had regressed in version 1.1.1, and now has been restored (#1106).

• The function better validates the zlim argument, to avoid crashes on misspecified axis limits if log.z = TRUE (#1140).

• The function now always returns invisibly the list of plot parameters, independently of the value of the output argument. This change made that argument redundant, so it was removed. This should have no impact on users; if specified, it will be ignored (#1142).

• The function stops with an error instead of crashing if the input contains only 1 data point (#1150).

• The function reports a message instead of throwing a warning when adding a line with negative value if log.z = TRUE (#1164, #1168).

• The summary text and line labels now scale better at high cex values (#1170).

• The function no lonver crashes if central.value is a non-positive value when log.z = TRUE (#1173).

• Some graphical artifacts related to the 2-sigma bar, which would appear for extreme settings of the central.value argument have been fixed (#1194).

plot_RLum.Analysis()

• Setting records_max to a value larger than the number of curves no longer generates spurious legend entries (#1017).

• The function now supports plotting of RLum.Data.Image objects contained in its records (#1028).

plot_RLum.Data.Image()

• The function now supports the ... argument mtext (#1031).

• The positioning of axis ticks and labels in raster and contour plots of low-resolution images has been improved, and pixel numbering starts consistently from 1 (#1191).

• The function gained support for the arguments digits and scientific. Moreover, it is possible to specify multiple plot titles in main. Thanks to @Zink-Antoine for the contribution (#1197).

plot_RLum.Data.Spectrum()

• The function doesn’t crash anymore if bg.spectrum is used and ylim specifies an interval that doesn’t contain any background channels (#1019).

• The function doesn’t crash anymore if the value of frames given is too large when plot.type = "multiple.lines" (#1021).

• The legend text now scales better at high cex values (#1068).

• The function now makes better use of the plot area by setting smaller margins and reducing the spacing between axis elements (#1070).

read_PSL2R()

• The function gained a pattern argument to refine the list of files to read when a directory path is specified (#1099).

read_SPE2R()

• The function no longer crashes if frame.range is misspecified (#1138).

subset.Risoe.BINfileData()

• The function also updates the .RESERVED slot (if present) to keep only the elements selected (#1089).

verify_SingleGrainData()

• The function crashed on empty RLum.Analysis-class objects. Now a warning and NULL is returned.

Other changes

• The ExampleData.portableOSL file has been updated with surface coordinates (#1095).

• Documentation for all example data (including synthetic datasets) is now more easily findable from RStudio (#1166; thanks to @DirkMittelstrass for reporting).

Changes in version 1.1.1 (2025-09-11)

New functions

• correct_PMTLinearity(): A helper function to correct luminescence signals measured with a PMT for count linearity (#920).

Removed functions and deprecations

• Functions calc_Kars2008() (defunct since 0.9.26), Analyse_SAR.OSLdata() (since 1.0.0), PSL2Risoe.BINfileData() (since 1.0.0) and Second2Gray() (since 1.0.0) have been removed from the package. Their functionality can be found in functions calc_Huntley2006(), analyse_SAR.CWOSL(), convert_PSL2Risoe.BINfileData() and convert_Second2Gray(), respectively.

• Function get_Risoe.BINfileData() has been removed as it was not used and provided no benefits to the user (#945).

Bugfixes and changes

add_metadata<-()

• The function throws an error if trying to assign a NULL value instead of silently ignoring it (#946).

analyse_Al2O3C_ITC()

• The function returns early if the fitting of the dose response curve fails (#979).

analyse_Al2O3C_Measurement()

• The function allows to specify the cross_talk_argument as a numeric vector of length 3, as was already documented (#930).

analyse_SAR.CWOSL()

• The function crashed if any of the curves had NA as its recordType (#867).

apply_CosmicRayRemoval()

• The method argument of smooth_RLum() was not reachable via ... as the apply_CosmicRayRemoval() already has an argument called method. Now the ... argument is called method_smooth_RLum and works as expected (9b27467).

• The function crashed if method = “Pych”andMARGIN = 1` were specified. The function now better validates its inputs and no longer returns invisibly but normally (#987).

calc_AliquotSize()

• The legend text now scales better at non-default cex settings (#849).

calc_Huntley2006()

• If the user set the n.MC argument, this was also used in the fitting of the dose response curve. This was not the expected behaviour, and resulted in poor performance (a 2x slowdown for n.MC = 10000). This regression was introduced in version 1.0.0 (#867).

calc_MaxDose()

• The function crashed if sigmab was set to a very small value (#898).

calc_MinDose()

• The function crashed when bootstrapping with bs.M = 1. Now the bs.M parameter is silently reset to 2 in that case, although such low values are discouraged as they may trigger a warning during loess fitting (#900).

