========= Changelog ========= 3.7 === * Add the new class Array to handle data like numpy arrays * Add a new module for doing mathematics: mathematics.py * Apply new convention for selectLines, rmsLines and calLines: one range per line instead of a list of ranges 3.6 === * Implement feature requested in redmine #1888 * Fix bug due to error precision (cf redmine #1888) and refactor 3.5 === * New feature: plotBestModel can plot selected lines (redmine #1429) 3.4.2 ===== * Fix log_mode issue 3.4.1 ===== * New feature: The best physical model can be saved with the function saveBestPhysicalModel 3.4 === * Force ratioAtCutOff < 1 and reducePhysicalParam > 1 * Fix database access * Fix a possible issue when reducePhysicalParam is an integer 3.3.5 ===== * Bugfix: geometry is incorrect in the LineAnalysis Panel in MCMC (redmine #1358) * Fix the computation of the dof * Fix initComponentsForMCMC * Fix log_mode issue with initComponentsForMCMC * Write the value of the collisioners with the exponential format in the output file * Fix collision files issue in Radex with MCMC * Bugfix: plotting the best model with telescope crashes * New feature: Selection of telescopes with ranges of lines * Bugfix: Collisioner value is not handled correctly with log_mode in RADEX with MCMC * Bugfix in MCMC with Radex * Rename rangesNotExcluded -> rangesSelected 3.3 === * Fix a bug: OverflowError when math.log10(0) * Allow Range("*", "*"). * Use lower case letters for the column names of the output file * Remove outputConcatFile * Replace: (fileName, extensionName, suffixName) -> inputFile * New feature: selectedLines accepts "*-*" as a key * Massive renaming: notExcludedLines -> selectedLines, rangesNotExcluded -> selectedRange * weightLines is not provided by the user anymore but. All its values are set to 1.0 * In MCMC, the chi2 value passed to compute the chi2Reduced is not divided by the number of points anymore * Remove function: displayNbOfModels. Generalize request of confirmation * Display a confirmation panel if warning is True and a message in the console if False * Bugfix: the interacting is not set correctly in MCMC * Display a warning when some lines selected are not used because they don't have enough pooints * Divide chi2 per line by number of points rather than by user-defined weight * Hardcode chi2Max to 1e20 to always allow determination of the best parameters * Fix computation of chi2 and reduced chi2 * Explain clearly the issue when no optimal param can be determined * Fix MCMC with RADEX model * Cancel division by the total weight in the computation of the reduced chi2Min * Fix computation of the reduced chi2Min * Check if the provided ranges are valid * Better warning message before the computation * Write the value in the output file not the log10 of the value (in MCMC only) * Sort the parameters ("nmol", "tex", "tkin", "fwhm", "size", "vlsr", "iso") * Define a more flexible the way for the user to select the lines with ranges and indexes * Fix computation of the reduced minimal chi2 displayed per model in the console * Print a message telling the user if lines were not taken into account * Bugfix: select the lines with enough points to prevent irrelevant computations * WIP: Determine which line must be used or not * Include continuum in the computation of the chi2 * Bugfix: chi2 computed correctly taking into account the weight when the whole spectrum is selected * Change initial minimal chi2: 1e9 -> 1e16 * Implement isoUnique in MCMC * Bugfix: Divide chi2 by weight * Bugfix: weight is used correctly in the computation of reduced chi2 * Bugfix: weightLines is passed to computeChi2Reduced * Weight the chi2 per line * Make reducePhysicalParam optional * Introduce weightLines * Make cutOff and ratioAtCutOff mandatory * Modify the computation of the reduced chi2 taking into account the number of independent points. Introduce the ratioAtCutOff * Add iso to alpha in MCMC * Check calLines, rmsLines and notExcludedLines * Set a new way of defining rmsLines * Fix bug with reducePhysicalParam * The user can define a reducing factor for every parameter * Move cutoff to the function computeChi2MinUsingMCMC * Set new parameters: cutoff for fac and calLines (calibration of the lines) * Remove alphaMaximal and alphaOptimal from Component * Change the way fac and alpha are calculated * Fix initialization of the computation * Make bestParameters identical in RG and MCMC * Add the keyword lineNumber for selecting the line to be plotted * Improve data output and display message * Bugfix: Nmol and Vlsr are correctly placed in the output file * Use temp instead of tex * Simplify the launch of module * Remove useless prints and prints the number of model the same way as RG 3.2 === * Rename the script to LineAnalysisScripting (RG and MCMC included) * Add Plot.py * Implement plotRates * Bugfix adding initializations for lists: variations, rates, thetas, iterIndexes * Add alphaMaximal, alphaOptimal, chi2Optimal keywords for MCMC * Fix the computation of chi2 with the functions extracted from ScriptLineAnalysisRG * Removes chi2min and keeps chi2Min * Refactor: use Plot to plot the scatter plot for the rates * MCMC is fully functional * Bugfix: plotBestModel works properly when run from the command line 3.1 === * New feature: new syntax for notExcludedLines * notExcludedLines and rangesNotExcluded -> notExcludedLines * Implement conversions to MHz and km/s (FrequencyScale) * Implement convertion to MHz from km/s, cm-1, micrometer * Remove keyword unit and delta that will be contained in Range * Add Range class * Formalize the calculation of the number of models in fct of isoUnique * Define new rules for exclusions * Allow to set both excludedLines and rangesNotExcluded at the same time * Bugfix: Prevent the division by zero when the dof = nbPointsUsed for the chi2MinReduced * Set keyword outputFile * Set keyword inputDir and outputDir instead of workingDir * Rescale ranges when necessary * Finish implementation of the conversion user inputs into the correct unit * Use the vlsr of the data for the shift rather than the vlsr of the user * Rename: freqsNotExcluded -> rangesNotExcluded, freqsExcluded -> rangesExcluded * A different tuningBand can be applied from plotBestModel * Fix chi2 computation using vlsrRange * Take into account the deltaV and vlsr in the chi2 computation on a selected frequency range * Take into account the user start index (1) instead of the logic start index (0) 2.9.9.3 ======= * Set warning message when the freqs specified are not subsets of the speciifed lines * Bugfix: freqsExcluded is determined properly * Set warning message when all the lines are excluded * Set warning message when freqsNotExcluded and freqsExcluded are both defined * Fix issue with freqsNotExcluded and freqsExcluded * Fix MCMC * Bugfix: Fix bad condition when more than 1 freq to be not excluded are provided * Improve theta, alpha parameters... * Add ScriptLineAnalysisMCMC * Feature implemented: Don't take into account collision parameters in LTE mode * Fix telescope check * Remove n (H2 density) from component * Add tbg and n as new extra parameters in the components * Fix a bug when no chi2 < chi2Min was found * plotBestModel and plotChi2Contour return a variable * Add quantun numbers: jup, kup, lup, mup * Display automatically the ContourPlotProduct in CASSIS standalone * Fix header display in the output file * Replace "_" with "-" in the collisioners in the output file * Collisioners start with a lower letter in the output file * Bugfix: interacting = True with many components * Display the detail about the number of values per parameter if warning is True * Change UserInputs.telescopes -> UserInputs.telescope * Check the telescopes in the inputs of the user * Check the continuum file existence * Bugfix: If min == max et nstep != 1, nstep must be set to 1 * Add new keyword iMin and set the title of the outputFile * the temp parameter is renamed tex or tkin regarding the selected model * Implement concatenation of the input data files for the plot of the best model * New feature: warning keyword * New implementation for isoUnique * Filter only the models where the isotopic ratios are similar * Introduce overSampling option in plotBestModel * Fix condition on freqsNotExcluded * Replace freqsExcluded with freqsNotExcluded * Fix the number of lines to be processd * excludedLines and notExcludedLines are handled properly * Radex model works * plotBestModel accepts the moltag instead of an index * Add geometry and interacting parameters in the components * Improve display of the information in the console * Introduce collision parameters for RADEX * Use specific module to prevent error precisions * Bugfix with log_mode 2.9.9.2 ======= * Bug fixes * Implementation of the MCMC method * Implementation of the RG method * Initial version