• The function crashed when all rows in the input data set contained NA values (#915).

calc_Statistics()

• The computation of the weighted median is now correct, while before it corresponded to the simple (unweighted) median (#905).

fit_CWCurve()

• The function hanged on particularly small datasets as it tried to fit too many components for the available data (#953).

fit_DoseResponseCurve()

• The object returned now contains an additional .De.raw column to store the calculated De value computed by the fitting function “as is”, without setting meaningless results to NA. The De and .De.raw columns differ only for mode = "interpolation", where the first sets the De to NA if negative, while the latter doesn’t. It is then up to the user to decide what to do with those values, bearing in mind that they may be arbitrary when negative (#957).

• The message reported for fit.method = "QDR" now states correctly whether the fit succeeded or failed (#961).

• The computation of De.MC, De.Error and HPDI for mode = "interpolation" has been modified to account correctly for possible negative De values resulting from the fit (#963).

• Previously, a single NA value in the Monte Carlo results prevented the computation of Highest Density Intervals (HPDI). This limitation has been removed, and now HPDIs are reported in more cases (#976).

• The columns of the results$De data frame are now reported in a different order; an additional “Mode” column reports the value of the mode argument; columns meant for internal use have been moved to the end and their names are now prefixed with . (#974).

fit_LMCurve()

• The automatic correction of the x-axis limits when log = "x" is specified and the lowest value is set to 0 threw the correct warning but actually did nothing (38e4324).

fit_OSLLifeTimes()

• The function ensures that the signal_range argument doesn’t contain negative values instead of crashing (#896).

get_RLum()

• The function doesn’t crash anymore on RLum.Analysis objects if using the record.id argument removes all available records (#873).

• An internal optimization increased the performance of the function. The difference is not perceivable on single calls to get_RLum(), but the change brings a visible speed up when merge_RLum() is called over a sufficiently large number of RLum.Analysis objects (#875).

merge_Risoe.BINfileData()

• The function gained argument verbose to allow disabling the output from read_BIN2R() and write_R2BIN() (#950).

plot_AbanicoPlot()

• The positioning of the y-axis label has been improved to be centred around the zero line (#847). The z-axis tickmarks and labels are better drawn at non-default cex values (#865).

• The minor grid lines were drawn incorrectly in a rotated plot (#849).

• Some plot elements didn’t scale correctly at non-default cex values (#861, #879).

• The frame argument is now respected also when the plot is rotated (#863).

• The weighted median is computed correctly when summary = "median" and summary.method = "weighted" are used (#905).

plot_DoseResponseCurve()

• We added support for the log graphical parameter, which can be used if the fit was obtained with mode other than "extrapolation" (#820).

• The normal curve drawn as part of the histogram plot could appear very jagged or even as a straight line, as it could happen that too few points were actually used when drawing the curve (#843).

• The legend symbols did not match anymore the plotted regeneration/dose points. This also affected functions such as analyse_SAR.CWOSL() and analyse_pIRIRSequence() (9ba54e65).

• The legend positioning can now be controlled via the ... argument legend.pos (#959).

plot_DRCSummary()

• The dose-response curve is now plotted as expected, after a regression in 1.1.0 caused the function to produce an empty plot (#969).

plot_DRTResults()

• Points didn’t scale correctly at non-default cex values (#879).

• Option summary = "weighted$median" is now officially supported (#905).

plot_Histogram()

• Option summary = "median.weighted" is now officially supported (#905).

plot_KDE()

• Outlier points didn’t scale correctly at non-default cex values (#879).

• The weighted median is computed correctly when summary = "median" and summary.method = "weighted" are used (#905).

plot_RadialPlot()

• Option summary = "median.weighted" is now officially supported (#905).

plot_Risoe.BINfileData()

• Input arguments are validate more strictly to avoid unexpected crashes in case of misspecification (#964).

• The x-axis label for TL curves now reports temperature rather than time, consistently with the data shown (#971).

plot_RLum.Analysis()

• The legend text now scales better at non-default cex settings (#854).

plot_RLum.Data.Curve()

• The function gained a new logical argument interactive that enables interactive plotting of curves using plotly::plot_ly(). It requires the suggested package 'plotly' to be installed (e4746eaa).

plot_RLum.Data.Spectrum()

• Types image and contour gained more control of the contour line labels through the argument labcex.

• For plot types "image" and "multiple.lines" the ... logical argument legend was added.

• Plot type "image" further gained a legend with support through ... for legend.pos, legend.horiz and n_breaks to control the number of colours in the graphic.

read_BIN2R()

• The FNAME metadata field is no longer left empty if the BIN-file didn’t specify one, but it’s populated with the BIN-file name without extension. This is the behaviour that was present up to version 0.9.26, but version 1.0.0 had regressed it (#928).

read_XSYG2R()

• Add support for the new function correct_PMTLinearity() (#920).

smooth_RLum()

• Add support for the Poisson smoother of Carter et al. (2018), which can be accessed by setting method = "Carter_etal_2018". This is helpful to ensure that the dark-background counts signals measured by a photomultiplier tube follow a Poisson statistic, and smooths non-conforming values according to an average over four neighbours (#921).

trim_RLum.Data()

• The function crashed if both values specified in the trim_range argument exceeded the number of channels available (#871).

use_DRAC()

• Report the message produced by the DRAC server more reliably in case of error (#982).

Changes in version 1.1.0 (2025-06-11)

New functions

• remove_SignalBackground(): A user-friendly method to subtract background signals from various curves in RLum.Analysis objects without resorting to lapply() loops. Depending on the record type, the function identifies pairs of curves; for instance, if in a sequence, a TL curve is immediately followed by another TL curve, the second curve is recognised as the background signal, subtracted using merge_RLum() and subsequently removed from the object (if desired). Alternatively, a set of potential background curves can be specified.

• remove_RLum(): This function further completes the set of methods that can handle and modify RLum-class objects. It operates on RLum.Analysis objects or a list of such objects to remove unwanted records from datasets. Although the function calls get_RLum() and relies on its functionality, the new implementation facilitates a more logical workflow and analysis pipeline.

• .as.latex.table(): Converts RLum.Results objects where suitable to LaTeX ready tables, for instance, objects produced by use_DRAC(). The function has been present in the package as an internal function for many years; now it is exported and better linked to make it discoverable.

New datasets

• RF70Curves is a new dataset consisting of two IR-RF curves measured with the RF70 protocol. This new dataset provides a more realistic example for analyse_IRSAR.RF().

Breaking changes

• Function get_Quote() is no longer exported, but remains available as an internal function. This is unlikely to affect any user of the package, as the function was only meant to report a random quote at startup (#644).

• In the functions fit_DoseResponseCurve() and plot_GrowthCurve(), the fit.method option LambertW was replaced by the more correct term OTOR.

• Argument cex.global has been removed from plot_DoseResponseCurve() (and consequently also from plot_GrowthCurve()), and if set it will be silently ignored. Users can set the cex graphical parameter via ... in its place (#831).

• The fit.method and fit.advanced arguments of function fit_LMCurve() have been removed. The default fitting method has been changed from port to LM, and support for the port algorithm has been removed. From now on, argument fit.method is silently ignored, unless fit.method = 'port' is used, in which case a deprecation warning is thrown (#793).

• The fundamental physical constants used in the package (such as Boltzmann constant, Planck constant, etc.) have been uniformed to those reported in the NIST Reference on Constants, Units and Uncertainty. This may affect the numerical results of the following functions: calc_FastRatio(), calc_Huntley2006(), calc_SourceDoseRate(), calc_ThermalLifetime(), convert_Activity2Concentration(), convert_Wavelength2Energy(), fit_CWCurve(), fit_LMCurve(), fit_SurfaceExposure(), fit_ThermalQuenching() (#693).

Bugfixes and changes

analyse_baSAR()

• The function crashed if only one aliquot was kept (#834).

analyse_FadingMeasurement()

• The function crashed if the number of Lx and Tx curves was not equal when structure = c("Lx", "Tx"), which is the default. The check that the number of points within each pair of curves has also been improved, and the function now produces more helpful error messages (#616).

• The function tests are now less rigorous for different Lx and Tx sizes. While they should match, numerical rounding issues in the data returned by the measurement devices could previously result in rejection of records, although this had no actual meaning for the data analysis.

analyse_IRSAR.RF()

• The legend and subtitle texts now scale better at non-default cex settings (#803).

• The printing of progress bars and messages to the terminal can now be controlled via the ... argument verbose. Previously this could only be done via the txtProgressBar argument (which is still supported), but the new option makes the interface consistent with most other functions (#805).

• The mtext and cex options are respected if method = "None" (#807).

• The residual plot correctly respects the logarithmic transformation of the x-axis when log = "x" or log = "xy" are specified (#814, #825).

• The function now deals correctly also with input objects containing multiple curves (#816).

• The residual indicator rectangle, which is drawn when method is either SLIDE or VSLIDE, is also drawn when log = "x" (#821).

• The plotting of the density is silently disabled if n.MC = NULL, which avoids a spurious warning (#823).

analyse_pIRIRSequence()

• The function now respects the cex graphical argument, and its plot output has been subtly improved thanks to various fixes that have occurred especially in plot_DoseResponseCurve() (#831).

analyse_portableOSL()

• The function now returns an error if mode is something other than "profile" or "surface".

• The mode argument was not respected when operating over a list of objects (#673).

• The function crashed when using mode = "surface" if the plotting limits were too tight and left only one point (#675).

• The check on the validity of the signal.integral argument has been improved, and now it occurs only once, so at most one warning is raised if the argument is set to a value exceeding the valid range (#678, #680).

• The function returns a clearer message when no x-coordinates were measured and mode = "surface" was used (#682).

• The cex and type graphical parameters can now configured via the ... argument (#684).

• The expected sequence pattern of the input object is validated more thoroughly to avoid crashes in case of misspecification (#687).

• The graphical ... argument contour never produced a meaningful contour plot due to an internal error (#686). Along, arguments contour_nlevels and contour_col are now supported through ... to better control the number and colour of contour lines.

analyse_SAR.CWOSL()

• Add support for fit.method = 'OTORX' following the changes in fit_DoseResponseCurve(); this change includes the new argument dose.points.test that is only of use in combination with the OTORX fit.

• Add new graphical output if the measurements were single grain measurements, in such case a disc with the position and grain number marked in shown. This plot replaces the IRSL curve check plot, but only for single grain data (#797).

• The rejection criteria plot was rewritten and now provides an easier to grasp visual feedback (#797, #798).

• The IRSL/Single Grain panel swapped place with the rejection criteria panel; the plot numbers remained unchanged to avoid regression.

• More code optimisation for better readability (#802)

analyse_SAR.TL()

• A check on the sequence.structure argument ensures that a “SIGNAL” entry has been specified (#779).

calc_AliquotSize()

• The new argument sample_carrier.diameter allows to specify a value for the diameter of the sample carrier, which up to now was hardcoded to the very common 9.8 mm size (#623).

• Several graphical parameters can now configured via ... arguments, so that the plot appearance can be fully customized (#671).

calc_AverageDose()

• A bug in the implementation prevented the default plot settings from being modified via ... as advertised. Now custom settings are respected (#658).

calc_FiniteMixture()

• The function doesn’t crash anymore if the n.components argument specifies non-consecutive values (#691).

• The function sometimes failed to plot some of the densities when the number of components was set to a value of 8 or more (#704).

• The density plots would not always be coloured completely, but especially for high values of sigmab there would be an unfilled area at the base of the densities (#706).

• If the very first iteration over the components during plotting was skipped, then the function crashed as a quantity computed only in that iteration was not available. This happened for very specific combinations of sigmab and n.components (#708).

• Another crash occurred when height of the largest density curve could not be estimated due to the presence of too many NA values in the intermediate computations (#710).

• The check for significance of each component added to the model has been corrected to be more statistically rigorous (#703).

• Blank gaps appearing to the sides of the proportion of components plot for high number of components have been removed (#713), as well as extra slices appearing due to rounding errors (#715).

• The plot can be better configured via the new plot.criteria argument to control whether the statistical criteria curves should be drawn. Moreover, support for the ... options has been added: cex to control the overall scaling, main.densities, main.proportions and main.criteria to set the subplot titles (#717).

• Plots are now generated even when results contain NA values, as they in general don’t affect the plot. However, when that happens we report it in the plot subtitle (#718).

calc_Huntley2006()

• Support was added for multicore computations via the cores argument, and for the nls-fitting control options maxiter and trace.

• The fitting of the simulated curve with the GOK model has been made more robust: when an initial fit of the model fails, the fit is attempted again with 10 different values for D0 and the best fit is used. This should reduce the number of occasions in which the error message “Could not fit simulated model curve, check suitability of model and parameters” is reported (#660).

• The function crashed if all simulated Lx/Tx values were identical and approximately zero, which could happen if the rhop argument was set to a large enough value (#725).

• An error message has been improved so that it doesn’t suggest setting the ‘fit.bounds = FALSE’ argument if it has already been set (#729).

• The computation of the x-axis limits has been improved to avoid having too much unused horizontal space, especially for mode = "extrapolation" (#731).

• The scaling of the plot can now be controlled via the cex argument (#735).

• The plot margins are set more precisely and avoid the summary text to be cut off (#737).

calc_MinDose()

• The function now warns if the number of bootstrap replicates is too low to perform the loess fitting (#721).

calc_OSLLxTxRatio()

• The function returned a warning for wrong integral settings for the Tx curve even if no Tx curve was provided.

• The function does not check any more of different object types for Lx.data and Tx.data but validate objects for allowed types (this should have no user-visible effects).

convert_Concentration2DoseRate()

• The function validates its input values more thoroughly (#613).

extract_IrradiationTimes()

• The function tries a little bit harder to extract the correct duration of TL steps, rendering the data output hopefully a little bit more intelligible (#651).

• The function gained a new argument called return_same_as_input, with default value of FALSE. If set to TRUE, the input object (usually an RLum.Analysis object or a list of them) is returned with updated info elements for IRR_TIME and TIMESINCEIRR. This makes the RLum.Analysis object compatible with functions that explicitly search for those two objects, such as those in the 'OSLdecomposition' package (#752).

fit_DoseResponseCurve()

• The function now allocates less memory for storing intermediate values (#610).

• Add initial support for OTORX fitting following Lawless and Timar-Gabor 2024 (#677). The code implementation follows the Python reference by jll2 with an addition for an allowed offset parameter a set if fit.force_through_origin = FALSE. This also enables to support mode = "extrapolation" (thanks to John Lawless for his input).

• The code of the function was optimised in several places to improve code readability and reduce redundant calls.

• The models for EXP, EXP+LIN, EXP+EXP and GOK are now available in C++. This cut the required computation times in half in benchmark scenarios. More importantly, this performance scales with the number of Monte Carlo runs.

fit_EmissionSpectra()

• Fix crash when attempting to plot a frame with non-positive counts (#761).

fit_LMCurve()

• If the user asks logarithmic scaling in the y-axis, using either log = "y" or log = "xy", this is now ignored when plotting residuals and component contributions (#755).

• The plot has been slightly reworked to reduce the cases of “margin too large” errors and to work better at high settings of cex (#757).

• Missing values in the input data are now silently removed (#759, #763).

• The plotting of residuals, component contributions and legend can now be disabled, and the legend position can be controlled (#785).

• The error computed when using option fit.calcError = TRUE is now returned correctly, instead of being left to NA (#789).

• Argument bg.subtraction also accepts the option "none" to disable background subtraction even if values.bg is provided (#795).

fit_SurfaceExposure()

• It is now possible to specify different values for the light attenuation coefficient mu when a list of input data is provided (#667).

fit_ThermalQuenching()

• The trace option, which can be specified via the method_control argument, is now respected.

• The model now get internally reformulated into a mathematically equivalent expression that is easier to fit. This should reduce the number of occasions when the function fails to find a valid solution and improve the uncertainty estimation (#696).

get_RLum()

• The argument subset can now be provided as a character that represents a logical expression. Before, it always required a logical expression, but this may lead to odd effects due to the early evaluation happening in R and might not be wanted. Providing subset as a character is now a viable workaround in those situations.

import_Data()

• The function uses the most common formats (BINX and XSYG) before trying all others.

merge_RLum.Data.Curve()

• The function does no longer stops for differing channel resolutions, but it does issue a warning. The user is responsible for the consequences.

plot_DetPlot()

• The logic for multicore support was incorrect, which resulted in always starting a parallel cluster even when multicore = FALSE (#742).

plot_DoseResponseCurve()

• The response curve always tries to the get the 0 point in the mode interpolation and alternate (#677).

• Minor graphical polish to limit overplotting and also plot a density curve for the L_n/T_n signal.

• If mode = "alternate", the message that the equivalent dose could not be fitted is no longer shown.

• Argument cex.global has been removed and will be silently ignored (#831).

plot_DRCSummary()

• Add support for fit.method = 'OTORX' following the change in fit_DoseResponseCurve() (#677).

plot_DRTResults()

• The summary and legend texts now scale better at non-default cex settings (#765).

• Argument given.dose is better validated against misspecifications, and setting it to 0 is equivalent to leaving it at its default NULL value, which corresponds to avoiding data normalization (#767, #799).

• The function crashed when multiple inputs were used with boxplot = TRUE and all preheat values were identical (#769).

• Plot title, summary text and legend are now better positioned (#773, #774, #781).

• A bug in the way colours and symbols are assigned to plot points has been fixed (#777).

• The horizontal axis doesn’t include anymore a spurious extra tick that didn’t correspond to any aliquot (#783).

plot_GrowthCurve()

• Add support for fit.method = 'OTORX' following the change in fit_DoseResponseCurve() (#677).

• Argument cex.global has been removed and will be silently ignored (#831).

plot_Histogram()

• The function now doesn’t produce warnings when the input consists of a single-column data frame, as it assumes that the De error is 10^-9 (#744).

• The right margin is now smaller when errors are not plotted, as there is no need to leave space for the standard error axis (#748).

• The summary text now scales better at non-default cex settings (#750).

plot_KDE()

• The function validates its input values more thoroughly (#635).

• Setting summary.pos to one of “left”, “center” or “right” resulted in the summary table not being visible (#642).

• Argument output has been removed, and it will be ignored if set (#700).

plot_RadialPlot()

• The function validates its input values more thoroughly (#639).

• The legend text now scales better at non-default cex settings (#746).

plot_RLum.Data.Curve()

• We added comprehensive support for base R plotting arguments utilised by plot.default() and par(). This enhancement ensures that all available arguments are fully supported (#646).

• The function gained a new logical argument named auto_scale. When set in conjunction with either xlim or ylim, this argument automatically adjusts the plot range to align with the corresponding settings for xlim or ylim. For instance, if a user intends to plot OSL curves but initially selects the xlim range c(10:30) to examine the background, the initial count values may be excessively large, resulting in limited visibility. With the introduction of the auto_scale option, the ylim values are automatically adjusted to compensate for this scenario. The auto_scale argument is also accessible through plot_RLum.Analysis() and plot_RLum() (#646).

plot_RLum.Data.Spectrum()

• The channel-wise background subtraction was essentially broken if a background spectrum with the same setting was provided. The function always calculated the arithmetic mean. This was fixed, and the manual updated accordingly.

• If a background spectrum was provided, the behaviour of the plot output was sometimes hard to understand without knowledge of the underlying code. This behaviour was improved and now ylim will also affect the background spectrum if bg.channels = NULL (the default).

• The function could crash if column names were missing and the 1bg.spectrum` argument was used (#726).

read_TIFF2R()

• The argument file can now be provided as a list or a character vector.

• The function gained a new logical argument called merge2stack that can be used if file is either a list or a character vector of length > 1. If set to TRUE, images are combined into one image stack.

read_XSYG2R()

• A redundant computation has been removed, and the function is now marginally faster on files where the detector is not a spectrometer (#753).

• Prepare import for an updated version of LexStudio2 (>=v2.31.1) where the horizontal hardware binning finally works after >10 years, but the changes introduce a lot of NaN values that would cause a crash of the function.

• Minor code refactoring for a small speed boost if spectrometer measurements are imported.

scale_GammaDose()

• Argument plot_singlePanels now works as documented, that is it produces all plots in a single page when set to FALSE (default), and one plot per page when set to TRUE (#698).

sort_RLum()

• The sorting mechanism for RLum.Analysis objects has been enhanced. It now enables sorting based on multiple parameters, including sorting all available info_elements in a prioritised manner (#606, #620).

• Sorting now works on a list of RLum.Analysis objects. If the list contains elements of a different type, they are passed through unchanged. The output is again a list (#620).

structure_RLum()

• The function now returns a less messy data frame because it encapsulates .pid and info as lists within the data frame. The function is primarily used internally to facilitate a rapid exploration of RLum.Analysis object structures. However, the change may potentially break existing code in extremely rare circumstances.

template_DRAC()

• The function now throws messages instead of warnings for wanted coercions; however, it will be hard on failed coercions that would cause use_DRAC() to fail.

use_DRAC()

• The function now checks for DRAC specific URL parts if a custom URL is provided; this avoids long searches for unspecific errors.

• Due to the internal masking of submitted values, the initial row order got mixed up; regression from #438. The order is maintained and the row index corrected; in other words, the masking should now be again invisible to the user.

verify_SingleGrainData()

• The function crashed if an object originating from read_XSYG2R() contained positions in its info field (#740).

write_R2TIFF()

• The function now supports the export of more than one image slice.

Internals

• The internal function create_UID() has undergone further optimisation for speed, resulting in a significant performance improvement. Additionally, it now generates a hash value with a consistent length instead of a time stamp with a random number.